Aquamark 3 in B3D Reviews
- Thread starter Dave Baumann
- Start date
OpenGL guy
Veteran
Most lighting is done through lightmaps with the exception of dynamic lighting.GameCat said:
3dcgi said:
Uhhhh, duh! Forgot about that - its the BSP tree's. Quake3's "rooms" are designed very nicely to cut off in relatively small sections for the bsp tree's to handle. The larger, wide open expanses in the ET scene aI'm using are probably far less BSP friendly.
bloodbob said:
How? We tried, but couldn't get anywhere.
Yeah I was asking if it was possible you could do fraps. I didn't download it being modem and all. Just thought it would be more relivant as opengl benchmark if you do then RWTC with modem cards cause of the higher triangle throughput. I guess its wait till the final is out.
I've been trying out FS2002 for another review but by the time you've lowered all the settings to reduce/remove the CPU dependency, the graphics are so pants that anything above a GF4 copes easily with the workload - shoving the frame rate back to the CPU again.Slides said:
I tried IL2-Sturmovik a while back too but didn't like the fact that (a) the demos ran forever and (b) none of them are particularly GPU intensive. Despite repeated attempts, I just couldn't get a decent demo of my own to include enough frames with a heavy graphics workload.
Reverend said:
Whoa, I strongly disagree here! Not only do we need to test both DX and GL game engines (drivers still matters a lot), but I would also regard it as a big mistake to dump a popular engine as the Unreal2 or the Quake3 (in whatever present incarnation).
If you're hellbend on using Aquamark, I would drop Serious Sam, but I would much prefer to hold out for HL2 instead of including Aquamark.
Beyond that I agree with Rev that in the future HL2 and DOOM3 might be close to all we need, but let's see how things pan out at that time.
www.lo-mac.com
You can try this do for a flight simulation. I think it is fairly GPU intensive. I can't run it, so I don't know.
You can try this do for a flight simulation. I think it is fairly GPU intensive. I can't run it, so I don't know.
At some point Beyond3D has to make up its mind on where its focus is, and for whom it exists. The benchmarks used (as with most sites) do not describe gaming in general, but a very small subset that are not CPU limited and which uses particularly advanced techniques.
In short, the sample is not representative of the population, in fact it is so unrepresentative that it is difficult to find such gaming benchmarks at all! Arguably completely impossible if you are not prepared to accept strong ties to particular IHVs and the bias that should reasonably introduce.
So what's the point? What, exactly, do you want to achieve?
Until that is made clear, benchmarking will only produce numbers without meaning. Mental masturbation for equipment freaks.
Personally, I'd prefer if B3D simply ditched gaming altogether and made clear that the focus is on technology rather than applications. I fail to see that it is even theoretically possible to make a very useful job on game graphics benchmarking. Transferability of results was possible when fillrate was the overwhelming factor, but going forward the waters will seemingly get progressively murkier (which may be just as well, as fillrate is a pretty boring story to tell over and over).
In short, the sample is not representative of the population, in fact it is so unrepresentative that it is difficult to find such gaming benchmarks at all! Arguably completely impossible if you are not prepared to accept strong ties to particular IHVs and the bias that should reasonably introduce.
So what's the point? What, exactly, do you want to achieve?
Until that is made clear, benchmarking will only produce numbers without meaning. Mental masturbation for equipment freaks.
Personally, I'd prefer if B3D simply ditched gaming altogether and made clear that the focus is on technology rather than applications. I fail to see that it is even theoretically possible to make a very useful job on game graphics benchmarking. Transferability of results was possible when fillrate was the overwhelming factor, but going forward the waters will seemingly get progressively murkier (which may be just as well, as fillrate is a pretty boring story to tell over and over).
What he says makes one hell of a lot of sense.Tim Sweeney said:
horvendile
Regular
Slides said:
Damn, that's some nice screenshots!
Maybe all modern flightsims look like this now, but I haven't played any (graphically) modern games for a long time...
Entropy said:
It's not that I can't see the point, but running say ShaderMark2 just won't be enough to meassure GPU-performance right now. We are still playing mainly DX7 (Doom III isn't even out yet!), we have very few DX8 level games and besides Tomb Raider and HL2 a major DX9 title could be a year away. Dull, but true.
Having said that I'm all for beyond3d being more tech than other web review sites by using syntetic benchmarks that meassure to-the-metal shader performance. But this is already done to some extent.
Excellent point!Entropy said:
The following is copied from the Mission:
beyond3D said:
From the above statement and from my impression of the site I would extract the following points:
1) Honest view of 3D architectures:
A truthfull description of various elements of 3D architectures that goes beyond marketing fluff. Pipeline arrangement, shader configuration or simple fillrate numbers spring to mind.
This is mostly achieved through synthetic benchmarks (often written by forum members! ) that test very specific parts of the 3D pipeline.
2) Honest view of 3D technology:
IMO this covers technology like texture filtering and AA techniques, shader technology (compilers, languages, implementation etc.) but also precision issues (in the old days 16bit vs 32bit, now FP16/32 vs FP24).
This could be achieved with the usage of only synthetic benchmarks, but beyond3D seems to like to get real world applications/games in the mix as these things vary from software to software (read: application specific optimization) so a larger sample gives a better understanding of the actual usage of said technologies.
3) Honest view of 3D software:
This is IMO closely tied with 2).
How does the software use the technology?
How does the architecture (implementation of the technology) deal with the software?
Currently most 3D software are games. There is of course the professional 3D market (content creation, CAD, etc.) but we are talking consumer devices here and SPECviewperf seems to be the only widely available benchmark.
As Longhorn is said to require DX9 level 3D hardware this may change in the future, but as of today if you want to evaluate a card's real world capability you have to benchmark games. :?
4) Honest view of 3D hardware:
IMO this is an often overlooked point.
I would like to see more of the following:
Display output quality tests (BNC, DVI, S-Video)
Noise level testing
Different features like downclocking in 2D mode, automatic overclocking, etc. and a honest evaluation of their implementation.
So after breaking it up like this I come to the following conclusion:
You need both synthetic and realworld benchmarks to cover everything that's stated in Beyond3D's mission. Currently I have the feeling that there is a very good balance between the different testing methods, and I get a pretty well-rounded review.
As for Aquamark 3?
I have no idea.
These points are of course based on my opinion and I may be wrong, so feel free to correct me.
What i really miss is diversity. All or most of the benchmarks are FPS and/or synthetic. Aldo there are solid reasons against it, there are enough reasons too to add benchmarks of Flight-sims, Race Sims and other games (simcity 4?). to name a few: image-quality, stability and afcource the performance overall.
for example: if i knew Simcity 4 scrolls better (faster) over the map with a Geforce Fx, it could be my number one choice. OR if i knew GP4 runs better (in any way) on a Radeon then that card could be the winnner for me.
for example: if i knew Simcity 4 scrolls better (faster) over the map with a Geforce Fx, it could be my number one choice. OR if i knew GP4 runs better (in any way) on a Radeon then that card could be the winnner for me.
Cyberon said:
I agree.
I know that this is a CPU review but Ace's hardware did a really good review of the Ahtlon XP 3200+ some time ago where it beat the 3.2 GHz P4 in f.e GP4 and also in 2-3 other games. Then we'll look at f.e Tom's 3200+ vs 3.2 GHz P4 review where he used Quake 3, Commanche 4 and some other games. All in which the P4 is faster.
Tom's conclusion is:
Sure, the P4 3.2 GHz is overall the faster CPU but it's definitely not as much better as what his review indicates and i think that a comparision with Ace's review shows that a really good review (be it CPU or graphics card) needs diversity.
