955 C3 And Water

[Cecil]

Replaced my X2 550 with a 955 C3 today. On a gig 790X-UD4P with 4gb OCZ 1150s. Water loop is all swiftech, Apogee GTZ, drive unit modded to MCP-355 with XSPC top, MCR-220 with res.

At stock, got 30C idle, 43C load across all four cores. Bumped vcore from 1.4 to 1.45 (1.44 after drop) and got 3.8 stable, 47C load on all cores. Have 15min left of OCCT at 3.9 1.47Vcore.

Benches, screen shots, and higher clocks to come.

[Cecil]

Here is the screenshot for 3.9. 4 is giving me issues, but Ill get it.

[BeepBeep2]

I just happen to be on a MA790X-UD4P with 4 gigs of OCZ (1.8v) 1150. :P Anyway, nice clocks...but I've got to speculate. Those temps do seem to be kind of high, my 965BE on a Sunbeam CCF has nearly the same temps.temps. Though, I need 1.50 (1.49v) to get 3920 stable. I'm also having trouble with 4 Ghz. :P

I'm thinking about moving to water in the future but I'm also curious on how a Megahalems with two Panaflo 38mm 2500RPM 107 CFM's would do. That costs around $120 with AMD mounting bracket, and I can't fit a large rad in my case.

[Cecil]

I just happen to be on a MA790X-UD4P with 4 gigs of OCZ (1.8v) 1150. :P Anyway, nice clocks...but I've got to speculate. Those temps do seem to be kind of high, my 965BE on a Sunbeam CCF has nearly the same temps.temps. Though, I need 1.50 (1.49v) to get 3920 stable. I'm also having trouble with 4 Ghz. :P

I'm thinking about moving to water in the future but I'm also curious on how a Megahalems with two Panaflo 38mm 2500RPM 107 CFM's would do. That costs around $120 with AMD mounting bracket, and I can't fit a large rad in my case.

My chip's stock vcore is 1.4. Takes 1.5 BIOS (1.488 load) at 3.9. When I had the loop on I was running 1.55V. The temps were high because I let it go overnight, and my PC is on the floor about 5 ft from a register vent and the heat was up to high that night.

I have since switched to a Corsair H50 for ease of use, and have lost nothing speed wise so far. 3925Mhz is still stable (have gotten it down to 1.5 as said above) and load is around the 56C mark. Im going to keep going till I get 62-65C load.

So if you are just trying to get max stable speeds, anything more then a top end air cooler or Corsair H50 is overkill.

[BeepBeep2]

Hmm, at 1.30v and stock speeds I load at 44c. (sigh)

It seems I could get something a little better, in time.

[gOtVoltage]

Post up a CPU-Z of the CPU and Ram Settings along with Temps....

(Your C3's will use way less volts than i am)

Currently running Unlocked with 4gigs ..JWhy just for fun with ALL four slots..
I know this Mobo has somes flaws and one of them is trying to go above the 269/270htt mark..

4100mhz+ with the 550 and 790X-UD4P...(using Water Also)

550@x4 4000mhz/2600nb =1.56vcore with 1x4gigs,
550@x4 4000mhz/2600nb =1.52vcore with 2x1gigs,

More Ram = More stress on DDR2 , especially when using Four slots on this Mobo.

I wish i had C3...= less volts for the same thing. Although ill know ill still push the Volts,hahaa


The NB and ram freqency\tweaking comes into play on DDR2 more so than DDR3, It takes a little More juice on all my Chips aswell once i go above a certain CPU freqency . C3 probally wont make huge difference on DDr2 except for Higher NB/HT. Were talking DDr2 here,Honestly i dont know if they really made a huge improvement over its IMC except for dual channel mode in four slots maybe. Im pretty shure it still takes more tweaking DDR2 compared to the onchip DDr3's IMC.

Simple to Try,,
Make shure the PCIE is set @ 100

The Object for now using ram @ 1100 or even lower ,,

First,, Make shure HT and NB is @ 10x(default)

Second ,, you want to use 250htt

Third ,,Set the CPUxmulti to get 4000mhz/2500nb
(Or Lower CPU at first if you are uncomfortable)

Keep the NBx10 and HTx10 @ 250...Why This will bring out any instability on the MOBO or CPU Even @ 250htt..99% it should boot and work %100..
Done this with several UD4P's ..My friends like to game and have me tweak/clock thier PC's, there lazy lol

Fourth,, Make shure your Ram timing is all on Auto(for Now)
it should set 5-5-5 15/18..what ever your rams Jdec is.

Fith,,CPUvid
use Voltage of 1.48v-1.5v

Sixth,, stock(default NBvid)

Seventh,, Check Temp in Bios Save then Reboot. Go back into Bios after Reboot Double check Voltages and CPU temp.. If all looks good Boot to Win7/Vista what ever.

On C3 you should be able to do this with a good Aircooler no problem.

Especially with 4x1gigs or even 2x2gig Dual Channel.


I usually Clock higher with 2gigs vs four gigs.
Maybe even See what ya get with two Gigs.. then try Again With Four Gigs .
Youll notice Less voltage when usually running 2gigs and High OC vs 4gigs and High Oc ..

Unless you are only using two sticks for either 4gig and 2gig. Thus it may be the same.. All depends on the chip too i guess.

Bios ATm i am using and some stuff ive seen
F7/F8 Bioses are the most stable for Higher HTT(hits wall @270)This will also Affect the I/O's and AHCI..I will eventually upgrade to C3 ..Waiting for a 550/C3 .
For now im just seeing what works Best and there should be another Bios popping up soon...The F9 worked but the F8/F7 seems best for now.

The last two weeks ive been running this 24/7...



The HTTlimit 435@StockxcoresX4mode 269/270htt max all chips same thing

The Screen below is done in Normal Mode Not Hybrid..Makes me believe that the Bios works the Same in Both Modes..

F7 F8 F9 hit the Same 269/270HTTwall with Variouse CPU's.while using HTT instead of just raising CPU Multi.
I usually use HTT and CPU xmulti to get a Nice OC


Now Hits Same Wall With X4cores




In the First Screen im using two different types of ram for 4gigs and its rock solid.....The Second Screen im only using 2gigs and i Always run the NB/SBchipsets one or two setting over stock..

NBchipset 1.3v,,SBchipset 1.3v...

Im using Hybrid ..
In (Normal Mode) id suspect there may be some HTT/Ram walls for you guys too....

So far tried Seven different chips
AII'sX3=X4Unlocked,,1x435
PHII'sX2=X4unlocked,,1x720be,,3x550be
Also 9600 Black and 9950 Black all seem to stop at 269htt....
I can run using multis up or way down on all chips and Clock past 4100+ on a few of the 550's(PHII 550...C2 . I think C3 should be Easier on this Mobo).

..Ive found using the 800 ram divider @ 250-268htt or lower with 4gigs and even 2gigs (4x1gig-2x2gig-2x1gig(Dual Channel) and (1x2gig single channel) works best.
If i had DDR2 that could do 1333 im shure 533divider or 800 would work fine. Unfortunatly weel never see 1333 ddr2.

NOTE:
Avoid using the (HT 8X multi )or you may have problems booting or unstability when setting Ram.. It can cause hard lock ups or no boot. Does this with all chips ive tried.I have used HTmulti x9up and HTmulti x7down with no hassle..

Im just waiting for a Uber Bios to pop up..I know this Mobo can handle things ... it just needs a super Bios Actually the Hybrid mode already is super for Unlocking

[BeepBeep2]

Hey gOtVoltage...

Why the NB Chipset and SB Chipset at 1.3v? I run the SB at 1.3v but the NB Chipset is at the stock 1.1v.

The best I can suffice even with 860 Mhz CAS 5 for 24/7 is 3890 @ 1.50v.

I've tried up to 4014 Mhz @ 1.56v and nothing is stable. On my best core (Core #1 or, core 2) I can validate it up to 4400 on 21c air. The rest of the CPU tops out at 4368. (Lucky one I can't replicate..in my sig)
I can run SuperPi 1M on core 1 (real core 2) @ 4330 with 1.575v also.

But I can't hold anything over 3.9 Ghz Prime95 stable.

My CPU-NB is strong, does 2.6 Ghz with 1.225v and 2.8 with 1.3v, then it takes 1.425v for anything over 2.9.

[gOtVoltage]

Let me dig through my Bios Pics...

Nb chipset 1.2v using 2gigs@1200mhz and 1.3v using 4gigs @ 1100-1150mhz on this Mobo.

SB chipset (1.28default) 1.3v on this Mobo to us more than 250htt .

Bios F7 and F8 F9 will all use the same Voltages..
Make shure you are setting NBchipset and SBchipset ..

CPUvid is CPU only
NBvid is for the IMC on CPU only ,,Thats not what im refering too.


NBchipset on the Mobo is supose to be with in 1.17v- 1.32v if you are Raising HTT or Overclocking Hard.. This Mobo Deafults to 1.17v which up to 2gigs and 3800 may be fine. Ive been running this Mobo At 1.3v from day one. Ive Clock Above 4000mhz 24/7.

If you Are Priming make Shure you are Not Cooking your PWM's. I will never prime a PHII Above 3800/3900mhz for more than a few hours ..Especially when im using above 1.55volts. These Mobos can not handle that fo rExtended use( 12hours)..Not even GD70...The PWM's will start to over heat and lose volts and other things that Cause erros ,although the CPU will be 100% stable..

Voltages with 2gigs @1200mhz ram..4gigs @ 1100-1150max.


Auto Ram setting and 533 divider for 1200mhz ram EPPsetting


Yeah i like this Mobo .. there are a few bugs and High HTT is one of them.. Also the HT and Ram settings can conflict too...

The main thing for C3 is how good its cooled.. If you can hit 4300..but not stable @ 3900 priming your Mobo can be problem not the CHIP. Prime is fine for a few hours but desinged for stressing with lower voltages ..Its most likely the Mobo cant hadle the heat. Even my M3A 79-Tdeluxe Primes stable up to a certain Voltage..Past that point i dont even bother because i kno i will ruin her.
I can run %100 stable all day@ 4100mhz +@1.56v .. with the voltage i use i will ruin this Mobo if i prime for more than 6-8 hours using 1.56v+ .Ive cooked one @ 1.54v after two days and one after 16hours @ 1.48v.. Priming is not the best thing to do for a Mobos power circiute.

[BeepBeep2]

Check that image out.

4306 @ 1.575 set volts, 1.568v.

The CPU-NB is at 1.40v and the RAM is running 2.3v set or 2.35v real. I'm just bored tonight so I thought I'd push CAS 4 on my old SuperTalent DDR2-800 4-4-3-8 2.2v kit.

Thats one core @ 4300, others are sitting around 2055 and the rest is told in the picture. I'm running the stock NB Chipset voltage and Stock SB Chipset Voltage.

And gOtVoltage:

If I run 3996 Mhz I am stable for the most part (real life stuff) @ 1.525v, if I try to up to 1.55v it gains me nothing. I fail Prime95 and OCCT within 10 seconds of them starting. 3900 @ 1.50v is stable and I tried 3900 @ 1.475 once and it BSOD'ed in gaming and stuff.

[Cecil]

Gotvoltage, what are you talking about? Your post made no sence at all.

First, more ram is more stress on the NB and IMC, not the ram itself. Also, 4 single sided sticks is the same as 2 double sided sticks.
Im using 2x2Gb OCZ 1150s, and they run fine around 1100 5-5-5-15, all subtimings as low as possible at 2.1V.
Second, C3 offers very little improvment, if any at all. The X2 550 C2 I had could unlock and be stable at 3850Mhz, and bench over 4Ghz easy. This 955 does the same, accept about 100Mhz higher for stable. I can easily bench over 4Ghz with ram maxed out. Also, 250HT is worthless, as it causes too much stress to bring NB speeds up which does nothing to help. Also, the ram would only be at 1000Mhz.

You also dont need any adjustments to SB or NB to get high speeds. At very high speeds a bump in NB may help, but Ive gotten past 4 with stock NB. It only wants to be cool, have about 1.5Vcore and +.1-.2 VTT and its fine. No need to complicate everything.

3925Mhz is 16hours prime stable with ram at 1080ish 5-5-5-15, vcore at 1.488 load, and +.1VTT.

[BeepBeep2]

Gotvoltage, what are you talking about? Your post made no sence at all.

First, more ram is more stress on the NB and IMC, not the ram itself. Also, 4 single sided sticks is the same as 2 double sided sticks.
Im using 2x2Gb OCZ 1150s, and they run fine around 1100 5-5-5-15, all subtimings as low as possible at 2.1V.
Second, C3 offers very little improvment, if any at all. The X2 550 C2 I had could unlock and be stable at 3850Mhz, and bench over 4Ghz easy. This 955 does the same, accept about 100Mhz higher for stable. I can easily bench over 4Ghz with ram maxed out. Also, 250HT is worthless, as it causes too much stress to bring NB speeds up which does nothing to help. Also, the ram would only be at 1000Mhz.

You also dont need any adjustments to SB or NB to get high speeds. At very high speeds a bump in NB may help, but Ive gotten past 4 with stock NB. It only wants to be cool, have about 1.5Vcore and +.1-.2 VTT and its fine. No need to complicate everything.

3925Mhz is 16hours prime stable with ram at 1080ish 5-5-5-15, vcore at 1.488 load, and +.1VTT.

Get your temps down below 50-51c on load and try to make that statement again. It's sortof the buzz around the AMD community that a bump in SB voltage will help CPU stability helps a tad. As for the CPU-NB voltage, Gigabyte boards tend to run low on the actual applied voltage and need anywhere from +.025 to .050/.075 more than other boards ie. the GD70 to reach the same speeds.

Four single sided sticks are NOT the same as two double sided sticks, because four single sided sticks still take up more memory banks and the IMC has to work nearly twice as hard to mantain speeds (it's controlling 4 sticks of ram).
I'm going to guess you are running into a temp wall...some guys get 4 Ghz on these easily with a modest 1.50v. (4.1 @ 1.54v and so on)

Also, be reminded that some chips just do not scale as well as others. The second core of my 965 C3 does benches over 4.3 Ghz while the other three cores don't prefer anything over 4.2-4.25. It's luck of the draw.

[Cecil]

I dont have many posts here, which makes it seem to long term members that I dont know this stuff. If you look in to me at EOCF and NBOC, and you will see I know what Im doing. I know all chips are different, and all cores as well. This isnt my first OC, by a long shot.

NB and SB at some point would need a change, but not untill they become the bottleneck. At 4Ghz, they arent, so there is no need to change them at that point. Also, you cant just snap a finger and lower temps. Ive used a high end water kit, and now a Corsair H50 and temps arent that different. Its just how these AMD chips are. 59C is fine, 50C would be better, but its just not happening. The water kit I had would probably get it down to about 54-55C, but its not worth the hassle. Its my gamer, so I dont care that much about it.

The point of this thread was just to share my expeirane with it. A lot of people feel the C3s are made of gold, when they really arent better then the C2s. There were some good cherry picked chips, but the majority are limited to about the same 3.9-4Ghz range the C2s were. I have three friends with them, and all of us are at about the same spot.

[vinister]

Why do the different programs display different CPU clocks?

I noticed during gaming everest shows my cpu slow down, all the way down to 2900mhz sometimes. Yet CPU-Z always stays at 3999 or 4000. CoolnQuiet is disabled and win7 is set to 'high performance' with cpu scaling turned off.

[BeepBeep2]

I dont have many posts here, which makes it seem to long term members that I dont know this stuff. If you look in to me at EOCF and NBOC, and you will see I know what Im doing. I know all chips are different, and all cores as well. This isnt my first OC, by a long shot.

NB and SB at some point would need a change, but not untill they become the bottleneck. At 4Ghz, they arent, so there is no need to change them at that point. Also, you cant just snap a finger and lower temps. Ive used a high end water kit, and now a Corsair H50 and temps arent that different. Its just how these AMD chips are. 59C is fine, 50C would be better, but its just not happening. The water kit I had would probably get it down to about 54-55C, but its not worth the hassle. Its my gamer, so I dont care that much about it.

The point of this thread was just to share my expeirane with it. A lot of people feel the C3s are made of gold, when they really arent better then the C2s. There were some good cherry picked chips, but the majority are limited to about the same 3.9-4Ghz range the C2s were. I have three friends with them, and all of us are at about the same spot.

I've seen guys doing 4.2 Ghz stable with no more than 45-46c on high-end water...

My Sunbeam CCF + Stock fan with TX-2 is doing 37c idle and 54c load @ 4 Ghz with 1.53v.

These C3's are quite a bit better than the C2's, when most C2's did not reach any more than 3.8 on air. Just look around int his subforum and you will find quite a broad range of results.
I know you know your stuff, your HWBot sig image tells more than enough.

http://www.xtremesystems.org/forums/...=239038&page=1

That threads got almost everyone's results in it.

[Cecil]

Ive seen a few at 4.2-4.5. But those are cherry picked, or early batches. The people I know, and myself had gotten them in the last month or so, and everyone of them has not gotten over 4Ghz stable. The X2 550 I had, the temps were the same or better, and took less or the same voltage as this does. I got 100Mhz more out of this stable so far, and I was expecting 400Mhz or so. So I am let down by it.

Im not saying they are worse then the C2s, or that they arent good chips. Just saying they arent what Ive seen them cracked up to be.

OT, I see you are from NE Ohio. Where at? I am in Akron, and know about 5 other OCers in the area. We are planning a LAN party soon.

[BeepBeep2]

Ive seen a few at 4.2-4.5. But those are cherry picked, or early batches. The people I know, and myself had gotten them in the last month or so, and everyone of them has not gotten over 4Ghz stable. The X2 550 I had, the temps were the same or better, and took less or the same voltage as this does. I got 100Mhz more out of this stable so far, and I was expecting 400Mhz or so. So I am let down by it.

Im not saying they are worse then the C2s, or that they arent good chips. Just saying they arent what Ive seen them cracked up to be.

OT, I see you are from NE Ohio. Where at? I am in Akron, and know about 5 other OCers in the area. We are planning a LAN party soon.

Btown XD
I'm 14 and I don't game much. Really, really passifist parents. No FPS'es in the house. At least, not anymore.

On topic:
I can do 4 Ghz + validations at the stock voltage and I hit a temp wall @ 55-56c nomatter what I do, even at 3.9 Ghz with 1.52-1.53v. I'm sure I could pull off 4 Ghz with a 2.7 Ghz NB 24/7 if my temps were just ~5c lower. I could reseat my CCF again but I'm just too lazy.

I took my rig outside a few weeks ago and got idle temps around 24-25c and got away with a 4.54 Ghz CPU-Z validation...I was doing Prime95 blend for around 10 minutes at 4 Ghz so the heatpipes wouldnt freeze up.
Considering I'm seeing alot of guys doing around 300-350 Mhz less 24/7 than what they can validate that would have meant 4.1-4.2 Ghz stable on the core at modest temps. (It's possible to get as low as 40c load with a monster WC setup)

EDIT:
For example I just validated this run with idle temps at 37c NB @ 2.78 and DDR2-1200:
http://valid.canardpc.com/show_oc.php?id=970560

[Cecil]

When I had my loop I was idle sub 30C, but with prime, it would slowly rise untill a couple hours later it hit the max.

All PH2s have a temp wall. With the x2 550 I had, it was at 52C where it was stable untill it hit 53C then Id get a BSOD. I haven't found that point with this one yet.

Ive got 4Ghz very benchable, and almost stable. I just dont have much time to mess with it lately, and its not a priority. Im sure Ill end up getting 4Ghz stable, and 4.2+ benchable.

[BeepBeep2]

True, and I think there is something wrong with my HSF mount. I'm doing 3706 in OCCT Large right now, I dont find any use doing Prime95 or OCCT for anymore than 2 hours...it's just going to end up killing the board.
The temp wall on this CPU is about 57c...I'm fluctuating between 48 and 50c at the moment an I'm 20 minutes into the test. I'll probably shut it off soon and call it stable.

I can bench 3DMark06 up to around 4175, then it just tanks at anything over that. I just find it funny that I can do around 3700-3750 on the stock voltage (reading as 1.39v) while I can't pass anything at 3900 with 1.5v+.

While it may not seem that there is a definitive temp wall on your CPU I'm sure you are on the borderline. I wouldn't be suprised if I get 3.8 stable on stock to 1.42-ish voltages by the end of the night, and I could probably do way better with better cooling. It's just frustrating.

[BeepBeep2]

Very weird, I just passed...then failed 9 minutes of OCCT Large...before I could only pass around 20 seconds @ 3920 1.525v. You guys have any ideas? Temps topped out at 58c.

[Cecil]

I run prime a minimum of 12 hours. There are times I pass OCCT and fail prime, and I notice that if I am close to stable, but not there yet, it usually fails prime just after 9 hours. Prime isnt going to kill your board either. The whole point of stress testing is to make sure the entire system is stable, so you dont end up killing something.

Anyway, at default 200HT the board runs 201.5. So I set it to 203, lowered my NB and HT multi to x9, lowered ram multi, and increased timings, increased vcore and vtt by .05, and bumped NB and SB .1. Still failed even after all that, just for 25Mhz more.

I did get 4.2 to boot and not have an issue for about an hour of normal use, so that should be benchable. Havent tried higher though.

[BeepBeep2]

Yeah, I just flashed F8 and I'm playing around with ACC...no more BSOD's on instability. I'm just getting "Error Detected of Core 2" and it's been Core 2 for the last three times it's failed. CPUNB is stock freq. with 1.275v,
and I'm trying to find ACC values to fit the slower cores.

And to the topic of Prime95 killing your board:

If you are running Prime95 on a Phenom II 965 C3 with a 4000 Mhz / 1.52v core speed and more than 2.2 Ghz NB Speed your TDP is already 200w...the board's power regulation section doesn't like supplying 200w for prolonged periods of time.

[afireinside]

Have you tried LinX? I make it a whopping 10 minutes into prime before my computer reboots I can pass 20 loops (hour) of linX and no errors in memtest so I run with it. Haven't had a stability issue in normal use. Was never a fan of P95.

[BeepBeep2]

I have before but not on this install.

OCCT and Prime usually fail for me within minutes, or don't fail at all. I dont remember what LinX did.

[Cecil]

Yeah, I just flashed F8 and I'm playing around with ACC...no more BSOD's on instability. I'm just getting "Error Detected of Core 2" and it's been Core 2 for the last three times it's failed. CPUNB is stock freq. with 1.275v,
and I'm trying to find ACC values to fit the slower cores.

And to the topic of Prime95 killing your board:

If you are running Prime95 on a Phenom II 965 C3 with a 4000 Mhz / 1.52v core speed and more than 2.2 Ghz NB Speed your TDP is already 200w...the board's power regulation section doesn't like supplying 200w for prolonged periods of time.

Gigabyte does fine handling that power. Thats why it has a 10 phase PWM. Its plenty to not have to worry. Ive run 400w i7 settings on the X58 without issue.

Have you tried LinX? I make it a whopping 10 minutes into prime before my computer reboots I can pass 20 loops (hour) of linX and no errors in memtest so I run with it. Haven't had a stability issue in normal use. Was never a fan of P95.

Lol you find something thats not stable, so you use another program and call it a day?
LinX is the best stress program for Intel, but sucks for AMD. 12+ hours of Prime is the best thing for testing AMD stability.

[crazydiamond]

cecil i disagree about c3 not better than c2,,, i personally gained ~200mhz and if you look at nvspec's most recent results with some 955's from newegg, some of his chips are getting 4.1ghz prime stable with only 1.43 vcore on air with normal ambients. i think there is a definate improvement over c2. i have noticed a lot of people on ddr2 arent able to get as good of clocks/volts, i wonder if that has anything to do with it.