WoW Performance Issues/Tweaks, Etc - Page 52

sp00n · 11/10/10, 7:09 PM

My guild leader also had this problem and noticed that altering the hardware sound acceleration setting (in Windows itself) fixed it for him. Some value lower than maximum, can't say for sure.
On the other hand, Blizzard in their known problems thread is recommending to make sure that hardware acceleration is set to maximum.

Also an update to this:

I've even tried to replace all combat sound effects with zero byte sound files and disabling all other sounds to be able to hear addons once again, only to find out out that currently more than half of the addon sound just isn't played at all (very noticeable with BigWig's emphasize function).

I've meanwhile replaced all those zero byte sound files with 'true' muted sound files (i.e. they just play a fraction of a second with total silence) and this seems to have fixed the sound dropouts most of the time.
I'm still trying to filter out only exactly the combat sounds though, which is a real PITA to do.

Another upside is, for the first time *ever*, I can hear the NPC speeches without a problem.

Qikk · 11/10/10, 11:58 PM

Someone finally posted on tech support forums suggesting doing just that. We'll see if it works next time I join a raid.

sp00n · 11/17/10, 6:11 PM

Originally Posted by sp00n

I've meanwhile replaced all those zero byte sound files with 'true' muted sound files (i.e. they just play a fraction of a second with total silence) and this seems to have fixed the sound dropouts most of the time.
I'm still trying to filter out only exactly the combat sounds though, which is a real PITA to do.

Another upside is, for the first time *ever*, I can hear the NPC speeches without a problem.

I have a relative stable basis for the muted combat sounds now, so I uploaded it onWoWInterface: Muted Sounds - combat only
It's not perfect though, I couldn't identify all combat sounds. If you find one and know which file it is, let me know and I'll update the rar.

I've also uploaded a version where ALL sounds are muted, which is basically the same as Bossmod Sound Fix (while game is muted). The only but important change is that I am using 'real' sound files that are muted instead of files with zero byte content. My addon sounds didn't work properly with these zero byte files, but are doing so just fine with the muted ones.
Download here: Muted Sounds - all

Amonfana · 11/18/10, 5:50 AM

I will give it a go in a raid later, thanks a lot sp00n.

edit: Just did weekly 10man and it worked nearly perfect. I only had some "scissory" and some "stompy" sounds (Rogue and Warrior related I guess). Don't have any of these classes, so I can't pinpont it closer, sorry and thanks again. Here is to hoping this won't break too soon.

Tinwhisker · 11/18/10, 8:45 AM

I gave it a shot last night in some mount runs we were doing and it worked very well. It was actually quite eerie at some points. I may actually modify it myself to put my own class sounds back in because I didn't realize how much it threw me off not to hear them; I don't know if I would be able to get used to that or not.

klüger · 11/24/10, 10:28 AM

Thought my computer had gotten superaids when I ported to sw to fly/ride through old world after the sundering. Went from 180fps in dalaran to ~10 while flying in old world.

Was running everything on ultra in 1650x1050, but had to crank it down to good to even be able to ride around decently.

I'm posting here to find out if you have any tips on things to turn off/down (like shadows) and what performance I should expec (and if my bottleneck is the vid-card like i think it is)

i5-760
4gb 1600 ram
hd5770
wow installed on a x25 gen2 ssd.

• EasirokThunderpants · 11/27/10, 1:26 PM

Originally Posted by klüger

Thought my computer had gotten superaids when I ported to sw to fly/ride through old world after the sundering. Went from 180fps in dalaran to ~10 while flying in old world.

Was running everything on ultra in 1650x1050, but had to crank it down to good to even be able to ride around decently.

I'm posting here to find out if you have any tips on things to turn off/down (like shadows) and what performance I should expec (and if my bottleneck is the vid-card like i think it is)

i5-760
4gb 1600 ram
hd5770
wow installed on a x25 gen2 ssd.

Video card is probably not the bottleneck. I believe the 5770 is dx11 capable, you could try enabling the dx11 mode in WoW to see if there is improvement.

Your i5 chip is going to struggle with ultra settings. I personally run a i7 940 extreme, and even that chip gets taxed pretty well by WoW with ultra settings. You can try tweaking the processAffinityMask cvar in your config.wtf file. This setting supposedly is not necessary anymore, but many people (myself included) see significant improvements by configuring the game to ignore hyperthreaded cores as well as core 0, which is typically running primary operating system processes. For my i7 940XM, the ideal value seems to be "84". Not sure what would be ideal for your i5, but there is an article on WowWiki about it.

Make sure your SSD drive is not overly full. Performance will increasingly deteriorate after it reaches 75% full. I presume you are running Windows 7 (to take advantage of TRIM) and are not running any defrag tools (these offer no benefit with SSD and only reduce the lifespan of the drive).

And just stating the obvious here, but make sure you have upgraded to latest ATI drivers for your video card. Also, install the Catalyst Application Profiles. There have been specific improvements made to the drivers and CAP for 4.0 WoW in the past two months.

Lastly, double-check that you have no antivirus or other resident stuff running while gaming.

For comparison, here are my specs below. I can run Ultra at about 50fps, and 30fps in raids. I prefer to turn a few things down though that are distracting to me, and end up 100-150 most of the time, 50fps in raids. I still stay above 30fps when frapsing a raid:

i7 940XM
8G 1333 ram
HD4870
Intel x25 gen2 (160G)
Win7 64-bit
(note: this is a laptop, not a desktop system)

Greymist1 · 11/28/10, 10:30 PM

Originally Posted by EasirokThunderpants

Your i5 chip is going to struggle with ultra settings. I personally run a i7 940 extreme, and even that chip gets taxed pretty well by WoW with ultra settings. You can try tweaking the processAffinityMask cvar in your config.wtf file. This setting supposedly is not necessary anymore, but many people (myself included) see significant improvements by configuring the game to ignore hyperthreaded cores as well as core 0, which is typically running primary operating system processes. For my i7 940XM, the ideal value seems to be "84". Not sure what would be ideal for your i5, but there is an article on WowWiki about it.

i5 doesn't automatically imply slower than i7, especially if it's desktop i5 vs. mobile i7.

It so happens that you two have the same physical chip design (45nm Lynnfield/Clarkfield quadcore) configured in different ways. The i5 760 has hyperthreading factory-disabled, operates at 2.8 GHz (nominal) to 3.33 GHz (max Turbo), and its on-die power controller (which regulates Turbo boost) is configured to limit power to 95 watts (that being the TDP rating of the chip). The mobile i7 940 extreme has HT enabled and 3.33 GHz max Turbo, but the nominal frequency drops to 2.13 GHz because it has to limit power to 55W.

Since WoW doesn't tend to use more than about 2 cores, his i5 should, if anything, be equal or better for WoW. Possibly lots better when anything is running in the background: the mobile i7 940 Turbo controller won't go above 2.4 GHz when more than 2 cores are active. That's the price you pay for quad core in a mobile CPU. Because of this fact, I doubt there are any benchmarks where his i5 would actually lose to your i7, even though it doesn't have HT.

P.S. Most SSDs (including Intel Gen2) have factory-reserved spare area (several gigabytes not visible to the OS) so you can never truly fill them, with the amount of spare area depending on how much performance loss the manufacturer decided to allow if the user tries to fill the drive to 100% capacity. Also, it's mostly a degradation of write performance, especially random writes. It doesn't hurt read performance much, if at all, and WoW's disk accesses are bound to be almost all reads, so WoW performance should not noticeably deteriorate if the SSD it's installed on is close to full.

Emolate · 12/01/10, 11:13 AM

I've been playing WoW a lot through Wine lately and it has been pretty good, though I still see better performance under Windows 7. (Under Windows 7 I use DX and have DX11 active, under Ubuntu I use OpenGL)

Many sites on installing or performance tuning for Wine on Linux installs are old/outdated or from people who base their performance on running around Undercity or standing in Dalaran. Are there any tidbits anyone can share on tuning their config in Wine or in Config.wtf to optimize for that environment?

Szynszyla · 12/02/10, 9:38 AM

I've recently got my hands on few intel dual core processors to test out the performance gains in WoW based on the model and specifics. I think it might help out people choose upgrades, and further i will have question to Phenom II users about performance. That's due to them probably having much better price/performance ratio compared to E8400, which prices are still hanging for around 90-120 euros at best.

Mind that i won't post any screenshots for now.

The CPUs i had to test were:
E2160 1MB cache- up to 3.6ghz (400x9) - couldnt go higher on FSB due to wall
E4300 2MB cache - up to 3.2ghz (400x8) on 1.55v with also skew top cap so the temps where insane on any cooling
E8400 6MB cache - didnt overclock for sake of similar clock results.
(The difference on this CPU's is pretty much only the L2 cache size)

System specs were
Win XP / Win 7, tested on both DX9/11.
Motherboard - GA-EP31-DS3L, northbridge active cooled, southbridge aswell. Good ventilation overall.
Ram - 1x2gb + 1x1GB GOODRAM CL5 DDR2 800mhz, CPU:MEM ratio 1:1 due to awkward motherboard behavior when not running synced FSB. Did not change the timings couse of awkward behavior aswell.
Graphic card - Sapphiron R4830 512MB DDR3 with R4850 chip (yes - production mistake) running 645/1040

Game settings / CCC
1680*1050
Good or High settings on main game bar, ultra with shadows on high. For sake of comparison, i've tried LOW settings on E2160 becouse of really bad FPS to compare.
CCC - 2 different settings, either optimal performance or balanced.

Testing places.
Dalaran in front of the bank
New orgrimmar in front of the leader building
FP between - vangeance to ulduar

Test #1 , E2160 at either stock 1.8ghz or 3.6ghz
Dalaran - pretty much ~15-25 FPS max no matter what graphic settings or CPU clock, . Pretty much no difference. Clear bottleneck on the CPU side.
New orgrimmar - even worse than above. Low settings - 20-23fps on 3.6ghz, good - around 13-14 fps during peak hours.
FP test - min. FPS ~12, avg ~110 on 3.6ghz.

Test #2, E4300, 3.2ghz gave about +9-10fps increase on the minimal FPS on the FP test compared to 2.7ghz. Overall +16FPS compared to E2160! from ~14 to 30
Dalaran - Able to squize around 45-60 FPS on good settings. Huge increase compared to same architecture E2160, just cache difference.
Orgrimmar - Even on stock clock, ~25-30FPS. At 3.2ghz on good settings ~40-55FPS. Ultra - ~30-35FPS with shadows on high. Again really big improvment over 1MB cache CPU.
FP test - Minimal 17FPS on stock clock, ~22-24FPS on 2.7ghz, ~30FPS on 3.2ghz. AVG/MAX FPS 145/280 peak at 3.2ghz.

Test #3, E8400
Dalaran - ~60-70FPS on ultra with shadows on high.
Orgrimmar - as above...
FP test - for some weird reason minimal FPS was bugging for me at it was usually at around 25-30fps. It's probably becouse of the drop in the frames when you click and enter the mount. I think its the only time where on any CPU i had FPS drops. AVG/MAX fps was at around 240/550FPS .

Conclusion - It seems like increasing CPU clock does not grant benefits or very low if your CPU bottleneck is cache. The results are different for CPUs with more than 2MB+ cache, where the clock starts to give benefits. Unfortunatly i was not able to O/C the E8400 to check the results at around 4GHZ, but the avg FPS on FP test and orgrimmar does tell a lot about overall compared to E4300 or poor 1MB E2160 which overclocks like mad.
I hope this information regarding how important CACHE SIZE is, not the CPU CLOCK for WoW performance will help you out on choosing the right upgrades.

If you don't get more FPS in certain situation with low settings, its pretty much the CPU holding the graphic card. It's been known for long, but i havent really seen such comparison yet regarding importance of cache in WoW.

I know that some of you won't agree that orgrimmar standstill test is not really important, but in general it looks like it should be one of the best raid enviroment tests how well does your CPU perform, especially with addons as it's really CPU based.

I also have a question like i said at the start of this post.
I'd like to get some information about Phenom II with 6MB L3 cache performance with similar graphic card as it's cheaper to get along with new motherboard and DDR3 than E8400. Also phenom II with 3 cores (likely to unlock the 4th one) is better suited for other games using more than 2 cores.

dr_AllCOM3 · 12/02/10, 10:16 AM

Originally Posted by Szynszyla

I know that some of you won't agree that orgrimmar standstill test is not really important, but in general it looks like it should be one of the best raid enviroment tests how well does your CPU perform, especially with addons as it's really CPU based.

I also have a question like i said at the start of this post.
I'd like to get some information about Phenom II with 6MB L3 cache performance with similar graphic card as it's cheaper to get along with new motherboard and DDR3 than E8400. Also phenom II with 3 cores (likely to unlock the 4th one) is better suited for other games using more than 2 cores.

The performance in Dalaran or Orgrimmar is very dependent on the amount of people there and your viewing direction. You can't really compare your results with other people's or other days.
Addons are another big performance killer.
Tests should be done in 25 man raids on a specific boss with equal (low, for pure CPU tests) graphic settings and no addons. Take the average and minimum FPS with a program like Fraps.

Szynszyla · 12/02/10, 10:36 AM

The raid enviroment tests arent close either becouse of things going on, people casting spells, different classes and point of view, another problem is switching CPUs for every raid and consequitive weeks needed to test. However I've got a plan to test E4300 and E8400 in ICC10 with friend today in raid with pretty much same system spec.

The amount of people in orgrimmar nowadays is bigger than at dalaran bank on my server and I'd say the few more people don't change the general idea of the FPS increase, I find myself that current orgrimmar is even bigger killer than when dalaran used to be, especially if you consider test results between E2160 and E8400 straight forward. E8400 is ofcourse 2nd technological step from core duo to c2d but direct comparison to E4300 at any clock is amazing in my opinion.

Maybe if I can get my hands on R1950pro DDR3 (shame 256MB so I won't be able to run high res tests) and make it work after blackout i could run additional tests to see how CPU/GPU is further effected by orgrimmar. Unfortunatly I won't be able to test it with E8400 so it would be comparison E4300+R1950 vs E8400+R4850.

It's even funnier I've been running that E2160 for over 3 years now, with E4300 on bro weaker PC next to me untill something came across my mind. I'm really looking forward for some response from Phenom II users as AM3 great O/C mobo + phenom + 4gb DDR3 is around 20-30 euros cheaper than E8400 + 4gb DDR2 and GA-(E)P31-DS3L motherboard.
Really hard choice atm for me what to choose as upgrade below 120 euros.

klüger · 12/03/10, 12:46 AM

The most fair benchmark (though it loses on reality) would be flight paths as I see it.

Say one up to island of quel danas from wherever the closest flightpath is (where there should be as little variance in # of people encountered as possible).

Only problem will be faction I guess, but don't most of us have a ~70 alt on another faction?

sp00n · 12/03/10, 4:58 AM

I used the bomb quest flight in Quel'Danas to benchmark various settings. There's a lot going on, multitude of mobs and fights between them.
Don't use /timetest or else the mobs won't be shown, which kind of defeats the purpose of this specific flight.

Also, I don't know if this really was the case, but I think I noticed that the frame rate changed as the time changed, i.e. when it got dark / sunset it suddenly changed notable, and this was the only explanation I could find (besides an error on my side with the settings, which I didn't notice).

Brekk · 12/03/10, 11:13 AM

Originally Posted by Szynszyla

System specs were
Win XP / Win 7, tested on both DX9/11.
Motherboard - GA-EP31-DS3L, northbridge active cooled, southbridge aswell. Good ventilation overall.
Ram - 1x2gb + 1x1GB GOODRAM CL5 DDR2 800mhz, CPU:MEM ratio 1:1 due to awkward motherboard behavior when not running synced FSB. Did not change the timings couse of awkward behavior aswell.

If you can't build a good system you really shouldn't be wasting your time with benchmarking. Especially when you try to establish conclusions about cache which is a minor enough hardware spec that even minor inconsistencies in your benchmarking could far surpass the effect of cache on your results.

The P31 is an ancient chipset, even by LGA775 standards. You have a bizarre ram setup, and a low power "factory freak" video card. This does not make for reliable Benchmarking.