580m ocing

Just so you know what I'm trying to do is OC GPU for practical purposes. Meaning I want to find that will work all the time. Not just be able to run a benchmark before it crashes or throttles.

Well the higher voltage in the vbios definitely solved my glitching problems. Before I was about to do 740 core max, now I can push it up to 800. At around 800 the temps get so high that I'm reaching 79C in a game and the GPU throttles. The GPU will go higher but it is not practical from temps and throttling perspective.

So, I lowered it to 780 core and I think it is doable to play high end games at this level and not get throttling, however the temps are still pretty high, around 77C while gaming.

At this point I looked at what I get from 740/1900, 68C vs 780/1900 77C. From the benchmark stats it looks like 1-2% difference in performance. Hmm I think it is not worth it. For such a small gain you're running much higher temps and you risk getting throttling from thermals anyway (because you're close to border line).

So I think I will be running stock vbios and OC that, unless someone can figure out how to pass 79C throttling mark or I can get a vbios with lower over volt than .92v. BTW .92V is just around 5% more then stock, it is interesting that it is making such a big difference in temperatures.

Just so you know I'm running my laptop on top of laptop cooler and my GPU is re-pasted with high quality compound. Thus most other people will get even worse temperature issues than I have.
My testing was in Crysis 2 DX11 ultra settings. I load game at the same saved point try to push the card. What seems to work well is going into armor mode and shooting water. All the splashing effects really strain GPU

 

Ok I have a solution for your problem, its an easy fix. Im not at home right now but when I get home I'll edit this post with directions ok? Also, when I overclocked mine from stock to 825 core and 1900 mem (I can't remember the shader clock) I got well over 35% performance increase on unigine and an average of 15-18 fps gain in games, while maintaing safe temps.
Temps were(after one hour on unigine:
Average: 69C
Max: 74C

P.S. - I did this from my cell, so please excuse any wrong spell corrections done by auto-correct.

 

Cool. Let me know what you have done. I would imagine it has to do something with Fan control. I didn't do any research into the fan behavior/throttling under load. Hopefully it doesn't mean running fans at 100% all the time . If it is Fan issue it would be nice to have something that just scales Fan speed differently than default.

 

i doubt 825 is stable in crysis 2 over an extended period

 

So I run into something interesting while testing Fan speed effect on temps in Furmark test.

It seems that setting Fans at max settings through HWiNFO64 (hw64_385_1363) disabled throttling at 79C.

So this test was on standard vbios, GPU at 740/1800. I run furmark without fan speed adjustment so it hit 79C and it throttled down to 74C at which point I kicked in Fan's to 100%. So the temps started to raise much slower and gradualy and it hit 79C,80C,81C,82C, at which point my screen just went dark and I had to do a hard reset. Notice I blew through throttling temp and nothing happend.

Anyone else notice similar behavior? It seems that adjusting fan controll disabled throttling at 79C. I didn't try to reproduce this yet, but I think I will try to run my Crysis 2 testing method at higher clocks with modified vbios.

What I also wanted to do is determine what is the built in Fan speed schedule based on temps. Anyone has this info or a tool that will work on m17x and show me the exact fan rpm.

 

Try to put the fan slider at full completely before you play Crysis 2. Might be able to crank out a full sesh without any throttling.

 

Did some more testing today.

Setup: overvolted vbios, Max fan speed set through HWiNFO64, Crysis 1080p DX11 ultra everything (Water shooting for high GPU strain in armor mode, it drops from 30fps to 11-12 when I do that)

I tried core at 800,810,820,830,840,850, memory set at 1900. Max temps were 75C up to 830 core, than it jumped to 76C.

At 840 and 850 I think I was getting occasional glitch which shows as a split second white line somewhere in the screen. It was so slight I'm not sure it actually happened. Other than that the game plays smooth without any issues. No crashes or any other abnormalities.

I didn't try pushing it past 850 core because I felt I might be playing with fire. It looks like 830 core might be a sweet spot in terms of stability and temps. Another thing is that it seems that the fan speed set by HWiNFO64 is much higher than what bios sets it to. Judging just form the noise it makes. Still don't have any tool to get the actual RPMs.

3D Mark Vantage score I got at 850/1900 GPU was P21377/GPU 17772 CPU 54606. Max temp 73C

So, beside the noise the HWiNFO64 max fan speed is nice to have to push this card further. I also measured the power consumption using kill a watt. Through the testing I maxed around 220 watts.

Another test was to run a FurMark at 740/1900 GPU, 100% fan speed, anIntelel burn in test at the same time with overvolted bios.

The GPU throttled at around 65C pretty quick too. The power usage hit 240 Watts at the time of throttling. It looked like the GPU throttle was triggered by power usage. Seems there is power throttle as well.

Anyway, I hope you can find this informatiousefulll, and please share your experience.

Thanks.

 

If it is throttling at 65c then you are not bypassing the Dell throttle.

65c on furmark is the Dell throttle. You should get 79c before throttling.

 

I don't think so. It will only happen if you hit your CPU very hard while doing the FurMark. Kind of a dooms day scenario if you will.

Try to run this IntelBurnTest v2.3 download from Guru3D.com and run FurMark at the same time.

I think it is power usage throttle because if you don't overload your CPU it won't happen. I got this idea from M18X forum where someone was talking about how the 330W power supply was more than enough for that rig with 580 in SLI, and someone else said not if you put a max load on GPU and CPU at the same time (it hits the max wattage). So I had to try it to see what happens on our rig. It made me dig in my basement for the kill a watt so I can see the power use.

Anyway, if it is possible to use M18X power supply for M17X, it would be interesting to see if the throttle would still happen.

Regardless, I'm not getting a throttling while at much higher clocks and decent CPU use as far game load is concerned. I'm also throttle free from the blind bios flashing method.

 

Hey GandalfTW, are u only using furmarkvantage to test for throttle. I'm starting to agree it seems to be a power load issue not so much a temp heat gpu thing.

Also remember Furmark is an application designed to stress the GPU by maximizing power draw well beyond any real world application or game. In some cases, this could lead to slowdown of the graphics card due to hitting over-temperature or over-current protection mechanisms. These protection mechanisms are designed to ensure the safe operation of the graphics card. Using Furmark or other applications to disable these protection mechanisms can result in permanent damage to the graphics card and void the manufacturer's warranty.

 

regarding power usage-throttle...at least the european ac´s are stuck at 240W max output.

since you said the system throttles at ~240W power consumption, i guess it all makes sense. just need a bigger AC ^^