GPU not being used 100%

[wolf2009]

so my HD4870 is not being used 100% in TF2. its only being used 44% max. is it because of my res 1280x1024 ? cpu is q9450 at 3.2Ghz.

 

[inf3rno]

On low resolutions games are more CPU dependent, At this resolution your CPU is doing most of work. Run the game at 1600X1200 and you ll feel the difference in GPU load. GPU usually takes charge from this resolution.

Edit: And why now "Diwali diyan lakh lakh wadhaiyan"?

 

[W1zzard]

tf2 is very cpu limited

 

[inf3rno]

tf2 is very cpu limited

I just described the common rule.

 

[wolf2009]

tf2 is very cpu limited

i just saw one of your reviews, and it confirms my suspicion. there is not much fps difference b/w 9800GT and HD4870 in TF2. TF2 is only using like 45-50% gpu.

 

[W1zzard]

i think we'll be dropping tf2 in the next complete rebench, and l4d probably too because its buggy as hell with new gpus.

btw: dont forget to use some kind of steam emulator with those games if you are looking to use those benchmarks for reviews. your fps data can be screwed big time by some steam auto updater thingie suddenly running, or the latest update improving fps by xx%.

 

[wolf2009]

i think we'll be dropping tf2 in the next complete rebench, and l4d probably too because its buggy as hell with new gpus.

btw: dont forget to use some kind of steam emulator with those games if you are looking to use those benchmarks for reviews. your fps data can be screwed big time by some steam auto updater thingie suddenly running, or the latest update improving fps by xx%.

thanks for the tip,

what kind of steam emulator do you mean ? how can you run those games without steam ?

btw i'm using a custom recorded gameplay timedemo , playing it with the demo player tool that can be accessed by Shift+F2 and measuring FPS with FRAPS, is the result accurate in terms of avg, min, max fps with that method ?

 

[W1zzard]

what kind of steam emulator do you mean ? how can you run those games without steam ?

google is your friend

btw i'm using a custom recorded gameplay timedemo , playing it with the demo player tool that can be accessed by Shift+F2 and measuring FPS with FRAPS, is the result accurate in terms of avg, min, max fps with that method ?

no idea if that approach is accurate, i never use it for benchmarking. You can use the +timedemoquit parameter to run a timedemo, save the results to some log file and quit. i noticed that a single run to warm up file system caches helps to get more reliable results

 

[wolf2009]

no idea if that approach is accurate, i never use it for benchmarking. You can use the +timedemoquit parameter to run a timedemo, save the results to some log file and quit. i noticed that a single run to warm up file system caches helps to get more reliable results

doesn't that approach try to play the demo as fast as possible ? if so wouldn't the min and max data by FRAPS be inaccurate ?

 

[Zehnsucht]

I made a quick observation during some gaming.
Settings: 1680x1050
4xAA
UT3 ~60-70% usage
Painkiller ~90-100% usage

Older games = more optimized?

 

[W1zzard]

doesn't that approach try to play the demo as fast as possible ? if so wouldn't the min and max data by FRAPS be inaccurate ?

yes, but it would load the gpu more which is the objective of a gpu review? your methodology might differ though, i guess there is something like a "demoquit" parameter too

 

[wolf2009]

yes, but it would load the gpu more which is the objective of a gpu review? your methodology might differ though, i guess there is something like a "demoquit" parameter too

thanks, will try tonight and report back.

 

[newtekie1]

I made a quick observation during some gaming.
Settings: 1680x1050
4xAA
UT3 ~60-70% usage
Painkiller ~90-100% usage

Older games = more optimized?

No, PainKiller just relies less on the CPU. UT3 has a lot more physic in it.

 

[wolf2009]

yes, but it would load the gpu more which is the objective of a gpu review? your methodology might differ though, i guess there is something like a "demoquit" parameter too

just tried that, this method had even less gpu usage. topped out at 35 now, approx average was 28.

 

[Steevo]

Some games also place a large demand on the memory of a graphics card, GTA4for example, I can get a few FPS more by bumping my memory up, but then I get artifacts.



W1zz, is there any way we could have a few benchmarks done with GTA4?

 

[W1zzard]

W1zz, is there any way we could have a few benchmarks done with GTA4?

nope. i did look into this last week and decided to boycott gta 4 due to retarded copy protection, drm and gay launcher.
maybe if they release a version that runs without internet access and launcher i'll revisit this.

gta 4 engine sucks anyway. i added hawx and dawn of war 2 to our benchmarks which will appear after the next full rebench

 

[Steevo]

nope. i did look into this last week and decided to boycott gta 4 due to retarded copy protection, drm and gay launcher.
maybe if they release a version that runs without internet access and launcher i'll revisit this.

gta 4 engine sucks anyway. i added hawx and dawn of war 2 to our benchmarks which will appear after the next full rebench

True about GTA4 DRM. It takes almost as much time to get the game started due the the launcher and associated BS as it does to load the damn thing.:shadedshu

I built my whole PC with the idea of being able to run GTA4 at high settings, and I can for the most part, still get a few stutters and the initial caching of the game is crap. The biggest complaint is the shitty looking shadows. Get in an area with lots of trees and it makes my system beg for mercy. Might have to get a 4890 to make up for it.

 

[wolf2009]

Some games also place a large demand on the memory of a graphics card, GTA4for example, I can get a few FPS more by bumping my memory up, but then I get artifacts.



W1zz, is there any way we could have a few benchmarks done with GTA4?

that is another game which doesn't use 100% gpu at 1280xx1024

 

[Studabaker]

thanks for the tip,

what kind of steam emulator do you mean ? how can you run those games without steam ?

btw i'm using a custom recorded gameplay timedemo , playing it with the demo player tool that can be accessed by Shift+F2 and measuring FPS with FRAPS, is the result accurate in terms of avg, min, max fps with that method ?

I know I'm super late here but to run COD4 MP without steam I just add --nosteam to the command line when I make a shortcut. I don't know what it's like for all the other games, as I don't really play any other games aside from the occasional demo.