View Full Version : 3DMark03 - damage control for the GeForce FX?
creedamd
02-06-03, 06:03 PM
Who wants to bet that the GFFX will kick the 9700 in the new 3dmark! This will be unbelievable if it does! Discuss!!!
Retrofire
02-06-03, 06:25 PM
I'm no engineer, but my thought is that the 128-bit memory bus will be as much a limiting factor in 3DMark2003 as it is in everything else.
The frustrating thing to me is that the GeForce FX is a technology spectacular hampered by comparatively poor memory bandwidth.
I have no doubt that in terms of raw processing power and capability, the GeForce FX will have no equal. At the same time, you can put a Ferrari engine in a Yugo and...Still have a Yugo. :(
ATi definately did an excellent job anticipating where the market was heading and putting a chip of less power out and designing a better method of feeding it.
Unfortunately GFX is trying to suck a golfball through a garden hose...And predictably looks like it will just end up sucking.
I'm still very anxious to see non-jaded real-world user reviews on the GFX though. :D
undercover
02-06-03, 06:29 PM
Lets have a sweepstake :)
I pick the GFFX to beat the 9700 in 3dM2003 by 25% overall at 1280x1024
Rumours are the retail version of the card will be much quiter fan wise, and I bet the drivers improve considerably.
What the chances that nvidia unveil "free" AA with newer drivers on the hardware at some point?! Now that would be impressive. I still think GF-fx's AA quality is fine. The ppl slating it seem to have no understanding of equal comparisons, and seem to be completley blind to me.
sytaylor
02-06-03, 07:22 PM
hmm i say 10% if its any more then theres serious foul play going on :eek2:
ASCI Blue
02-06-03, 07:45 PM
25% might be right 3dmark was designed to make nv cards look good.
undercover
02-06-03, 08:10 PM
Originally posted by ASCI Blue
25% might be right 3dmark was designed to make nv cards look good.
I'm sure Futuremark have no loyalties either way :)
gokickrocks
02-06-03, 08:29 PM
Originally posted by Nutty
Rumours are the retail version of the card will be much quiter fan wise, and I bet the drivers improve considerably.
What the chances that nvidia unveil "free" AA with newer drivers on the hardware at some point?! Now that would be impressive. I still think GF-fx's AA quality is fine. The ppl slating it seem to have no understanding of equal comparisons, and seem to be completley blind to me.
rumours of drivers improving is stating the obvious, all video cards have a chance to improve through drivers because they are never 100% efficient
chances of "free" AA - 0%
i would have to disagree on your last statement, but wont go off explaining why
undercover
02-06-03, 08:41 PM
Originally posted by gokickrocks
rumours of drivers improving is stating the obvious, all video cards have a chance to improve through drivers because they are never 100% efficient
From what JC said in reply to Reverends questions it sounds like the GFFX has a very flexible architecture that requires carefull use by the drivers to make the most of it. Reading between the lines I'd say the GFFX has, by it's very nature, much more room for performance gain via the drivers. I guess the Nvidia engineers will be looking at each game and app and determine where the bottlenecks are and how the drivers can push the hardware harder...
Only time will tell :)
Originally posted by Nutty
Rumours are the retail version of the card will be much quiter fan wise, and I bet the drivers improve considerably.
What the chances that nvidia unveil "free" AA with newer drivers on the hardware at some point?! Now that would be impressive. I still think GF-fx's AA quality is fine. The ppl slating it seem to have no understanding of equal comparisons, and seem to be completley blind to me.
Good to see someone with the same thinking as me. No one has said anything about Nvidias 4xs FSSA, and from what we have see so far, it looks better than 6x2 and 8xs, so thats obvious there are driver problems there.
I think the GF-FX will do well, but lets not forget, 3dmark will push todays cards, remember 3dmark2001 and the GF3/8500 and now look at it with the 9700/GF-FX. Still I cant wait to see what Im missing out on by not having a 9700 or GF-FX :(
Spliffstarr
02-07-03, 12:44 AM
either ways considering that it will be out in the begining of next week we will not have to wait very much longer to see how the two chips will perform..
what I've read was, that GF FX is actually significant SLOWER in 3dMark 2003 than radeon 9700pro...
... according to a german games-Magazin (GameStar)
SamuraiCatJB
02-07-03, 10:30 AM
I say, by product launch that nVidia will tweak the drivers to bench just above the 9700 Pro... I would say by 5%, but not more than 10%.... However, one has to remember that the R350 is scheduled for release shortly thereafter which will again top the FX (I expect).
I am actually hoping for a bandwidth/speed war. I think it is great that nVidia has competition with ATI and vice/versa. We can expect a slew of cards trying to top each other every 6 months (though some news reports say this may be shortened to 4 for a while -- that could prove interesting). :)
The_KELRaTH
02-07-03, 11:35 AM
Heh.. if you believe anything from the Inquirer there won't be an NV30 at this rate. 1st they said only 100,000 would be produced and now it's dropped to 10,000!!
By launch day the only sales could be just the review cards:D
Inquirer 7th Feb (http://inquirerinside.com/?article=7658)
digitalwanderer
02-07-03, 12:27 PM
I don't think ATI is gonna give up to nVidia on this one without a fight, their latest drivers that will be available soon (cat3.1, probably released the same day as 3dm2k3. ;) ) have some increased D3D performance...and it wouldn't shock me at all if'n it happened to run better on 3dm2k3 than the last set.
ATI ain't gonna give nVidia a damned inch, they're gonna make nVidia earn each and every win from here on out! :p
It's a different ballgame today, and I betcha the futuremark crew knows it too. ;)
Originally posted by creedamd
Who wants to bet that the GFFX will kick the 9700 in the new 3dmark! This will be unbelievable if it does! Discuss!!!
I doubt it. If anything it'll beat a R9700Pro stock speed by a whole 100 to 1000 points :rolleyes: And a overclocked R9700Pro... wooo baby watch out
:firedevil
ricercar
02-07-03, 02:25 PM
Originally posted by undercover
I'm sure Futuremark have no loyalties either way :)
FutureMark accepts "donations" from companies and listens to their "recommendations" for future benchmarking tests. Unfortunately, accepting such donations likely creates less indifference than say Consumer Reports, who refuse to accept any money from any company.
A demo developer I work with claimed that ATI made such a donation to ensure that the technologies in Radeon 9700Pro would be important to 3DMark 2003 benchmarks. While this doesn't mean the GeForce FX will perform poorly, I'll be looking to see whether the difference between GeForce FX and R9700Pro was larger in 3DMark 2003 than in 3DMark2001.
Hellbinder
02-08-03, 12:51 AM
Who wants to bet that the GFFX will kick the 9700 in the new 3dmark! This will be unbelievable if it does! Discuss!!!
Well being that the GFFX ultra just got canned.. good luck on this one. Although there will no doubt be a few peeps out there with the board that can try to overclock the hell out of it. But well see...
To bad the R350 is about to launch and whoup some serious GFFX non ultra ass... but that sthe wat these things go some times.. You guys can get some nice revenge 3dmarking this summer with the Nv35.
Originally posted by Hellbinder
Well being that the GFFX ultra just got canned.. good luck on this one. Although there will no doubt be a few peeps out there with the board that can try to overclock the hell out of it. But well see...
To bad the R350 is about to launch and whoup some serious GFFX non ultra ass... but that sthe wat these things go some times.. You guys can get some nice revenge 3dmarking this summer with the Nv35.
You really thing the GF-FX Ultra is canned? As far as I know, the chips are already being produced, so if they are not going to release the Ultra I guess those chips will go to the non Ultra. Now if Nvidia doesnt have a high end card, I'll bet that they will be bringing the NV35 out sooner ;)
BTW the R350 hasnt been announced yet, so we dont really know how far away it is!
creedamd
02-08-03, 11:05 AM
why wouldn't they can it? I would, I have seen a total of probably 25 people say that they would buy the gffx and half of probably don't have the money to go to the movies. Smart move by nvidia, let the graphics world ride for a while and concentrate on mobos.
sebazve
02-08-03, 06:18 PM
i doubt the nv30 will be faster than r300. Hhahahaha turn on AF and AA and it will smoke the nv30:afro:
Originally posted by creedamd
Who wants to bet that the GFFX will kick the 9700 in the new 3dmark! This will be unbelievable if it does! Discuss!!!
You may be right. It all depends on what percent of the score they decide to give to each test. Remember the big boost to 3DMark scores because of the score assigned to the pixel shader test? All they have to do is assign a larger percentage of the final score to the fillrate test and the GFFX will win.
Originally posted by Smokey
You really thing the GF-FX Ultra is canned? As far as I know, the chips are already being produced, so if they are not going to release the Ultra I guess those chips will go to the non Ultra. Now if Nvidia doesnt have a high end card, I'll bet that they will be bringing the NV35 out sooner ;)
BTW the R350 hasnt been announced yet, so we dont really know how far away it is! And now you have a whole slew of benchmarks to see what you're scores will be if you can clock 100Mhz over stock on the core and memory of a normal FX.
Not directed at you, but at tons of posters. Why do people assume Nvidia will be able to move up the launch of a core by 4 months without a problem? If it was that easy to do, more companies would just simply bring their stuff out sooner. Think about it logically, they started development on this thing likely with a 128-bit memory controller(looking at Nv's comments on the 256-bit memory over the past 6 months make me tend to think that NV30's successor was likely still a 128-bit like Nv30) and it's target release was Fall(likely Oct going by previous Nv fall releases). Now at some point they realized it needed a 256-bit interface to compete with R300 and went and changed it(from all rumors it seems likely to sport a 256-bit bus). Now changing it mid stream and still releasing it on time would take quite a bit of any "lead time" they had (ie time waiting to release new core while the old one is taking in the high end profits). So adding the month or two of chaging the memory controller and adding it to the 4 month moveup of the launch would point me to believe that NV theorhetically could release it 6 months before they planned to. Doesn't seem likely to me as they've seen what R300 brought to the table for 6 months. Another thing, Nv30 and Nv35 have been in the making for a long time...they had certain target performance planned(forget about the 9700 for a moment). Going by past NVx5 cores, they improved performance by 20% -30%. Why do people think that they had plans for more than that with NV35? The high end of their line usually is not too far off from its clocking limit so it would remain to be seen how much higher they'd be able to clock the GFFX equivilent of the4600. Plus they're trying to do all this at TWICE the pace(ie half the time)between generations. In other words, I'll believe it when I see it, and a lot of you are setting yourselves up for the same dissapointment as the FX left.
SamuraiCatJB
02-08-03, 07:30 PM
Originally posted by ChrisW
You may be right. It all depends on what percent of the score they decide to give to each test. Remember the big boost to 3DMark scores because of the score assigned to the pixel shader test? All they have to do is assign a larger percentage of the final score to the fillrate test and the GFFX will win.
or tests that use specific math in the shaders that is known to work better on the FX.... there are lots of ways he could leverage the data towards the FX... I hope he doesn't though.
Originally posted by SamuraiCatJB
or tests that use specific math in the shaders that is known to work better on the FX.... there are lots of ways he could leverage the data towards the FX... I hope he doesn't though.
True. They could use Cg compiled shaders that use instructions not supported by the 9700. Then the 9700 would have to make these calculations via the cpu which would also slow it down.
vBulletin® v3.7.1, Copyright ©2000-2012, Jelsoft Enterprises Ltd.