dslreports logo
 
    All Forums Hot Topics Gallery
spc
Search similar:


uniqs
3389
clawfury
join:2012-02-14
Rochester, MI

clawfury

Member

AMD new GPU and WoW...

Guys,
Quick question for you all. I built a new computer (i7-4770k water cooled, 16GB ram, SSD, and a MSI R9 290 4GB). As you can imagine, it's pretty snappy for most things - but WoW gives it fits. After tons of tweaking, compatibility settings, and *face desk* I finally got it to quit doing BSOD or red screen randomly - but the frame rate and hitching is still ridiculous. I never break 40fps. I've tried older Catalyst drivers, new Catalyst drivers, beta Catalyst drivers, tweaking settings & detail level, everything. It's ridiculous because my old 6950 can pull 100fps with Ultra in WoW...and this much faster computer can barely do half of that. Anyone else running a 290 card that knows a fix? Thanks!

Krisnatharok
PC Builder, Gamer
Premium Member
join:2009-02-11
Earth Orbit

Krisnatharok

Premium Member

Do you run any other games? It sounds like a bad card.

Your CPU isn't downclocking and bottlenecking anything, is it? (check UEFI or power save states and ensure the clockspeed ramps up when under load).

I assume you have already run prime95 on the CPU and it's stable, and run memtest on the memory and it passes?

Do what you can to rule out good components.

Cat5
join:2013-04-26

Cat5 to clawfury

Member

to clawfury
That's a beast of a setup for WoW, you should be having no issues at all. Sounds like a dodgy card to me. Do you have another card you can try in that system?
clawfury
join:2012-02-14
Rochester, MI

clawfury

Member

Cat, I agree - it's way overkill for WoW. I don't think its the card, though - because it runs Tomb Raider with 4xSSAO Ultra at 60 fps. I think it's a WoW driver issue. It had issues with most games, until I tweaked some settings. It also runs Bioshock infinite at highest settings with no issues.
clawfury

clawfury to Krisnatharok

Member

to Krisnatharok
Kris,
I had the CPU overclocked and passing prime 95 up to 4.6Ghz. I turned off the OC once I got the video card in, because it ramped up the box temp quite a bit, and I wanted to eliminate sources of stress when I was troubleshooting. I have not run memtest, I will do that. To be honest, the card gave me problems on almost every game for a while - I've tweaked settings and for the last week it hasn't crashed anything. I checked my crash logs, and every time it points to the 3rd party driver as the issue.
clawfury

clawfury

Member

Also, temperatures are stable - running GPU-Z and the processor version - the processor is on a Corsair H100i - max temp during gaming is like 62 C. GPU is running 60-65 C while gaming. CPU 30 C at idle, GPU also fine at idle. Even after a crash the GPU temp is not exceeding 70 C.

Goldheart
join:2002-06-09
Las Cruces, NM

1 edit

Goldheart

Member

Are you using directx-9 or 11 in WoW? Try the other mode and see what happens.

You may also wish to download the latest Directx from MS.

Also, have you tried full screen vs. windowed or vice-versa?

Another thing you can try, check your bios and make sure all power saving options are off.
clawfury
join:2012-02-14
Rochester, MI

clawfury

Member

I'm running DX11 - I will try DX9, and also downloading the latest, thanks. I've tried full screen vs. windowed, both of them did it. I will check the BIOS tonight, but I think power save options are already off. I'm afraid it's just a compatibility issue with the Catalyst drivers.

Talarn
join:2009-09-02
Clintwood, VA

1 edit

Talarn to clawfury

Member

to clawfury
I'm not a computer guy by no means, but I was watching a video and a guy was having similar problems. He changed something in his BIOS that was only allowing a single core of his CPU to run.

Edit: It was a powersave option.
clawfury
join:2012-02-14
Rochester, MI

clawfury

Member

Hmm. Interesting. I'll check that as well, thanks for all of the suggestions.

Goldheart
join:2002-06-09
Las Cruces, NM

Goldheart

Member

Google dxwebsetup.exe for the directx updater. It would only affect DX9 but can't hurt to try I suppose.
clawfury
join:2012-02-14
Rochester, MI

clawfury

Member

I assumed that with Windows 8, that DX11 would update itself automatically. Not the case?

Goldheart
join:2002-06-09
Las Cruces, NM

Goldheart

Member

Right Windows update will update DX11, however for DX9 there isn't an updater.

CableLane
@159.45.71.x

CableLane to clawfury

Anon

to clawfury
I would take a look at this article

»us.battle.net/support/en ··· eo-cards

Wow doesn't always like the most up to date driver for a video cards, I myself was having game crashes with the latest version of the nvida driver once I down graded all is well.
Hazmat
join:2007-07-10
Laveen, AZ

Hazmat

Member

said by CableLane :

I would take a look at this article

»us.battle.net/support/en ··· eo-cards

Wow doesn't always like the most up to date driver for a video cards, I myself was having game crashes with the latest version of the nvida driver once I down graded all is well.

According to the list, I am using newer drivers for my NVidea GeForce GTX660Ti than the recommended ones.

But I still have the same issue I have always had - the card appears to cease to function, and recovers, but I need to relog in.

Gets...dangerous...mid raid...

Tirael
BOHICA
Premium Member
join:2009-03-18
Sacramento, CA

3 edits

Tirael to clawfury

Premium Member

to clawfury
said by clawfury:

it runs Tomb Raider with 4xSSAO

I think you mean SSAA.
said by clawfury:

I had the CPU overclocked and passing prime 95 up to 4.6Ghz.

Define "passing". Prime95 is part of GIMPS, The Great Internet Prime Search. It was not originally meant to be a CPU test. It works, however, because it writes processes to all cores, which puts them at their max clock speed. It is used to find Mersenne Primes. So, unless you found a new Mersenne Prime (defined as (2^n)-1), you didn't "pass" anything.

If you are overclocking, you should have some of your C-States disabled. Your FPS issues are most likely not linked to the GPU, if the GPU is working in other games. The engine that WoW runs on (currently) was not made with the R9 (or 700 series Nvidia) GPUs in mind. The game depends a lot on CPU configuration (see this blue post, so I don't look like I am talking out of my ass).

If your CPU is attempting to fall into a C-state because it is only using a few cores, it may enter C3 on some of the cores. When Windows attempts to move threads around, it could cause problems. This is because C-states have exit latencies (time to return to an operating state) and cause the CPU voltage to go all caddywhompus when attempting to return to the previous clock as quick as Windows is asking it. This can cause several problems internally to the CPU (which is way more technical than what I have already explained). Suffice it to say, this is one of the main causes of BSODs when overclocking. Most people will tell you to disable C3-C6, which prevents the cores from powering down. C6 is especially bad because it can set the voltage of a core to 0 (I use the term 'core' to mean units inside a CPU, of which there are several).

TL;DR

Disable C3-C6 in your UEFI (I wish people and companies would stop calling it BIOS). Then, run Prime95 for 8 hours on the stock clock. If you don't BSOD, increase the clock by 100 MHz until you get to a clock that fails. Take the last clock you get that passes the previous test and run it for 24 hours. If it passes, good. If it does not, reduce the clock by 100 MHz. Continue doing this until you get a stable clock. Currently, the highest OC I have seen is 5.0 GHz on the 4770k. WIth yours running at 4.6 GHz, that is most likely where the problem lies.

Krisnatharok
PC Builder, Gamer
Premium Member
join:2009-02-11
Earth Orbit

Krisnatharok to clawfury

Premium Member

to clawfury
How long did you run prime95?

Goldheart
join:2002-06-09
Las Cruces, NM

Goldheart to clawfury

Member

to clawfury
I think we're all wondering if you have made any progress on this issue?

DarkLogix
Texan and Proud
Premium Member
join:2008-10-23
Baytown, TX

DarkLogix to Krisnatharok

Premium Member

to Krisnatharok
said by Krisnatharok:

How long did you run prime95?

That's one reason I like Ycruncher, you can have it do the same pi calc each time so it also verifies that the output is correct and not with any compute errors.
clawfury
join:2012-02-14
Rochester, MI

clawfury

Member

Sorry guys, didn't have time to get on and take a look. No, in fact, it went backwards. I tried scaling down to "high" - and that increased the frame rate, but it still hitches. Now it's crashing again - hard crash that requires a complete reboot. In frustration, I opened up the case - took off the top and side panels, I'm still suspicious it's a heat issue (it's a micro - ITX box). Since then (maybe an hour or two play time) it hasn't repeated - but the frame rates are still nothing to write home about. I still need to try memtest, and I'm thinking that my cabling in that small of a case is creating a heat issue - I might try recabling to create better air flow.
clawfury

clawfury to Tirael

Member

to Tirael
OK, just finished reading all of the comments. Tirael, it does make sense that this is a CPU heat issue - I DID have other games crashing, constantly - until I made some compatibility tweaks. Prime95 I let it run it's stress test, I think - it was like 10 minutes. I already took off the overclock - I was afraid of too much heat in that small of a box with a R9 290. I will go into the UEFI settings tonight and take a look. Like I noted just a minute ago, though - it's running fine with the panels off, which makes me think it is overheating due to the R9 290 in the box. I'm not sure why, though - it has a closed loop watercooler, and the actual temps look fine. Puzzled. I will try the memtest and prime95 again tonight, and review UEFI settings, thanks all! Sorry for late response.

CableLane
@159.45.71.x

CableLane to Hazmat

Anon

to Hazmat
That could be the problem if you are using a newer version than what is supported. You may want to try and down grade to the recommend version and see if that helps your issue.
clawfury
join:2012-02-14
Rochester, MI

clawfury

Member

Cable,
I've tried older stable drivers, the newest driver, and beta drivers :-/ They all seem to crash it, or cause fps issues. It performs in WoW like a 5 year old computer, it's ridiculous. If I had to do it again I would go Nvidia - these new R9 chips seem to cause trouble with almost everything I've tried, no game has ever played well out of the box. I don't see any heat issues on the monitors - but I still suspect it - I haven't had a single crash since I opened up the top/side.

Snuffbox
nice irl
Premium Member
join:2011-04-15
Milwaukee, WI

Snuffbox

Premium Member

If you think it's a heat problem it should be extremely easy to diagnose...

If you're overclocking, you should already be using coretemps or similar program.

For your GPU, there's another light weight GPU temp program: »www.gputemp.com/

Run a CPU & GPU temp monitor with the case fully assembled on load, what kind of temperatures are you seeing? Prime95 won't stress your GPU.

Goldheart
join:2002-06-09
Las Cruces, NM

Goldheart to clawfury

Member

to clawfury
Yeah its sounding more and more like a heat issue to me as well.
clawfury
join:2012-02-14
Rochester, MI

clawfury to Snuffbox

Member

to Snuffbox
said by Snuffbox:

If you think it's a heat problem it should be extremely easy to diagnose...

If you're overclocking, you should already be using coretemps or similar program.

For your GPU, there's another light weight GPU temp program: »www.gputemp.com/

Run a CPU & GPU temp monitor with the case fully assembled on load, what kind of temperatures are you seeing? Prime95 won't stress your GPU.

The odd part is - Core Temps is showing 45 degrees C roughly per core in full operation, when the crash occurs. The crash dump tool refers to the crash as caused by the dx****.sys file (3rd party Catalyst driver). However, the air coming out of the H100i water cooler radiator is extremely warm, the top of the case feels warm, and the fans are scaling up. I took the top off the case, and the side off - and there hasn't been a single issue since. I'm honestly puzzled - if it was heat, you would think the temp monitors would tell me.

Snuffbox
nice irl
Premium Member
join:2011-04-15
Milwaukee, WI

1 edit

Snuffbox

Premium Member

It's possible that temperatures are indirectly the issue. What PSU do you have specifically?

When heat picks up your CPU/GPU fans kick on to a more aggressive level and could be dipping your PSU below stable levels. So while temp monitors may still reflect reasonable temperatures, your computer could be unstable because of the extra draw from fans.

So, possible PSU failure?
clawfury
join:2012-02-14
Rochester, MI

clawfury

Member

Snuff,
I don't know - it's a 750W Corsair High Performance (gold rated) PS. I took the blu-ray drive out - to increase air flow, that didn't help - it is a small case (Corsair 250D). My next move is going to be rerouting and braiding cables / maybe getting a short cable set to try and increase air flow. I might just turn the base fan speed up as well.

Snuffbox
nice irl
Premium Member
join:2011-04-15
Milwaukee, WI

Snuffbox

Premium Member

If CPU and GPU temperatures are fine (below 80's) then heat isn't directly you're problem.

Especially if more graphically demanding games aren't causing the same FPS/Freezing issues?

Have you put your 6950 back in the machine?
clawfury
join:2012-02-14
Rochester, MI

clawfury

Member

My 6950 is in my wife's machine, so no. This was a brand new build. This weekend I hope to have time to un-do all of the cables and braid things neatly - the underside of the cables was extremely warm until I took the top off the case. I have integrated video on this machine - and that ran wow, albeit slowly, without a hitch and never a crash before.