M17 R2 OC Woes.....

So I finally ran wprime 1.55, apparently its 1024M is more rigorous than its newer version 2.04. Originally using a 26 multiplier and TDC/TDP 90/75 via throttlestop as thats what was possible using wprime 2.04, I got BSOD using wprime 1.55 with 8 threads going.

I reduced the multiplier to 25 and it passed wprime 1.55's 1024M test in 285.63 seconds with a max CPU temperature of 97 C.

I havent increased the voltage or overclocking in BIOS, also the DDR3 voltage is at 1.5.

With the GPU's, I got BSOD running 825/1200 in Furmark under a stability test. I went with an 825/1150 which passed Furmark with flying colors but had Crysis 2 freeze up on me, Im not sure if its throttlestop or the overclocked cards or whether they are now both competing for a limited amount of incoming wattage that is culprit. I reflashed the cards to 800/1100 and will run a stress test with Crysis 2, fingers crossed.

I think with the cards moderately overclocked and throttlestop also drawing more current, the problem may be a power supply issue, not in that the power supply is faulty, its just that the increased demand isnt something it can accommodate.

Per someones advice on TechInferno, I tried a combination of 3% BIOS OC, 75 mV, and a 25 multiplier TDP/TDC 90/75 and got BSOD in wprime 1.55. But not only that, the reported speed in CPU'z and throttle stop was fluctuating wildly, dropping as low as 900 Hz, and the FID was scaling back significantly as well. With just throttle stop and no BIOS OC or additional voltage, the reported speed is much more stable, fluctuating between 3.1 and 3.3 Mhz and the FID only scales back to low 24's (25x).

Im not shooting for a bench OC, I am trying to attain a solid, everyday use OC as high as possible but keeping the CPU temperatures in the mid 90's and the GPU temperatures in the low 80's.

It seems that its only with wprime 1.55 that I get such high temps. While playing Crysis 2 the CPU was in the low 80's, with the GPU's around 72 and 74.

Should I increase the voltage to the cards? Does anyone know of a stable, moderate OC for both the 920 and X-fire?

Also posting these questions in TechInferno

 

Wish I had more time to address your question as throughly as I would like, but I'm running from one job to the next so I only have a few min.

wPrime 1.55 is the standard version for benching so it is what I use to judge my temps on. A lot of the temp issues are heavyly influenced by ambient temps. First I run 100/85 for everyday use. I always have RealTemp running when gaming to monitor. My temps in Black Ops get up to the 90's. What you should do is remove the keyboard and run wPrime 1024 while your adjusting the screws. Monitor your temps and when the screws are adjusted properly, your temps at 90/75 should taper off. Put the keyboard back. I see a lot of systems that hit the thermal limit and throttle because the screws are not properly adjusted. When I had al screws all the way tight on the CPU heatsink, it caused the system to heat up more. So this should help with that issue. Pain, but worth the time. Makes more differance then good thermal paste.

For the GPU's, are you on stock voltage? If so then I would bump the voltage up to 1.1v from 1.05v. If your at 1.1v, then move to 1.15v. Thats as high as the voltage can be set, but the temps should still be good. Run a test with those changes and see what you get. I never overclock my GPU's unless I'm benching, but if the clocks are still an issue, try 808/1100. May help. Hope it works out for you.

Maybe I'll drop in later to pass along more info. Gotta run! Adios for now.

EDIT: Forgot to add, I run with a 75mV increase as well because thats the only way I can keep my system stable at 25x. Realistically, your system will rarely be pushed as hard as wPrime 1.55 pushes it, I don't remember if mine BSOD later during wPrime or not, but I have yet encounter any game throttling, overheating when running virus scans or anything of the sort.

 

Thanks for the advice! I know youre very savvy with overclocking. Im going to up the voltage from 1.05 to 1.1 on the GPU's, keeping them at 800/1100 and 75 mV in BIOS.

The reason I suspect its a power supply issue is that I know of others with slower CPU's that are running higher settings (825/1200 etc) on their GPU's without additional voltage and arent having the freezing, BSOD issues etc.

I will play around with the CPU screw tension later, the good thing is that there is minimal variance in heat across the cores, typically 1-2 C.

Do you think its a power issue?

 

Not really. I've seen the psu pull 270w benching. I think it May just be voltage. If it was psu relate the psu would trip and shut down. you'd have to unplug it for 10 sec to reset it.

 

If you did not raise the voltage to 1.1V that may be why it was fail @ 825/1200. I have mine at 825/1200 @ 1.1v.

 

I think you guys might be right, it may have been a voltage issue.

I upped the voltage from 1.05V and am currently at 825/1150 1.1V. Ditto for the CPU, (75 mV).

Doing a Furmark Stability run my GPU's were at 84 and 88 C.

The ambient temp is a bit warmer where I am (public library) and I had the laptop on a cubby type desk with heat accumulation behind it. Expect lower temperatures elsewhere.

I made another 3Dmarkvantage run and my CPU temp did hit a max of 99 C. And contrary to what I reported earlier, Core 0 was averaging 10-15 degrees hotter than cores 1,2 an 3. Is this to be expected?

Going to do a wprime 1.55 run soon. I hope it doesnt lock up in Crysis 2 as that seems to be the real gauntlet for overclocking.

Result: P17186

Result

If this thing is stable while gaming and I get GPU temps in the mid 80's and CPU temps in the high 80's then Im done.

 

I cant seem to run 825/1200 or 825/1150 on the GPU's even with 1.1V.

In Furmark, with 1920x1200, AAx2, etc, it failed and shut my computer down.

Before failure the GPU's temperatures were 82 and 86.

It failed the first time in 1024x700 resolution in about 6 minutes, that time only Furmark quit.

The 2nd run, this time 825/1150 1.1V, 1920x1200 it failed in about 3-4 minutes and restarted the pc.

Doing some 800/1100 runs now with 1.05 voltage. Thinking of using 1.1 volts to keep things stable.

I increased my TDP/TDC to 92/80 and it survived wprime 1.55's 1024M in 3:57.

Any other recommended settings?

 

mmmm che my sign, you may want to try the same settings.

It is my daily setup. I did how 5150Joker sugested when installing the xm chip, with the laptop opened and adjusting my heatsink screw by screw and monitoring temps. I cannot get any game to rise my core temps over 78 degrees, with 28 degrees room temp. Both cards are flashed 800/1000 and the higest temp while gaming is 81 degrees peak, mostly they stay between 70 and 77 degrees. I have no overvolt applied on gpu/cpu.

Edit: my tdp/tdc is 95/85

 

I run mine @808/1200 voltage original and let it run everyday with no probs at all, but when i tried raised up the voltage to 1,1 or 1,15 even with that clock above, I couldn't run even Vantage and any game,I dont know why...

 

This is amazing, I know of others as well who are getting away with similar speeds on stock or 1.1 voltage but the most I can muster is 800/1100 on 1.1V. I am going to try a flash of 820/1140 1.1V later, fingers crossed.

Have you passed Furmark's stability test using 1920x1200, AAx2, with the "heatwave effect"? I run mine to 10 minutes as it has survived this test for 6 minutes using 825/1150 1.1V when I got a BSOD. It seems 10 minutes is a minimum amount of time to really determine the stability.

I really need to adjust the CPU screws, what do you use to monitor the temperatures of the cores, Hwinfo32?

Are you stressing the CPU while adjusting? Because my idle temps are really tight, within a few degrees of each other across the cores.

So youre running a 26 multi on core 0, 25 for core 1, 24 for core 2 and 23 for core 3 all at 5% BIOS OC? WOW.

Earlier I tried all 25x, 3% BIOS OC, 75mV per the advice of someone on TechInferno and I got a BSOD running wprime 1.55.

Right now it seems pretty stable at 25%, no BIOS OC, and 75mV.

I noticed that in my case, under load core 0 is typically 10-15 degrees hotter than the others which remain within a degree or two of each other. They all have the same idle temp, which is around 47 C (75mV) with an ambient temp of ~65 F. But when stressed again Core 0 gets a lot hotter.

Are your cores doing the same? Would it be ok to set core 0 to 26x like you have if not? Or should I do the reverse, setting core 3 to 26x, core 2 to 25x etc etc?

Experimenting, I havent crashed wprime 1.55 yet with TDP/TDC 95/85 either, I have my fall back profile at 85/75 as the temps will get up to 99C with the foremost profile.

I have tried 825/1050 at 1.1V but got a BSOD in Furmark 1920x1200-AAx2 with the "heat wave" effect.

So then I upped the voltage to 1.15V and it crashed even faster! It doesnt seem to like 1.15V

So Im stable after 10 minutes in Furmark at 800/1100 but at 1.1V. The temps get up there too, after 10 minutes around 84 and 88 C.

I get much lower temps while gaming, after half an hour playing Crysis 2 (1920x1200 all settings "very high") with throttlestop at 25x TDP/TDC 95/85, 75mV and 800/1100 1.1V the CPU's highest temp was 88 degrees and the GPU's only registered 72 and 75 C. Great framerate!

Please let me know your temps across the cores for your throttlestop settings of 26x, 25x, 25x, 23x, and Im assuming that they correspond in order to core 0,1,2,3?

Again, I cant believe youre at 5% OC as well. I may play around with it again later and try your settings out of curiosity.

Sometimes, my wprime 1.55 1024M times actually increase while upping the TDP/TDC, is the normal? I dont know how to gauge an improvement, the only measure I have been using is if It actually passes wprime 1.55 and I dont get a BSOD.

 

yeah, I ran Furmark about 20 mins and the heat was not too hot, it's about 85/84 degrees for master and slave card ( with repaste thermal compound and retention mod) but I couldn't raise up the voltage at all, I tried with many clocks but no luck, it couldn't run anything about 10 mins and black screen afterward

 

Your temp spread on the cpu should be less dramatic after adjusting the screws. Also too much voltage on the gpu May be a bad thing. Try the same settings at stock voltage. Tour time on the cpu and getting higher because at 95*C your cpu is throttling. Again adjusting the screws should prevent that. Then you can run more of an overclock. Like 25 on all cores plus bios OC. Adjust the temps at idle first, Then adjust them under load while you run wprime 1024. Monitor the temps with throttle stop, real temp or hwinfo64 which is available on in the third party apps of T|I. Personally I use real temp just cuz it has throttle stop built in and I've always used it. Habit I guess

 

GREAT SUCCESS!!!

Ok I think Im getting close to a perfect tune here.

Using the following settings in throttlestop TDP/TDC 95/85 multi: 27,26,25,25, no BIOS OC, factory voltage it pulled 244.06 seconds in wprime 155's 1024M!

Max temp was 96 C across two cores, 94 on the others and it didnt get there until like 75% completion.

It scored a 8.002 second score in 32M.

Monitoring the FID it stayed at 24 almost all of the way through.

The starting temp was in the low 40's and the ambient temperature was ~65.

Next run, I tried a multiplier combination of 26,26,25,25.

It didnt do 244. seconds, although it came close at 255. or so, cant remember exactly. Monitoring FID it was dialing back to mid to low 24 multiplier around 50% completion.

It pulled 7.894 seconds in wprime 155's 32M! Very Nice!

I kept the latter multiplier settings as they seem more robust and its only in the latter half of 1024M do they start to throttle back to 24.xx.

The cards are stable now! I flashed them at 825/1140 1.1V and they pulled 35 FPS avg. in Furmark, settings: 1920x1200, AAx2, Post FX on. Temperature's were high though, 87 and 91 C after 10 minutes. The only curious thing is that they only register as 825/1125? I double checked the flash and they reflect as 825/1140 in Radion BIOS Editor? Weird.

Results in Crysis 2: Settings: 1920x1200, all Very High. Solid framerate, very little drop in framerate, usually only in open areas with a lot going on but still only a subtle drop in FPS.

After an hour or so, max temperatures were only 81 for the CPU and 74 and 77 for the GPU's.

Im very pleased with this, this is pretty much what I wanted.

Previously, my experience with adding voltage in BIOS (75 mV) has been much higher CPU temps. It would scale to 99 C before 50% completion in wprime 155's 1024M and could only attain 375 seconds.

Playing Crysis 2 the temps got up to 90 C.

My experience with throttlestop and BIOS OC at 3% has been an erratic clockspeed while running wprime 155, jumping around from 3200 to 2900 and even dipping down to 900 Hz? And I couldnt get away with BIOS OC without the additional voltage.

I think Im just going to leave it where it is.

I started playing Portal 2 tonight, the game is really funny!

Well anyways, thanks for all of the help with this, I think Im good to go.

Will post new 3DMark Vantage score soon.

 

I use Real Temp for core temperature while adjusting the screws and stress the cpu with wprime configured for 8 thread. (that you can launch from within Realtemp, it has a link for it) my settings are as follow:

4 cores: 23x140: 3220Mhz
3 cores: 24x140: 3360Mhz
2 cores: 25x140: 3500Mhz
1 core : 26x140: 3640Mhz

both gpu@ 800/1000 (flashed v-bios)

Here is a screenshot i took yesterday to post it here, it shows my idle and load temps:



I used wprime for the cpu and run a loop of 3 times vantage for the gpu, since i don't like furmark (there is a comment from an AMD guy somewhere that states that furmark can damage the cards. so for me is a no no). Those are temperature that i'm unable to reach during daily use of the pc (ex. gaming). I'm quite satisfied by them.

 

What gadget is that for the GPU's?

 

It is called gpu monitor, it works with rivatuner in background monitoring various things. The only thing i'm unable to set properly are the gpu/vram frequencies, rivatuner read them incorrectly. If you or somebody else know a way to properly set rivatuner for those readings (i think is something under the power user tab) it would be really apreciated.

The gadget allow for a lot of configuration on the way it looks. As you see it on my screenshot i configure it to be the way it looks now.

 

I need to make a correction to my previous post, my cards are actually flashed at 815/1125 1.1V.

I think the AMD guy might be right, I have gotten very high temps using Furmark, the last run was 88 and 91 C after 10 Minutes! But while playing Crysis 2 1920x1200 settings all "Very High" they were only 74 and 77 C.

Going to do a new vantage run soon.

OH BTW, I did a successful wprime 155 1024M run with multiplier settings of 27,27, 26,25 in 248. seconds! Not any better but the 32M run was 7.78 seconds! And its stable! Max temp for the 1024M run was 97 C, but it was performed with a warmer ambient temp than the previous runs.

I am presently on these settings, Crysis 2 seems a bit crisper, but with max CPU temp at 86 and GPU temps of 75 and 79, but I cant entirely attribute this to the new multiplier settings as again the warmer ambient temp.

I tried 5% BIOS OC, no additional voltage and settings of 27,26,25,24 and passed wprime 155 but got a BSOD within just a few minutes of playing Crysis 2.

I May try your settings of 26,25,24,23 later, but my experience so far has been BSOD pretty quickly with 5% and even 3% BIOS OC in Crysis 2 without additional voltage.

What has been your experience performance wise playing with the multipliers with no BIOS OC? I take it you have achieved better results with the OC? Vantage results with both?

Thanks for the tips on adjusting the screws, I need to do this as Core 3 is typically 7 degrees lower than the others. I take it that Im going to have to disconnect the X-Fire cable to accomplish this task? Any other tips before I try this out? Going to download RealTemp as well, thanks.

Is there a program to monitor FPS in games? I constantly hear reports of FPS changes while playing so Im assuming there is?

 

How are your temps higher than mine with a retention clip mod? and I am clocked higher than you and at the same voltage? My GPU's in furmark never get that hot.

 

I have no idea! That Furmark run's settings were 1920x1200 AAx2 Post FX on, not sure if this matters.

While playing Crysis 2 1920x1200 settings "very high" they have been 73 77C yesterday and 75 79C today but with warmer ambient temps.

It could be that I didnt do the best pasting job too, there are a few factors I can think of.

 

New 3DMark Vantage best!

Result

P18256

The everyday (not bench) settings I used are in my sig, in throttlestop I use another profile while browsing which is much less agressive with factory TDP/TDC.

Max CPU Temps: Core 0: 94, Core 1: 97, Core 2: 92, Core 3: 88 C

Idle CPU Temps: 42, 43, 42, 42 C

Max GPU Temps: GPU 0: 75, GPU 1: 78

Ambient Temperature: ~65 F

Im pretty stoked! My original Vantage score before playing with ThrottleStop and flashing my cards was P15400, this is a vast improvement!

My CCC settings are "use application profile". And beneath that I have most settings in the middle of the slider.

 

nice tuning you got there

Did you overvolt your vid cards or are they olding the clock with stock voltage?

 

sorry for the delayed reply, yes Im overvolting the cards and it took much persuasion to get them to 815/1125 at 1.1V.

I tried 825/1150 and get BSOD in Furmark around the 6 minute mark with the following settings: 1920x1200, AAx2, post fx on.

So I upped the voltage from 1.05 to 1.1 and stil BSOD.

I tried 1.15 and I get BSOD faster!

So I tried 815/1140 at 1.1V and its stable through Furmark after 10 minutes but the temp of GPU 1 was 91 C! (this was before I discovered the "real ultimate cooling mod": propping up the back of the R2 with a book or what have you)

The inexplicable thing is that my cards register 815/1125 and not 815/1140 as they reflect in Radeon Bios Editor. Very strange.

Anyhow, I think this is the most Im getting out of these cards, I know others have had luck with pushing them into the 900/1200 territory but not I.

 

900/1200 i believe is just for benchmark time, not daily use.

I'll keep mine at 800/1100, no overvolt required and the performance boost from stock is nice

 

Yes youre right, I think I was pretty stable at 800/1100 at 1.05v, cant remember if I was crashing in Crysis 2 or not, I may experiment with a return to this setting as the lower voltage is good for card longevity, although Im not hitting high temps now really (75-77 and now low 70's with "the book mod").