X48 Rampage Formula Preview.

First sorry for my bad english (french schools :p)

I have a rampage formula bios 0406 with a X3360 xeon CPU
Corsair pc8500 ram 2x1go dominator
The xeon cooling was a : Thermalright 120 Xtreme with a papst PWM

I was before on a crucial pc8500 8FE5 memory and it was very unstable and i changed for these corsairs

- ALL CPU tests Rockstable , OCCT CPU one hour

http://img376.imageshack.us/img376/3473/occt4bz3.jpg

But memtest on windows and OCCT ram very unstable , crash in less of 1 min...

http://img61.imageshack.us/img61/3907/occt2ht6.jpg

Also memtest with a boot cd was stable :

http://img364.imageshack.us/img364/8958/dsc00004so2.jpg

My temperatures was ok :

http://img67.imageshack.us/img67/332/tempkw7.jpg


I want to have my FSB450 in windows with strap 333 and 1071 mhz for the memory... But can't.

It was stable at FSB 445 , STRAP 400 , 1:1 with 1.51V vNB (very poor performances... 7239mb/s in everest read test)
FSB 445 , Strap 333 12:10 Unstable at 1.65vNB
FSB 450 , strap 400 , 1:1 : Stable only at 1.73 vNB !!!! and of course strap 333 always unstable



Noyau CPU 1.34 V
+3.3 V 3.28 V
+5 V 4.90 V
+12 V [ TRIAL VERSION ]
Mise en veille +5 V 4.97 V
FSB VTT 1.25 V
Noyau North Bridge 1.51 V
Noyau South Bridge 1.06 V
South Bridge PLL 1.52 V
DIMM 2.20 V
DIMM VTT 1.14 V


Memory :
CAS Latency (CL) 5T
RAS To CAS Delay (tRCD) 5T
RAS Precharge (tRP) 5T
RAS Active Time (tRAS) 15T
Row Refresh Cycle Time (tRFC) 60T
Command Rate (CR) 2T
RAS To RAS Delay (tRRD) 4T
Write Recovery Time (tWR) 14T
Read To Read Delay (tRTR) Same Rank: 4T, Different Rank: 6T
Read To Write Delay (tRTW) 8T
Write To Read Delay (tWTR) Same Rank: 11T, Different Rank: 5T
Write To Write Delay (tWTW) Same Rank: 4T, Different Rank: 6T
Read To Precharge Delay (tRTP) 5T
Write To Precharge Delay (tWTP) 14T
Precharge To Precharge Delay (tPTP) 1T
Refresh Period (tREF) 3120T
DRAM Read ODT 3T
DRAM Write ODT 6T
MCH Read ODT 8T
Performance Level 14
Read Delay Phase Adjust Neutral
DIMM1 Clock Fine Delay 8T
DIMM2 Clock Fine Delay 4T
DIMM3 Clock Fine Delay 8T
DIMM4 Clock Fine Delay 4T

What can i try? Change a paramater in my BIOS ? Or My ram? (corsairs bad with x48? Must i buy adata pc9600?) ? Or an issue with ma rampage and a should i do a RMA?

Quote:

---

Originally Posted by foxvengeor

First sorry for my bad english (french schools :p)

I have a rampage formula bios 0406 with a X3360 xeon CPU
Corsair pc8500 ram 2x1go dominator
The xeon cooling was a : Thermalright 120 Xtreme with a papst PWM

I was before on a crucial pc8500 8FE5 memory and it was very unstable and i changed for these corsairs

- ALL CPU tests Rockstable , OCCT CPU one hour

http://img376.imageshack.us/img376/3473/occt4bz3.jpg

But memtest on windows and OCCT ram very unstable , crash in less of 1 min...

http://img61.imageshack.us/img61/3907/occt2ht6.jpg

Also memtest with a boot cd was stable :

http://img364.imageshack.us/img364/8958/dsc00004so2.jpg

My temperatures was ok :

http://img67.imageshack.us/img67/332/tempkw7.jpg


I want to have my FSB450 in windows with strap 333 and 1071 mhz for the memory... But can't.

It was stable at FSB 445 , STRAP 400 , 1:1 with 1.51V vNB (very poor performances... 7239mb/s in everest read test)
FSB 445 , Strap 333 12:10 Unstable at 1.65vNB
FSB 450 , strap 400 , 1:1 : Stable only at 1.73 vNB !!!! and of course strap 333 always unstable



Noyau CPU 1.34 V
+3.3 V 3.28 V
+5 V 4.90 V
+12 V [ TRIAL VERSION ]
Mise en veille +5 V 4.97 V
FSB VTT 1.25 V
Noyau North Bridge 1.51 V
Noyau South Bridge 1.06 V
South Bridge PLL 1.52 V
DIMM 2.20 V
DIMM VTT 1.14 V


Memory :
CAS Latency (CL) 5T
RAS To CAS Delay (tRCD) 5T
RAS Precharge (tRP) 5T
RAS Active Time (tRAS) 15T
Row Refresh Cycle Time (tRFC) 60T
Command Rate (CR) 2T
RAS To RAS Delay (tRRD) 4T
Write Recovery Time (tWR) 14T
Read To Read Delay (tRTR) Same Rank: 4T, Different Rank: 6T
Read To Write Delay (tRTW) 8T
Write To Read Delay (tWTR) Same Rank: 11T, Different Rank: 5T
Write To Write Delay (tWTW) Same Rank: 4T, Different Rank: 6T
Read To Precharge Delay (tRTP) 5T
Write To Precharge Delay (tWTP) 14T
Precharge To Precharge Delay (tPTP) 1T
Refresh Period (tREF) 3120T
DRAM Read ODT 3T
DRAM Write ODT 6T
MCH Read ODT 8T
Performance Level 14
Read Delay Phase Adjust Neutral
DIMM1 Clock Fine Delay 8T
DIMM2 Clock Fine Delay 4T
DIMM3 Clock Fine Delay 8T
DIMM4 Clock Fine Delay 4T

What can i try? Change a paramater in my BIOS ? Or My ram? (corsairs bad with x48? Must i buy adata pc9600?) ? Or an issue with ma rampage and a should i do a RMA?

---

G.Skills work best with this MoBo look at the other systems & you will see that alot people are using them

+1 for the G. Skills.

ok i will try to echange my corsairs with gskills... PC8500 PI or PK series?

after 24 hours of playing around I got my rampage running stable at 490mhz FSB with my q6600 g0 , 7x multi, 3.43ghz, 1.424v vcore real, bios = 1.43875. anything above 450fsb takes a little bit of work to get stable. i found the most important things were NB voltage, DRAM clock skews, GTL voltage, NB/CPU ref, AI clock twister, and the secondary memory timings also play somewhat of a role. RFC = 60 works great for me with my Geil 2x2gb. I found PL 9 is adequate when using 333 strap, its timings are pretty tight as is. 333 strap doesnt like high FSB VTT. 490 runs happily with 1.32v fsb vtt and 1.53v NB, NB ref 0.56x or 0.60x, CPU gtl ref 0.63x, CPU PLL = 1.58v. DRAM skews, Channel A = +200ps, Channel B = +350ps. Setting up these skews correctly made going above 450fsb a breeze with 333 strap.

Quote:

---

Originally Posted by foxvengeor

First sorry for my bad english (french schools :p)

I have a rampage formula bios 0406 with a X3360 xeon CPU
Corsair pc8500 ram 2x1go dominator
The xeon cooling was a : Thermalright 120 Xtreme with a papst PWM

I was before on a crucial pc8500 8FE5 memory and it was very unstable and i changed for these corsairs

- ALL CPU tests Rockstable , OCCT CPU one hour

http://img376.imageshack.us/img376/3473/occt4bz3.jpg

But memtest on windows and OCCT ram very unstable , crash in less of 1 min...

http://img61.imageshack.us/img61/3907/occt2ht6.jpg

Also memtest with a boot cd was stable :

http://img364.imageshack.us/img364/8958/dsc00004so2.jpg

My temperatures was ok :

http://img67.imageshack.us/img67/332/tempkw7.jpg


I want to have my FSB450 in windows with strap 333 and 1071 mhz for the memory... But can't.

It was stable at FSB 445 , STRAP 400 , 1:1 with 1.51V vNB (very poor performances... 7239mb/s in everest read test)
FSB 445 , Strap 333 12:10 Unstable at 1.65vNB
FSB 450 , strap 400 , 1:1 : Stable only at 1.73 vNB !!!! and of course strap 333 always unstable



Noyau CPU 1.34 V
+3.3 V 3.28 V
+5 V 4.90 V
+12 V [ TRIAL VERSION ]
Mise en veille +5 V 4.97 V
FSB VTT 1.25 V
Noyau North Bridge 1.51 V
Noyau South Bridge 1.06 V
South Bridge PLL 1.52 V
DIMM 2.20 V
DIMM VTT 1.14 V


Memory :
CAS Latency (CL) 5T
RAS To CAS Delay (tRCD) 5T
RAS Precharge (tRP) 5T
RAS Active Time (tRAS) 15T
Row Refresh Cycle Time (tRFC) 60T
Command Rate (CR) 2T
RAS To RAS Delay (tRRD) 4T
Write Recovery Time (tWR) 14T
Read To Read Delay (tRTR) Same Rank: 4T, Different Rank: 6T
Read To Write Delay (tRTW) 8T
Write To Read Delay (tWTR) Same Rank: 11T, Different Rank: 5T
Write To Write Delay (tWTW) Same Rank: 4T, Different Rank: 6T
Read To Precharge Delay (tRTP) 5T
Write To Precharge Delay (tWTP) 14T
Precharge To Precharge Delay (tPTP) 1T
Refresh Period (tREF) 3120T
DRAM Read ODT 3T
DRAM Write ODT 6T
MCH Read ODT 8T
Performance Level 14
Read Delay Phase Adjust Neutral
DIMM1 Clock Fine Delay 8T
DIMM2 Clock Fine Delay 4T
DIMM3 Clock Fine Delay 8T
DIMM4 Clock Fine Delay 4T

What can i try? Change a paramater in my BIOS ? Or My ram? (corsairs bad with x48? Must i buy adata pc9600?) ? Or an issue with ma rampage and a should i do a RMA?

---

Just so you know, 70C is way too hot for that CPU! Especially if you have a Thermalright Ultra 120 eXtreme. I'm not sure whether that's idle temps or load temps (I'm hoping that's load temps), but your CPU temps are way too high.

As well as I don't think it's safe to be running more than 1.45V on the NB, if I remember correctly.

It's the load temp. The Fan was auto regulated by the motherboard ! at 72° ! Because my papst PWM can go to 3000 RPM and when i'm on full load he is at 1600 RPM only... It's Qfan... :rolleyes:. I had the same temp @stock but the fan was at 1100RPM max...

I've just test FSB 470 * 6 = CPU @ stock. OCCT test RAM : unstable. CPU : Unstable.

And FSB 410*8.5 RAM @ 546 mhz. OCCT test RAM : stable.

I really think that my rampage have a serious issue with FSB...

try setting NB ref to 0.60x, 490mhz fsb needs around 1.55 - 1.57v NB. Also if DRAM clock skews aren't set correctly, it will either lock up, come up with ridiculous amount of errors in memory or fail to post. 333 strap works fine at high FSB if you understand how to tame it. Basically just loosen all memory timings and go from there. 333 strap timings are really tight, PL 9 at 333 seems equivalent to PL 7 at 266 strap. Using PL 8 with 333 strap and 490 fsb, I managed to corrupt some Windows system files! I've gotten memory 100% stable at 490 fsb, but am lost for reasons why heavy CPU activity is locking the system up even with 1.45v Vcore and 1.60v cpu pll, 0.62x cpu gtl+ ref. This is when the mch/cpu skew adjustment feature the P45 boards have would be helpful.

Memory Timings
CAS Latency (CL) 5T
RAS To CAS Delay (tRCD) 5T
RAS Precharge (tRP) 5T
RAS Active Time (tRAS) 18T
Row Refresh Cycle Time (tRFC) 60T
Command Rate (CR) 2T
RAS To RAS Delay (tRRD) 6T
Write Recovery Time (tWR) 14T (Bios WR=7)
Read To Read Delay (tRTR) Same Rank: 5T, Different Rank: 6T
Read To Write Delay (tRTW) 9T
Write To Read Delay (tWTR) Same Rank: 13T, Different Rank: 5T
Write To Write Delay (tWTW) Same Rank: 5T, Different Rank: 6T
Read To Precharge Delay (tRTP) 5T
Write To Precharge Delay (tWTP) 14T
Precharge To Precharge Delay (tPTP) 1T
Refresh Period (tREF) 3120T
DRAM Read ODT 3T
DRAM Write ODT 6T
MCH Read ODT 8T
Performance Level 9
Read Delay Phase Adjust Neutral
DIMM1 Clock Fine Delay 11T
DIMM2 Clock Fine Delay 2T
DIMM3 Clock Fine Delay 10T
DIMM4 Clock Fine Delay 2T

Thats roughly the memory timings that I got to work flawlessly at 490 fsb, using 1:1 divider, 980mhz ddr. Very loose but didnt get around to fine tuning the secondary/tertiary timings as yet, well only a few of them.

With those loose timings, everest cache test was showing ~ 8800MB/s read, ~8700MB/s write, 8100MB/s copy, 57ns latency at 980mhz.

Alright, so i have my system nice and stable at 400fsb with 1.43v to the nb and 1.3v to the fsb, memory running at 1066 with common performance level of 6, now when i try to go up to 450 i cannot even get prime to pass for 1 minute, fails within 30 seconds, this is prime blend. I have tried Nb voltage of 1.65 and fsb voltage of 1.5v, still fails within 30 seconds, I have lowered memory to 900mhz so it runs 1:1 divider with common performance level of 10 and it still fails just to see if its the memory at fault, now memtest passes with no problems, its prime blend that fails, prime blend is fine at 400fsb trd of 6 and low voltage settings, but 450fsb is a bit hard to attain, any ideas??? im on 0403 bios.

Just want say that I've tired - Crucial pc8500 8FE5, OCZ Reaper pc8500 (4gb) and now G.skill F2-8500CL5D and the G.Skill have been by far the easiest ram to configure... They are even stable at auto settings

Im now a proud member of the 8gb G.skill club :D But I need ideas to cool 'em - I have trouble fitting in a fan suggestions are greatly appreciated

Quote:

---

Originally Posted by Eurasianman

As well as I don't think it's safe to be running more than 1.45V on the NB, if I remember correctly.

---

Could you link where you found that information?

It's safe to run whatever voltage you need for NB provided you keep it's temps below 50-55c load off top of my head. Otherwise they tend to fail rapidly.

Make sure there is a fan on the chipset to cool it down.

I have taken care of the chipset, its the ram modules im worried about, they get pretty hot to touch.

i've been using a dominator ram cooling module for a long while, they fit the part pretty well, and have kept a couple sets of ddr2 alive for > 1yr at 2.4v daily

Quote:

---

Originally Posted by Takle

Just want say that I've tired - Crucial pc8500 8FE5, OCZ Reaper pc8500 (4gb) and now G.skill F2-8500CL5D and the G.Skill have been by far the easiest ram to configure... They are even stable at auto settings

Im now a proud member of the 8gb G.skill club :D But I need ideas to cool 'em - I have trouble fitting in a fan suggestions are greatly appreciated



Could you link where you found that information?

---

try this memory fan

works great for me on my G.Skills

\


http://www.performance-pcs.com/catal...ducts_id=20910

Quote:

---

Originally Posted by mikeyakame

It's safe to run whatever voltage you need for NB provided you keep it's temps below 50-55c load off top of my head. Otherwise they tend to fail rapidly.

---

with 1.43v it idles at 50-52c loads around high 50's.........maybe thats my problem???, i have a fan on my antec 900 pointing to the side which helps kind of.

With the 3 year warranty its all good :) (at least for the rampage)

Quote:

---

Originally Posted by GAR

with 1.43v it idles at 50-52c loads around high 50's.........maybe thats my problem???, i have a fan on my antec 900 pointing to the side which helps kind of.

---

lap & remount the cooler
Q9450\mine is at 3.36 Ghz volts 1.228 & temps run 54-49 under prime/blend idles at 32
MX-2 thermal compound

Foxy-

Your English is a lot better than my French. I wish I could blame it on my schools.

Your settings with the FSB seem to indicate it as the culprit, especially if you've already pushed the voltage up. People have been getting Corsair to work on this board, but the exact model numbers need to checked.

Lower your NB voltage. I've heard of instances were the NB didn't like voltage too high.


To everyone-

Can the Rampage support any 775 socket chip? I'm wondering if my memtest86+ errors are because of my e8400. A buddy lent me a Pentium E2160 to see if it could pass memtest86+. Do you think the CPU could be the reason it won't pass? Cache, maybe?

mm yeah language school was horrible :D

I've already test to find my FSB max with the rampage... At coef 6 , CPU@Stock. And the ram at the lowest level as possible always strap 400.

I've test FSB 475 vNB 1.67 vFSB 1.48. Can't boot !
FSB 470 vNB 1.6 to 1.67 : can boot but errors at 4 seconds max

According to french forum i had a Volt Damage on my rampage??! And i must do a RMA... And my revendor propose an change with a P5 premium.... With 100 euros of différence. Or with a new rampage. i'm hesitating. Sav or not? Rampage or P5E5? I will receive my Gskill tomorrow for test before.

Quote:

---

Originally Posted by Takle

Could you link where you found that information?

---

Don't have a link. I just don't see how going from 333 FSB to 400 FSB+ on the NB could require going from 1.25V to 1.50V+... but that's just me.

Quote:

---

Originally Posted by mikeyakame

It's safe to run whatever voltage you need for NB provided you keep it's temps below 50-55c load off top of my head. Otherwise they tend to fail rapidly.

---

:rofl:
I don't even have my NB voltage that high and I'm already at 50C-55C load! Thank God for QX and unlocked multiplier.

Quote:

---

Originally Posted by GAR

with 1.43v it idles at 50-52c loads around high 50's.........maybe thats my problem???, i have a fan on my antec 900 pointing to the side which helps kind of.

---

I don't think that's your problem. I'm running less voltage and and still getting similar temps at load and remain stable.

Quote:

---

Originally Posted by Shadowthor

With the 3 year warranty its all good :) (at least for the rampage)

---

Some of us voided the warranty when removing the stock TIM. In a way, it shouldn't since the TIM ASUS used sucked major cajones!

Quote:

---

Originally Posted by Rickpatbrown

Foxy-

Your English is a lot better than my French. I wish I could blame it on my schools.

Your settings with the FSB seem to indicate it as the culprit, especially if you've already pushed the voltage up. People have been getting Corsair to work on this board, but the exact model numbers need to checked.

Lower your NB voltage. I've heard of instances were the NB didn't like voltage too high.


To everyone-

Can the Rampage support any 775 socket chip? I'm wondering if my memtest86+ errors are because of my e8400. A buddy lent me a Pentium E2160 to see if it could pass memtest86+. Do you think the CPU could be the reason it won't pass? Cache, maybe?

---

Agreed about the NB voltage. As far as memtest goes, using a different CPU won't change anything. I've been reading that memtest gives errors on the X48 regardless of what RAM you use. I just use Prime95 and F@H SMP to test my overall system stability :D

Quote:

---

Originally Posted by foxvengeor

mm yeah language school was horrible :D

I've already test to find my FSB max with the rampage... At coef 6 , CPU@Stock. And the ram at the lowest level as possible always strap 400.

I've test FSB 475 vNB 1.67 vFSB 1.48. Can't boot !
FSB 470 vNB 1.6 to 1.67 : can boot but errors at 4 seconds max

According to french forum i had a Volt Damage on my rampage??! And i must do a RMA... And my revendor propose an change with a P5 premium.... With 100 euros of différence. Or with a new rampage. i'm hesitating. Sav or not? Rampage or P5E5? I will receive my Gskill tomorrow for test before.

---

Are you referring to the RAM slots being overvolted? I know my board likes to add around ~.04V to the RAM voltage :mad:

No. With the load calibration enabled if i put 2.20 i had 2.20 :shrug:

Everest tends to report much higher core temperature than RealTemp and SpeedFan. In Everest, my core is ~71 C under load and RealTemp reports ~65 C. I recalled that these two programs use different thermal constant to calculate core temperature. I think the general consensus is that RealTemp is the more accurate one.

I am also using PWM fan (Nexus) with Thermalright Black 120 heatsink. In order to ramp up the fan speed under QFan, you need to use the Performance Mode profile under the bios's CPU fan control menu.

My Nexus PWM fan can go full speed under load.

The temps i was talking about were my NB temps, im not sure if everyone is talking about the same.............my cpu is fine, idles around 28-32c per core, load is very cool as well.

Yes, I am talking about CPU core temperature difference between Everest and RealTemp.

I have similar high NB temperature at ~60C under benchmarking. However I set my NB voltage at 1.53V. The NB voltage setting depends on the PL level you select. Mine is PL10 with FSB=496MHz. I found this configuration to be quite stable (though I tried PL9 at one time and it is fine too).

An excellent NB voltage vs. performance article can be found at http://www.anandtech.com/showdoc.aspx?i=3208&p=5

Can someone tell me if I can use two different video card in this board?
I want to hook up my 9800GTX to windows via D-sub, and a different series Nvidia card or an ATI card to Mac OS via DVI.

Pushed a pair of 2gb Corsair Dominator PC2-8500C5D ver 1.2 all the way to 1233.2mhz STABLE @ 2.18v today :D

Quote:

---

I've been reading that memtest gives errors on the X48 regardless of what RAM you use

---

Those guys are doing something wrong. There are so many settings on this board . . .

I was much more impressed with memtest86+ v2.01. This is the one from http://www.memtest.org/. The one from memtest.com is memtest86 v3.4 (notice there is no plus sign). memtest86 v3.4 had my CPU listed as pentiumIII. That shows you when the last time they updated it was.

Anyway. I figured out my problem. It was from a Saitek back lit keyboard being plugged into a Dtek USB hub :shrug:. The problem is reproducible. I need to do some more testing, but it doesn't seem to have anything to do with the Rampage or the RAM or the E8400 (so sad, I was hoping for a chance to RMA cause 3.9Ghz at 1.4v sucks).

This FAQ was very helpful in figuring out my problem, especially the part about getting rid of all unecessary components. You'd think common sense would have told me to do this, but it didn't. http://forum.x86-secret.com/forumdisplay.php?f=14 If you have any questions about memtest86+, this is the place to go. The guys seem very dedicated to solving any problems that may occur.

The good things about this whole ordeal is that I really got to learn about the settings on the board. "DRAM controller reference voltage" still alludes me as well as the GTL settings. I haven't messed with them yet.

On a side note- I realized that the power connector for the CPU has a cap over half of it. I originally had a four pin PSU connector plugged into it. Tonight, I took the cap off and plugged in the eight pin connector. The difference is two more yellow (+12v) wires. Could this make a difference in anything? Is this for quad cores maybe? Maybe I can get some more out of my E8400.

Quote:

---

Originally Posted by Rickpatbrown

On a side note- I realized that the power connector for the CPU has a cap over half of it. I originally had a four pin PSU connector plugged into it. Tonight, I took the cap off and plugged in the eight pin connector. The difference is two more yellow (+12v) wires. Could this make a difference in anything? Is this for quad cores maybe? Maybe I can get some more out of my E8400.

---

yer gonna find that things will run much better useing the eight pin connectors as i have the same psu on both my R/F MoBo's & have not had the problems some have had ,I've been useing the 8 pin all along on both builds , get back to us on what u fibd out

Quote:

---

Originally Posted by foxvengeor

No. With the load calibration enabled if i put 2.20 i had 2.20 :shrug:

---

Load Line Calibration only refers to the CPU voltage, no other voltage IIRC.

Quote:

---

Originally Posted by hsinhow

Everest tends to report much higher core temperature than RealTemp and SpeedFan. In Everest, my core is ~71 C under load and RealTemp reports ~65 C. I recalled that these two programs use different thermal constant to calculate core temperature. I think the general consensus is that RealTemp is the more accurate one.

---

From my understanding, all Core 2 CPUs Tj Max is 95C. I believe Everest and Real Temp uses 105C as the Tj Max. If you look at Core Temp, you'll see that the program now lists the Tj Max for Core 2's as 95C

Quote:

---

Originally Posted by hsinhow

Yes, I am talking about CPU core temperature difference between Everest and RealTemp.

---

Noted above.

Quote:

---

Originally Posted by alexlyc

Can someone tell me if I can use two different video card in this board?
I want to hook up my 9800GTX to windows via D-sub, and a different series Nvidia card or an ATI card to Mac OS via DVI.

---

I believe you can depending on the OS. Best way to find out is by testing it... if you can.

Quote:

---

Originally Posted by Rickpatbrown

Those guys are doing something wrong. There are so many settings on this board . . .

I was much more impressed with memtest86+ v2.01. This is the one from http://www.memtest.org/. The one from memtest.com is memtest86 v3.4 (notice there is no plus sign). memtest86 v3.4 had my CPU listed as pentiumIII. That shows you when the last time they updated it was.

Anyway. I figured out my problem. It was from a Saitek back lit keyboard being plugged into a Dtek USB hub :shrug:. The problem is reproducible. I need to do some more testing, but it doesn't seem to have anything to do with the Rampage or the RAM or the E8400 (so sad, I was hoping for a chance to RMA cause 3.9Ghz at 1.4v sucks).

This FAQ was very helpful in figuring out my problem, especially the part about getting rid of all unecessary components. You'd think common sense would have told me to do this, but it didn't. http://forum.x86-secret.com/forumdisplay.php?f=14 If you have any questions about memtest86+, this is the place to go. The guys seem very dedicated to solving any problems that may occur.

The good things about this whole ordeal is that I really got to learn about the settings on the board. "DRAM controller reference voltage" still alludes me as well as the GTL settings. I haven't messed with them yet.

On a side note- I realized that the power connector for the CPU has a cap over half of it. I originally had a four pin PSU connector plugged into it. Tonight, I took the cap off and plugged in the eight pin connector. The difference is two more yellow (+12v) wires. Could this make a difference in anything? Is this for quad cores maybe? Maybe I can get some more out of my E8400.

---

As far as my experience has gone, USB hubs always seem to give users more issues. As far as the 4 pin and 8 pin CPU power connector. The 8 pin power connector is there for CPUs that are rated for 130 WATT use such as the Intel Extreme Edition CPUs. Although, it won't hurt to have it utilized when using a non extreme edition CPU. If anything, it might help with stability.

Quote:

---

Originally Posted by Eurasianman

Load Line Calibration only refers to the CPU voltage, no other voltage IIRC.


From my understanding, all Core 2 CPUs Tj Max is 95C. I believe Everest and Real Temp uses 105C as the Tj Max. If you look at Core Temp, you'll see that the program now lists the Tj Max for Core 2's as 95C


Noted above.


I believe you can depending on the OS. Best way to find out is by testing it... if you can.


As far as my experience has gone, USB hubs always seem to give users more issues. As far as the 4 pin and 8 pin CPU power connector. The 8 pin power connector is there for CPUs that are rated for 130 WATT use such as the Intel Extreme Edition CPUs. Although, it won't hurt to have it utilized when using a non extreme edition CPU. If anything, it might help with stability.

---

BTW, realtemp tj max is 95 its coretemp that has tj max at 105.

Quote:

---

Originally Posted by Eurasianman

As far as the 4 pin and 8 pin CPU power connector. The 8 pin power connector is there for CPUs that are rated for 130 WATT use such as the Intel Extreme Edition CPUs. Although, it won't hurt to have it utilized when using a non extreme edition CPU. If anything, it might help with stability.

---

This seems especially true when driving the voltages higher than spec...

Quote:

---

Originally Posted by Eurasianman

As far as the 4 pin and 8 pin CPU power connector. The 8 pin power connector is there for CPUs that are rated for 130 WATT use such as the Intel Extreme Edition CPUs. Although, it won't hurt to have it utilized when using a non extreme edition CPU. If anything, it might help with stability.

---

as most of just about everyone who buys the R/F MoBo is doing O/C'ing & putting their systems into the realm of the Intel QX9xxx Xtreme chips anyway & is on this Xtreme Systems Forum for that reason .

The Asus Rampage Formula is made for the Xtreme use of your CPU .

YOU would be wise to use the 8 pin power connector ,
use a highend psu that has a clean stable 12 volt power single rail rated in the stabilty of 80%+ range of ripple free 12volts & a deadedcated 8 pin connector
use proven memory for the X48 chipset,

alot of reported instabitly of this MoBo by persons has been due to the above reasons of ppl Not having the above !
as some ppl have no problems at all while other ppl have problems from the get go!

all this is " I M H O " as it's your MoBo do as you want .

a follow-up to my X48 chipset water block mod for the MPC-X38 -DD block (no longer sold , you now have to use the DD universal X38) , i have recieved the
" NEW "
Danger Den MPC- 975X chipset block/with the BRASS TOP option
\ >> It is the SAME basic block as the DD MPC -X38 chipset block with the BRASS TOP option(no longer sold) / except the holes in this thin plexi hold down are set for the Intel 975x chip set >> my mod of the X38 thin plexi hold down will work for this block also
this is I M H O a great block that is a high flow / high heat transfer , water block that can , with this mod use all of the 4 holes in the R/P MoBo 's PCB to hold down this block firmly & squarely on the X48 chipset
all other blocks in the market at the moment(that I know of) only use 2 screws to hold the chipset on the R/F X48 chipset & some (ie; EK,DD Universal ) don't fit squarley on the X48 & use slider type holes to hold the block on to the chipset

i'll do some pix of the different blocks & show how easy this mod really is

I'll also be doing a BRASS hold down mod (for both of my R/F MoBo's)
as i have now in my hands some thick brass stock that will replace the thin plexi hold down on both of the out dated MPC-X38 & the new Intel 975X water blocks to fit the X48 Rampage Formula
this 4 hole / 4 screw == Brass mod will replace the plexi hold down with a Blinged Uppped Brass Polished custom fit for this Kick-Ass water block

Excuse my english and my little knowledge on O/C matters.
This is my first post...

Does anyone has a Rampage Formula with Q9300 on it?
I am very new to O/C (2 weeks student) and I am trying to do a start.
Is it possible for someone with similar PC configuration to post his BIOS settings in order for me to have something to depend on...

My PC configuration
- Case: Cosmos S
- M/B: Asus Rampage formula (403)
- CPU: Q9300
- CPU Coller: Scyth Ninja (AIR)
- Memory: Transcent AXE (1066, 5-5-5-15)
- GPU: Shapphire 3870X2
- HD: 2x Raptor 76GB (RAID 0)
- OS: Windows Vista Ultimate x64

Any help for my begining is really appreciated...
Thanks in advance.

I would read over the overclocking guide in the forum somewhere. Post VID, what are you trying to overclock to?

Quote:

---

Originally Posted by jimpoeng

Excuse my english and my little knowledge on O/C matters.
This is my first post...

Does anyone has a Rampage Formula with Q9300 on it?
I am very new to O/C (2 weeks student) and I am trying to do a start.
Is it possible for someone with similar PC configuration to post his BIOS settings in order for me to have something to depend on...

My PC configuration
- Case: Cosmos S
- M/B: Asus Rampage formula (403)
- CPU: Q9300
- CPU Coller: Scyth Ninja (AIR)
- Memory: Transcent AXE (1066, 5-5-5-15)
- GPU: Shapphire 3870X2
- HD: 2x Raptor 76GB (RAID 0)
- OS: Windows Vista Ultimate x64

Any help for my begining is really appreciated...
Thanks in advance.

---

try these to start
AI o/c tuner ; manuel
CPU level up ; set @ qx9770
fsb feq ; 400
PCIE feq ; 100
DRAM feq ; 1003 use what you memory is rated at ( u might have to boot twice to get this up)mine is ddr1000 if yours is ddr1066 use that )
DRAM ; manuel & if the bios don't set the timings ,then set them are per specs of your memory ( mine are 5-5-5-15)
comand rate ; 2T
AI twister ; moderate
everything else set on AUTO & i do mean everything
use this as a start point & see what ya get from there u can screw with voltages

Quote:

---

Originally Posted by bigslappy

try these to start
AI o/c tuner ; manuel
CPU level up ; set @ qx9770
fsb feq ; 400
PCIE feq ; 100
DRAM feq ; 1003 use what you memory is rated at ( u might have to boot twice to get this up)mine is ddr1000 if yours is ddr1066 use that )
DRAM ; manuel & if the bios don't set the timings ,then set them are per specs of your memory ( mine are 5-5-5-15)
comand rate ; 2T
AI twister ; moderate
everything else set on AUTO & i do mean everything
use this as a start point & see what ya get from there u can screw with voltages

---

Thank you very much bigslappy.

What about the voltages (CPU, NB, SB)?
Do I have to set them at the lowest prices shown in the manual ?

Quote:

---

Originally Posted by jimpoeng

Thank you very much bigslappy.

What about the voltages (CPU, NB, SB)?
Do I have to set them at the lowest prices shown in the manual ?

---

see what u get with the above setting & go slow upping voltages as u go check yer O/C with Prime95 on blend keep notes so u can go back to what u set that worked

Memtest86+ only gives errors on X48 when your timings are out of whack between mch & ram, or ram / mch clocks even a little out of sync at higher fsbs. IF you are getting errors I've found the solution is to tweak the dram clock skews, and lower the common Performance Level by 1 sometimes 2, then fine tune it by enabling or disabling Phase Pull In for mch/dram clock phases. As far as my understanding goes based on trial and error, this allows u to set Read Delay very finely for MCH-DRAM and DRAM-MCH clock signalling. It can be used to compensate for either side being somewhat out of skew, above 450 fsb , PL < 9 and > 333mhz AGTL+ reference clock (fsb strap) this is very useful. Even slightly out of sync MCH-DRAM/AGTL+ clocking phases above 450 fsb seem to wreak all kinds of havoc with quad cores.

RF does need more flexible CPU GTL Reference adjustment, it's a major hindrance to fine tuning the FSB. The fixed ratio adjustments do help to a point, but there are large FSB holes at 400+ mhz that even these can't fix. Between 450 - 470 FSB lies a large hole which can only be explored with a very high relative AGTL+ reference voltage. Too much ground/power bounce going out of reasonable boundaries. At 500mhz FSB there lies another decent size hole.

I've been trying to get CPU clocks to sync correctly with MCH clocks at 495mhz FSB (333mhz agtl ref.)at 7x multi on my Q6600 G0, boots fine, runs fine too, until you heavily load the CPU with the likes of Prime or Linpack then within 10-15 seconds complete system freeze, not even a BSOD, requires full power cycle and the LCD poster continues displaying the last display until you boot again. If RTC is locking like that, the only conclusion I can muster up is massive clock synching difference that even a high CPU Phase Loop Lock voltage fails to contain. I pushed PLL upto 1.8v or so which did improve the situation somewhat, but that is kind of going beyond normal bounds. Might just be a symptom of the CPU also. It loves high FSB's but not high clocks, will get around to playing with a penryn some day.

I'll try find some more time this weekend to further look into it. Adjust one or two settings, test, rinse and repeat. I've had some very interesting results and conclusions to say the least.

Hi. I've been reading through this whole thread and while I've learned an awful lot, I feel that I'm still lacking the knowledge I need for a establishing a stable overclock on my system (seen in my sig)

I've been seeing such varying voltage settings for what I perceive to be similar set ups that I'm getting a bit confused.

I'm wondering if someone could point me to a guide that provides some detailed descriptions of what each voltage setting will effect in my overclock and what ranges would be reasonable for my setup, I'd really appreciate it. I've tried a couple things and I've not really had what I would call success.



I'm hoping for what I consider reachable with my components: 533MHz FSB x7.5 multi at 1:1 for 1066 memory



My CPU is under water with a D-Tek Fusion in a single block loop with a PA120.3.

My NB and GPU are under water with an EK on the NB and a DD on the 8800GTX and that's flowing through an Innovatek Konvekt-o-matic Ultra with a crossflow fan cooling it.


I'm not looking for all of the answers to be spoonfed (although that would be great too), I'm willing to read, read and read even more. I'm just finding it difficult to navigate many of the settings on the Extreme Tweaker menu of the BIOS ie:

• DRAM CLK Skew on Channel A/B
• Ai Clock Twister
• Ai Transaction Booster
• Performance Level
• CPU PLL Voltage
• SB 1.50 Voltage (why are there two SB voltage controls?)
• Has the LLC debate ever been resolved? Should I enable or disable? I have 4B pencil lead on the indicated cap for vDroop mod but, I'm not 100% sure why.
• Should I disable both Spread Spectrum?

I know to disable everything on the CPU Configuration screen and I've read Graysky's Guide HERE and it explained a lot. But, his guide doesn't touch on any of the Memory Timing tweaks that I've seen mentioned in this thread and his BIOS screens don't mention many of the settings I've indicated above.

As I've said, I'm not looking for the magic wand solution, I'd just like to be pointed in the right direction for decoding some of these settings that still remain puzzling for me.



Huge thanks to all that have posted their information thus far. I'm definitely learning a lot from the posts.

Even bigger thanks to those who can help me further. I'm looking forward to learning more.

Wow, I think this is probably the stupidest question ever but here it goes...

Also, this is assuming the person in the post above me is running all SATA drives...

How can you run 7 HD's when the Rampage only has 6 SATA ports?

If one of the drives is IDE, then completely disregard this question...

Quote:

---

Originally Posted by hartigan

I'm not looking for all of the answers to be spoonfed (although that would be great too), I'm willing to read, read and read even more. I'm just finding it difficult to navigate many of the settings on the Extreme Tweaker menu of the BIOS ie:

• DRAM CLK Skew on Channel A/B
• Ai Clock Twister
• Ai Transaction Booster
• Performance Level
• CPU PLL Voltage
• SB 1.50 Voltage (why are there two SB voltage controls?)
• Has the LLC debate ever been resolved? Should I enable or disable? I have 4B pencil lead on the indicated cap for vDroop mod but, I'm not 100% sure why.
• Should I disable both Spread Spectrum?

---

Here's what information I can give you based on experience/intel chipset documentation reading.

DRAM CLK (Clock) Skew Channel A/B

No trace lengths on boards are ever perfect, this gives delay ( - ) and advance ( - ) adjustment in PS (picaseconds) for Memory clock signals on both Channel A/B, ie. dimm A1 / B1 would make up one Dual Channel set. Trace lengths will vary within a tolerance from board to board, but usually Channel B will need a bit more adjustment than A since its farther from the MCH. Advance or Delay depends on your particular board, it can be either really. As you increase DRAM frequency, there becomes a smaller frame where the clock signals have to be in phase with each other.

Ascii square wave example.

|__|--|__| <--- Channel A Clock Signal on Board X
_|--|__|-- <--- Channel B Clock Signal on Board X

Any command MCH sends, will be received at Channel B 1/2 a clock later, so when memory sends data back to MCH, Channel A data will arrive before Channel B. In a small window at high frequency, that 1/2 a clock might be longer than the MCH can wait for data to return. In that situation MCH will either re-request data or forward incomplete memory request to CPU. If this data held an address pointer to another location in memory, the CPU may send a request to write data at the incomplete or damaged memory address, and then you have get anything from scrambled return data to system crash if for example this damaged pointer address happened to fall inside another applications private memory area.

Skew adjustment allows u to make some corrections to this, to improve stability at tighter timings / higher clocks / higher fsb.

DRAM CMD (command) Skew Channel A/B (when CMD Rate is set to 1N you can access these)
Basically similar to clock skew adjustment, but extra fine adjustment to timing of commands for channel A/B on top of clock skew adjustments. As FSB speed is raised, MCH clocks may become too far ahead or far behind DRAM clocks, and as such commands to DRAM may be sent too early or too late. This allows one to delay or advance Command timings for either channel.

Performance Level (tRD MCH)
Performance Level is basically Read Delay for CPU/MCH & MCH/DRAM. On either side of MCH there will be a difference between clock signal timing. Lower Read Delay (0 -> PL, ie approaching 0, (n < n + 1) where n is lower read delay, n + 1 is higher) defines smaller window of latency for point A to B or B to C, Higher Read Delay ( PL -> n, IE approaching n which is a positive integer ( n > n - 1) where [n - 1] is lower read delay, and n is higher read delay) defines larger window of latency for command / return data. Lower RD (PL) means higher memory bandwidth, higher RD means lower memory bandwidth. Reason being with lower latency Read Delay more data can be throughput per clock cycle. Lower RD also makes MCH work alot harder and require more voltage to switch between power (+) and ground (-) to raise clock signal then switch to ground to terminate it, this makes your basic square wave on AGTL+.

CPU PLL (Phase Loop Lock)
Voltage overhead above CPU vCore used to keep CPU/FSB clock signals in phase (alignment), it's somewhat of a vCore burst / spike limit to put it simply, it comes into play under sudden load/throttling conditions. Rampage uses by default CPU PLL Voltage 125% of vCore, 1.50v. Higher PLL voltage allows CPU to momentarily overshoot voltage to correct clock phases, but from what I understand this is exactly why Intel warns not to use too high of PLL voltage as it may damage or fry a CPU.

SB 1.5V ref voltage.
Simply a reference voltage which all southbridge based clock signals are based on. PCI/PCI-E/IDE/SATA/etc.

CPU / PCI-E spread spectrum.
According to what I've read from Intel docs, these basically take a mean frequency across associated clock frequencies and use as reference for. I personally disable both as they affect RTC/PCI-E/CPU address/CPU data clocks which can make an OC very unstable.

AI Transaction Booster
Auto intelligently adjusts PL / phase. Manual you can set PL / phase on Ch. A/B. Total phases = DRAM + AGTL Boot Clock Divider (FSB STRAP Divider ie. 1:1, 5:4). Pull in phase gives you fine control of read delay (PL) for each active phase on Ch. A/B. Consider this fine tuning of Read Delay on each Clock phase. Common performance level is the base level, for read delay from CPU and to DRAM.

AI Clock Twister
Intelligently adjusts timings tighter/looser to get most throughput per clock based on current PL and timings between DRAM and MCH. As far as I understand this is based on the chipsets previous adjustments / learned values as far as timing / clock / cmd skews goes. For stronger it tunes based on minimal window of latency, for lighter in maximal window of latency. I find strong is very good, stronger if you have excellent RAM dimms that can take a low PL, high frequency, timings, etc. Stronger will turn up memory errors in large amounts in Memtest86+ if even some mem chips can't handle the tighter timing set.

Like i said this is part experience / part tech docs. This is pretty much as much as I can tell you, hope it helps. Anybody else feel free to correct / add their own knowledge of these to this. I'd like to know more myself.

Quote:

---

Originally Posted by Mr. HT

Wow, I think this is probably the stupidest question ever but here it goes...

Also, this is assuming the person in the post above me is running all SATA drives...

How can you run 7 HD's when the Rampage only has 6 SATA ports?

If one of the drives is IDE, then completely disregard this question...

---

I'm using a HighPoint SATA Raid Card as well. Just didn't feel the need to list it. :up: It's running my two 500GB drives because they're "DATA" drives and I want to have the ability to move them from PC to PC when I upgrade so I don't have to worry about what controller is on the motherboard I would be potentially moving to.


Mikeyakame: Thank you TONS for the reply. I'll be back with questions, I'm sure. I just need time to read, digest and test out all of the information you've provided.

In addition to all of the great information you've provided, I guess I'm also looking for how each of the settings would benefit my o/c one way or the other. For example, how to I diagnose whether or not bumping a setting one way or another is going to be the thing I need in order to fix my lack of stability.

IE: What feedback, in terms of instability, would I get that would tell me, "Hey, you need to bump your CLK Skew down one ps or up one ps."?

Quote:

---

Originally Posted by mikeyakame

CPU / PCI-E spread spectrum.
According to what I've read from Intel docs, these basically take a mean frequency across associated clock frequencies and use as reference for. I personally disable both as they affect RTC/PCI-E/CPU address/CPU data clocks which can make an OC very unstable.

---

This is not well explained :)
Normally the clocks are pure single frequency signals with very low sideband noise.
On a spectrum scope they appear as a vertical line very thin and clean. As on spectrum scopes monitor the X-axis represents frequency and the Y-axis signal's voltage, a very high and clean vertical line means a clean and powerfull clock signal on a single frequency.

This is the best clock that enables the digital components to work at their best.

Unfortunately this very clean and high signal could overcome the limits written in FCC and EC rules about EMC. If this occurs, the system can not obtain the FCC/EC mark and can not be sold.

As these rules measure the EMC level with a quasi-peak filter, it is better to use a clock not so clean. In the spectrum scope it will be no more a vertical line, but a very thin gaussian curve. It means that all the energy of clock is not on one frequency, but is spread a little around the central frequency of signal.

This is very easily done by modulating in frequency the clock signal.

This trick fools the quasi-peak detector of receiver used to test the EMC and, even if not perfect, allows the normal working of the device.

As in normal offices or houses environments the EMC is not so important, it is better to disable the spread spectrum of clocks to have the best performances in overclock.

I hope I have explained a little this difficult argument :)

Ok. I figured there was some simple solution that I was ignorant about but thanks for responding to my stupidity...

Here is my first try to O/C my Q9300 with the Rampage formula (as I said some posts above I am only 2 weeks student, don't be harsh).

I uploaded to rapidshare some screenshots from AsusProbe, GPUZ, CPCZ, coreTemp, RealTemp) and please I need your help to make it safe/better/stable :) ...

RapidShare.com Link:
http://rapidshare.com/files/12894067...__1st_try_.rar

- What do u think about the 3 different temps I have (AsusProbe, RealTemp, CoreTemp)? The difference in metrics is quit big...what should I consider accurate?
- Why my NB and SB look red in the Probe? How can I reduce their temps? What should be ok for SB and NB temps and voltages?
- Why coreTemp shows temp only for 2/4 of CPUs? Is there any problem?
- What are your suggestions about everything :) (how to configure voltages, memory, BIOS, etc).

I USE AIR COOLING...
The voltages are set to auto...

Excuse my amature questions, i still have a lot to learn...

Brama and Mikeyakame-

Thank you for taking to time to explain some of these things. It is so had to find information about all these settings. I'm guessing that you guys have some formal education in this stuff? The problem is that the explanations that I can understand don't tell me what I want to know. The articles that tell me what I want to know are way over my head.

It is also very difficult because this stuff is so spread out over the web.

I guess the easiest thing would be for me to eat your brain :yepp:. . .mmmmmmmm, brains.

(smiley added to not freak you out :))

Quote:

---

Originally Posted by jimpoeng

Here is my first try to O/C my Q9300 with the Rampage formula (as I said some posts above I am only 2 weeks student, don't be harsh).

I uploaded to rapidshare some screenshots from AsusProbe, GPUZ, CPCZ, coreTemp, RealTemp) and please I need your help to make it safe/better/stable :) ...

RapidShare.com Link:
http://rapidshare.com/files/12894067...__1st_try_.rar

- What do u think about the 3 different temps I have (AsusProbe, RealTemp, CoreTemp)? The difference in metrics is quit big...what should I consider accurate?
- Why my NB and SB look red in the Probe? How can I reduce their temps? What should be ok for SB and NB temps and voltages?
- Why coreTemp shows temp only for 2/4 of CPUs? Is there any problem?
- What are your suggestions about everything :) (how to configure voltages, memory, BIOS, etc).

I USE AIR COOLING...
The voltages are set to auto...

Excuse my amature questions, i still have a lot to learn...

---

RealTemp is proven to be the best That Asus software is Sh*tware most of delete it or never load it & use Everest, HW monitor , real temp ,prime95
post yer clock you got .i'd want to buy that rapid share s/w

Quote:

---

Originally Posted by Rickpatbrown

Brama and Mikeyakame-

Thank you for taking to time to explain some of these things. It is so had to find information about all these settings. I'm guessing that you guys have some formal education in this stuff? The problem is that the explanations that I can understand don't tell me what I want to know. The articles that tell me what I want to know are way over my head.

It is also very difficult because this stuff is so spread out over the web.

I guess the easiest thing would be for me to eat your brain :yepp:. . .mmmmmmmm, brains.

(smiley added to not freak you out :))

---

I Ditto ^^^^^^ that feeling !!
But keep up the great work & we low brain fuctions ppl really are trying to understand all this & thanx to you'all many times over
the learning curve is steep but i'm up for it !

Quote:

---

Mikeyakame: Thank you TONS for the reply. I'll be back with questions, I'm sure. I just need time to read, digest and test out all of the information you've provided.

In addition to all of the great information you've provided, I guess I'm also looking for how each of the settings would benefit my o/c one way or the other. For example, how to I diagnose whether or not bumping a setting one way or another is going to be the thing I need in order to fix my lack of stability.

IE: What feedback, in terms of instability, would I get that would tell me, "Hey, you need to bump your CLK Skew down one ps or up one ps."?

---

Once you reach the limits of your current DRAM frequencies on sane voltages, and limits of timings, skewing the clocks correctly will let you push farther on this bound at the same or less voltage. Basically it will boot or it wont with advance or delayed skews at a DRAM frequency greater than the limit without any skew adjustment. It may also get rid of persistent errors, I use Memtest86+ to see what kind of results clock/cmd skew adjustment has for a given value. If i know the DRAM frequency turns up errors, adjusting the skews either way will either make these errors become fewer to none or increase even more. I use tests 5/6/7 for the most part. These seem to find errors in memory quickly. You press 'c' for configuration, '3' for test select, 'x' for individual test, then '0' to go back to testing.

Saves running repeated tests which you know pass. 32bit integers, memory moving and random number generation are all critical of bit for bit correct read/writes.

Quote:

---

Originally Posted by Brama

This is not well explained :)
Normally the clocks are pure single frequency signals with very low sideband noise.
On a spectrum scope they appear as a vertical line very thin and clean. As on spectrum scopes monitor the X-axis represents frequency and the Y-axis signal's voltage, a very high and clean vertical line means a clean and powerfull clock signal on a single frequency.

This is the best clock that enables the digital components to work at their best.

Unfortunately this very clean and high signal could overcome the limits written in FCC and EC rules about EMC. If this occurs, the system can not obtain the FCC/EC mark and can not be sold.

As these rules measure the EMC level with a quasi-peak filter, it is better to use a clock not so clean. In the spectrum scope it will be no more a vertical line, but a very thin gaussian curve. It means that all the energy of clock is not on one frequency, but is spread a little around the central frequency of signal.

This is very easily done by modulating in frequency the clock signal.

This trick fools the quasi-peak detector of receiver used to test the EMC and, even if not perfect, allows the normal working of the device.

As in normal offices or houses environments the EMC is not so important, it is better to disable the spread spectrum of clocks to have the best performances in overclock.

I hope I have explained a little this difficult argument :)

---

Awesomeness. Thanks for explaining the principles behind spread spectrums! I knew what they did but not how they did it.

Quote:

---

Originally Posted by mikeyakame

I use tests 5/6/7 for the most part. These seem to find errors in memory quickly. You press 'c' for configuration, '3' for test select, 'x' for individual test, then '0' to go back to testing.

Saves running repeated tests which you know pass. 32bit integers, memory moving and random number generation are all critical of bit for bit correct read/writes.

---

Great overall information. So, to sum up, most of what you were discussing regarding CLK Skew, CMD Skew and all other items involving RAM can be addressed using MemTest. Thereby, avoiding the Windows boot altogether until the settings are successfully stable according to MemTest.

Is there any way you can give me a similar diagnostic tool to really help me peg exactly what the best FSB/Multi/Voltages (NB, SB, SB 1.5, PLL, Vtt, etc.) would be for my setup as well.

You've been AMAZING with all of the information you've provided thus far and I know it takes a lot of your time to do so. If you want to take a break and get back to me when convenient, I understand. You've really aided me in addressing my RAM based questions and I genuinely appreciate it.

Thanks also to Brama for the great information on Spread Spectrum.

I know I'm compiling all of this information into my own little guide of the various settings I'm confronted by with this ASUS BIOS. All of the options for tweaks to my system are fantastic to have available, they're just rather daunting when addressed as a whole.

Thanks again for all the help and enjoy your weekend!!

ADD: What is the best ver. of MemTest86+ to use? The one at MemTest.org (V2.01) or the one at MemTest86.com (V3.4)? I've read that it's better to use the one from MemTest.org (V2.01) but, I'd like a second opinion. THX

First off . . .

Sorry for the double post.


Alright . . .

I've tried to set up a configuration I thought would work and I get an error in MemTest86+ v2.04

Here's the settings I'm using:

Code:

---

AI Overclock Tuner.......................................  Manual
CPU Ratio Setting.........................................  7.5
FSB Strap to North Bridge..............................  333MHz
FSB Frequency............................................  500MHz
PCI-E Frequency..........................................  100MHz

DRAM Frequency..........................................  DDR2 - 1002MHz
DRAM Command Rate....................................  AUTO
DRAM CMD Skew on Channel A........................  AUTO
DRAM CMD Skew on Channel B........................  AUTO
DRAM CLK Skew on Channel A.........................  AUTO
DRAM CLK Skew on Channel B.........................  AUTO
DRAM Timing Control.....................................  AUTO
DRAM Static Read Control..............................  AUTO
AI Clock Twister..........................................    LIGHT
AI Transaction Twister.................................    AUTO

CPU Voltage...............................................    1.35000V
CPU PLL Voltage..........................................    1.52V
Northbridge Voltage.....................................    1.63V
DRAM Voltage.............................................    2.00V
FSB Termination Voltage...............................    1.40V
South Bridge Voltage....................................    1.050V
SB 1.50V Voltage.........................................    1.50V
Loadline Calibration......................................    Disabled
CPU GTL Voltage REF...................................    AUTO
NB GTL Voltage REF.....................................    AUTO
DRAM Controller Voltage REF.........................    AUTO
DRAM Channel A Voltage REF........................    AUTO
DRAM Channel B Voltage REF........................    AUTO

CPU Spread Spectrum.................................    Disabled
PCI-E Spread Spectrum...............................    Disabled

---

The result I get in MemTest86+ can be seen below.


I know there must be some settings that I can change to help fix this. I'm just unable to narrow down specifically what they are.

I'd really appreciate any help.

TIA.

Quote:

---

Originally Posted by hartigan

First off . . .

Sorry for the double post.


Alright . . .

I've tried to set up a configuration I thought would work and I get an error in MemTest86+ v2.04

Here's the settings I'm using:

Code:

---

AI Overclock Tuner.......................................  Manual
CPU Ratio Setting.........................................  7.5
FSB Strap to North Bridge..............................  333MHz
FSB Frequency............................................  500MHz
PCI-E Frequency..........................................  100MHz

DRAM Frequency..........................................  DDR2 - 1002MHz
DRAM Command Rate....................................  AUTO
DRAM CMD Skew on Channel A........................  AUTO
DRAM CMD Skew on Channel B........................  AUTO
DRAM CLK Skew on Channel A.........................  AUTO
DRAM CLK Skew on Channel B.........................  AUTO
DRAM Timing Control.....................................  AUTO
DRAM Static Read Control..............................  AUTO
AI Clock Twister..........................................    LIGHT
AI Transaction Twister.................................    AUTO

CPU Voltage...............................................    1.35000V
CPU PLL Voltage..........................................    1.52V
Northbridge Voltage.....................................    1.63V
DRAM Voltage.............................................    2.00V
FSB Termination Voltage...............................    1.40V
South Bridge Voltage....................................    1.050V
SB 1.50V Voltage.........................................    1.50V
Loadline Calibration......................................    Disabled
CPU GTL Voltage REF...................................    AUTO
NB GTL Voltage REF.....................................    AUTO
DRAM Controller Voltage REF.........................    AUTO
DRAM Channel A Voltage REF........................    AUTO
DRAM Channel B Voltage REF........................    AUTO

CPU Spread Spectrum.................................    Disabled
PCI-E Spread Spectrum...............................    Disabled

---

The result I get in MemTest86+ can be seen below.


I know there must be some settings that I can change to help fix this. I'm just unable to narrow down specifically what they are.

I'd really appreciate any help.

TIA.

---

I dont think you will get any board thats gonna do 500fsb with a quad and 8gb of mem.

I just got my TRUE black today and decided to apply some AS5 to the NB/SB.
I thought it's gonna be tough to remove the assembly since I have been using my RF for almost a month now. But it was actually very easy, the NB part came right off when I unscrew the screws on the NB, SB is a bit tricky but doable. What surprise me is that the NB/SB heatsink doesn't actually attach to the mosfet heatsink part (pg 3 of this thread said it's all in one piece).
http://img76.imageshack.us/img76/2943/img1770vx7.jpg
http://img65.imageshack.us/img65/8127/img1773jw7.jpg
http://img71.imageshack.us/img71/2540/img1774iz0.jpg
http://img59.imageshack.us/img59/3927/img1772qh4.jpg
http://img78.imageshack.us/img78/1332/img1771dm9.jpg
Can someone tell me what the heck are all these, should I clean both materials out? Do I have to keep the thin plastic layer whatever that is???
http://img59.imageshack.us/img59/6545/img1777su9.jpg

id go ahead and say no, should just be able to apply as5 to the Southbridge with a razor full coverage once u clean it. It was like that for the last X38 i did. I haven't gotten around to doing my own yet, need to buy more AS5.

Hartigan,

Thats a hell of a lot of errors so earn in. I've never even seen such severe errors before!

To start with, try getting it working at < 450mhz fsb. Until you get that stable, don't even bother with higher using 2 x 2gb dimms.

NB voltage seems a lot too high. I can run 495mhz fsb with 1.53 - 1.55v.
FSB Termination also too high, try dropping to 1.36v.
CPU PLL might be too low, I don't have all that much experience with Penryns to say yes or no. Others who do may be able to give u more advice on that.
DRAM voltage too low, I dont know what your recommended voltage is from G.Skill but I can imagine it'd be at least 2.2v at DDR2-1066 for 2gb dimms.
Disable static read control.
Enable AI Transaction booster, set common performance level to 10. and leave all pull in disabled.
AI clock twister = Moderate or Strong
CPU GTL Ref, try 0.62 or 0.63
NB GTL ref, try 0.60 or 0.63
DRAM Command Rate, 2N
DRAM timing control, enable.

CAS = 5
RAS to CAS = 5
RAS PRE time = 5
RAS ACT Time = 16 - 18
RAS to RAS = 4 - 5
Ref Cycle Time = 60+
Write Recovery Time = 7
Read to Pre time = 4
Read to Write Delay (Same/diff) = 9
Write to Read Delay (Same) = 4
Write to Read Delay (Diff) = 6
Read to Read Delay (Same) = 4
Read to Read Delay (Diff) = 7
Write to Write Delay (same) = 4
Write to Write Delay (diff) = 7
Write to Pre Delay = 15
Read to Pre Delay = 6 - 7
Pre to Pre Delay = 2
All Precharge to Activate delay = 7
All Precharge to Refresh delay = 6

See if that helps any. Thats pretty loose timings, but it'll help eliminate memory errors first and foremost, once you get the frequency stable at loose timings then you can proceed to tighten them up, increase PL, tighten clock twister, etc.

I added AS5 but temp doesn't seem to drop a bit :shrug:... NB still idle around 54 degree w/o any fan at stock voltage (haven't put my side panel fan yet).

UPDATE:
I realize I wasn't actually running at stock voltage. NB voltage auto setting actually runs at 1.616V, so temp was high at idle.
Right now NB is idling at 42c @ 1.39v with two asus blower fans @ 70&#37; fan speed,

So you've got something to compare to. I have a 40x40x20mm fan sitting on top of NB heat sink via custom cable tie arrangement and NB Core Voltage = 1.44V real, bios is 1.43 or 1.45V, temp runs around 38c idle. It's about 4-5c cooler than it was without a fan. I need to put AS5 on. I'm sure once you put a fan on it, the AS5 will make a lot more difference.

Quote:

---

Originally Posted by GAR

I dont think you will get any board thats gonna do 500fsb with a quad and 8gb of mem.

---

Erm... that's not necessarily true.

http://www.xtremesystems.org/forums/...d.php?t=187947

Quote:

---

Originally Posted by Zucker2k

Erm... that's not necessarily true.

http://www.xtremesystems.org/forums/...d.php?t=187947

---

Yeah i had mine up there with 8gb of 1066 but that's the QX9770=R/F board of mine & it's down right now gettin' watered up .. i'll post pix later today when i get the chipset brass hold down mod done & I'll take suggestions/comments on how i should run the tubing ---
soo far Vtec V2 cpu , modded DD MPC block X48 , EK for the mosfets & the rest , waiting on the 4870X2 card for the Vid Gonna do C/F with those full block cover maybe ? of GPU blocks ?? don't know till i get cards in hand will use my 3870 till then on air

& no i didn't document it with a screen shot was just messing around not really knowin' what i was doing .. sure wish i had kept the setting wrote down BUt will do that in the future .We will see if I can do it agin ..when i get it all watered up & with dual awaki's & dual GTX480's on a blower with a 50 f ambiant air temp running thru the GTX480's

Quote:

---

Originally Posted by bigslappy

Yeah i had mine up there with 8gb of 1066 but that's the QX9770=R/F board of mine & it's down right now gettin' watered up .. i'll post pix later today when i get the chipset brass hold down mod done & I'll take suggestions/comments on how i should run the tubing ---
soo far Vtec V2 cpu , modded DD MPC block X48 , EK for the mosfets & the rest , waiting on the 4870X2 card for the Vid Gonna do C/F with those full block cover maybe ? of GPU blocks ?? don't know till i get cards in hand will use my 3870 till then on air

& no i didn't document it with a screen shot was just messing around not really knowin' what i was doing .. sure wish i had kept the setting wrote down BUt will do that in the future .We will see if I can do it agin ..when i get it all watered up & with dual awaki's & dual GTX480's on a blower with a 50 f ambiant air temp running thru the GTX480's

---

bigslappy, you have to be the main hardware guy on XS. Nice systems you got there.

Quote:

---

Originally Posted by bigslappy

sure wish i had kept the setting wrote down BUt will do that in the future .We will see if I can do it agin ..

---

anxiously awaiting those settings. I'm still havin' a bear of a time getting this locked down. I used the settings suggested by mikeyakame and I'm not getting anywhere. I'm definitely not blaming mikey. I realize it's a hit or miss type thing and especially challenging when you're not even physically in front of the PC.

I can't even get past DET DRAM with the following settings.

Code:

---

AI Overclock Tuner.......................................  Manual
CPU Ratio Setting.........................................  7.0
FSB Strap to North Bridge..............................  333MHz
FSB Frequency............................................  500MHz
PCI-E Frequency..........................................  100MHz

DRAM Frequency..........................................  DDR2 - 1002MHz
DRAM Command Rate....................................  2N
DRAM CMD Skew on Channel A........................  AUTO
DRAM CMD Skew on Channel B........................  AUTO
DRAM CLK Skew on Channel A.........................  AUTO
DRAM CLK Skew on Channel B.........................  AUTO
DRAM Timing Control.....................................  Manual

CAS# Latency...............................  5
RAS# to CAS# Delay......................  5
RAS# PRE Time.............................  5
RAS# ACT Time............................  18
RAS# to RAS# Delay.....................    5
REF Cycle Time............................    65
WRITE Recovery Time...................    7
READ to PRE Time........................    4
READ to WRITE Delay (S/D)...........    9
WRITE to READ Delay (S)..............    4
WRITE to READ Delay (D)..............    6
READ to READ Delay (S)................    4
READ to READ Delay (D)................    7
WRITE to WRITE Delay (S)............    4
WRITE to WRITE Delay (D)............    7
WRITE to PRE Delay.....................    15
READ to PRE Delay.......................    7
PRE to PRE Delay.........................    2
ALL PRE to ACT Delay...................    7
ALL PRE to REF Delay....................    6

DRAM Static Read Control..............................  Disabled
AI Clock Twister..........................................    Moderate
AI Transaction Twister.................................    Manual

Common Performance Level..............10
Pull-Ins.........................................Disabled

CPU Voltage...............................................    1.47500V
CPU PLL Voltage..........................................    1.58V
Northbridge Voltage.....................................    1.63V
DRAM Voltage.............................................    2.10V
FSB Termination Voltage...............................    1.40V
South Bridge Voltage....................................    1.050V
SB 1.50V Voltage.........................................    1.50V
Loadline Calibration......................................    Disabled
CPU GTL Voltage REF...................................    0.63x
NB GTL Voltage REF.....................................    0.63x
DRAM Controller Voltage REF.........................    AUTO
DRAM Channel A Voltage REF........................    AUTO
DRAM Channel B Voltage REF........................    AUTO

CPU Spread Spectrum.................................    Disabled
PCI-E Spread Spectrum...............................    Disabled

---

I know the specs on the G.Skill are 2.0-2.1V so I'm at the top end of spec. Other than that, I can't figure out why these settings won't boot.

The following adjustments allow me to POST but not boot:

CPU Voltage..................... 1.35000V
CPU PLL Voltage............... 1.50V
North Bridge Voltage......... 1.65V
FSB Termination Voltage.... 1.42V
CPU GTL......................... 0.67x
NB GTL........................... 0.67x



I'm concerned about going any higher with the Vtt because I've read in this thread that above 1.40V is dangerous for 45nm CPUs. I'm not too happy about the 1.65V for NB either, but it's what allows me to POST.

The modded settings also allow Memtest86+ V2.01 to Pass with flying colors. I just can't boot Vista.

Any ideas?


EDIT:

OK. I found THIS post and I've tried these settings for my voltage

Here's my new BIOS Config:

Code:

---

AI Overclock Tuner.......................................  Manual
CPU Ratio Setting.........................................  8.0
FSB Strap to North Bridge..............................  333MHz
FSB Frequency............................................  475MHz
PCI-E Frequency..........................................  100MHz

DRAM Frequency..........................................  DDR2 - 1002MHz
DRAM Command Rate....................................  2N
DRAM CMD Skew on Channel A........................  AUTO
DRAM CMD Skew on Channel B........................  AUTO
DRAM CLK Skew on Channel A.........................  AUTO
DRAM CLK Skew on Channel B.........................  AUTO
DRAM Timing Control.....................................  Manual

CAS# Latency...............................  5
RAS# to CAS# Delay......................  5
RAS# PRE Time.............................  5
RAS# ACT Time............................  18
RAS# to RAS# Delay.....................    5
REF Cycle Time............................    65
WRITE Recovery Time...................    7
READ to PRE Time........................    4
READ to WRITE Delay (S/D)...........    9
WRITE to READ Delay (S)..............    4
WRITE to READ Delay (D)..............    6
READ to READ Delay (S)................    4
READ to READ Delay (D)................    7
WRITE to WRITE Delay (S)............    4
WRITE to WRITE Delay (D)............    7
WRITE to PRE Delay.....................    15
READ to PRE Delay.......................    7
PRE to PRE Delay.........................    2
ALL PRE to ACT Delay...................    7
ALL PRE to REF Delay....................    6

DRAM Static Read Control..............................  Disabled
AI Clock Twister..........................................    Moderate
AI Transaction Twister.................................    Manual

Common Performance Level..............10
Pull-Ins.........................................Disabled

CPU Voltage...............................................    1.38750V
CPU PLL Voltage..........................................    1.58V
Northbridge Voltage.....................................    1.55V
DRAM Voltage.............................................    2.10V
FSB Termination Voltage...............................    1.32V
South Bridge Voltage....................................    1.050V
SB 1.50V Voltage.........................................    1.50V
Loadline Calibration......................................    Disabled
CPU GTL Voltage REF...................................    0.63x
NB GTL Voltage REF.....................................    AUTO
DRAM Controller Voltage REF.........................    AUTO
DRAM Channel A Voltage REF........................    AUTO
DRAM Channel B Voltage REF........................    AUTO

CPU Spread Spectrum.................................    Disabled
PCI-E Spread Spectrum...............................    Disabled

---

This allows me to boot into Windows but prime 95 fails Core 1 immediately.

I feel like I'm getting close to a very nice O/C with tolerable voltages. I just need to tweak it and I don't know what exactly to tweak in order to get it stable.


Sorry for posting two full BIOS settings but I'm trying to provide as much diagnostic information for anyone willing to help me.

Thank you!!

Quote:

---

Originally Posted by hartigan

anxiously awaiting those settings. I'm still havin' a bear of a time getting this locked down. I used the settings suggested by mikeyakame and I'm not getting anywhere. I'm definitely not blaming mikey. I realize it's a hit or miss type thing and especially challenging when you're not even physically in front of the PC.

I can't even get past DET DRAM with the following settings.

---

The X3350 is a Q9450 variant for the server right? If so, there's your answer. It's almost impossible to run these chips at those fsb rates even with extreme cooling; unless you got a cherry chip. I can go all the way 488 (stable) but it takes too much VFSB, and VNB to get stable. If you really got balls, give it 1.5+ on VFSB and 1.7+ on the NB. I run my fsb at 1.5 for 3.8Ghz.

Quote:

---

Originally Posted by Zucker2k

The X3350 is a Q9450 variant for the server right? If so, there's your answer. It's almost impossible to run these chips at those fsb rates even with extreme cooling; unless you got a cherry chip. I can go all the way 488 (stable) but it takes too much VFSB, and VNB to get stable. If you really got balls, give it 1.5+ on VFSB and 1.7+ on the NB. I run my fsb at 1.5 for 3.8Ghz.

---

Thank you for the reply! I really appreciate everyone's help with getting me through is. I'm hoping this documentation of my progress will help others with this board, the G.Skill RAM and an X3350

And yes, the X3350 is a server variant of the Q9450. I chose this CPU because I figured the server variant is designed to handle more extreme heat and voltage situations than the desktop variety.

Ballz, I do not have. After reading the Anandtech article regarding 45nm chips and Vfsb beyond 1.4V I'm definitely hesitant to go or 1.5+. I just don't have the $$ for that.


Therefore what would you suggest as a comfortable voltage config. for the 475x8 that I'm trying to run? With my luck, there's no way I got a cherry chip but, I don't think it's unreasonable to expect 475x8 out of this chip.

I can get Windows to boot and 1M Pi runs fine but 32M kicks an error. I'm just not experienced enough to know where to go from here.


ALSO . . . I have a Q6600. I figured the 12MB cache on the X3350 would be better than the 2x4MB on the Q6600 but I'm wondering whether I'd be able to get more performance out of the Q6600. What are your thoughts on the Q6600 vs. the X3350? Should I stick with the X3350 because it's 45nm?

Quote:

---

Originally Posted by hartigan

Thank you for the reply! I really appreciate everyone's help with getting me through is. I'm hoping this documentation of my progress will help others with this board, the G.Skill RAM and an X3350

And yes, the X3350 is a server variant of the Q9450. I chose this CPU because I figured the server variant is designed to handle more extreme heat and voltage situations than the desktop variety.

Ballz, I do not have. After reading the Anandtech article regarding 45nm chips and Vfsb beyond 1.4V I'm definitely hesitant to go or 1.5+. I just don't have the $$ for that.


Therefore what would you suggest as a comfortable voltage config. for the 475x8 that I'm trying to run? With my luck, there's no way I got a cherry chip but, I don't think it's unreasonable to expect 475x8 out of this chip.

I can get Windows to boot and 1M Pi runs fine but 32M kicks an error. I'm just not experienced enough to know where to go from here.


ALSO . . . I have a Q6600. I figured the 12MB cache on the X3350 would be better than the 2x4MB on the Q6600 but I'm wondering whether I'd be able to get more performance out of the Q6600. What are your thoughts on the Q6600 vs. the X3350? Should I stick with the X3350 because it's 45nm?

---

I was stuck at 3.5/3.6 Ghz for a while, then 2 things changed. First was the 0406 bios. I tried 0403 but it was slower than 0219 so I tried 0406 and went faster immediatly. Then I saw a post in this thread linking to an OCZ forum thread on R/F bios settings and some of thier advice and settings really helped me. I now am using DRAM clock skew (advance 300ps A & B) Ai transaction booster (8, use lowest # that will boot) and am keeping volts low. CPU 1.35v, DRAM 2.14v, NB 1.49v. Memtest 86+ shows no errors and Everest benches are Read 9434, Write 10092, & Copy 9761MB/s. Stick with the Xeon.

Nice write speeds there! At the moment with 480FSB on Q6600 and 1152mhz DRAM 5-5-5-15-3-50-2N-PL7 I'm getting 9572mb/s READ, 8658MB/s WRITE, 9375MB/s COPY, 53.4ns Latency. I set AI twister to Moderate till I completely stabilize memory timings and down adjust some voltages, it was crossing 10000MB/s on one of them also with Twister Stronger, but no good for fine tuning stability.

The perfect skews for my board are 50ps Delay Ch. A, 200ps Advance Ch. B, Dram REF -10mv, Dram Ch. A Ref V DDR_REF, Dram Ch. B Ref +10mv. Its working really good like that at 2.26v Vdimm with 4 1GB sticks.

Hartigan, If you are getting one core failing, your clocks are either out of sync or you need more vNB/vCore/GTL. Only way I got around random cores failing was figuring out the correct skewing values for the DRAM Clocks, but even 50ps is too large an adjustment. I needed around 10-15ps more delay for synch to be near perfect, had to find other ways to delay the MCH / DRAM clocks. There are plenty once you realise each timing / voltage can have a minute or major effect on clock timings between domains. tRFC is your friend if you figure out the right value. For your 4 x 2gb, I'd dare say push it up to 70 at most. If you time tRFC correctly no matter how large it is, the performance gains are more than all the other sub memory timings being tight combined.

mikeyakame
So you've got RF rev. 1.01G right? Can you confirm? Is the revision number imprinted on the PCB or sticker under which it's still 1.00G?

Just double checked, it is actually 1.00G. It was shipped from Taiwan mid april from what I can tell. Had 0308 Bios on it when i got it. Must be going blind! I was sure I saw 1.01G when I put it in!

Quote:

---

Originally Posted by mikeyakame

So you've got something to compare to. I have a 40x40x20mm fan sitting on top of NB heat sink via custom cable tie arrangement and NB Core Voltage = 1.44V real, bios is 1.43 or 1.45V, temp runs around 38c idle. It's about 4-5c cooler than it was without a fan. I need to put AS5 on. I'm sure once you put a fan on it, the AS5 will make a lot more difference.

---

Thats quite impressive - I've added 2x40mm fans on the Northbridge and i get 40c idle at stock voltage. I also added 3x120mm Scythe Ultra Kaze 2000 to my p182 and lapped my TRUE which allowed me to drop my vcore significantly :up:

Think i should try what Abducted did
http://www.xtremesystems.org/forums/...&postcount=427

Since it seems like there a big difference in the mount quality of NB/SB heatsink.

I've had a few peeps in here pm me for my settings... I've had so many toys to play with I didn't have the chance. Here are my settings for 3.8Ghz stable.

CPU =
CPU PLL = 1.584v (1.50v set in bios)
FSBT = 1.520v (yes, most people would say this is high, but this is very much cpu dependent, yours may need less).
DRAM = 2.00v (1.940v set in bios)
NB = 1.650v
SB = 1.136v (1.100v set in bios - this could help for ich stability especially if you have a lot of stuff hooked up ie. I have 4 HDDs in raid 0, lots of peripherals on USB hubs)
SB 1.500v = 1.664v (1.600v set in bios

DRAM Skew = Advance 350 (for both A/B) - 300 is fine too, I noticed no difference between the two
DRAM Static Read = Enabled, this has little effect on this particular configuration so you can set to Disabled or AUTO
Ai Clock Twister = Stronger


Now some pictures.....

Quad 32M of Superpi



Everest screenie showing timings and latencies as well as performance level

http://www.hostthenpost.org/uploads/...cb4fd4c410.jpg

Everest RAM Bandwidth

http://www.hostthenpost.org/uploads/...82207d8177.jpg

Thanks.

Richard.

I've had a few peeps in here pm me for my settings... I've had so many toys to play with I didn't have the chance. Here are my settings for 3.8Ghz stable.

CPU =
CPU PLL = 1.584v (1.50v set in bios)
FSBT = 1.520v (yes, most people would say this is high, but this is very much cpu dependent, yours may need less).
DRAM = 2.00v (1.940v set in bios)
NB = 1.650v
SB = 1.136v (1.100v set in bios - this could help for ich stability especially if you have a lot of stuff hooked up ie. I have 4 HDDs in raid 0, lots of peripherals on USB hubs)
SB 1.500v = 1.664v (1.600v set in bios

DRAM Skew = Advance 350 (for both A/B) - 300 is fine too, I noticed no difference between the two
DRAM Static Read = Enabled, this has little effect on this particular configuration so you can set to Disabled or AUTO
Ai Clock Twister = Stronger


Now some pictures.....

Quad 32M of Superpi

http://www.hostthenpost.org/uploads/...333a6600c7.jpg

Everest screenie showing timings and latencies as well as performance level

http://www.hostthenpost.org/uploads/...cb4fd4c410.jpg

Everest RAM Bandwidth

http://www.hostthenpost.org/uploads/...82207d8177.jpg

Thanks.

Richard.

Thank you so much for providing that information!! Very useful.

I was wondering if you could post a screenshot of your Computer>Sensor screen in Everest. I want to compare my voltages being reported to what you have.

One thing that's really weird is my WRITE to READ Delay (S) is set to 11T in my BIOS but Everest is reporting it as 3T.


I was also wondering how you set your tREF in the BIOS. I can't find the setting anywhere.

Here's what I have running currently, trying to match your settings best I can:

Code:

---

AI Overclock Tuner.......................................  Manual
CPU Ratio Setting.........................................  8.0
FSB Strap to North Bridge..............................  333MHz
FSB Frequency............................................  475MHz
PCI-E Frequency..........................................  100MHz

DRAM Frequency..........................................  DDR2 - 951MHz
DRAM Command Rate....................................  2N
DRAM CMD Skew on Channel A........................  AUTO
DRAM CMD Skew on Channel B........................  AUTO
DRAM CLK Skew on Channel A.........................  Advance 300ps
DRAM CLK Skew on Channel B.........................  Advance 300ps
DRAM Timing Control.....................................  Manual

    CAS# Latency............................... 5
    RAS# to CAS# Delay...................... 5
    RAS# PRE Time............................. 5
    RAS# ACT Time............................ 15
    RAS# to RAS# Delay..................... 3
    REF Cycle Time............................ 45
    WRITE Recovery Time................... 14
    READ to PRE Time........................ 5
    READ to WRITE Delay (S/D)........... 8
    WRITE to READ Delay (S).............. 11
    WRITE to READ Delay (D).............. 5
    READ to READ Delay (S)................ 4
    READ to READ Delay (D)................ 6
    WRITE to WRITE Delay (S)............ 4
    WRITE to WRITE Delay (D)............ 6
    WRITE to PRE Delay..................... 14
    READ to PRE Delay....................... 5
    PRE to PRE Delay......................... 1
    ALL PRE to ACT Delay................... 7
    ALL PRE to REF Delay.................... 6

DRAM Static Read Control..............................  Disabled
AI Clock Twister..........................................    Stronger
AI Transaction Twister.................................    Manual

    Common Performance Level..............8
    Pull-Ins.........................................Disabled

CPU Voltage...............................................    1.41250V
CPU PLL Voltage..........................................    1.50V
Northbridge Voltage.....................................    1.65V
DRAM Voltage.............................................    2.10V
FSB Termination Voltage...............................    1.52V
South Bridge Voltage....................................    1.100V
SB 1.50V Voltage.........................................    1.60V
Loadline Calibration......................................    Disabled
CPU GTL Voltage REF...................................    AUTO
NB GTL Voltage REF.....................................    AUTO
DRAM Controller Voltage REF.........................    AUTO
DRAM Channel A Voltage REF........................    AUTO
DRAM Channel B Voltage REF........................    AUTO

CPU Spread Spectrum.................................    Disabled
PCI-E Spread Spectrum...............................    Disabled

---

I'm able to get Prime95 to run for 4 minutes until Core 4 fails.

If you have any suggestions as to what will help, I'm open to anything. I'm still unsure as to how to set the tREF.

Thanks again for sharing your settings with us!! It has been very useful.

Quote:

---

Originally Posted by hartigan

Thank you so much for providing that information!! Very useful.

I was wondering if you could post a screenshot of your Computer>Sensor screen in Everest. I want to compare my voltages being reported to what you have.

One thing that's really weird is my WRITE to READ Delay (S) is set to 11T in my BIOS but Everest is reporting it as 3T.


I was also wondering how you set your tREF in the BIOS. I can't find the setting anywhere.

Here's what I have running currently, trying to match your settings best I can:

Code:

---

AI Overclock Tuner.......................................  Manual
CPU Ratio Setting.........................................  8.0
FSB Strap to North Bridge..............................  333MHz
FSB Frequency............................................  475MHz
PCI-E Frequency..........................................  100MHz

DRAM Frequency..........................................  DDR2 - 951MHz
DRAM Command Rate....................................  2N
DRAM CMD Skew on Channel A........................  AUTO
DRAM CMD Skew on Channel B........................  AUTO
DRAM CLK Skew on Channel A.........................  Advance 300ps
DRAM CLK Skew on Channel B.........................  Advance 300ps
DRAM Timing Control.....................................  Manual

    CAS# Latency............................... 5
    RAS# to CAS# Delay...................... 5
    RAS# PRE Time............................. 5
    RAS# ACT Time............................ 15
    RAS# to RAS# Delay..................... 3
    REF Cycle Time............................ 45
    WRITE Recovery Time................... 14 (3)
    READ to PRE Time........................ 5 (3)
    READ to WRITE Delay (S/D)........... 8
    WRITE to READ Delay (S).............. 11 (3)
    WRITE to READ Delay (D).............. 5
    READ to READ Delay (S)................ 4
    READ to READ Delay (D)................ 6
    WRITE to WRITE Delay (S)............ 4
    WRITE to WRITE Delay (D)............ 6
    WRITE to PRE Delay..................... 14
    READ to PRE Delay....................... 5
    PRE to PRE Delay......................... 1
    ALL PRE to ACT Delay................... 7 (6)
    ALL PRE to REF Delay.................... 6

DRAM Static Read Control..............................  Disabled
AI Clock Twister..........................................    Stronger
AI Transaction Twister.................................    Manual

    Common Performance Level..............8
    Pull-Ins.........................................Disabled

CPU Voltage...............................................    1.41250V
CPU PLL Voltage..........................................    1.50V
Northbridge Voltage.....................................    1.65V
DRAM Voltage.............................................    2.10V
FSB Termination Voltage...............................    1.52V
South Bridge Voltage....................................    1.100V
SB 1.50V Voltage.........................................    1.60V
Loadline Calibration......................................    Disabled
CPU GTL Voltage REF...................................    AUTO
NB GTL Voltage REF.....................................    AUTO
DRAM Controller Voltage REF.........................    AUTO
DRAM Channel A Voltage REF........................    AUTO
DRAM Channel B Voltage REF........................    AUTO

CPU Spread Spectrum.................................    Disabled
PCI-E Spread Spectrum...............................    Disabled

---

I'm able to get Prime95 to run for 4 minutes until Core 4 fails.

If you have any suggestions as to what will help, I'm open to anything. I'm still unsure as to how to set the tREF.

Thanks again for sharing your settings with us!! It has been very useful.

---

What RAM do you have? Some of your settings are off. To get a higher TREF, you'll have to revert to bios 0308. 0219 also has a higher TREF. Under hardware, what voltages is your bios reporting?

Quote:

---

Originally Posted by Zucker2k

What RAM do you have?

---

I'm using the RAM in my sig. The 1066MHz G.Skills

Quote:

---

Some of your settings are off. To get a higher TREF, you'll have to revert to bios 0308. 0219 also has a higher TREF. Under hardware, what voltages is your bios reporting?

---

So I should downgrade the BIOS to 0308 for the tREF. No problem.

Here's the voltages as reported by BIOS in the Hardware Monitor section:

CPU Voltage = 1.392V
CPU PLL Voltage = 1.568V
FSB Termination Voltage = 1.504V
DRAM Voltage = 2.144V
North Bridge Voltage = 1.648V
South Bridge Voltage = 1.136V
SB 1.5V Voltage = 1.648V
DRAM Controller Voltage REF = 1.072V
3.3V Voltage = 3.296V
5V Voltage = 5.040V
12V Voltage = 12.208V


I'll get back with any changes after I downgrade my BIOS.

THANK YOU for all of your help!! I can't wait until I have this thing stable and can stop bothering you guys.

EDIT: Updated Voltages to what I have set up now.

EDIT (2): So far, with 0308 BIOS, the updated voltages and mem timings fixed as instructed by Zucker2K, I've gone the longest on Prime95 so far. I've gone 20 minutes with no errors. :clap: I'm going to let it run overnight and hopefully I can sleep through the anxiety of wanting to check how it's doing every 20-30 minutes :wth: :lol:

I'll report how it does in the morning.

EDIT (3): Well, it BSODed about an hour into the Prime95 run. I can't complain about that, it's MUCH better than anything I've run before. I'm going to try bumping that Vtt up to the full 1.520 you suggested, Zucker2K. Hopefully that will be what I need.

EDIT (4): With Vtt at 1.520V, I BSODed in about 10-15 minutes. Not sure where I can tweak this to get that full stability.

Hey guys, I am considering getting this MB, but don't know which RAM to get for it.

Here are some options please tell me which you think will be better

1 - OCZ Reapers 4GB kit PC8500 (OCZ2RPR10664GK)

2 - Corsair PC8500 4GB kit (TWIN2X4096-8500C5DF)

3 - Team Group PC8500 4GB kit (TXDD4096M1066HC5DC-D)

I'm in Australia and these are the best options at the place I'm buying from. Which would be the best compatible kit from these? Or would I be better off getting PC6400 RAM and overclocking it?

I've got a question about this board since i'm considering this to replace my P5K-E/Wifi-AP which has the Sata bug apperantly.

Does this board have any problem's with WD's Velociraptor and Samsung's F1 1TB drives, since my Raptor reinitializes every once in a while making the system freeze for a few seconds but it passes WD's diag.

I plan to run this board with my Q6600 that does 3,2 at default voltage and 400FSB which is what i'm also looking for in this board, higher is ofcourse better but lower is not ;)

Btw the P5K-E is going to be replaced also because i want propper crossfire support.

Thanks for any input guys!

I'm getting NOWHERE with my Rampage Formula/QX9650. I can up the multi to 10 and it works fine, but even increasing the bus 7 MHz to 340 Vista will not load. I've tried upping voltages to insane levels just to see if it helped and it doesn't seem to. I had this same CPU and memory at 420 x 10 on my Maximus Formula 12 hours + prime stable.

Since it will generally POST I'm wondering if maybe I have a Vista issue. I swapped from the Maximus without a re-install and recently went from an 8800GTX to HD4870. Removed the nvidia drivers, did driver cleaner, etc using the 8.6 hotfix catalyst drivers. Since the video card swap there are some issues. Sometimes it won't go into standby (S3) as set in the power properties. Manually putting it into standby works fine. Sometimes it will crash coming out of standby. I guess I'll hang on until the official 8.7 catalysts come out and do a re-install. In the meantime running at 3.33 GHz. Hopefully reinstalling Vista will get my OC back.

On a side note the Maximus was RMA to Asus and they have return shipped. It'll be interesting to see what shows up. It seems like the X48 should be > X38, so there must be some other issue with the OC.

Quote:

---

Originally Posted by Clump

I'm getting NOWHERE with my Rampage Formula/QX9650. I can up the multi to 10 and it works fine, but even increasing the bus 7 MHz to 340 Vista will not load. I've tried upping voltages to insane levels just to see if it helped and it doesn't seem to. I had this same CPU and memory at 420 x 10 on my Maximus Formula 12 hours + prime stable.

Since it will generally POST I'm wondering if maybe I have a Vista issue. I swapped from the Maximus without a re-install and recently went from an 8800GTX to HD4870. Removed the nvidia drivers, did driver cleaner, etc using the 8.6 hotfix catalyst drivers. Since the video card swap there are some issues. Sometimes it won't go into standby (S3) as set in the power properties. Manually putting it into standby works fine. Sometimes it will crash coming out of standby. I guess I'll hang on until the official 8.7 catalysts come out and do a re-install. In the meantime running at 3.33 GHz. Hopefully reinstalling Vista will get my OC back.

On a side note the Maximus was RMA to Asus and they have return shipped. It'll be interesting to see what shows up. It seems like the X48 should be > X38, so there must be some other issue with the OC.

---

What clocks are aiming for? Please post your settings.

Quote:

---

Originally Posted by Zucker2k

What clocks are aiming for? Please post your settings.

---

I'd like to get the same OC I had on the Maximus, 420x10. At this point I'll settle for being able to up the bus speed at all.

I've tried leaving everything on auto, manually setting voltages and timings and combinations. I tried Hartigan's settings and upped all of the voltages to the first setting where they go yellow. The result is the same - crash on Vista load. Keep in mind that all of this is with a bus speed of 340 and a 9x multi. That's what leads me to thinking it's a Vista issue.:shrug:

Quote:

---

Originally Posted by Clump

I'd like to get the same OC I had on the Maximus, 420x10. At this point I'll settle for being able to up the bus speed at all.

I've tried leaving everything on auto, manually setting voltages and timings and combinations. I tried Hartigan's settings and upped all of the voltages to the first setting where they go yellow. The result is the same - crash on Vista load. Keep in mind that all of this is with a bus speed of 340 and a 9x multi. That's what leads me to thinking it's a Vista issue.:shrug:

---

What are your current settings?

Quote:

---

Originally Posted by Zucker2k

What are your current settings?

---

Currently everything on Auto, except:
CPU Multi - 10.0
V DIMM - 2.10

Running OK with these settings. Going back to 9.0 multi and raising FSB from 333 stock to 340 = crash loading Vista at CPU voltage from auto to 1.7.

It's BIOS 0406 Beta BTW.

Quote:

---

Originally Posted by Ice009

Hey guys, I am considering getting this MB, but don't know which RAM to get for it.

Here are some options please tell me which you think will be better

1 - OCZ Reapers 4GB kit PC8500 (OCZ2RPR10664GK)

2 - Corsair PC8500 4GB kit (TWIN2X4096-8500C5DF)

3 - Team Group PC8500 4GB kit (TXDD4096M1066HC5DC-D)

I'm in Australia and these are the best options at the place I'm buying from. Which would be the best compatible kit from these? Or would I be better off getting PC6400 RAM and overclocking it?

---

everyone who has the G.Skills is very happy with them & the board is happy also .. soo there have been problems with the OCZ & corsair but then agin some are happy with them but I don't know which ones of each will work BuT the G.Skills do work quite well & those being the DDR2-1000 & DDR2-1066 2x2 gb's alot of us are running 8 gb's of these
PM me if u want the part #'s
the problem is not the board per say but the Intel X48 chipset email all the above Memory MFG's & ask about compatabilty of their memory with the Intel X48 chipset any good MFG'er will advise what will work of their's & what will not .... if they can;t that tells you something
hope this helps
this is a great board

Quote:

---

Originally Posted by Clump

Currently everything on Auto, except:
CPU Multi - 10.0
V DIMM - 2.10

Running OK with these settings. Going back to 9.0 multi and raising FSB from 333 stock to 340 = crash loading Vista at CPU voltage from auto to 1.7.

It's BIOS 0406 Beta BTW.

---

Okay, try this for starters:

Extreme Tweaker
Ai Overclock Tuner : Manual
OC From CPU Level Up : AUTO
CPU Ratio Control : Manual
- Ratio CMOS Setting : 10
FSB Frequency : 400
FSB Strap to North Bridge : 400
PCI-E Frequency: 112
DRAM Frequency: DDR2- 1066
DRAM Command Rate : 2T
DRAM CMD Skew on Channel A: AUTO
DRAM CMD Skew on Channel B: AUTO
DRAM CLK Skew on Channel A: 300
DRAM CLK Skew on Channel B: 300
DRAM Timing Control: AUTO (This is same as):

Primary: 5-5-5-15-3-AUTO (for 52)-6-3

Secondary: 8-3-5-4-6-4-6

Tertiary: 14-5-1-6-6

DRAM Static Read Control: Disabled
Ai Clock Twister : Moderate/Strong
Transaction Booster : Manual
CPL: 6

CPU Voltage :
CPU PLL Voltage : 1.50 - 1.56
North Bridge Voltage : 1.49v - 1.55v
DRAM Voltage : 2.1ov
FSB Termination Voltage : 1.300 - 1.36v
South Bridge Voltage : 1.100v
Loadline Calibration : Disabled
CPU GTL Reference : AUTO
North Bridge GTL Reference : x67
DDR2 Channel A REF Voltage :
DDR2 Channel B REF Voltage :
DDR2 Controller REF Voltage :
SB 1.5V Voltage :

NB LED Selection : NB Volt
SB LED Selection : SB Volt
CPU LED Selection : CPU VOlt
Voltiminder LED :

CPU Spread Spectrum : Disabled
PCIE Spread Spectrum : Disabled

With the Loadline Calibration disabled, if you haven't done the pencil mod, you'll need about .1v to make up the difference in vdrop (note, not vdroop).

Goodluck,

Richard.

Quote:

---

Originally Posted by Clump

I'm getting NOWHERE with my Rampage Formula/QX9650. I can up the multi to 10 and it works fine, but even increasing the bus 7 MHz to 340 Vista will not load. I've tried upping voltages to insane levels just to see if it helped and it doesn't seem to. I had this same CPU and memory at 420 x 10 on my Maximus Formula 12 hours + prime stable.

Since it will generally POST I'm wondering if maybe I have a Vista issue. I swapped from the Maximus without a re-install and recently went from an 8800GTX to HD4870. Removed the nvidia drivers, did driver cleaner, etc using the 8.6 hotfix catalyst drivers. Since the video card swap there are some issues. Sometimes it won't go into standby (S3) as set in the power properties. Manually putting it into standby works fine. Sometimes it will crash coming out of standby. I guess I'll hang on until the official 8.7 catalysts come out and do a re-install. In the meantime running at 3.33 GHz. Hopefully reinstalling Vista will get my OC back.


On a side note the Maximus was RMA to Asus and they have return shipped. It'll be interesting to see what shows up. It seems like the X48 should be > X38, so there must be some other issue with the OC.

---

look this has worked for me more than once as i had the same problem !
1st SAVE every thing on the drives !!!!


I'd would clear the CMOS (switch the CMOS lever on the mobo to off & then remove battery for an hour, replace battery then move switch back & at that point do a fresh install of the Vista ) this has happened to me & i did this this way & Boom every thing worked fine , my Asus P5E MoBo would do the same thing soo i've been down this road a few times ...

Hi Guys,

A quick question, has anybody come across an issue with quads over 500Mhz FSB??

The reason I ask is, I can boot my Q6600 @ 520 X 7 with 1.61v on the NB. The problem is when I do this I only have 2 core's available. They were prime small fft stable too for just over an hour :p:. As soon as it boots, on the VGA BIOS screen it reads core Count: 2. Is this just a weird bios bug or am I missing something? As soon as I drop back to 490 it's back to all 4 cores.

Quote:

---

Originally Posted by Lurker246

Hi Guys,

A quick question, has anybody come across an issue with quads over 500Mhz FSB??

The reason I ask is, I can boot my Q6600 @ 520 X 7 with 1.61v on the NB. The problem is when I do this I only have 2 core's available. They were prime small fft stable too for just over an hour :p:. As soon as it boots, on the VGA BIOS screen it reads core Count: 2. Is this just a weird bios bug or am I missing something? As soon as I drop back to 490 it's back to all 4 cores.

---

520 with a quad! No wonder you're having issues. Try more NB volts + more VFSB. Also, you may need more vcore.

Quote:

---

Originally Posted by hartigan

EDIT (4): With Vtt at 1.520V, I BSODed in about 10-15 minutes. Not sure where I can tweak this to get that full stability.

---

Go two steps up on the NB. What temps are you getting on the NB on load? If it fails, back down on VFSB to previous setting, and set both GTLs manually to x67.

Quote:

---

Originally Posted by Zucker2k

520 with a quad! No wonder you're having issues. Try more NB volts + more VFSB. Also, you may need more vcore.

---

Cheers, I pushed it about as far as possible voltage wise. 1.69 NB, 1.49 vcore. Since I'm still on air I can't push it to the red zone, no matter how many fans I throw at it :explode2:. The case of the missing cores continues....

Might be time to take the 'plunge' with a h2o setup (Bad pun intended ;) )

Here's a pic for those interested in such things;

http://img353.imageshack.us/img353/4389/520x7rh6.jpg

lurker is that the newest ver. of realtemp ? as it's showing the wrong cpu try installing it agin & check settings ,test

turn yer Vista side bar OFF when testing

Quote:

---

Originally Posted by bigslappy

lurker is that the newest ver. of realtemp ? as it's showing the wrong cpu try installing it agin & check settings ,test

turn yer Vista side bar OFF when testing

---

no probs, it was just a quick screenie, back on 460 X 8 and everything reads fine. It seems to go a bit mental over 500 FSB :D.

Quote:

---

Originally Posted by Zucker2k

Okay, try this for starters:
<SNIP>
Goodluck,

Richard.

---

I really hate it when I outsmart myself. I sat down to try the settings that Richard reccomended and when I got to the CPU voltage realized that I wasn't up to 1.6V, I had gone "up" to 1.16V. :rolleyes:

No wonder it wouldn't boot! I'm running at 400x10 now. Did a quick prime 95 run and was stable for 10 minutes or so. Now to some tweaking.

I don't think I ever made that noobish mistake before.:rofl:

Thanks for the help!:up:

Quote:

---

Originally Posted by Subsequence

I've got a question about this board since i'm considering this to replace my P5K-E/Wifi-AP which has the Sata bug apperantly.

Does this board have any problem's with WD's Velociraptor and Samsung's F1 1TB drives, since my Raptor reinitializes every once in a while making the system freeze for a few seconds but it passes WD's diag.

I plan to run this board with my Q6600 that does 3,2 at default voltage and 400FSB which is what i'm also looking for in this board, higher is ofcourse better but lower is not ;)

Btw the P5K-E is going to be replaced also because i want propper crossfire support.

Thanks for any input guys!

---

I run this board with two WD Velociraptors and two of the new WD Black 32Mb 1.0TB drives without problem.

The Rampage Formula is a very stable platform and should easily overclock your Q6600 to 3.2Ghz

My E8500 will do 4.2Ghz at 525Mhz FSB on air with the Rampage...

Quote:

---

Originally Posted by bigslappy

everyone who has the G.Skills is very happy with them & the board is happy also .. soo there have been problems with the OCZ & corsair but then agin some are happy with them but I don't know which ones of each will work BuT the G.Skills do work quite well & those being the DDR2-1000 & DDR2-1066 2x2 gb's alot of us are running 8 gb's of these
PM me if u want the part #'s
the problem is not the board per say but the Intel X48 chipset email all the above Memory MFG's & ask about compatabilty of their memory with the Intel X48 chipset any good MFG'er will advise what will work of their's & what will not .... if they can;t that tells you something
hope this helps
this is a great board

---

I can't get GSkill from the place I'm buying the RAM, also GSkill is a bit scarce here in Australia. Which would you choose out of the ones I've listed?

I will have to track down the email and send each company an email. So it's the X48 chipset not the RAM that is causing problems? I thought I also read somewhere that the PC8500's don't work too well on X38, X48 and that PC6400 works better?

Quote:

---

Originally Posted by Zucker2k

Go two steps up on the NB. What temps are you getting on the NB on load? If it fails, back down on VFSB to previous setting, and set both GTLs manually to x67.

---

I'm hitting 40*C with the North Bridge at 1.696V under load. So far . . . it's stable with that setting but, I've only run p95 for 10 minutes. I'll update more with that later.

One thing that is weird (and I don't think I can fix it) is that my Core 0 and Core 1 always report 10 degrees lower than Core 2 and Core 3. It doesn't matter what monitoring prog. I use (CoreTemp, RealTemp, Everest, etc.).

Also CoreTemp and Everest report 63* and 73* for cores 0,1 and 2,3 respectively. Whereas RealTemp reports them as 53* and 63* From what I've read, that has something to do with the TJ Max settings being different.

I'm comfortable with the temps reported in RealTemp (53* and 63*) as long as they are the most realistic temps. I just hope that I'm not surpassing the 70* limit mentioned for these processors.

If anyone has any thoughts on my 10* discrepancy between cores, I'd be curious to hear them. From what I've read, the best diagnosis is a stuck thermistor on the core that has 0 and 1 on it.



Thank you very much for your help Zucker2K!! I owe you.

I cant get anything over 400fsb stable, overall i am disappointed if this is all i can achieve, is there something i am doing wrong? i am on bios 0403

Quote:

---

Originally Posted by hartigan

I'm hitting 40*C with the North Bridge at 1.696V under load. So far . . . it's stable with that setting but, I've only run p95 for 10 minutes. I'll update more with that later.

One thing that is weird (and I don't think I can fix it) is that my Core 0 and Core 1 always report 10 degrees lower than Core 2 and Core 3. It doesn't matter what monitoring prog. I use (CoreTemp, RealTemp, Everest, etc.).

Also CoreTemp and Everest report 63* and 73* for cores 0,1 and 2,3 respectively. Whereas RealTemp reports them as 53* and 63* From what I've read, that has something to do with the TJ Max settings being different.

I'm comfortable with the temps reported in RealTemp (53* and 63*) as long as they are the most realistic temps. I just hope that I'm not surpassing the 70* limit mentioned for these processors.

If anyone has any thoughts on my 10* discrepancy between cores, I'd be curious to hear them. From what I've read, the best diagnosis is a stuck thermistor on the core that has 0 and 1 on it.



Thank you very much for your help Zucker2K!! I owe you.

---

No problem mate. Yes the difference between core temp and real temp is the 10C in TJMax. You can change that in settings.ini file which you can locate in the coretemp folder. Make sure this line is exactly like this: TjMaxOffset=-10; The temp difference is common to most 45nm chips. Hope things prove better this time by way of stability.

Quote:

---

Originally Posted by GAR

I cant get anything over 400fsb stable, overall i am disappointed if this is all i can achieve, is there something i am doing wrong? i am on bios 0403

---

I apologize for not getting back to you asap. I did try to post settings once only to find the server was under going an upgrade. Anyway, since you're on air, temps are going to be your biggest problem. What FSB/Clocks are you looking at?

Quote:

---

Originally Posted by Zucker2k

No problem mate. Yes the difference between core temp and real temp is the 10C in TJMax. You can change that in settings.ini file which you can locate in the coretemp folder. Make sure this line is exactly like this: TjMaxOffset=-10; The temp difference is common to most 45nm chips. Hope things prove better this time by way of stability.



I apologize for not getting back to you asap. I did try to post settings once only to find the server was under going an upgrade. Anyway, since you're on air, temps are going to be your biggest problem. What FSB/Clocks are you looking at?

---

I was hoping for atleast 450fsb, ive tried several different settings posted in this thread, with no luck tho, i would love to get 475, but i can settle for 450, thanks for the help.

Can you list the settings you have tried?