Forum

SnowiE's timedemo test (Updated 25th Jan 2011)

Created 24th June 2010 @ 15:24

Add A Reply Pages: « Previous 1 ... 3 4 5 ... 12 Next »

kira!

xen
don\\\'t

Quoted from RareSpycrab

Guys i just realized the reason for all the low fps’s. You guys must be alt tabbing out while the demo is running. It drops your fps down to the low 20’s

**Reserved for results**

i play in a window

RareSpycrab

Quoted from kira!

[…]
i play in a window

So do i but the second i alt tab my fps goes to shit. You can see it if you window your browser small enough so you can see your fps meter. The second you alt tab it will turn red.

freshmeatt

‹Con›

So I did a few modifications and ran the whole process once again, i.e. from -autoconfig to setting affinity and stuff.

My best result:

CPU: Intel P4 3.2GHz @ 2.65GHz (HT enabled)
RAM: 1.5GB DDR RAM
GRAPHICS CARD: Nvidia GeForce 7600GS

RESOLUTION: 1024×768
WINDOWED/FULLSCREEN: Windowed

Draw viewmodels: On
FPS Config Used (if any): m0re’s highfps
Kept nvsvc32.exe running this time, didn’t touch affinity settings, mat_queue_mode “-1”.
Average FPS over 3 demo playthroughs: 51.51.

I’m going to try some other mods tomorrow, like other affinity settings, OC’ing CPU and mem, getting deeper into RivaTuner D3D settings.
Good night, ladies and gentlemen.


Last edited by freshmeatt,

Tobyy

TEZC
Pixie <3

CPU: AMD Phenom II 955 Black Edition
RAM: 4GB OCZ DDR3
GRAPHICS CARD: ATI 5770 1GB

RESOLUTION: 1440×900 (also running 2nd monitor)
WINDOWED/FULLSCREEN: Windowed, noborder

Video Advanced Settings (Options > Video > Advanced) –
Model Detail: High
Texture Detail: Very High
Shader Detail: High
Water Detail: Reflect All
Shadow Detail: High
Colour Correction: Disabled
Antialising: 8xMSAA
Filtering Mode: Anisotropic 16X
Wait for vertical sync: Disabled
Motion Blur: Disabled
Field of view: 90
Multicore Rendering: Enabled
High Dynamic Range: Full + Bloom
Direct X Level: 9.0

Draw viewmodels: Off
FPS Config Used (if any): None

Average FPS over 3 demo playthroughs: 116.1

Result with Chris’ max quality config: 89.26 fps average, so not a massive drop


Last edited by Tobyy,

octochris

(0v0)

Quoted from Koeitje

[…]

No, because standard deviation says nothing about how low it drops. Or, well, lets put it this way: it says nothing about the lowest drop (you could guess the probability of a very low fps figure showing up, but that’s pretty much useless). And that’s what matters. 150fps with a standard deviation of 10fps can still drop to 30fps and fuck up that meatshot that you just had to hit to save the round. Min fps would have shown your config doesn’t give you a solid 60+ fps.

It says about the variance, which is what you seem to care about, or what you seem to think is the most important thing to avoid. I agree that variance is the most important thing to try to avoid. As for min FPS, it doesn’t mean jack shit, there are plenty of reasons why you might get a short freeze which is not determined by your hardware which would incorrectly end up displaying your min FPS at a value that is not representative of the result.

jgmaster

BM

Quoted from freshmeatt

So I did a few modifications and ran the whole process once again, i.e. from -autoconfig to setting affinity and stuff.

My best result:

CPU: Intel P4 3.2GHz @ 2.65GHz (HT enabled)
RAM: 1.5GB DDR RAM
GRAPHICS CARD: Nvidia GeForce 7600GS

RESOLUTION: 1024×768
WINDOWED/FULLSCREEN: Windowed

Draw viewmodels: On
FPS Config Used (if any): m0re’s highfps
Kept nvsvc32.exe running this time, didn’t touch affinity settings, mat_queue_mode “-1”.
Average FPS over 3 demo playthroughs: 51.51.

I’m going to try some other mods tomorrow, like other affinity settings, OC’ing CPU and mem, getting deeper into RivaTuner D3D settings.
Good night, ladies and gentlemen.

You should try Chris’ fps configs.

freshmeatt

‹Con›

Quoted from jgmaster

[…]

You should try Chris’ fps configs.

Read my post on first page.

RareSpycrab

Wow after reserving the space i forget all about it. Delete this please


Last edited by RareSpycrab,

metalpiss

Lol I just did the timedemo again with the affinity from core0 removed and got an average of 220.39 fps…That’s a 23% improvement.

EDIT: I’m an idiot, looks like the multi-threading cvars are set to 0 in Chris’s config, so basically I didn’t had multicore there (though in the options it was enabled). With those cvars enabled and affinity thingy I now get 250.21 average…and I started from 178,95 lol.


Last edited by metalpiss,

Koeitje

AUTOBOTS

Quoted from octochris

[…]

It says about the variance, which is what you seem to care about, or what you seem to think is the most important thing to avoid. I agree that variance is the most important thing to try to avoid. As for min FPS, it doesn’t mean jack shit, there are plenty of reasons why you might get a short freeze which is not determined by your hardware which would incorrectly end up displaying your min FPS at a value that is not representative of the result.

Variance is not a problem as long as it stays above a certain level (your fps can vary lots between 100 and 200fps, wont make a damn difference). Short drops, however, are a problem. Consistently having a low min fps in several timedemo tests is not something that can be caused by many reasons, unless your pc is filled with loads of shit and starts to run SETI at random moments. I am not talking about a min fps of 25 in 1 out of 25 benchmarks, I am talking about consistently having a low min fps. Average fps runs wont show consistent dropping to low fps at certain moments in a midfight, those 0.5 seconds where all explosives hit and ragdolls go off are easily offset by the seconds nothing is happening. Yes, it will be lower than running around on an empty map, but it will still not give an accurate representation of actual performance. I don’t think anyone gives a damn about average fps, I do think everybody doesn’t want to see their fps drop below a certain point at any given moment in the game.

Inuy

#sadboys

Quoted from Koeitje

[…]

Variance is not a problem as long as it stays above a certain level (your fps can vary lots between 100 and 200fps, wont make a damn difference). Short drops, however, are a problem. Consistently having a low min fps in several timedemo tests is not something that can be caused by many reasons, unless your pc is filled with loads of shit and starts to run SETI at random moments. I am not talking about a min fps of 25 in 1 out of 25 benchmarks, I am talking about consistently having a low min fps. Average fps runs wont show consistent dropping to low fps at certain moments in a midfight, those 0.5 seconds where all explosives hit and ragdolls go off are easily offset by the seconds nothing is happening. Yes, it will be lower than running around on an empty map, but it will still not give an accurate representation of actual performance. I don’t think anyone gives a damn about average fps, I do think everybody doesn’t want to see their fps drop below a certain point at any given moment in the game.

http://nobodyputsbabyinahorner.files.wordpress.com/2009/11/1238157980_scanners_-_head_explosion.gif

Koeitje

AUTOBOTS

Quoted from Inuy

[…]

http://nobodyputsbabyinahorner.files.wordpress.com/2009/11/1238157980_scanners_-_head_explosion.gif

http://tweakers.net/ext/f/HsCEKwRAM4zNLTlbuP6eAe14/full.gif

octochris

(0v0)

Quoted from Koeitje

Variance is not a problem as long as it stays above a certain level (your fps can vary lots between 100 and 200fps, wont make a damn difference).

Then to balance out the fact that such extremes could very likely be anomalous, the best setup for such an thing would be:

Average FPS:
Minimum FPS:
Stdev:

That tempers the potential innacuracy of such an extreme, at least.

With that said, however, Fraps is commercial (outside of the 30sec trial), so it’s not really feasible. I don’t know of any other benching tools which could do that.

hleb

CPU: AMD athlon 64 x2 4400+@overclocked to 2.7 per core
RAM: 2 gb ddr2
GRAPHICS CARD: ATI HD 2600 XT

RESOLUTION: 1024×768
WINDOWED/FULLSCREEN: Full

Video Advanced Settings (Options > Video > Advanced) –
Model Detail: High
Texture Detail: High
Shader Detail: High
Water Detail: Reflect All
Shadow Detail: High
Colour Correction: Disabled
Antialising: none
Filtering Mode: trilinnear
Wait for vertical sync: Disabled
Motion Blur: Disabled
Field of view: 90
Multicore Rendering: Enabled
High Dynamic Range: Full + Bloom
Direct X Level: 9.1

Draw viewmodels: Off
FPS Config Used (if any): None

5125 frames 60.556 seconds 84.63 fps (11.82 ms/f) 8.224 fps variability

crzfst

CPU: Intel Q9400
RAM: 4GIG DDR2 RAM
GRAPHICS CARD: Nvidia GTS 250 1GB

RESOLUTION: 1280×1024
WINDOWED/FULLSCREEN: FULLSCREEN

Video Advanced Settings (Options > Video > Advanced) –
Model Detail: High
Texture Detail: High
Shader Detail: High
Water Detail: Reflect World
Shadow Detail: Medium
Colour Correction: Disabled
Antialising: None
Filtering Mode: Bilinear
Wait for vertical sync: Disabled
Motion Blur: Disabled
Field of view: 100
Multicore Rendering: Enabled
High Dynamic Range: None
Direct X Level: 8.1

Draw viewmodels: Off
FPS Config Used (if any): Some random config i found on etf2l some time ago http://www.mediafire.com/?qg2mmqzymmk
Average fps over 3 demo playthroughs: 153.28

Add A Reply Pages: « Previous 1 ... 3 4 5 ... 12 Next »