Post Your Framerate Benchmark Here

[Iplaygoats]

On sindragosa i have gotten as low as 3. i tend to carry a speed pot for when i feel i need to run just a bit faster to not die.

[Cilraaz]

On sindragosa i have gotten as low as 3. i tend to carry a speed pot for when i feel i need to run just a bit faster to not die.

I had no real drop on Sindragosa herself. I only experience a drop when we're mass AE'ing the two packs of whelps before her two minis.

[Cilraaz]

Nobody else going to post a benchmark? I think I'll sticky this for a while in the hopes that people with other setups will post as well.

Anyway, here's mine from tonight.

• System Specs:
  
  • Intel Core i5 750 @ 3.8GHz (200MHz base clock, 19x multiplier)
  • ATi Radeon HD 5770 @ stock speeds (850MHz core, 1200MHz VRAM)
  • G.Skill DDR3 4GB CL9 @ 1600MHz (200MHz base clock, 2:8 multiplier)
• WoW Settings:
  
  • 1920x1080
  • 24-bit Color
  • 8x Multisampling (Edge Detect forced from drivers, 16x AF forced in drivers)
  • Ultra, minus half shadows
  • V-Sync Disabled
  • N/A
  • Maximized Windowed Mode
  • All extra effects enabled
• Benchmark Setup:
  
  • ICC - 25-man
  • Partial Instance - First four, Crimson Hall, Dreamwalker, and Sindragosa trash
  • No extenuating circumstances
• Benchmark Results:
  
  • Min: 15 fps
  • Max: 114 fps
  • Avg: 54 fps

Low point was during Marrowgar. The very end was bouncy because of the Sindragosa trash (constant in and out of combat). Just before that was bouncy for god only knows what reason. Probably trash bouncing in and out of my view during Dreamwalker. I usually stare at Dreamwalker and heal her, the BM pet, and the raid.

[None]

I haven't done any raiding for a few months now, but this was with /timetest on a flight from Vengeance Landing to The Shadow Vault:

* System Specs:

• AMD Phenom II x4 955 @ 3.4ghz (Summer made me lower it, I'll redo the same test at 3.8ghz eventually)
• Nvidia GTX470 @ Stock
• G.Skill DDR3 4GB CL8 @ 1600mhz

* WoW Settings:

• 1920x1080
• 24-bit Color
• 8x Multisampling
• Ultra, full shadows
• Vertical Sync Disabled
• Maximized, full screen
• All extra effects enabled

* Benchmark Setup:

• Vengeance Landing - The Shadow Vault
• /timetest locks the camera at a set angle.

* Benchmark Results:

• Min: 43 fps
• Max: 246 fps
• Avg: 83.695

Here's what WoW said after the /timetest:

And here's what the FRAPS log said:
Frames, Time (ms), Min, Max, Avg
38849, 464176, 43, 246, 83.695

Line Graph:

Original Image URL

I did this particular flight because it flies by Dalaran and goes through Icecrown and Howling Fjord which I consider the two most detailed zones. If I start raiding again in Cataclysm I'll do it again while I raid if I upgrade again.

[vesseblah]

• System Specs:
  
  • AMD Phenom II x3 720 BE @ 2.8GHz (stock speed, 14 x 200MHz, stock cooler)
  • Nvidia GTX 460 768MB (stock speed, 675MHz core, 1350MHz shader, 1800MHz memory, Nvidia ForceWare 258.96)
  • Kingstone Valueram 2x2GB CL9, 1333MHz stock speed
• WoW Settings:
  
  • 1920x1080
  • 24-bit Color
  • 8x Multisampling
  • Ultra, shadows at '3' (0-5 scale)
  • V-Sync Disabled
  • Maximized Windowed Mode
  • All extra effects enabled
• Benchmark Setup:
  
  • Toravon the Ice Watcher - 25-man (just one boss +20 seconds of idling before and after)
• Benchmark Results:
  
  • Min: 18 fps
  • Max: 67 fps
  • Avg: 34 fps
  • Median: 29 fps
• Temperatures
  
  • Amb: 24C
  • CPU: 53C max during WoW, 39C idle.
  • GPU: 67C max during WoW, 30C idle.

Will do longer benchmark with ICC25 when possible.

[B@nj0 P3do]

what program/addon do you use to record the fps?

[vesseblah]

Fraps saves nice clean excel file. Turning it into graph takes 2 minutes.

[tlitd]

Cilraaz/None/Vesseblah : finally, some serious benchmarking attempts. Thanks.

A few suggestions :
1. choose a standard (I think that None's idea has the best potential; benchmarking a raid is prone to unnecessary/unwanted fluctuations - raid size, raid comp, encounter)
Furthermore, a given standard should also be free of extra variables. Examples :
- define and adhere to a standard testing mode (minimized window, FS window, FS)
- remove the variability that comes with addons / addon packs, stick to the default UI
2. properly characterize the testing conditions
- be sure to include any relevant driver-level information (driver version, settings such as AA mode/SSAO/force mipmaps/maximum prerendered frames)
- be sure to include in-game settings (AF/MSAA/graphics quality level/shadows), relevant CVars (MaxFPS/M2Faster/gxfixlag/ffxGlow/gxApi/textureCacheSize/gxTextureCacheSize), more obscure CVars (shadowinstancing/fixedfunction/componentCompress)
3. also test the non-obvious decisions (e.g. in-game MSAA versus driver-level MSAA/SSAA/_other implementation_)

[vesseblah]

Cilraaz/None/Vesseblah : finally, some serious benchmarking attempts. Thanks.

A few suggestions :
1. choose a standard (I think that None's idea has the best potential; benchmarking a raid is prone to unnecessary/unwanted fluctuations - raid size, raid comp, encounter)
Furthermore, a given standard should also be free of extra variables. Examples :
- remove the variability that comes with addons / addon packs, stick to the default UI

Problem with None's benchmarking is that doing timed flights on set path will tax almost exclusively GPU making it GPU benchmark. It has only nominal connection to expected raid performance.

Also since 25-man raids are the only reliable way of benchmarking, turning off addons is not really an option. Most of my raiding today is either various pugged ICC25 runs on alts or LK25HM tries. Alt run might be possible to do with clean UI, but not really LKhm.

The good thing about benchmarking raids with slight variance to addons used would give best and most honest numbers people could expect to see when they're raiding with addons themself, but it would require lot more data points than the few so far. What we could standardize is certain WoW settings, which would make comparison easier.

[tlitd]

doing timed flights on set path will tax almost exclusively GPU making it GPU benchmark (I).
It has only nominal connection to expected raid performance.(II)

I. Almost, but not quite. (viewdistance/shadows)
II. That's precisely the point. =]
One can make a "generic" GPU (well, mostly) benchmark, or a much more complex system benchmark. Of course, the latter actually gives an idea of how the behemoth (hardware+software) actually performs. Standardizing a system benchmark is a hell of a task, though.
On the other hand, niche benchmarks (None's) have the advantage of "comparing apples to apples". And this is no small feat.

[Cilraaz]

So why not do both a timetest and a real world test? Next time I'm in game, I'll run a timetest using the same flight points as None did.

[None]

4.0ghz test:

* System Specs:

• AMD Phenom II x4 955 @ 4.0ghz
• Nvidia GTX470 @ Stock
• G.Skill DDR3 4GB CL8 @ 1600mhz

* WoW Settings:

• 1920x1080
• 24-bit Color
• 8x Multisampling
• Ultra, full shadows
• Vertical Sync Disabled
• Maximized, full screen
• All extra effects enabled

* Benchmark Setup:

• Vengeance Landing - The Shadow Vault
• /timetest locks the camera at a set angle.

* Benchmark Results:

• Min: 44 fps
• Max: 250
• Avg: 95.463

Here's what WoW said after the timetest:


And fraps minmaxavg:

Min,
44,
Max,
250,
Avg,
95.463

Graph:

Original Image URL

[vesseblah]

• System Specs:
  
  • AMD Phenom II x3 720 BE @ 2.8GHz (stock speed, 14 x 200MHz, stock cooler)
  • Nvidia GTX 460 768MB (stock speed, 675MHz core, 1350MHz shader, 1800MHz memory, Nvidia ForceWare 258.96)
  • Kingstone Valueram 2x2GB CL9, 1333MHz stock speed
• WoW Settings:
  
  • 1920x1080
  • 24-bit Color
  • 8x Multisampling
  • All on ultra
  • vsync Disabled
  • Maximized Windowed Mode
• Benchmark Setup: Vengeance Landing to Shadow Vault /timetest, all addons disabled, first person view, UI disabled with alt-z
  
• Benchmark Results Fraps minmaxavg (WoW /timetest) in FPS
  
  • Min: 38 (4)
  • Max: 228 (245)
  • Avg: 66 (65)

---------- Post added 2010-09-07 at 05:02 PM ----------

Did few extra benchmarks using the exact same settings and hardware as above in Cataclysm beta 12857. I would draw the conclusion from these that the DX11 mode will suffer notable framerate hit in the old content but is pretty even (few fps behind) in new areas with improved water effects. Probably/hopefully in the release version DX11 version will be equally fast as DX9 if not faster.

Vengeance Landing to Shadow Vault /timetest, all addons disabled, first person view, UI disabled with alt-z. (This route is slightly different from the one in current live 3.3.5)

• Benchmark Results Fraps minmaxavg DX9/DX11 (WoW /timetest DX9/DX11) in FPS
  
  • Min: 15/22 (15/9)
  • Max: 178/135 (219/167)
  • Avg: 64/51 (64/44)

Orgrimmar to Gadgetzan through revised areas /timetest, all addons disabled, first person view, UI disabled with alt-z.

• Benchmark Results Fraps minmaxavg DX9/DX11 (WoW /timetest DX9/DX11) in FPS
  
  • Min: 33/25 (12/12)
  • Max: 141/108 (149/112)
  • Avg: 54/46 (55/47)

[Cilraaz]

Let's make sure we're keeping all of the images to large thumbnails at most. No more than 800px wide.

[None]

My graphs were roughly 700px wide D:

Anyways, any more i5/i7 benchmarks?

[Cilraaz]

My graphs were roughly 700px wide D:

URL: http://img835.imageshack.us/img835/9150/graph2.png
Image width: 1245px

[None]

Not to drag this off topic, but where's that number from? 1,360px × 723px are the dimensions I get when opening it up in Photoshop and from Firefox.

It's fine though, they were very large. It was a joke

[Cilraaz]

Not to drag this off topic, but where's that number from? 1,360px × 723px are the dimensions I get when opening it up in Photoshop and from Firefox.

It's fine though, they were very large. It was a joke

I got it from Chrome's "inspect element" option.

[None]

I got it from Chrome's "inspect element" option.

Does the image stretch past your 1920x1080 resolution or is it too short? That's strange that you're seeing something else.

http://img835.imageshack.us/img835/9150/graph2.png
Try clicking on the direct link and looking at the tab it opens, it should have the image size.
I get the same 1360x723px in Chrome as well.

<img style="-webkit-user-select: none; cursor: -webkit-zoom-out; " src="http://img835.imageshack.us/img835/9150/graph2.png" width="1360" height="723">

That's what I get from Inspect Element in Chrome.

Edit: Durp, I see now.

[Cilraaz]

Ah, mine was auto-shrinking (no idea why on 1920x1080, but whatever).