In Game Benchmark Thread

[EastCoasthandle]

I always wondered how some review websites were able to show line charts of in game performance. I stumbled upon how when configuring my Fraps time demos and it's really easy to do (I will explain below). In any case games like UT3, BF2 single player (BF2 bot changer), CS:S, TF2, Crysis, COD4 SP/MP, etc.

Here is how you do it:
-Open Fraps and tab over to FPS
-Enable Stop benchmark automatically after 250 seconds
-Enable MinMaxAvg, Frametimes and FPS
-Start game
-When ready start the benchmark
-When complete go to C:\fraps\benchmark\*fps.csv (file)
-Open the file with Excel or similar program
-Click on the bar graph icon to convert the data to a line chart
-Follow the instructions to create the line graph.
-Once created you can right mouse click the fore/back grounds of the chart to change color, etc.
-Post your specs

3DMark06
Return to Proxycon

Average 40.478 FPS

FireFly Forest

Average 48.852 FPS

Canyon Flight

Average 54.086 FPS

Deep Freeze

Average 65.340 FPS



COD4: The BOG
1680x1050
4xAA/16xAF (AKA slider to the far right)
All other settings set to it's maximum

Min 17
Max 108
Avg 50.514




DiRT
1680x1050
All settings enabled except shadow as max shadow settings do not work with this video card. Therefore the next best setting was used.
Map: Avelsbachring, German 2 laps
Peugeot 307 vs Subaru Impreza WRX STI Spec-C
Used keyboard to maneuver

In game play
Min 32
Max 47
Avg 40.952

Replay (using the same camera angle)
Min 35
Max 47
Avg 41.164

Disparity is the result of:
-In game cut scenes at beginning and end of race that are not found in replay
-Differences in frame rate appear within margin of error.

 

[EastCoasthandle]

Anyone willing to give it a try?

 

[Hawk1]

I'll give it a go, but cant until next weekend (work has me swamped this week). Does it matter what game? Was going to try BF2142. Is SP play a must (havent done that since, oh, the first five minutes I owned it), or can I do it with MP?

 

[EastCoasthandle]

I'll give it a go, but cant until next weekend (work has me swamped this week). Does it matter what game? Was going to try BF2142. Is SP play a must (havent done that since, oh, the first five minutes I owned it), or can I do it with MP?

No it really doesn't matter. It can be any game be it SP or MP.

I also made a correction to the OP. The file name is *FPS.csv

 

[pepsi71ocean]

seem interesting i'll give it a go as soon as modern marvels is over

 

[MikeJeng]

Radeon 9800

Pentium 4

1 GB RAM



Crysis: .2 FPS (As I said before, I'm not joking)

Bf2: 15 fps

UT3: 7 FPS

Counter strike Source: 20 FPS

Orange box: 14 FPS

Cod4: Unplayable

Portal: 9 FPS

Crysis menu: 30 fps. What the aksdjfklsa

 

[EastCoasthandle]

Radeon 9800

Pentium 4

1 GB RAM



Crysis: .2 FPS (As I said before, I'm not joking)

Bf2: 15 fps

UT3: 7 FPS

Counter strike Source: 20 FPS

Orange box: 14 FPS

Cod4: Unplayable

Portal: 9 FPS

Crysis menu: 30 fps. What the aksdjfklsa

Hmm, what resolution are you playing? Have you tried 1024x768?

 

[MikeJeng]

Hmm, what resolution are you playing? Have you tried 1024x768?

I am playing source games at 1024x768 and COD4, Crysis and UT3 are at 800x600 on all low.




Also, I don't officially own UT3. I just tried it.

 

[EastCoasthandle]

If you ever decide to upgrade you can look at my system specs just below my user name to get an idea of what to get (besides the gpu as it's end of line EOL).

This is CS:S at 1920x1200 4xAA/16xAF

 

[pepsi71ocean]

Nothing much more to say..

Does those seem like ok frame rates?

Resolution max resolution, at 60Hz
all settings maxed out

 

[EastCoasthandle]

^^Heck yeah! That's pretty good. What scene/map did you play? One thing you show everyone is a better picture of how well the game plays with your setup.
------------------------------------------------
Below you will find UT3:
-Map Suspense
-1680x1050
-31 bots
-all settings maxed

 

[pepsi71ocean]

The one that deal with D-Day i think the 3rd mission. It was on expert mode as well. And i noticed a frame drop when i zoomed in with the scope.

Either tonight or tomorrow i'll have Axis and Allies, Sim City 4 World In Conflict and Rainbow 6 dome.

I would like to do a comparison between everything stock and where i am right now, that should bee an interesting study.

 

[EastCoasthandle]

Great!

 

[EastCoasthandle]

Here is another example
BF2 Karkand SP with 56 bots
1280x1024 (this game does not scale properly at higher resolutions)
UAV and Artillery strikes were used
game.lockfps 400 was used... "~"

 

[d3athviru5]

UT3 , Shangri La ... with 31 bots .. running at 1280*1024 , 8x CSAA , 16x AF , maxed settings ..

Graph has only the first 250s of the benchmark .. the full details are below ..

Frames : 16219 , Time : 323264 ms
Min : 36 , Max : 66 , Avg : 50.173

 

[EastCoasthandle]

deathvirus_me,

Nice score...it looks like you have the 62 FPS lock on. You can disable it by:
-C:\Documents and Settings\yourname\My Documents\My Games\Unreal Tournament 3 Demo\UTGame\Config
-edit UTEngine.ini
-find the:
[Engine.GameEngine]
bSmoothFrameRate=TRUE <--Change this to FALSE
MinSmoothedFrameRate=22
MaxSmoothedFrameRate=62 <--Change this to 150
bClearAnimSetLinkupCachesOnLoadMap=TRUE

-save the file
-restart the game (you may have to start the game twice to get it to work). You should see higher frame rates at the menu using fraps.

 

[EastCoasthandle]

WiC
1680x1050
2xAA/2xAF
All settings max for DX9 game play
Benchmark was used

41 FPS average



Quake 4
1680x1050
All settings Maxed
Mission: Escort Strauss to a command station to turn off liquid cool room

60 FPS cap or else game doesn't work correctly.



UT3: Shangri La 31 bots
1680x1050
All settings max'd

Min 50
Max 131
Avg 79.723

 

[EastCoasthandle]

The purpose of these graphs is to prove more information of how a PC is doing in a game besides min, max and avg that we normally find. IMO this provides you with a much clearer picture of performance vs. how high/low frame rates are for a second or 2. Some believe min. frame rates are important. When you create your own chart you will clearly see that min doesn't become important unless the dips are more frequent then normal. One or two depends doesn't detract you from the game's immersion. It's the consistency of those dips that do. IMO consistent dips below 30 FPS followed by peeks of 60 FPS can be cause for concern. The quake 4 chart is IMO the most ideal frame rate you would want in any game. It's the test were I found the frame rate that consistent. IMO Q4 is the best optimized game for both gpu/cpu I've seen to date.

 

[pepsi71ocean]

hey sorry for the late reply, but i ran into a case over clocking mania, i have the stock scores and i'll work on my over clocked scores when i get hone from work today. I'll add them to this post when i finish my stock scores.

 

[EastCoasthandle]

hey sorry for the late reply, but i ran into a case over clocking mania, i have the stock scores and i'll work on my over clocked scores when i get hone from work today. I'll add them to this post when i finish my stock scores.

No problem

 

[Hawk1]

Well here's my results from last night:

BF2142: Camp Gibralter MP (63 players)
All settings max except 2xAA

x1950xtx @ 675/1017
Other specs to the left

Resolution: 1920x1200 w/widescreen fixer enabled (true ingame FOV)
Edit: This was in XP Pro SP2

Min. 38
Max 94
Avg 73

Will try Crysis this week (have to do a reinstall). On my 12x10 monitor, I could run it smooth w/left side on High and right on Medium, so I'll see how 19x12 does.

 

[EastCoasthandle]

that's pretty good for that game!

 

[Hawk1]

that's pretty good for that game!

Well, I mostly play sniper, so I don't get into the frontline battles, but yeah, I was pleased with the performance (just got my 24", so I will have to try different settings on Crysis etc. to see what is playable).

 

[EastCoasthandle]

Dirt
1680x1050
all settings maxed (except shadows as it will not work max with this card)
Adaptive AA enabled in CCC

In Game
Min 29
Max 46
Avg 41.239

Replay
Min 36
Max 48
Avg 42.296

Difference between In game and replay is 1.057 FPS. Now look at the numbers, then look at the chart. You can clearly see that the numbers make the frame rate difference worst then the actual results in game vs replay. The truth of how this game played is in the chart, not the numbers. That's why min/max and average really has no tangible meaning in and of itself unless you also see the progression that lead to that result. The difference at the beginning and end of the game was do to a mini cut scene in game that's not found in the replay. One could naturally assume that the min of 36 FPS was common frame rate however the chart clearly show it's not. It's only found for a fraction of the time while gaming. Therefore, the difference between the min of 29 in game and 36 in replay have no real tangible meaning for this game. It's only 1 piece of a puzzle. The whole picture can be seen using the graph. The same goes with the other DiRT chart I create. In all showing pattern which lead me to this conclusion for this game and some of the in game maps I've played so far.

Most of the charts presented in this thread so far clearly show that the min frame rate is not as common as some people thought. That it only represents a fraction of the game play. The same goes with the max frame rate. Therefore, as you look at the numbers and look at the graph the numbers clearly don't make a whole lot sense. IMO, this is because you look at the numbers you only see min, max and average which are simply 3 pieces of a 150+ piece puzzle. When you assemble the pieces together and form a graph it's clear that the min and max is nothing more then dips and spikes found within that time frame and are part of the game play, not the average for the game play.

So that leaves us with the average. Why isn't the average jiving with the chart? Well, it's suppose to account for the entire experience. However, because it's symbolized in double digit form you don't get the whole clear picture of how the game progressed. Instead of seeing something similar in the graph showing a line averaging around 41 or 42 FPS you see a much different picture which includes the spikes and dips normally associated with most games you play.

However in all, the difference between in game play and replay are very similar looking at the chart.
Thoughts?

 

[Hawk1]

Thought I would give Supreme Commander a try tonight:

Map: Setons Clutch
4 player(2 teams) - 3 AI Normal
500 unit max

x1950xtx 675/1017

Resolution: 1920x1200

Min 12
Max 63
Avg 24.84

Hadn't played this game in a while, so thought I'd try a quick skirmish with a 4 man easy AI (this is my only RTS game so I pretty much suck and didn't want to get my ass kicked early on hard AI). I ran the bench near the end (final battle) and it finished just before the AI's guy went nuke with pretty much everyone having lots of units on the board (although I was looking at a birds eye view fairly close). Despite the low min/Avg I found the game/viewing very playable (but maybe its just my shit perception of FPS).