Source Engine Video Stress Test

[CrazyCrusher]

I found this to be interesting, Given how many sites differ from one anothers story. below is a copy and past of part of the review, click on the link if interested. if you mods feel this subject has been molded over feel free to close it, I shall take know offence :)


It's important to realize that the Video Stress Test included with the current Counter-Strike: Source beta is not a benchmark of Half-Life 2. It's not a benchmark of the new Counter-Strike game, either. The VST is simply a synthetic 3D benchmark using the Source engine. It's a flyby through several "rooms" meant to showcase select features of the game engine, but it does not put the same load on your hardware that playing any Source engine based game would. It may give us a relative picture of graphics performance in Source-based games, but you can't look at benchmark numbers from the VST and think that's what you'll get from the games.

the two tests are really not comparable. The Doom 3 benchmark is a recording of actual gameplay in a shipping game. This is a synthetic test that just happens to use a real game engine, and from an as-yet unreleased game at that. It's a bit like looking at an Unreal Tournament 2003 flyby demo, before UT2003 even hits stores. This may be a sneak peek of relative performance in Half-Life 2 and other Source engine based games, but then again, it may not.

http://www.extremetech.com/article2/0,1558,1640456,00.asp

[cusimar9]

Where can I download the test from? It looks cool...

[MJCfromCT]

You have to download the beta of cs:s in order to use the stress test.

[Bigjakkstaffa]

You are right in that Benchmarks such as the VST should not be taken as the performance gospel. Indeed 99% of all Demo's or Benchmarks won't give you a fully accurate performance indicaton. mostly because they are synthetic in nature.

Pointing to Doom III's time-demo's (As with all other time-demo's) however as reliably relaying in-game performance is also a little off the ball, as while it does contain genuien in game sequences, the amont of work goign on behid the scenes is significantly reduced. Essentially time-demo's are just a test of how quickly your graphics card can re-draw a video sequence on screen, creature AI etc. is predetermined where in game it woudl be handled by the system, sound is missing etc etc.

The only really full valid "benchmarks" are those in which a level is played through, in real time, on the various test-beds and an average FPS reading taken via Fraps etc. calculated and even then there will be flaws in the results.

As for VST, it is just that, a STRESS test rather than a benchmark, most of what appears in the VST (from what i've seen at least) will not appear in such density in the average HL2 level, afterall the VST is essentially a timedemo of the E3 2003 techdemo levels from the alpha-build. That said probably all of the whizzo tricks shown in the VST will be encountered in playing HL2 and as such at least it shows how various cards will handle such effects (the water effects in the Alpha - which are fairly commonplace in the harbour area - absoloutley kill my FPS).

As with all benchies, the VST will not give a wholly accurate figure of HL2 on your system, however it will give a decent indication. Afterall, lookit 3dMark 03, that applications validity as a performance measuring tool has been more or less destroyed over the past year, yet oen of the first things we do when we get a new card is go and run a benchmark with it ;)

--Jakk:t