+ 1Quote:
Originally Posted by kuebk
Printable View
+ 1Quote:
Originally Posted by kuebk
We definitely need someone that can MOD the BIOS and add some more fun stuff to the BIOS.:yepp:
These boards have significant potential but need bios to unleash their power.
I'm having a hard time getting it stable at 4.05GHz, DDR 1199 and tRD 6 with Prime95 and UT2004 together.
Everything runs fine, not a single sign of instability. The problem is that just before it's running for four hours it becomes totally unstable and reboots like someone is pushing the reset button.
I tried with giving it more FSB Termination Voltage but it didn't help. I tried with a little more NB Voltage but still the same. I tried with changing tREF to 4680T or changing the CPU and NB GTL Voltage Reference but it doesn't help.
I can run 100 passes of LinX stable and my memory is 1000% MemTest for Windows stable.
The only thing I didn't try is changing the PCIE Frequency. I'll give it a try with 110MHz to see if that fixes something.
The only thing that I find odd is that if I start a 3D application (UT2004) my memory frequency starts to fluctuate between 588MHz and 602MHz in MemSet.
Did anyone else have this problem when running a 3D application?
A-Grey - try the highest possible tREF. About memtest I would rather prefer to use dos version instead.
I'll try that tomorrow. Changing the PCIE Frequency didn't help.Quote:
Originally Posted by kuebk
I also use MemTest86+. I know it's stable with MemTest86+ but I did test it with MemTest for Windows to be really sure that my memory is running stable.
MemTest86+ can test much more memory than memtest.
Anyway I found out that with the highest possible tREF my ram is more stable and faster - too bad that you can't set it in bios so you could test ram using memtest86+.
I'm curious to see if changing the tREF to max has any impact on stability as well. I was under the impression that setting the tREF to max was only used to give lower latency whilst benchmarking and not generally used in everyday use. Out of curiosity what are you using to change the tREF memset or bar_edit?
Memset.
Something is seriously wrong. I've been testing with Prime95 in custom mode to change the test times to 3 minutes and I can't run it for more than an hour.
It doesn't fail in Prime95 it just crashes. With FSB termination Voltage at 1.14V it just crashes Prime95. With higher FSB termination Voltage it crashes earlier and not just Prime95 it crashes and reboots.
I've tested my memory with tREF at 16383T and it's stable.:up:
Now I'm going to try to run Prime95 and UT2004 together with tREF at 16383T to see if that fixes the problem.
I think that I've found the solution for my problems.
Prime95 V25.11 64 bit is buggy and causes these crashes.
It's running fine with Prime95 V25.9 64 bit.:up:
So changing tREF gave you nothing?
No, it didn't fix the problem.Quote:
Originally Posted by kuebk
that's odd the P95 is the problem because the stress test i ran with UT2004 was with 25.11 and it worked perfectly fine.
You're right Prime95 V25.09 ended up with a crash too but only later.Quote:
Originally Posted by leo27
I'm going to try with Intel Chipset Version 9.0.0.1008. I've updated them recently to the latest version and maybe this wasn't a good idea.
Hey Guys - I've recently switched to Vista 64-Bit from Windows XP 32-Bit and I'm having some issues when I run certain benchmarks (in particular SuperPI and Prime95 (Windows 64 v25.9 build 4). When I used to run these benchmarks in Windows XP and got an error the program itself would show an error message of some sort, (in SuperPI it would say something like "there was a rounding error" and in Prime95 it would show that one or more cores failed). Well in Vista 64-Bit I'm getting messages from the operating system saying that the program needs to close (even though there are no messages from the programs themselves)? Windows shows these errors as being "APPCRASH" and gives me the option to either "Close the Program" or "Connect To Internet for Possible Solutions To the Problem". Is this simply the way these benchmarks work in Vista 64-bit if/when there is a stability issue? In any case guys thanks very much for any and all help you can give (I have indeed searched for help on this problem both here at XS Forums and on Google with very little luck)?
I use vista x64 as well. Usually when there is an error P95 will say a core has failed. If your system is really unstable you could get the occassional BSOD or restart. i remember having a similar problem with P95 that you are describing and i think i fixed it be downloading a different version of P95. Try P95 x64 25.11 and see if that helps.
Thanks Leo27 for taking the time to reply, in hindsight I should have known to try a different version first (before posting my question)! I just finished completely re-installing Vista 64-bit and I'm gonna try Prime95 25.11 and hope for the best. Thanks again!
no problem let us know how it goes
also A-Grey
I'm using Intel Inf 9.0.0.1011 and havn't had any problems on XP or Vista x64. In fact a device called ATK0110 ACPI Utility only showed up with this version of the intel inf. I know that this is related to an ASUS feature but the point is this is the only inf i've used that found this device.
I'm currently using Intel Chipset V9.0.0.1008. I was looking for V9.0.0.1011 in a zip file but I can't find it.Quote:
Originally Posted by leo27
-------------------
Executing the Intel Chipset Driver V9.0.0.1011 from the Command Prompt with this command -OVERALL should do the trick to update from V9.0.0.1008 to V9.0.0.1011.
You can find the Intel Chipset Driver V9.0.0.1011 at X-drivers.com.
The funny thing is that Intel Chipset V9.0.0.1008 isn't mentioned in the X-drivers.com X48 section and it's the only version that you can download from Intel's official downloadpage.:confused:
It is a bit wierd that it can't be found on intel's website but 9.0.0.1011 works so i don't question it.
9.1.1.1015 are latest from station-drivers.
Finally stable at 4.05GHz and DDR 1199 with tRD 6.:D
This is with Intel Chipset Version 9.0.0.1011 and tREF at 16383T.
http://i431.photobucket.com/albums/q...D6-LinX-BI.jpg
http://i431.photobucket.com/albums/q...D6-Prime95.jpg
Very nice! (read with borat accent) :D
Pretty nice A-Grey but is that memtest86+ stable? I think rather not.
Anyway I'm kinda confused about your vMCH, mine board wants at least 1.45v for 450fsb, 600mem, pl6 and I'm on c2d.
I don't know if it's MemTest86+ stable with tREF at 16383T. I know that it's MemTest for Windows stable and LinX doesn't give me errors so I think it's OK.Quote:
Originally Posted by kuebk
About vMCH. Maybe it's the temperature. My NB always stays under 40° Celsius with my watercooling setup. Or I'm lucky that it doesn't need more voltage to run at FSB 450MHz.
25 loops are not enough to say it's stable, few days ago linx at mine pc crashed after 150 loop out of 200.
Looks like heat issues.Quote:
Originally Posted by kuebk
I'm not sure what caused it. MCH is cooled by hr-05 with 120 fan which also throws fresh air on mems. TRUE on cpu.
I can run 100 loops in LinX without problems. That's more than enough for me.Quote:
Originally Posted by kuebk
25 loops is maybe not enough to say it's stable but if it isn't stable Prime95 is going to fail too.
I think it's rather my ram, probably wants more power.
Would it be worth me running my ram at a tREF of 16383T? I've already run HCI memtest at this speed and it passed 300%. Is there any benefit to this(my settings are on page 116)?
Better scores in everest memory benchmark. Apart from that? Not much.Quote:
Originally Posted by leo27
So would it be recommended to change to the higher tREF?
Does anyone tried to put 2x1GB + 2x2GB = 6GB on this board? I am planning to get one Transcend aXe 1200 2x1GB kit for some benchmarks and wondering if it is posible to run stable 6GB setup.
I see no reason not to use it when you don't have stability problems with tREF at 16383T.Quote:
Originally Posted by leo27
It's possible. I know someone that uses 6GB on the Rampage Formula.Quote:
Originally Posted by OLDcomer
im using 2x Corsair Dominator CM2X2048-8500C5D and 2x Corsair Dominator CM2X1024-8500C5D without issues. However i intentionaly bought same model/brand so i dont have any issues...Quote:
Originally Posted by OLDcomer
I use Adata 1066 2x2GB and Kingmax 1066 2x1GB in dual channel. Xeon X3350 8x450, mem 6:5 (DDR2 1080 & TRD7) - works like a charm.Quote:
Originally Posted by OLDcomer
Here's an other picture running Prime95 for 8 hours with tREF at 16383T.
http://i431.photobucket.com/albums/q...D6-tREF163.jpg
I did a custom blend test of Prime95 with a test time of 3 minutes. I did this to run more FFT's of different sizes. It finds problems much faster like that.
I think that I can say that it's stable with these settings.:yepp:
Anyone is still using 0219 bios? How to change multiplier using it? I couldn't find any option to change it.
you have to use the + and - keys on the keyboard
Q9650@4.185GHz - G.SKILL F2-9600CL5D-4GBPI@1239MHz - tRD 6 - tREF 16383T- BIOS 0902 :cool:
http://i431.photobucket.com/albums/q...RD6-tREF16.jpg
No I haven't because I haven't found any option which would let me change the multiplier.Quote:
Originally Posted by Breathless
Congratulations, A-Grey! Great bandwidth.Quote:
Originally Posted by A-Grey
Thanks OLDcomer. I've passed the 10K and latency under 50ns.Quote:
Originally Posted by OLDcomer
My board loves this Q9650. :D
http://i431.photobucket.com/albums/q...tRD6-tRE-1.jpg
A-Grey does setting tREF to 16k gives any noticable profit in stability?
It's difficult to say. The memory is 100% stable with tREF at 16383T but it's also stable with the BIOS default 3120T.Quote:
Originally Posted by kuebk
When I was using the Intel Chipset Device Software V9.1.1.1015 (for ASUS motherboards :confused:) and had those crashes, it could run longer stable before crashing with tREF at 16383T.
That makes me believe that using 16383T for higher memory speeds is better than using the BIOS default 3120T.
I think that BIOS 0902 is the only BIOS that makes it possible to run with 16383T without stability problems. At least for me with the G.SKILL F2-9600CL5D-4GBPI. I've tried it with BIOS 0803 and it wasn't stable.
There's a new Intel Chipset Device Software Version available for download from Intels downloadpage.
Intel Chipset Device Software V9.1.1.1019
WOOW!:yepp: thats great scores mate with such a low volts. That cpu you have there realy:bananal:Quote:
Originally Posted by A-Grey
Finally we can see some great results with our "old and forgotten" mobos :D
That was a few days ago. Today I tried to run Prime95 blend and it fails within a few minutes. With a room temperature of 28° Celsius it isn't the coolest day to do some stress testing with Prime95 at 4.185GHz.Quote:
Originally Posted by greg.m
Back to 4.050GHz and run Prime95 blend. There's still something wrong. It runs fine for hours stable at 4.05GHz but sometimes I get a BSOD - Machine_Check_Exception while running Prime95.
I'm going to try with disabling Execute Disable Bit because I think that this is causing these BSOD's.
Not this is causing them, I have it disabled and sometimes I'm getting them too.
This also happens when I try to lower the CPU Voltage with one step in the BIOS. My system crashes while running LinX after a few stable passes. This is very strange because there is no sign of instability before the BSOD.Quote:
Originally Posted by kuebk
I don't think that my PSU is the problem because all voltages are stable. I don't really know what is going on. :confused:
Higher vCore could be the solution for the MACHINE_CHECK_EXCEPTION.
Here's what I've found: http://www.ocforums.com/showpost.php...5&postcount=10http://www.ocforums.com/showpost.php...5&postcount=10
If this is the solution to make these BSOD's go away than we can skip all these long linpack stress tests and move to Prime95 after a few passes linpack stable.
A-Grey do you have LLC enabled?
i think that it has something to do with vtt(att least for me)I had similar problems and 1-2 steps higher vtt solve the problem.Quote:
Originally Posted by A-Grey
kuebk - I don't mean to speak for A-Grey but I'm pretty sure he has LLC "Disabled", at least he did during some recent stability testing. He referred me to some BIOS settings he posted (when running his Q9650 @ 450Mhz x 9.0x) which I printed out and just checked. I had the same results as you greg.m with a small increase in Vtt voltage, really made a world of difference. With that said what do you currently have your Vtt set at?
@greg.m
I'll give it a try with vtt 1.16V real.
@kuebk
No, LLC isn't enabled. It doesn't work with my board.
@Roger_D25
Do you know how much watts you need to run at 4.05GHz and your GTX295?
A-Grey set VTT to 1.24 in bios.
It was set at 1.24V in the BIOS. The board undervolt it to 1.14V.Quote:
Originally Posted by kuebk
I'm going to try with 1.26V in the BIOS.
Disabling Execute Disable Bit made it a little better. Prime95 crashed but Windows was still running.
I've tried with 1 step higher vtt and it fails Prime95 after 1 hour and 8 minutes.Quote:
Originally Posted by greg.m
To see if it's the CPU voltage, I lowered it one step in the BIOS and BSOD - MACHINE_CHECK_EXCEPTION after 10 minutes.:shakes:
A-Grey - I dont' know the exact wattage needs but I've had no problems with my current Thermalright 850w. With that said I do have a "P3 Kill A Watt" meter which can measure volts, current, and watts with reasonable accuracy. I'll set it up later tonight and figure out how many watts my PSU is pulling from the outlet (granted this isn't perfect but it will give an indication of what my system is pulling). With my hardware I can't specifically measure just the CPU and GPU (only what the whole system is pulling). I remember the number being less than I thought it would be when I first tried the "Kill A Watt" meter out. Will post back in a few hours after I get home from class.
The reason I'm asking this is because my PSU is an OCZ EVOStream 600W and it's probably not enough to run at 4GHz+ and a BFG NVIDIA GeForce GTX 285 H2OC.Quote:
Originally Posted by Roger_D25
It doesn't really matter if it isn't enough. I'm going to buy an FSP Booster X3 at the same time for the BFG NVIDIA GeForce GTX 285 H2OC. Than I'm going to have 300W for the GFX Card and 600W for everything else.
All I have to do now is save some money.:D
I highly doubt that 600w is not enough for such config.
I'm not going to try without the FSP Booster X3. I'd rather spend the extra 22 Euro to buy a FSP Booster X3 than risking to blow up my PSU.Quote:
Originally Posted by kuebk
I need at least 575W according to the Antec PSU calculator and there could be voltage peaks around 620W. Stability would suck without the extra 300W that the FSP Booster X3 can give.
Increasing the CPU Voltage with 2 steps in the BIOS fixed the MACHINE_CHECK_EXCEPTION BSOD problem.Quote:
Originally Posted by kuebk
This is very strange. I didn't expect to need more CPU Voltage after being stable for 100 passes in LinX.:confused:
I bought today Q9550. After your posts I'm afraid if the PSU will handle my configuration. Of course after CPU OC.Quote:
Originally Posted by A-Grey
LOL'd @ 600w not handling quad oc with 4870.
I couldn't wait. I've just ordered an EVGA GTX 285 Hydro Copper and a FSP Booster X3.:D
Thats awesome A-Grey, what card is the GTX285 replacing? I didn't know much about the FSP Booster until just checking it out, the performance of that PSU is very impressive for its size! I assume your going to use it for the GTX285 (or for something else)? Earlier today before going to class I setup the "Kill A Watt" meter to measure how many watts my PSU was pulling from the outlet. I didn't have much time to play around but I ran Prime95 25.11 Small FFT's along with 3DMark Vantage's "Jane Nash" benchmark which is the most power hungry of the Vantage benchmarks. Below is a photo of my setup (to give you guys an idea of what my system is composed of) and a photo of the meter while those benchmarks were running. The wattage peaked at about 491watts (but becaues of the camera delay it captured a lower number), lol! By the way my system was only running at 3600Mhz (400Mhz x 9.0x @ 1.25v and GTX295 @ 650/1418/2200 on stock volts). I'm still trying to get my setup stable at 4050Mhz (I can boot at 4050Mhz but as soon as I run Prime95 my system BSOD's). As soon as I get things stable I'll post the overclocked power consumption.
http://img171.imageshack.us/img171/7761/002lw.th.jpg
http://img169.imageshack.us/img169/8668/003fo.th.jpg
I have no idea why you guys are buying such a useless things like FSP Booster X3, better would be to sell current psu and buy better one.
Finally my prime stable 480x9 results:
http://img193.imageshack.us/img193/5...klippprime.jpg
http://img29.imageshack.us/img29/8766/testsu.jpg
I bit high volts but it works:D
Excellent work greg, that is very impressive despite the high voltages! That is the highest stable FSB I've seen on this board while running a quad-core on air/water :clap: I can see your voltages in the screenshot but what were your actual "BIOS" settings for the following voltages if you don't mind sharing?
VCORE:
FSB VTT:
NB:
I'm using an ASUS EN7800GTX@513MHz. It's good enough for what I'm doing but I want it replaced for a new one with a full coverage waterblock to coolQuote:
Originally Posted by Roger_D25
the VRMs and VRAMs too.
Do you need 1.25V for 3.6GHz at full load? You're going to need at least 1.37V for 4.05GHz and you'll need it for your GTX 295.
I'm not really sure about that. Did you read the review?Quote:
Originally Posted by kuebk
With voltage fluctuation of only 0.4% when placed under 80% load, the Booster X3 provides more stable rails than a lot of high-end power supplies which is testament to FSP considering the size of the unit.
Your going to see some huge performance increases when you install that GTX285, granted there is nothing wrong with the 7800GTX but it can't compete with a GTX285 even at 513Mhz! From the testing I've been able to do so far I don't think the Q9650 needs 1.250v to run stable at 3600Mhz but I haven't found the right combination of settings/voltages to get the VCORE down yet (I'm too busy trying to get 4050Mhz stable).
EDIT: The reviews I read on the FSP Booster all said essentially the same thing, that little booster is quite impressive given its size and cost. The rails are very solid and stable and its probably cheaper than selling your current PSU and upgrading to a more expensive and more powerful PSU? In any case make sure to keep us updated on how it works for you if you get the time.
Whoops:sonic:... Well done mate.:clap:Quote:
Originally Posted by greg.m
The GTX 285 is 4 X faster than my old 7800GTX.Quote:
Originally Posted by Roger_D25
I'm not sure if it's going to be an EVGA GTX 285 Hydro Copper. I've got a mail from my computer dealer saying that they don't have them right now and he doesn't know when they are going to be available and it's going to cost more than I thought it would be.:shakes:
I can get a MSI N285GTX HydroGen that can be delivered really soon but it isn't the card that I want. I'd like to have an EVGA GTX 285 Hydro Copper or a BFG GTX 285 H2O.:yepp:
Thanx mate:) Here is bios settings:Quote:
Originally Posted by Roger_D25
The problem i have is that the last core doesnt work as fast as the others do:mad: and its temp sensor is stuck@lowest 41c, doesnt go lower. I would rma my cpu but im thinking the possibility to receive one that doesnt clock as well as this piece of s..t:rofl:Code:Ai Overclock Tuner [Manual]
CPU Ratio Setting [AUTO]
FSB Strap to North Bridge [AUTO]
FSB Frequency [480MHz]
PCIE Frequency [105MHz]
DRAM Frequency [1204MHz]
DRAM Command Rate [AUTO]
DRAM CLK Skew on Channel A/B [AUTO]
DRAM TimingControl [Manual] @ 5-5-5-15 everything else auto
DRAM Static Read Control [Disabled]
Ai Clock Twister [Moderate]
Ai Transaction Booster [Manual]
Common Performance Level [07]
Pull-In of CH A/B all disabled
CPU Voltage [1.41250]
CPU PLL Voltage [1.54V]
North Bridge Voltage [1.69V]
DRAM Voltage [2.00V]
FSB Termination Voltage [1.50V]
South Bridge Voltage [1.100V]
SB 1.5V Voltage [1.55V]
Loadline Calibration [Enabled]
CPU GTL Voltage Reference [0.62X]
NB GTL Voltage Reference [0.67X]
DRAM Controller Voltage REF [Auto]
DRAM Channel A/B Voltage REF [Auto]
CPU Spread Spectrum [Disabled]
PCIE Spread Spectrum [Disabled]
CPU Clock Skew [Delay 100ps]
NB Clock Skew [Normal]
CPU Ratio Setting [AUTO]
C1E Support [Disabled]
CPU TM Function [Disabled]
Vanderpool Technology [Disabled]
Execute Disable Bit [Disabled]
Max CPUID Value Limit [Disabled]
I asked for a RMA for my Q9650 because it was one of the worst that you can find and had a stuck core temp at 39° Celsius on the fourth core.Quote:
Originally Posted by greg.m
Now I've got a really good one but it also has a stuck core temp at 40° Celsius on the fourth core.:confused:
Just wondering is there anyone that doesn't have a stuck sensor on the fourth core of his Q9650?
Join the club guys, one of my cores is also stuck at 42C (no matter what it goes no lower than that)! I have thought about RMA'ing it many times but like greg.m mentioned I'm worried I'd get some POS replacement chip (and it would mostly like have a stuck temp sensor anyway), lol! A-Grey did you RMA your first chip through Intel or through the retailer you bought the chip from? I have no experience with Intel or their warranty service but I imagine its quite slow? By the way greg.m thanks for taking the time to post all your settings, that is a huge help for me!
Something else ive learnt about this cpu is that after 60c-65c fails. So icant linx test my current settings because temps will hit red - 65c-70c - and probably cpu will fail. Watercooling is the only way for these speeds...Quote:
Originally Posted by A-Grey
My retailer asked for the RMA. All I did was waiting to receive the new one and give the old one back.Quote:
Originally Posted by Roger_D25
Is it the last core in RealTemp that is stuck at 42° Celsius?
That was nothing, forgot to mention that i use 0902bios as wellQuote:
Originally Posted by Roger_D25
:yepp:Quote:
Originally Posted by A-Grey
dreadful 1.3 vider instead
I see your settings, great clocks but what about temps during stresstesting..Quote:
Originally Posted by OLDcomer
You can't have it all.:DQuote:
Originally Posted by OLDcomer
Look @ CPU Watts:shocked::shocked:
Linx:
http://img34.imageshack.us/img34/2013/skrmklipphh.jpg
Prime small ffts:
http://img34.imageshack.us/img34/9559/skrmklipp2.jpg
OMFG:shocked:
The reason that the FSP Booster X3 is so cheap is that it only comes with 2 x 6 pin power cables.
There's also the FSP Booster X5. This one can give you 450W and can be used with 6 or 8 pin power cables.
A Grey - It is indeed the last core in CoreTemp that is stuck at 42C!
greg.m - I've been told that LinX puts more stress on the CPU but I didn't realize it was that much of a difference over Prime95!
It's strange that we all have a stuck sensor around 40° Celsius with our Q9650 and our ASUS Rampage Formula.Quote:
Originally Posted by Roger_D25
OLDcomer hasn't got a stuck sensor around 40° Celsius. He isn't using a real ASUS Rampage Formula.
LinX does indeed put a lot more stress on the CPU than Prime95 Small FFT's does and that's why I didn't expect to have to increase the CPU Voltage to run Prime95 stable. I don't understand why I still have this MACHINE_CHECK_EXCEPTION BSOD while I'm 100 passes stable with all available memory in LinX.
I can't use LLC Enabled and the CPU Voltage drop is huge under load with a Q9650. That could be the problem that I have to increase the CPU Voltage with 2 steps in the BIOS.
The ASUS Rampage Formula does also need more CPU voltage than other boards do. Is this caused because of the 8 phase power? :shrug:
Is there some genius among us that can explain all these strange things?
Mine e8500 is also stucked at 40*c - that's the lowest temp showed by this board, but I haven't experienced similar behaviour while using e6750.
Do you think its possible guys that the sensors on our motherboards are causing this "Stuck Sensor" problem or is it more likely to be the CPU or sensor reading software? Granted many people have stuck sensors on their Intel chips but you brought up a good point A-Grey that we all have this issue on the Rampage (with both dual and quad core chips)? I've run an E8400, E8500, E8600, Q9550, Q9450 and my current Q9650 on this motherboard over the last year and the only chip without a stuck sensor was the E8400 (which was an ES chip)? I'm pretty sure its got nothing to do with the motherboard but I'm curious what you guys think?
As I already said I haven't got any problem with 65nm cpu.
Aircooled, 22C ambient.Quote:
Originally Posted by greg.m
http://img166.imageshack.us/img166/6508/9x470.th.jpg
My first attempt with Q9550 OC :)
http://img9.imageshack.us/img9/8719/q95504ghz13v.jpg
My first attempt with Q9550 OC :)
http://img9.imageshack.us/img9/8719/q95504ghz13v.jpg
Not bad for your first attempt.:up:Quote:
Originally Posted by Burdoror
I think this is good result - specially with the CPU Voltage :) In my E8400 I did stable 24/7 4.2Ghz, but with 1.4V :)