Nvidia NV3X hardware too slow for HL2 and Doom3?

[Ferocious]

Gabe Newell and John Carmack have a word or two....



click

 

[Corporate Thug]

how many times are we gonna see this???

 

[First]

For HL2, yes, it looks as if it's true from the HL2 benchmarks we've run.

 

[Mingon]

Do you think its a problem with the coding rathan the hardware

*cough* 5200 review *cough*

 

[BenSkywalker]

I wonder if there was some careful editing of Carmack's reply. He has made what is going on with DooM3 and why fairly clear. The optimal path for the NV3X boards to run is their proprietary/native path as running ARB_2 forces everything to run at a higher level of precission then is beneficial for the most part(with a small handful of exceptions). When D3 is running ARB_2, it is running at a higher level of precission then the R3xx core boards and higher then what Carmack had planned on(as he had been requesting FP16 support which the NV3X has, the R3xx is forced to run in FP24 which has some benefit on occasion). Based on his comments, the 5900 is going to be faster then the 9800Pro running the settings that he has been wanting support for for the last couple of years(dating back to the Quake3 days I can recall him stating the need for FP16). There is no meaningful benefit to running FP24 over FP16, but the ATi boards don't support the lower level of precission so they lose out.

The reverse seems to be true for Half-Life2(although Newell hasn't been as explicit as Carmack has). In order to run the shaders in HL2 the 5900 is going to be forced to run in FP32 which has a significant impact on performance. There is no meaningful benefit to running in FP32 over FP24, but the nV board don't support the lower level of precission so they lose out.

 

[Richdog]

I had'nt seen this article yet, it's a good one and he seems a sensible writer instead of a "fan boy" to me. :beer:

 

[McArra]

Originally posted by: BenSkywalker
I wonder if there was some careful editing of Carmack's reply. He has made what is going on with DooM3 and why fairly clear. The optimal path for the NV3X boards to run is their proprietary/native path as running ARB_2 forces everything to run at a higher level of precission then is beneficial for the most part(with a small handful of exceptions). When D3 is running ARB_2, it is running at a higher level of precission then the R3xx core boards and higher then what Carmack had planned on(as he had been requesting FP16 support which the NV3X has, the R3xx is forced to run in FP24 which has some benefit on occasion). Based on his comments, the 5900 is going to be faster then the 9800Pro running the settings that he has been wanting support for for the last couple of years(dating back to the Quake3 days I can recall him stating the need for FP16). There is no meaningful benefit to running FP24 over FP16, but the ATi boards don't support the lower level of precission so they lose out.

The reverse seems to be true for Half-Life2(although Newell hasn't been as explicit as Carmack has). In order to run the shaders in HL2 the 5900 is going to be forced to run in FP32 which has a significant impact on performance. There is no meaningful benefit to running in FP32 over FP24, but the nV board don't support the lower level of precission so they lose out.

 

[Mem]

Nvidia NV3X hardware too slow for HL2 and Doom3?

I`m sure it will run at 800x600 with no AA/AF just fine ,like all things it depends on the CPU,Ram, resolution settings etc you try to run the game in,besides hardcore gamers will be buying the latest and fastest video cards just before the release of these two games.

 

Originally posted by: Evan Lieb
For HL2, yes, it looks as if it's true from the HL2 benchmarks we've run.

Evan, do you think its because HL2 uses Pixel Shader 2.0? I dont know if it does. But if so, could you force the application to use PS 1.4. I am hearing that there is no difference between the two except "how" it is done. If you have the time, and the curiosity to try it, please do. Otherwise, just wait for nvidia's det 5 drivers I guess. That might be more worth the time.

Thanks,

GorillaMan

 

[Johnbear007]

just wait and see how it runs on what you have when it releases.

 

[Mingon]

I wonder how much the $5,000,000 helped ati's case 'oh go on, just patch it later'

 

[nRollo]

Oh no! If it is, the USA will fall off the planet into space!

 

[Lonyo]

Originally posted by: Rollo
Oh no! If it is, the USA will fall off the planet into space!

ATi's based in Canada

 

[nemesismk2]

Originally posted by: Johnbear007
just wait and see how it runs on what you have when it releases.

 

[bjc112]

Originally posted by: Lonyo

Originally posted by: Rollo
Oh no! If it is, the USA will fall off the planet into space!

ATi's based in Canada

 

[Pete]

Originally posted by: Evan Lieb
For HL2, yes, it looks as if it's true from the HL2 benchmarks we've run.

Evan, will you be posting HL2 benchmarks in the next review you publish? Care to name a date?

Originally posted by: Lonyo
ATi's based in Canada

Yeah, but the part of ATi that matters, ArtX, is located next door to nVidia in Santa Clara.

 

[edmundoab]

yeah I remember seeing that article before posted here recently.

 

I dont think there ever was a voodoo3 1000. Voodoo2 1000 yes.
Just admiring your 3dfx collection there.

 

[MrC4]

I dont think there ever was a voodoo3 1000

The Voodoo3 1000 was an OEM card with a core clock speed around 125 if I remember correctly. Since it was clocked so low, a few of these cards were made without even a heatsink!!

 

[SickBeast]

Yeah, but the part of ATi that matters, ArtX, is located next door to nVidia in Santa Clara.

Pfft...ATi was in its full glory long before the existence of ArtX. It's too bad the nvidianites still don't know how to produce a 2D image.

 

[Ruroni]

Originally posted by: Richdog
I had'nt seen this article yet, it's a good one and he seems a sensible writer instead of a "fan boy" to me. :beer:

Really or you're being sarcastic?

I thought otherwise when I saw statements like

"What?s disturbing for us, however, is that typically 5900 Ultra?s are priced incredibly higher than 9800 Pro?s; our last check around the web have the 9800 Pro?s starting at $290.00 while a 5900 Pro will run you closer to $400.00 for quite a bit less"

That doesn't sound very professional (disturbing? ATI's 256MB solution goes for MORE than NVidia's 256MB solution, I think disturbing is a strong word), nor was it extremely accurate, for at the time it was written the cheapest you could get a 9800pro was 297 closer to 300 (although currently, it seems to have gone down a bit; but after this review) and they rounded off, down to 290; where as the 5900 ultra could be found for 391 (closer to 390) and yet for this case they rounded upwards.

And when I read

If you?re still ?stuck in the past? and think that ATI is plagued with driver issues, than go ahead and keep your head stuck in the sand like an ostrich, buy a 5900 Ultra and then start crying when your pals are smoking your ass in games like Half Life 2 and Halo because they?re running ATI hardware.

That too seems a little baised. What's wrong with people feeling more secure with NVidia's drivers, who have been more reputable for longer? Granted, I'm not saying that one shouldn't be open minded about it, but to tell the reader to "go ahead and keep your head stuck in the sand like an ostrich", maybe that's a little insulting and belittling, to say the least; It's a taunt.

And to tell you "start crying when your pals are smoking your ass in games like Half Life 2 and Halo because they?re running ATI hardware." That's REALLY pushing it. I seriously doubt that if someone has a Radeon 9800XT and someone else has a TI4200, the person with the 9800XT has a better chance at winning in a game, because of the graphics card they have. I think that would really depend more on actual gaming skills. Sure hardware makes something of a difference, but not that much, and the performance difference between a 9800 and a 5900 surely isn't going to be enough to have "your pals are smoking your ass in games like Half Life 2 and Halo". I doubt the difference between a 9800 and TI 4200 is going to be enough to make up for actual skills.

So clearly we can see this was an article written by a fanboy. They're clearly swearing by ATI as if they've gone to bed together. I'm not saying that there isn't truth in what they say, but I'd take it all with a grain of salt. I don't know if this review or anything written in it can be taken to heart. I seriously doubt if many of the alledge statements by the alledge source are entirely accurate, if not fictitious (to include the statements from the game developers and the statements and lack there of from the manufacturers themselves.)

I'm not saying it's all lies, I'm simply stating that it would probably be prudent to bear an open mind as we sift through these marketting techniques, reviews, and oppinions.

I personally am inclined towards believing that ATI is in fact a better purchase because it appears (from what I've read) that image quality is better with ATI's cards. But I have to make sure I don't fall prey to the mud flinging and marketing strategies out there.

 

[Richdog]

I see where you're coming from Ruroni his terminology wasn't the best i admit, but he also states that he's a "performance fan boy" as opposed to an "ATI fanboy", something i can identify with. And while his statements were not the most professional by any means, and while some of the language was a little strong and off the wall for my taste he makes some very valid points. I think it's just a case of over-zealousness and passions running high without maintaining a clear head, his heart was in the right place but the brain had trouble putting it into decent wording, lol.

From a "true" performance perspective ATI is the stronger competitor, it needs no special routines or driver-trickery to get insane framerates like current Nvidia cards do, and as previously stated, ATI's Pixel Shader 2.0 performance is far stronger, a factor which will undoubtedly become clear and very relevent when the latest and greatest generation of games begin to be released.

I'm with you, it's important to keep an open mind and think for yourself when it comes to forming opinions on graphics cards, if Matrox brought out a card tommorrow that smoked both ATI and Nvidia in terms of performance (however unlikely that may be) i'd rave on about that. Graphics cards should be praised on their merits irrespective of brand-name, all i'm interested in is which card performs the best while sacrificing nothing in terms of visual quality and features. I currently own a Geforce 3 Ti200 which has served me brilliantly up until now and believe my next upgrade will be an ATI card, most likely a 9700 non-pro as for the price/performance ratio nothing can touch it.

These as always are but my opinions and if anyone has any problems with it feel free to correct me or argue any point (reasonably).

Peace.

 

[1ManArmY]

Why not get a 9800 NP, best bang for the buck?

 

[Richdog]

Pricing structures are FAR different here in the UK. We pay 50% more than you at least for a comparabel piece of hardware, p|sses me off no end...

 

[First]

Originally posted by: Pete

Originally posted by: Evan Lieb
For HL2, yes, it looks as if it's true from the HL2 benchmarks we've run.

Evan, will you be posting HL2 benchmarks in the next review you publish? Care to name a date?

Well, I won't be posting HL2 benchmarks first, but Anand just revealed a few interesting HL2 tidbits an hour ago.

Btw, sorry for the massive video card delay, we've been going through major administrative and editor related changes lately (as you can see by our now daily articles).