3D Mark 03 out!!!!
- Thread starter Nebuchadnezzar
- Start date
GraphixViolence
Newcomer
I think you're missing the point. This benchmark is using graphical techniques that will be used in upcoming games. The makers have consulted with a wide range of hardware and software developers to ensure that this is so. The benchmark is telling you that if you want to play games that look like that, you are going to need a much more powerful system (especially on the graphics side) than what you have. Note that your system may be scoring a fraction of what you'd get with a 9700 Pro, but at the same time it's also scoring several times higher than what you'd get with a GF4 MX, or (heaven forbid) integrated graphics. And we all know how many people out there have low-end systems like that.Gollum said:
As for when games like that will start to come out, that's all going to depend on how soon people upgrade. If everyone sits there contently with their DX7 and DX8 hardware, then game devs are not going to see any point in making games that look like the 3DMark03 tests. On the other hand, DX9 hardware is rapidly becoming affordable (well under US$200 now), and Futuremark had to create a benchmark that would still be relevant 1-2 years from now
You may be right about current games, but how can you say that about future games? Do you really think the programmers at Futuremark suck compared to most game developers out there? At least this benchmark has hopefully prepared you for the nasty shock you're going to get when the first real DX9 games start popping up. Those of us with DX9 hardware jost hope those of you without start upgrading soonGollum said:
Joe DeFuria
Legend
Heh...
Actually, I think the benchmark might also prepare those WITH CURRENT DX9 HARDWARE.
In other words, in games that really exploit pixel shading in general, even today's DX9 hardware that "supports it", might be too slow for it to be effective in a game.Note that this is nothing really new. IHVs routinely add "features" to their cards for marketability...even if when those features are used to an good extent, performance slows to a crawl.
I mean, the original GeForce has enough "features" to properly run Doom3 with all effects....but it's too damn slow for it to be practical. So by default, it doesn't run with full visual quality.
Doom3 is really a "DX8" level type renderer, and the best DX8 cards will probably just run it "sort of OK" in terms of performance. It looks very much like it's going to take these high powered DX9 cards like R-300 and NV30 to run a game (Doom3) that fully exploits DX8 features with what we'd call good performance.
3DMark is forward looking. It's just more confirmation to me that today's DX8 hardware isn't up to the task (performance) of running a game that fully exploits what DX8 shaders can do.
Nebuchadnezzar
Legend
Crusher said:
I win.
What machine do you have??? A Rage128 ?
Freak'n Big Panda
Regular
I got 43 with a GF2MX classic and a Athlon 900.... So it woouldn't have to be that bad, maybe a GF2MX 100/200? Still bad but not as bad as a Rage128
Missing the point requires there to be one, and I kind of think the DX8 game benchmarks in 3DM03 are pointless at this time. I didn't really want to get involved into a huge argument about this as I respect the Futuremark crew a lot. As I was saying I am aware that the benchmark performs and looks as it does intentionally. I also think it was good move to make it more GFX card dependent instead of CPU limited, commendable! Futuremark wanted a benchmark that could be used for probably at least 2 years, understandable as a program as fully featured and comprehensive as this takes a lot of work and effort. I think however that they overachieved and the final result has become a two-edged sword. Now they're basically telling 90%+ of the gamers that their current rigs aren't good enough to be getting any kind of decent performance from anything but DX7 games. Even those that have one of the more up-to-date DX8 cards are basically screwed yes?GraphixViolence said:
Why even have DX8 tests if they perform miserably on any DX8 card out there? Surely not all DX8 games will perform that horrible, there are and will continue to be loads of new games that won't make use of all the kinds of ultra-performance hungry stencil shadowing and other features these game tests use. There's plenty of room between using all those extremes and still looking and performing great! Most games of the next 12-18 months are most likely to be significantly less hardware-hungry than those tests because they won't use all those features at once, Doom 3 certianly isn't every other game out there. When looking further into the future I don't need to be a psychic to know my current card won't play the games of 2005 very well, and I certainly don't need a 180MB gypsie to tell me what I already know!
Only a short while ago people were demanding the move to DX8 class hardware, all off a sudden even those with DX8 class hardware are supposed to feel like they're holding back the progress of 3D graphics? Is it only me or is there something wrong with this picture?
The same could have been said for DX6, DX7 and DX8 hardware before. There is a continuous slow stream down from the high-end to the low end and it takes a couple years for a certain level of hardware to reach the kind of mainstream penetration that is required to make it a viable platform for games. No benchmarking tool or game is going to dramatically change or speed up the pace of that process, get used to it. By the time true DX9 games are around a R9700Pro isn't going to cut it anymore either, if you think you're buying for that future you're deluding yourself, the past has prooven countless times already that it doesn't work. Actually you only have to look at 3DM03, even R9700Pros don't exactly achieve stellar performance in the DX8 benchies...
So now its us DX8 card-owning folks that's to blame? If it wasn't for that wink I'd take that DX9 card of yours and shove it where the sun don't shine, hehe...You may be right about current games, but how can you say that about future games? Do you really think the programmers at Futuremark suck compared to most game developers out there? At least this benchmark has hopefully prepared you for the nasty shock you're going to get when the first real DX9 games start popping up. Those of us with DX9 hardware jost hope those of you without start upgrading soon
I'm not saying the crew's programmers suck, but I do think they went over the top in trying to put up an impressive looking show. They hopped on the train of people trying to follow Carmack's lead, yet failed to deliver to some degree. In case you're wondering why I'm saying this, yes I did fiddled with the leaked Doom 3 Alpha and yes, it performed FAR better than any of those game tests. Once tweaked a little bit it never even touched single digits at 1024x768x32, most of the time it ran in the 20-30 fps range on the very same system these tests achieve only 3-5 fps. Not great but quite a difference.
I guess I could sum my opinion up by saying that IMHO the DX8 game tests in their current form are at least 6-12 months too early to be considered close to a realistic scenario even for a future-proof benchmark...
Joe DeFuria
Legend
I would just like to say Gollum, that I agree with you to an extent. I think having BOTH DX8 tests being that intensive (with all the stenciling / shadows, etc.) was a bad decision by Future Mark. I do think it is important that they have at least one DX8 test like that though....
GraphixViolence
Newcomer
No offence, but I'm going to suggest once again that you are missing the point of the benchmark. You are seeing a poor frame rate in the two tests that require DX8, and interpreting this as meaning that you will not be able to run existing DX8 games smoothly on your hardware, which of course is false. Compare the graphics in either test to any DX8 games you have right now, and I think you'll agree that the tests look far better. The point is that these tests are making comprehensive use of DX8 shaders on practically everything in the scene, while most DX8 games currently on the market use shaders for just one or two specific effects. What the benchmark is actually telling you is that if you try to run a game that looks like this on your PC, it will run slow. And this is a perfectly valid and useful piece of information for someone deciding on what card to buy today, because games will actually look like this in 1-2 years. Maybe even sooner if Doom 3 arrives on schedule.Gollum said:snipped for brevity
Now they're basically telling 90%+ of the gamers that their current rigs aren't good enough to be getting any kind of decent performance from anything but DX7 games. Even those that have one of the more up-to-date DX8 cards are basically screwed yes?
Why even have DX8 tests if they perform miserably on any DX8 card out there? Surely not all DX8 games will perform that horrible, there are and will continue to be loads of new games that won't make use of all the kinds of ultra-performance hungry stencil shadowing and other features these game tests use. There's plenty of room between using all those extremes and still looking and performing great! Most games of the next 12-18 months are most likely to be significantly less hardware-hungry than those tests because they won't use all those features at once, Doom 3 certianly isn't every other game out there. When looking further into the future I don't need to be a psychic to know my current card won't play the games of 2005 very well, and I certainly don't need a 180MB gypsie to tell me what I already know!
Hey, don't shoot the messenger. I just assumed that since you seem like someone who's interested in keeping their PC up to date, that you had already accepted the fact that everything you buy will be obsolete in about a year. Just like that GeForce4, which came out exactly one year ago.
In case you hadn't noticed, the 9700 Pro is already 6 months old, meaning its life as a cutting edge product is already half over. Maybe even more if the R350 comes out soon. Why would Futuremark release a new benchmark that even a 6 month old product could breeze through at 60 fps? We already have 3DMark 2001 if that's what you're looking for. And if you don't think that a visually stunning and technologically advanced benchmark like 3DMark03 isn't going to encourage a lot of people to upgrade, think again.
Just because Carmack can squeeze out these kinds of frame rates and still create some amazing graphics, I wouldn't count on other game developers to do just as good a job. Remember that Doom 3 includes an optimized GeForce3/4 path, while 3DMark03 is using straight DirectX so that fair comparisons can be made between different architectures. We can hope that upcoming games follow Carmack's lead and include robust fallbacks for older hardware, but I doubt most other developers will put the same level of care and attention into their engines.
You're welcome to your opinion, but I respectfully disagree. This benchmark is giving you a reasonable look at how different products are going to stack up when Doom 3-like games hit the market. In other words, the new DX9 cards are going to signigicantly outperform cards from earlier generations.
Well I found part of my problem, apparently I had 4x FSAA and 2x Anisotropic enabled. I disabled both of those, and set the mipmap detail back to "normal" and ran it again. It certainly didn't look as nice, but my score did go up a little. My new score is... 30. Yep, I don't get it either.