Doom 3 Benchmarks at [H]

[gururu]

Originally posted by: DAPUNISHER

Originally posted by: Malladine
My opinion of HardOCP has risen greatly...awesome article.

Personally, I feel THG and [ H ] have both always been solid overall, they just have a tendency to interject there own subjectiveness into their work. It's one of the things I really miss about Anand being far more hands on around here. He is like a Vulcan when it comes to evaluating things, and I always went away from reading his work with a sense of "Now I know, and without anyone trying to influence me one way or the other"

I like all three sites. Hardocp is probably my favorite only because I've found their new attitude on benching cards refreshing. I wonder how they got the 'IN' with this article and Carmack. It is certainly a resounding endorsement.

 

[DAPUNISHER]

I wonder how they got the 'IN' with this article and Carmack. It is certainly a resounding endorsement.

People can say what they want about Kyle, but if you are looking to sell ice cubes to Eskimos he's the guy to call. He is the P.T. Barnum of the harware scene

 

[Evdawg]

my pants are rising from the dead.. me oh so excited

punisher was the one who kinda convinced me to buy a 6800gt... punisher u are oh so smart

 

[reever]

Originally posted by: PowderBB3D
ATI != OpenGL.

Looks playable to me.....

 

[DAPUNISHER]

Originally posted by: Evdawg
my pants are rising from the dead.. me oh so excited

punisher was the one who kinda convinced me to buy a 6800gt... punisher u are oh so smart

You had already made up your mind, I just provided the external validation you were looking for

 

[biostud]

Originally posted by: LTC8K6
okay, phray showed me where I missed the line about the newer ATI drivers. Interesting that even drivers 2 versions newer than the cat4.7's were not able to close the gap with NV's OGL performance.

Now, what happened to the frame rate cap? Wasn't it Carmack who said it would be capped at 60fps or was I dreaming that?

Maybe it's disabled in the timedemo?

 

[aldamon]

Originally posted by: gururu
I wonder how they got the 'IN' with this article and Carmack. It is certainly a resounding endorsement.

Both id and HardOCP are located in Texas. That might have helped. Who knows?

 

[LTC8K6]

Maybe it's disabled in the timedemo?

Well, that would make sense. I wish they had talked about it though.

 

[Childs]

Originally posted by: LTC8K6

Maybe it's disabled in the timedemo?

Well, that would make sense. I wish they had talked about it though.

I thought ingame the framerate was going to be capped at 60FPS, but for benchmarking it could be uncapped. They probably scrapped the whole cap after the hardcore "professional" gamers complained.

 

[Pete]

Comparing these new benchmarks to the previous ones, it looks like iD chose a massively more complex demo to test: the 5900U back then scored twice as high as the 5950U now in 10x7 0x0 Medium Q (in a faster system with newer drivers). I'd be surprised if performance will be even lower in single player game with more AI, as Kyle said.

As to the "60fps cap," again, iD wanted to cap the physics engine to 60Hz to prevent the physics sweet spots in Quake 3 (where certain framerates allowed you to jump higher and farther, due to rounding errors). The video card will be able to render higher than 60fps, but it just won't show any more info than at 60fps (if the physics engine is capped, then onscreen action isn't updated faster than the cap). iD may have dumped the engine cap, though that's a separate issue from the framerates discussed in HOCP's article.

 

[Regs]

Resident Evil and Parasite Eve were the last of the remaining few great single player games. And I played those games more than once or twice before I tossed them.

 

[ai42]

I wonder how they got the 'IN' with this article and Carmack. It is certainly a resounding endorsement.

Remember Kyle Bennett is out of Plano, TX (which is where I am) and John Carmack is out of Garland just a few miles down the road. Also consider that Kyle has been doing his harware workshop at QuakeCon every year (and formerly CPL) for as long as I can remember. So I'm sure they've met (I've met then both at QuakeCon on a couple occasions) so it isn't that far of a strech for Kyle to be real nice to John and work out some sort of deal to see the game early (which from his previous articles I think Kyle has visited Id quite a few times).

 

[WicKeD]

O'DOYLE RULES!!!!!!

 

[VirtualLarry]

Originally posted by: DAPUNISHER

One other thing, independent of the "cheats", is that I saw mentioned that the D3 "demo1" benchmark, doesn't run the game AI/physics, only the 3D engine (and sounds?). So actual in-game framerates will necessarily be *slower* than those shown. Additionally, since the demo appears to be CPU-bound at 1024x768 no AA/AF levels, on the 6800/X800 cards, then it appears that D3 may require quite a bit of CPU power, perhaps moreso than graphics-card power. (Since the 3D engine alone is already sucking up 100% of the CPU power of the systems that they tested those cards on - once you start playing the actual game, frame rates are bound to drop lower. How much lower remains to be seen.)

You are forgetting what both you and I quoted from that article, Larry

we were playing DOOM 3 on a 1.5GHz Pentium 4 box with a GeForce 4 MX440 video card and having a surprisingly good gaming experience.

If a Willy can handle it the average 2ghz+ XP and P4 NW system will only make it better

Well, I think that you missed the point of the first comment that I made that you quoted above. I wasn't trying to suggest (at least, not intentionally) that one wouldn't be able to obtain playable framerates; rather, that the framerates shown by the benchmark numbers, should not be the ones that actual players expect. Ie., if you want to have a steady 30fps in-game, then you will need *more* capable hardware, than the benchmark shows for 30fps. What with all of the comments in this thread about how well people's existing hardware seemingly did on the benchmarks, I thought that might be valuable to point out.

I have a sneaky feeling, that for those with slower CPUs, the game might actually run the physics engine with a much slower "tick rate", and just interpolate the results. Same deal with the graphics engine, I would be willing to bet that lesser cards won't be able to see the realtime dynamic lighting, projected shadows, etc. So while you may be able to get a "surprisingly good gaming experience", that doesn't mean that the gaming experience will be "as good" as on recommended hardware. (Well, that's kind of a given, I suppose.) My take is that the gaming experience on older hardware will be "surprising", because it's not complete crap. Which in the end, is definately a good thing.

Also, I'm pretty sure that the supposed frame-limiter was disabled in the benchmark mode, I could have sworn that I've read that somewhere already. It makes sense, since if the limiter is tied to the physics engine, and the physics/AI aren't run during the benchmark, then that means that it doesn't have to be limited likewise either.

 

[SickBeast]

Originally posted by: Megatomic
HardOCP didn't even do the tests, id did the tests and shared the results with them. I hope a broad selection of platforms is used for official reviews.

That's right, if HardOCP had performed the tests, the Radeons would have come up on top.

 

[spaceman]

where are the budget benches?
my ti42008x and my gf ti200 128GS want to know

 

[Cat]

Originally posted by: VirtualLarry

Well, I think that you missed the point of the first comment that I made that you quoted above. I wasn't trying to suggest (at least, not intentionally) that one wouldn't be able to obtain playable framerates; rather, that the framerates shown by the benchmark numbers, should not be the ones that actual players expect. Ie., if you want to have a steady 30fps in-game, then you will need *more* capable hardware, than the benchmark shows for 30fps. What with all of the comments in this thread about how well people's existing hardware seemingly did on the benchmarks, I thought that might be valuable to point out.

I have a sneaky feeling, that for those with slower CPUs, the game might actually run the physics engine with a much slower "tick rate", and just interpolate the results. Same deal with the graphics engine, I would be willing to bet that lesser cards won't be able to see the realtime dynamic lighting, projected shadows, etc. So while you may be able to get a "surprisingly good gaming experience", that doesn't mean that the gaming experience will be "as good" as on recommended hardware. (Well, that's kind of a given, I suppose.) My take is that the gaming experience on older hardware will be "surprising", because it's not complete crap. Which in the end, is definately a good thing.

Also, I'm pretty sure that the supposed frame-limiter was disabled in the benchmark mode, I could have sworn that I've read that somewhere already. It makes sense, since if the limiter is tied to the physics engine, and the physics/AI aren't run during the benchmark, then that means that it doesn't have to be limited likewise either.

1. The whole point of Doom's engine is produce nearly identical results across all the rendering paths. The dynamic lighting, and stencil shadows will be used in all 4 paths.

2. The physics sim runs at 60Hz, no more, no less, regardless of CPU. This has been explicitly stated.

3. The timedemo is described in the article as being an unusually intensive example of the game.

 

[Killrose]

Interesting the info made about overclocking and having possible problems. does this mean that cards like the X800XTpe and 6800Ultra may actually have to be declocked to run the game at full detail?

I mean that generally nvidia's Ultra series cards are nothing more than overclocked versions of an exsisting core. Could be interesting.

 

[Gamingphreek]

Originally posted by: GeneralGrievous

Didn't you forget to mention that the 5950 performed roughly the same as the 9800 XT?

Wasn't Doom 3 supposed to be

A: The focal point of the NV3x architecture
B. The 5xxx's saving grace? Given a lack of HL2 performance anyway.

Personally I'm about ready to puke over the non stop multi player focus in so many games. I for one will immensley happy if there is NO multi player in Doom 3, and that they spent ALL of the time making a great single player game that actually has a good STORY.

My perspective is that you can play through a single player game once or twice, maybe 3 times at most, then you toss it. You can play multiplayer games for much longer. We have Unreal Tournament for that but Doom 3 could have provided a different feel. You can only have so much fun abusing computer AI again and again.

Well the point that you are referring to when the 9800XT is tied or winning by a bit is when the game is CPU bound. Therefore it is expected for one to lead the other within a couple percentage points because it is within the margin of error. However when everything becomes GPU bound you see the 5950U pull away.

Also i fealt that this was one of the better reviews from |H|. Usually they have a very apparent ATI bias IMO. However it was not apparent at all and seemed to be very even with a bit towards Nvidia but everyone has a little bias whether you think so or not, no one is completely objective (one thing i learned in Journalism class last year )

Also ATI does too optimize for games. How in the hell do you think they do so well in HL2. Both have optimizations however it does extremely poorly on NV3x architecture as ATI and Valve are working hand in hand. That does not mean that Nvidia hardware is horrible at HL 2 it just means that it isn't performing as good as ATI. THe final release will tell us much more than can be put into words right now.

-Kevin
-Kevin

 

[LTC8K6]

Now we just need to know what the minimum framerates are.

Who slows down the most? :beer:

 

[KoaTech]

9800pro looks to be ok, as long as we throw in a nice 3ghz 800fsb with dual channel gig ram and raptor hard drives!

 

[GeneralGrievous]

Well the point that you are referring to when the 9800XT is tied or winning by a bit is when the game is CPU bound.

How is the game CPU bound when the NV40/R420 is getting ~70 fps at the same settings on the same rig?

 

[Gamingphreek]

Take a guess why else would both cards be performiung almost exactly the same. Ummmmmmm.... CPU bound buddy. Of course there will be small differences as the environment can change, therefore it is within the margin of error.

-Kevin

 

[CrazyHelloDeli]

Im so glad I didnt settle for an X800Pro because I just couldnt wait for the 6800GT. ATI Really owes the community a revamped Openl GL driver.

 

[ss284]

Originally posted by: GeneralGrievous

Hey General, if the guy who bought your XT for 700 bucks is a Doom Fan, he should be commiting Harey Carey right about now. (not sure how to spell harey carey? LOL)

He might be. And if he is? I guess he'll have to live with only having the 3rd best card to run this game as opposed to the first. I personally wouldn't pay $55 for what is looking more and more like a solely single player game. By the time all the modders get done with making this a truly viable multiplayer game, it'll probably drop in price a bit.

btw, what happened to your bold statement that the 5900s would crush the 9800s in this game?

Deus Ex was single player, and it was one of the best fps games of all time. HL1 multiplayer was pretty bad as well; it was the mods that made it fun to play online. ID is making it really easy to mod the game, so I doubt itll be long before a good mod come out.

In the end, if the 20-25 hours of gameplay are incredible, its totally worth the 50 dollars for the game.

-Steve