What games do you want to see in TPU CPU reviews?

[cdawall]

Cad said he is willing to rebench some of his chips to get us some gaming results in the CPU testing he does. What games does everyone want to see? Requirements are an automated benchmark and it has to show a good amount of change no sub-1FPS differences.


So far we have

• ARMA II
• GTAIV
• Crysis
• Dirt Showdown

Please comment on the games already chosen and on games you think should be tested!

 

[erixx]

Sniper Elite 2
BF3
FSX

PD: Dirt Showdown, why? I thought it was a misbirth and not up to Dirt series level.

 

[btarunr]

Dave could be doing CPU reviews, so his selection could include CPU-limited titles (such as Starcraft II), as well as games that are cutting-edge, which benefit from multi-core and new instruction-sets (such as Battlefield 3), but not so much the "in-betweens" (console-ports).

 

[cdawall]

PD: Dirt Showdown, why? I thought it was a misbirth and not up to Dirt series level.

Dist shows a decent difference with minor CPU changes just ran it on all low/off 5045X1050 @3.6 and 3.84 only changing mult and there is a 7FPS difference in that alone for average FPS.

 

[EarthDog]

Id like to see this testing across a couple of resolutions... Show a severely CPU limited res like 1280x1024, up to 1920x1200 or 2560x1440 if he has a monitor to support it.

 

[Sir B. Fannybottom]

The only problem with some of these is that it's hard to have an exact same run every time which could mess with results

 

[cadaveca]

Dave could be doing CPU reviews, so his selection could include CPU-limited titles (such as Starcraft II), as well as games that are cutting-edge, which benefit from multi-core and new instruction-sets (such as Battlefield 3), but not so much the "in-betweens" (console-ports).

That's why i use Shogun2 already. However, it's also very sensitive to memory performance, and this can be affected from motherboard to motherboard.

Id like to see this testing across a couple of resolutions... Show a severely CPU limited res like 1280x1024, up to 1920x1200 or 2560x1440 if he has a monitor to support it.

I will not test multiple resolutions. I test 1920x1080 only, as for now, that's the "standard" for 16:9(HDTV).


I have 2560x1600, and 3x1920x1080 monitors too.

 

[cdawall]

The only problem with some of these is that it's hard to have an exact same run every time which could mess with results

Gotta run them a couple of times and get an average.


here is a list of all games with built in benchmarks

Games with built-in benchmarks
DroneZmark ( Built-in Benchmark )
FarCry ( 2 pre-recorded Timedemos, record/playback timedemo available )
Crysis ( 1 pre-recorded GPU Timedemos, 2 pre-recorded CPU Timedemos, record/playback timedemo available )
Devil May Cry 4 ( Built-in Benchmark )
Call Of Juarez ( Built-in Benchmark )
Doom III ( pre-recorded timedemo, record/playback timedemo available )
Quake IV ( record/playback timedemo available ) [ I think there's also a dedicated benchmarking application from HardwareOC available ]
Prey ( timedemo recording and playback capabilities )
S.T.A.L.K.E.R. : Shadow Of Chernobyl ( timedemo recording and playback capabilities )
Enemy Territory: Quake Wars ( record/playback timedemo available )
Call Of Duty ( record/playback timedemo available ) [ or use BenchEmAll's timedemos and automated features ]
Call Of Duty 2 ( record/playback timedemo available )
Call Of Duty 4 ( record/playback timedemo available )
Tomb Raider: Angel Of Darkness ( BenchEmAll )
The Chronicles Of Riddick: Escape From Butcher Bay ( dedicated benchmark application available, goes by the game's name followed by the word benchmark )
F.E.A.R. ( Built-in Benchmark )
Unreal Tournament ( record/playback timedemo )
Unreal Tournament 2003 ( record/playback timedemo & BenchEmAll )
Unreal Tournament 2004 ( record/playback timedemo & BenchEmAll )
Unreal Tournament 3 ( record/playback timedemo & Dedicated Benchmarking Utility )
Counter-Striker: Source ( Built-In BenchMark )
Half-Life 2: Lost Coast ( Built-In Benchmark )
Lost Planet ( Built-In Benchmark )
World In Conflict ( Built-In Benchmark )
Company Of Heroes ( & Opposing Fronts ) [ Built-in Benchmark ]
Supreme Commander ( Built-In Benchmark ) * not 100% sure, my memory....woes
x2: The Threat [ Built-in Benchmark ]
x3: Reunion ( Built-In Benchmark )
Comanche 4 ( Built-In Benchmark )
Serious Sam II ( record/playback timedemo )
PT Boats: Knights Of The Sea ( Built-in Benchmark )
BattleField 2 ( record/playback timedemo & HardwareOC benchmarking utility )
BattleField 2142 ( record/playback timedemo & HardwareOC benchmarking utility )
Half-Life 2 ( Built-In Benchmark & Benchmarking Scripts available )
Painkiller ( BenchEmAll )
Halo: Combat Evolved ( BenchEmAll )
Splinter Cell ( BenchEmAll )
Return To Castle Wolfenstein ( record/playback timedemo & BenchEmAll )
Unreal II: The Awakening ( BenchEmAll )
Quake III Arena ( BenchEmAll )

(thanks ashen)

 

[cadaveca]

"Time demos" are not acceptable, nor is titles using "bench'emall".

Benchmark must either be self-contained, or a repeatable built-in that anyone that has the title has access to by default. A tool to run the benchmark like Crysis for FarCry or Sleeping Dogs use is fine.

Remember that I do one review a week, sometimes two, and as such, I do not have unlimited time to run benchmarks(I am not an employee of TPU). All testing must show actual performance differences between CPUs, even if the difference is only a multi or two.

 

[Peter1986C]

The Metro 2033 benchmark with PhysX enabled and running on the CPU could be a candidate. It is an intensive scene (lots of action) and the PhysX could load the CPU even more. The behaviour of the NPCs is exactly the sajme each and every time.

@cdawall: if Far Cry refers to Far Cry 2: that benchmark is not fully constant, because IIRC some NPCs react slightly different between runs.

 

[cadaveca]

@cdawall: if Far Cry refers to Far Cry 2: that benchmark is not fully constant, because IIRC some NPCs react slightly different between runs.

That's fine, I'l lrun tests 5 times, and average the results. Common practice when it comes to reviews, actually, altohugh I dunno that everyone runs tests 5 times like I do

 

[EarthDog]

I will not test multiple resolutions. I test 1920x1080 only, as for now, that's the "standard" for 16:9(HDTV).


I have 2560x1600, and 3x1920x1080 monitors too.

Oh sorry, I missed the point of this thread if thats the case... my bad.

 

[cadaveca]

Oh sorry, I missed the point of this thread if thats the case... my bad.

What tests, not HOW those tests are run. There are many different sites that review dstuff, fortuantely, so I am not eager to change my testing perspective just yet. If all sites review stuff in the same manner, there is no point to having multiple reviews out there...


I won't falsely create bottlenecks for CPU or GPU to affect test scores. Let me put it THAT way, so you understand my stance on this.

few peopel gmae @ 1280x1024, and likewise with 2560x1600 and eyefinity. 1920x1080 is that ocmmonality that keeps things on an even keel, and then performance can also transfer to those that use HDTVs as monitors, or are interested in HTPC usage.

Higher-res create GPU bottleneck...lower res created CPU or memory bottleneck. You need to be able to distinguish what that bottleneck is...exactly what it is...memory? CPU? CPU BUs? or perhaps PCIe bandwidth? Too many factors to account for.


cdawall asked for other game tests. I'm more than willing to add 4-5 more games to my testing routine. I'm not about to completely change how I do the reviews though, at least, not at this juncture.

Maily, adding another resolution exponetially increases the number of tests I must run. Say 40 tests..two resolutions is now 80 tests...3 resolutions, 120 test...I simply do not have the time for that. If you find someone who does, let me know, I'd love to offer them some review work.

 

[EarthDog]

Agreed... A hair more than 1/4 of STEAM users use 1920x1080 (27%). 29% use 1280x1024 or 1366x786 (10% and 19%) with another 10% coming in at 1680x1050... ALL less than 1920x1080. Above 1920x1080 is a mere 4%. The majority of users are less than 1920x1080 but that is what I would test as well.

http://store.steampowered.com/hwsurvey?platform=pc

 

[cadaveca]

The majority of users are less than 1920x1080 but that is what I would test as well.

1920x1080 is also the maximum that any single VGA can really handle. I've had 2560c1600 for many years(since October 2007), and still no single card gives the performance I need for that res, so it's still a GPU bottleneck = testing not gonna be done by me.


With these new drivers from AMD, it looks like i might do another re-bench and then debut new scoring in the new year. I'll probably move down to a single GPU at that point, but whaich one will depend on what I can get for free.

Even dual HD 7950's isn't quite enough for 2560x1600...but damn do I wish it was!

 

[TheMailMan78]

I wish BF3 had a standardized bench.

 

[cadaveca]

I wish BF3 had a standardized bench.

me too, since i have near 500 hours in it playing!

 

[EarthDog]

1920x1080 is also the maximum that any single VGA can really handle. I've had 2560c1600 for many years(since October 2007), and still no single card gives the performance I need for that res, so it's still a GPU bottleneck = testing not gonna be done by me.


With these new drivers from AMD, it looks like i might do another re-bench and then debut new scoring in the new year. I'll probably move down to a single GPU at that point, but whaich one will depend on what I can get for free.

Even dual HD 7950's isn't quite enough for 2560x1600...but damn do I wish it was!

Your standards are high I suppose. I rock 2560x1440 and can run BF3 all day long on Ultra.

 

[tacosRcool]

me too, since i have near 500 hours in it playing!

how do you have the time to play that much?!?! I'm not even at 200 hours yet (more like 150 hours)

 

[cadaveca]

how do you have the time to play that much?!?! I'm not even at 200 hours yet (more like 150 hours)

I don't have a job. Reviews for TPU is hte closest I've had in the past two years, actually. BF3 has been out for nearly a year now(i think october 25th?), so that's just a wee bit more than an hour a day...and I do play daily.

Your standards are high I suppose. I rock 2560x1440 and can run BF3 all day long on Ultra.

I get slowdowns to under 60 FPS even with dual 7950's. anything less than 60FPS has a very differnet visual sharpness to it that I have been seeking for ages. Typically, I'll turn down settings to get that, even, and AA is the first to go.

 

[EarthDog]

Try SSAO instead of HBAO and FXAA instead of the native to "ultra" MSAA.

And christ, yeah your standards are HIGH! With those settings though, Im not sure I see less than 60FPS to be honest... single card (680 Lightning at stock)

 

[TheMailMan78]

Try SSAO instead of HBAO and FXAA instead of the naive to "ultra" MSAA.

Thats not balls deep. Dave only knows nuts or nothing.

 

[cadaveca]

Try SSAO instead of HBAO and FXAA instead of the naive to "ultra" MSAA.

I do not like the quality of the image.

 

[qubit]

Unreal Tournament 2004. I just wanna compare the performance of today's graphics with those around when this game was on the TPU gaming benchmarks. One can really get a sense of just how far we've come with comparisons like these.

 

[cadaveca]

Unreal Tournament 2004. I just wanna compare the performance of today's graphics with those around when this game was on the TPU gaming benchmarks. One can really get a sense of just how far we've come with comparisons like these.

Ok, so I guess I'll ahve to add "Works in Windows 7/8 64-bit" to the requirements list?





Good idea though, and that was actually an app I have looked at in the past, but it doesn't always work right in Windows7.