GTX 680M erratic usage

[RWNinja]

Hey guys, I have a Clevo P170HM (Sager NP8170) that i recently upgraded with a GTX 680M. The seller I purchased it from was Eurocom, which as I'm to understand has some effect on the outcome. Basically the issue I'm having is that my GPU usage is pretty strange. I play Guild Wars 2 and World of Tanks. The one thing I noticed is that even with overclocked rates, my gpu usage fluctuates anywhere from 40% in large battles to 99% and FPS ranges from 15-120fps in GW2. All the while, the clocks are changing as well.

Now, when I go to use NvFlash through the dos bootable USB method, it says it can't identify the pci subsys vendor, even with the stock flash from eurocom. It was a drop in card and worked from the start, once updated to the latest Bios and EC from Sager themselves.

I'm uncertain if this is an issue that can be resolved, however I welcome any insight into this matter. I have kept the clocks stock for the time being to allow the card to settle in and for me to get an average idea of the temps it will produce, though I am open to doing what needs to be done.

I ordered some Gelid Extreme Thermal Compound and I'm currently using IC Diamond 7(not quite a fan of it over AS5).

My processor is a 2860QM, so I'm also aware there is plausibility that the bottleneck is my CPU.

Other HW :Intel SSD, 8GB Kingston HyperX.

Thanks in advance!

[RWNinja]

I should also make note that while looking at a non-graphic intensive area, my gpu usage goes up to 99%, do a 180 and look in a more intensive area, and it drops to 71%. I'm pretty confused, shouldn't it be the other way around? Also, clocks remain the same @719MHz as reported by MSI Afterburner. (I've also tried setting power setting to performance in Nvidia Control Panel and disabling vysnc in game and in the control panel.

[Jezz3r]

I also experienced similar problems with gpu usage. I have i7-3630qm and from my knowledge cpu is not to blame given that my osd says cpu usage is about 30% (i play mechwarrior online). But i have lots of issues with my pc so the cause of my gpu usage going up and down randomly may be totally different.

[12Bravo]

<dd class="aboutme_right" id="profilefield_value_5">I have been experiencing the same things as Jezz3r with MW:O plus it seems on the main menu screens and the pre-game ready screens my gpu is at about 95% load on avg and while playing the game somewhere around 70% avg. I've tried pretty much everything I can think of to normalize things but haven't figured it out yet. Have a GTX 680m and a 3920xm coming in in the next week or so, so I am hoping that is what will solve it for me.

My current specs:

NP9170

i7 3630QM

GTX 675MX

16GB RAM

Crucial 512GB SSD

1.5TB HDD

</dd>

[RWNinja]

So far it sounds almost entirely like it's just underutilization on either the drivers side of things or on the game engine. My temps never go above 72 degrees and usually hover around 58 in GW2. Darn. Hopefully someone will chime in and be able to help us out!

[Jezz3r]

Well it's not only MWO for me My Vga dont wanna render anything harder than this, i only tested far cry 3 on high details with a video_tdr_failure BSOD after a minute and that was with and without slight oc of the gpu. Im starting to think about a total format and tweaking my card from square 1. Or 2, i've already flashed the vbios

[RWNinja]

I was thinking of a complete reformat as well. I had a 6990m and windows keeps going "help with error ATI/AMD GPU". Uhhhh....Haven't had one in two weeks champ....I have tried the Overvolt MSI Vbios and that seems to make no difference with my clocks. I can only go to 902, and even so after a while it brainfarts and goes to 324MHz..aka stock low usage clocks. I don't really want speed, I just want stability. I had a 7970m and it died within two days of ownership so any ATI card is out. lol.

When you NVflash do you use nvflash -4 -5 -6 to force vbios over the subsys vendor? I was curious as to whether or not that may take a role in it.

- - - Updated - - -

Also, as for the video_tdr_failure you're having, have you tried manually updating DirectX and PhysX? That might make a difference.

[Jezz3r]

Manualy updating direct x? You mean what, through runtime from Ms website? About nvflash, i use -6 command as stated in the guide here on the forums flashes with no problems in few seconds. Im experiencing the same issues also, i mean the gpu clock going up and down sometimes, but to me fixing the tdr_error and gpu usage thing is crucial maybe today or tomorrow im gonna buy an external hdd to backup my stuff and try a total reinstall, since my system started acting weird anyways after win 8.1 update.

Ps. Is the 8.1 update worth it?

[RWNinja]

Manualy updating direct x? You mean what, through runtime from Ms website? About nvflash, i use -6 command as stated in the guide here on the forums flashes with no problems in few seconds. Im experiencing the same issues also, i mean the gpu clock going up and down sometimes, but to me fixing the tdr_error and gpu usage thing is crucial maybe today or tomorrow im gonna buy an external hdd to backup my stuff and try a total reinstall, since my system started acting weird anyways after win 8.1 update.

Ps. Is the 8.1 update worth it?

Not really sure about 8.1. I had windows 8 for a while and it drove me batty so I went back to 7. As for DX yes, manual update from MS website. As for NVflash I have been using -6 as well, but it's strange. I know that PXXXHM users needed to perform an EC and Bios update, but if I uninstall all Geforce drivers, screen goes blank and beeps, then I reboot and it says I have no display adapter xD. So far, so good.

Also Update: I bit the bullet and reformated after realizing both of my HDD's somehow filled with random crap over the past 2 years. It -has- gotten better, but the difference is marginal. The biggest problem that seems to have corrected itself in this format is the temperature issue. This beaut' runs cold as ice now. On idle at desktop around 34 degrees, in GW2 about 58, and in WoT about 64-70.

Once my Gelid Extreme comes in, I'll probably stuff an OV Bios under the hood for the GTX.

- - - Updated - - -

TDR is Timeout, Detection, and Recovery. It happens when the Video Driver fails to respond to the system after a certain period of time. My guess is it'd be cured with a fresh format, unless the card is at fault. At the same time, if you don't want to lose everything you could try this:

1) Boot into windows normally

2) Control Panel: Uninstall all Nvidia drivers

3) reboot into safe mode using F8

4) run Driver Cleaner and get rid of all Nvidia crap in the registry.

5) reboot normally and install new driver

6) reboot again and cross fingers while drinking a beer!

[Joo]

@RWNinja: have you tried installing older drivers?

situation you described above made me somewhat unsettled now as i've just bought 680m from eurocom myself. long 3 weeks of waiting ahead

[RWNinja]

I have installed older drivers and the one I settled on was 334.89 with a modded inf. It works, but I've come to determine that alot of games can't harness the GTX's power. Especially games built on the old DX9 platform. I did, however, manage to get a stable 1006 MHz/2250MHz overclock on it. I've since backed it down since performance gain in games is negligible.

Let me clarify, though. In games like World of Tanks, the usage stays at full bore usually. in terms of Mhz. But the percent usage drops constantly. I've chalked it up to a poorly optimized game engine that the russians decided to not make any better due to rolling out a new game engine soon.

Metro: LL I'm usually at 80-90 fps so it's solid.

Eurocom wasn't a bad seller, and it was a worthy upgrade from a 6990m. Once you are ready to install the GTX680M you need to get the latest bios and EC firmware from Sager (if that's your reseller). Update EC, then bios.

One thing that made a huge difference was unparking my processors cores, allowing it to harness more power from the GPU. BIG difference there.

Overall, in conclusion, most games just aren't optimized enough to withstand the awesome that is the GTX 680. I run almost any other game on near max settings. I'm just trying to determine if I possibly got a bum card. Worst case I buy an EM platform and migrate everything over minus the CPU.

[Joo]

hmm, i see other people complaining about low gpu usage and therefore low fps in wot and gw2 too. that is mainly due to these games are not optimized to utilize 8 cpu threads. so the problem seems to be not with your 680m after all :] by "unparking processors cores" you mean that you flashed bios with increased tdp values, right? as for em platform @Prema got 780m somewhat working with it and will probably make modified vbios available to the public in the near future.

[RWNinja]

I noticed that. Those two games are poorly optimized. Sadly, so is Titanfall. All these games are being produced without the capacity to use multi-core support for the most part. As for unparking the core, it was more a set of registry tweaks instead of TDP values. However, I'm open to any suggestions in that.direction

I thought about attempting a minor overclock on the 2860QM (honestly, the warmest this thing gets with some cooling mods I've done is 64 degrees.)

I'll likely upgrade to 16GB of ram to offset some of it. (not that many games can even take over 4g).

I've also dipped into Nvidia Inspector and used a couple modded command lines to a set of clocks to force pstate2 constantly. It works, but it's a shady workaround.

The hottest both of these have gotten in WOT are CPU: 64 degrees and GPU: 71 degrees.

I know WoT is expecting a serious overhaul with 9.0 switching to the havok physics engine supplemented by the U3 engine for GFX. That should take a serious load off of my GPU.

[Jezz3r]

this is how situation looks after total format and win 8 reinstall, and trying everything again (different drivers, overclocking, turning off tdr etc.) im helpless against an unknown enemy. as all can see - GPU and CPU temperatures are totally fine, so i figure throttling is not an option here