The Game Rundown: Finding CPU/GPU Bottlenecks, Part 1

[lafery]

Great idea for an article. Thanks!

This makes me wonder if it would be possible to do something alike and find out what are the best configuration when doing large PvP battles in some MMORPGs - those always bring down my systems to their knees. What's the bottleneck in these scenarios? Network? Video card? Number of cores?

I would love to find that out!

 

[Guest]

Yeah please test the games @ 1440x900.
That's a way more common resolution than 1920x1200...

 

[rhino13]

This is such a great article.
This is why I read Tom's.
Thanks so much for another superlative!

 

[LORD_ORION]

This looks like useful information, but your CPU performance testing is deeply flawed. You need to learn how to use performance counters. Simply because CPU utilization is a X% means nothing.

eg: For starters System\Processor\Queue Length Counter: Processor activity itself only indicates that the resource is being used, not that maximum use of the resource is a problem. However, a long, sustained queue indicates that threads are being kept waiting because a processor cannot handle the load assigned to it.

In short, it ultimately doesn't matter if CPU utilization is 99.9%. It might or might not be a processor bottle neck without more indepth understanding of performance monitoring.

 

[precariousgray]

I am not certain how practical it would be to do so, but I'd like to see a game such as Final Fantasy XIV included in part two, or three if there is one. Might as well have this be a continuing series and plow through every game people play, eh?

 

[warezme]

Why didn't you also test the 460's in SLI?

 

[agawtrip]

include racing games please, and arma 2, lost planet 2

 

[nforce4max]

[citation][nom]adonn78[/nom]Its True GTA4 can use 1.68GB of Video RAM at 1920x1200. And up to 1.8GB at higher res. The game can get pretty choppy unless you got a 2GB video card and a quad core. They even fixed the shadows so that they use less VRAM but it still craves GPU memory.[/citation]

The only way that I managed to get the game to run smooth was to move the bloated pagefile to a new SSD.

 

[sohei]

test the configuration on Civilization 5

anyway a cpu is not just for gaming maybe you run something else in background when you gaming

 

[Sorax]

Great article. This is what I come to Toms for.

 

[opt1x]

This is one of the best articles I've read. Tom's Hardware always produces amazing articles like this, and I take them into consideration when making upgrading decisions.

Please in part II add Starcraft 2! That would greatly benefit me and I'm sure a ton of others!

 

[adder1971]

This is quite possibly the best article on this subject I have ever seen. GTA IV is a CPU beast! Really need to get a quad to feed my GTX 465. (was wondering why the frame rates seemed lower than expected)

 

[fausto]

this is the type of article i like to see. Anandtech dropped the ball in recent times.

 

[coldmast]

I love you Toms Hardware! Finally seeing how different games behave and use quad-core, so much for games not using all the cores, the grand-theft auto benchmark was maxing out the system! Also for part 2 is toms going to be using their (crazy) custom SCII bench?

 

[coldmast]

[citation][nom]bikermicefrmars[/nom]All the tests should have also been done at 1440x900![/citation] Toms was picking a base resolution that most gamers will be after on a new purchase, and reflecting the bottleneck on various games. I would like to see processor scaling @ 1920x1200, with 480s single & SLI, next to 5870 single & crossfireX; but that would take months.

 

[funtastic]

++++++
Excellent article. Leaves many users wanting more though. I speak on behalf of millions when I ask Tom's to please sticky a database of many more popular games showing if it's more dependent upon; cpu 1-4core / gpu speed vs memory size/ sli vs xfire etc....
I mean, it'd be nice to go to a page where I see stats based on the game instead of only the cpu/gpu reviews. Maybe even have drop down boxes selecting a specific cpu/gpu setup which reflect stats instead of combing through. This article is a step in the right direction for gaining memberships.

 

[L0g1c]

What software did Tom's Hardware use to see CPU / GPU Load percentages. I'd like to do my own testing.

 

[wolfram23]

I would LOVE to know how you figure out the amount of VRAM being utilized!!

 

[pale paladin]

great article. It's nice to see the scaling with different cores. good job guys and gals on this one.

 

[gamerk316]

I would have liked to see Process Explorer used to get the thread count for each game...

 

[techonnapkins]

The BC2 diagnosis really drives home the point of more focus needed on minimum framerates. I overclocked my e8500 to over 4GHz and was still hitting hiccups in physics-heavy scenes. When I upgraded to a Thuban, it ran smooth as butter.

 

[dndhatcher]

I would like to have also seen a comparison of a mainstream HDD like a WD Black 1TB FALS model vs a good SSD instead of a raptor for everything. I hear people frequently claim that moving a game to the SSD reduced choppiness. It would be helpful to see if some games do more frequent disk access and benefit more from an SSD than others.

 

[c_shura]

Great article. Right now I clearly know I don't need to upgrade my E8400.

 

[cnox]

Are we ever going to see this same set of tests on AMD platform? I'm really interestd in understanding what the differences betwen the i7 and penom II architecture has on gpu/cpu balance. Obviously a GHZ isn't a GHZ between the cpus, so why do they run differently...

 

[bphill561]

I hope part 2 includes Starcraft 2. Great article so far.