This review obviously has some silly mistakes in it. Is it going to get fixed? It looks like a bit of a joke right now. Gives a bad impression of your site.
Will the 5900ULTRA review ever be fixed?
- Thread starter Ady88
- Start date
GTaudiophile
Lifer
- Oct 24, 2000
- 29,767
- 33
- 81
WaltC over at Rage3D had this to say in response to this article from theinq:
Sites like the Inquirer simply don't understand a fundamental fact:
ATi never got an opportunity to look at, much less tune up an OpenGL driver, for the nVidia-paid-for Doom III demo which was recently released to a tiny number of sites under very tight security.
This demo was originally 100% assembled by nVidia itself for the sole purpose of promoting its $500 nv35 reference design due out in 2-3 months...(which is one reason ATi never even knew it existed)... While Carmack states he didn't "like" the original nVidia demo, and changed it--Carmack judiciously stops short of revealing whether nv35 products performed better or worse after his changes. (Which I found pretty amusing.) He says something inane about nVidia having "guts"--but does not state what kinds of "changes" he made.
I also think the "high-security" comments surrounding the made-by-and-for nVidia DIII demo are pretty funny as well. Is the worry about security legitimate, or is it more of a worry the Demo would get out and people would see what a kludge it was?
I mean, here you've got a Doom III demo which *nobody* can use, ever, as an actual demo of Doom III (no one's allowed to download it), and I wonder why released screen shots wouldn't have suited ID's Doom-III promotional needs just as well.
But--oh--I forgot. The purpose of this demo was not to promote Doom III, but to promote nv35...! So that's why ATi was never informed of it, and why you can't download it to run it yourself and make your own judgements. I do find it odd that Carmack thinks the demo is good enough to bench nv35 on but not good enough for public distribution. Very odd if not revealing situation.
Sites like the Inquirer simply don't understand a fundamental fact:
ATi never got an opportunity to look at, much less tune up an OpenGL driver, for the nVidia-paid-for Doom III demo which was recently released to a tiny number of sites under very tight security.
This demo was originally 100% assembled by nVidia itself for the sole purpose of promoting its $500 nv35 reference design due out in 2-3 months...(which is one reason ATi never even knew it existed)... While Carmack states he didn't "like" the original nVidia demo, and changed it--Carmack judiciously stops short of revealing whether nv35 products performed better or worse after his changes. (Which I found pretty amusing.) He says something inane about nVidia having "guts"--but does not state what kinds of "changes" he made.
I also think the "high-security" comments surrounding the made-by-and-for nVidia DIII demo are pretty funny as well. Is the worry about security legitimate, or is it more of a worry the Demo would get out and people would see what a kludge it was?
I mean, here you've got a Doom III demo which *nobody* can use, ever, as an actual demo of Doom III (no one's allowed to download it), and I wonder why released screen shots wouldn't have suited ID's Doom-III promotional needs just as well.
But--oh--I forgot. The purpose of this demo was not to promote Doom III, but to promote nv35...! So that's why ATi was never informed of it, and why you can't download it to run it yourself and make your own judgements. I do find it odd that Carmack thinks the demo is good enough to bench nv35 on but not good enough for public distribution. Very odd if not revealing situation.
My problems with the review has nothing to do with the Doom3 demo. What I'm mainly reffering to mainly is the problems surrounding the Splinter Cell portion of the 'review'. I thought it was a well known fact that Splinter Cell does not yet support FSAA, as Ubisoft has stated this before. They plan to release a patch that will fix this problem but are yet to do so. It is only pointed out in this Review that the ATi cards have a problem running SC because of graphical errors, yet this problem exsists with both nv and ati cards. It is also obvious that from the benchmark results that the nv cards are running no AA at all while the Ati cards are.
Ady
I agree with your conclusion that "the anandtech review was not done properly."
Seemed very half baked and I was dissapointed in both the lack of games benched and lack of drivers tested.
The SC scores were about 30-40% lower than my 9700 scores with the cat 3.4s-I don't know what they did wrong but it is certainly conspicious.
Rogo
I agree with your conclusion that "the anandtech review was not done properly."
Seemed very half baked and I was dissapointed in both the lack of games benched and lack of drivers tested.
The SC scores were about 30-40% lower than my 9700 scores with the cat 3.4s-I don't know what they did wrong but it is certainly conspicious.
Rogo
And then you have the Quake3 scores for the 5900 staying at 223fps (link?) at all 3 resolutions while the 5800U and 9800pro (and all other cards) frames rates go down drastically at the higher resolutions. Nothing short of ridiculous. Anand should question stuff like this because it surely can?t be right.
It always possible that the 5900ULTRA is still CPU limited, while the other cards are not. It's definitely something you would check out though and make sure you had everything right.
The rage3d argument is just plain Fanboi, who was it that originally leaked the beta doom III ? To try to excuse the failure of it to run because of opengl drivers is a fault of ATI - their opengl drivers have always been poor
Mem
Lifer
- Apr 23, 2000
- 21,476
- 13
- 81
LOL..you have to be pretty dumb to go to a VERY PRO-ATI site for a review or comment on Nvidia cards.
I don`t want to start a flame war but go to neutral sites for reviews,it`s common sense
.Getting back on topic,you won`t see proper reviews until retail models are reviewed on websites ie ,Leadtek,Gainward,MSI,Asus etc so you`ll have to wait until late June or later for any retail FX 5900 reviews.
seriously though, do any of the reviews so far make sense? I mean, the doom3 bench has to be taken with a grain of salt. then there's the unresolved 3dmark and nvidia driver issue. not to mention that every site had conflicting results with splinter cell. and also the point you mention with SC FSAA.
the other thing that annoyed me was that sites were given the 'privelege' of running a formulated D3 demo with their 5900U reviews. its like the sites felt honored to be handed the grace of Mr. Carmack. makes me wonder if they felt they owed nvidia something. too silly. and its not just Anands review.
where's aquamark when we need it so bad.
the other thing that annoyed me was that sites were given the 'privelege' of running a formulated D3 demo with their 5900U reviews. its like the sites felt honored to be handed the grace of Mr. Carmack. makes me wonder if they felt they owed nvidia something. too silly. and its not just Anands review.
where's aquamark when we need it so bad.
Mem
Lifer
- Apr 23, 2000
- 21,476
- 13
- 81
I think the Doom 3 benchmarks was just to give us a rough idea on how it will perform,I`m sure when the retail FX5900 reviews are out we will see better drivers etc,remember the drivers will be updated and improved down the road.
Bottom line is we have only seen a reference FX5900 previewed with drivers that can be improved one way or another.
Bottom line is we have only seen a reference FX5900 previewed with drivers that can be improved one way or another.
If you talk to anyone in the Nvidia PR department they are very confident that the 5900 Ultra will be a huge success. The 5900 Ultra with the DetonatorFX 50 drivers is said to deliver amazing performance. ATI had better get tweaking their drivers or the R9800 Pro is dead in the water, oh I forget ATI only make unstable drivers which don't work fully.
Originally posted by: nemesismk2
If you talk to anyone in the Nvidia PR department they are very confident that the 5900 Ultra will be a huge success. The 5900 Ultra with the DetonatorFX 50 drivers is said to deliver amazing performance. ATI had better get tweaking their drivers or the R9800 Pro is dead in the water, oh I forget ATI only make unstable drivers which don't work fully.
Thought They (NV PR dept) had the same if not better opinion of 5800 (aka flow FX).
Wasn't it hailed as the 2nd coming of new dawn of video cards?
Id hardly think that someone from a site dedicated to worshipping a single manufacturer can be considered unbiased. No, this is not a flame. Its common sense
I forget where business turned fair.
Why should iD give ATi a chance to optimize for that demo?
Business is fair? Heh. I didn't know that.
Just like 2003, just cuz nVidia does it and beats ATi (who is on the futuremark board) suddenly things have to be fair?
Why should iD give ATi a chance to optimize for that demo?
Business is fair? Heh. I didn't know that.
Just like 2003, just cuz nVidia does it and beats ATi (who is on the futuremark board) suddenly things have to be fair?
Originally posted by: blahblah
Originally posted by: nemesismk2
If you talk to anyone in the Nvidia PR department they are very confident that the 5900 Ultra will be a huge success. The 5900 Ultra with the DetonatorFX 50 drivers is said to deliver amazing performance. ATI had better get tweaking their drivers or the R9800 Pro is dead in the water, oh I forget ATI only make unstable drivers which don't work fully.
Thought They (NV PR dept) had the same if not better opinion of 5800 (aka flow FX).
Wasn't it hailed as the 2nd coming of new dawn of video cards?
I didn't get the impression that the nvidia pr department was especially impressed with the FX5800, they are alot more confident in the FX5900 and hopefully their confidence is justified.
I personally can't wait to see what XFX have planned for the FX5900 range, I've used a few (and sold a few hundred FX5200 when most retailers didn't even have ANY!) of their products and I've been impressed with the build quality and general reliability of their products.
being on the futruemark board doesnt help that much when nvidia cheats in their drivers and benchmarking now does it?
there has been a thread on Nvidia's cheating at the benchies so im not really sure if i want to beleive anybodys reviews other than the ones over at extreme tech cause they use their own benchmarking system and not the same exact games that nvidia has optimized/cheated on
You don't remember the "biggest contribution to 3d graphics in 10 years" quote? They hyped the heck out of the card. Right up until the nv35 was released. At that point they changed their tune and said the nv30 was a failure and it's shortcoming were fixed in the nv35.
That is not what this topic is about though. This topic is about having glaring errors in your review and having them still there a week later even though the errors were pointed out on the first day the review was published. It's a pretty well known fact that Splinter Cell has problems with any card using Multisample anti-aliasing. That includes both the r300/r350 and the nv30/nv35. Yet Anand dedicated a whole section of his review to point out that r300 has driver problems because it has graphics anomolies in Splinter cell. These anomolies are a known issue with Splinter Cell because of Anti-aliasing and would be present in the nv35/nv30 if he actually had AA working on those cards with Splinter Cell. AA was not on for the FX cards in the Splinter Cell AA benchmarks. The scores and the lack of graphics glitches both show that. This has been brought to his attention since the first day the review was posted. A week later and he still has erroneous information in his review.
As far as the quake 3 scores, Anand's is the ONLY review that I have seen where the NV35 has the same scores across all resolutions.
Back to Splinter Cell. Originally Ubisoft said they would fix this in a patch. They have since confirmed this will not be fixed in splinter cell.
The Doom3 and Splinter Cell benches both need fixing. It seems that AA/AF were not enabled in those games on the 5900U and were on the 9800P.
Beyond3D
Doom3 bench thread
Splinter Cell bench thread
Upadated "no cheat 3DMark 330 scores"
Poor 5900U IQ
Seeing the massive FPS hit the 5900U takes in UT with AA/AF in the [ H ] review, but none in Doom3 (FPS actually went up), there is something very wrong.
We need to have some better testing done @ review sites. Most I've seen lately are just poor. Benches have to be run with settings that give as close to the same actual IQ as possible. This does not mean equal IQ settings. A card benching higher with poor IQ while the other card has great IQ is not good testing, never mind with cheats enabled.
Beyond3D
Doom3 bench thread
Splinter Cell bench thread
Upadated "no cheat 3DMark 330 scores"
Poor 5900U IQ
Seeing the massive FPS hit the 5900U takes in UT with AA/AF in the [ H ] review, but none in Doom3 (FPS actually went up), there is something very wrong.
We need to have some better testing done @ review sites. Most I've seen lately are just poor. Benches have to be run with settings that give as close to the same actual IQ as possible. This does not mean equal IQ settings. A card benching higher with poor IQ while the other card has great IQ is not good testing, never mind with cheats enabled.
kylebisme
Diamond Member
- Mar 25, 2000
- 9,396
- 0
- 0
Originally posted by: Mem
LOL..you have to be pretty dumb to go to a VERY PRO-ATI site for a review or comment on Nvidia cards.
I don`t want to start a flame war but go to neutral sites for reviews,it`s common sense
lol guys, nice use of ad hominem.
well it would have been better of WaltC was actually affiliated with rage3d and not just making a post on their forums, but all the same it was a telling display of character the way you attacked the credibility of his argument without any regard to the argument itself.
Originally posted by: TheSnowman
Originally posted by: Mem
LOL..you have to be pretty dumb to go to a VERY PRO-ATI site for a review or comment on Nvidia cards.
I don`t want to start a flame war but go to neutral sites for reviews,it`s common sense
lol guys, nice use of ad hominem.
well it would have been better of WaltC was actually affiliated with rage3d and not just making a post on their forums, but all the same it was a telling display of character the way you attacked the credibility of his argument without any regard to the argument itself.
pwned
i was about to post that i didnt think eh was actually affiliated with rage3d, but i wasnt sure lol
Originally posted by: TheSnowman
lol guys, nice use of ad hominem.
well it would have been better of WaltC was actually affiliated with rage3d and not just making a post on their forums, but all the same it was a telling display of character the way you attacked the credibility of his argument without any regard to the argument itself.
Okay, I'll bite.
Ah, yeah... this whole Doom3 thing was news to the ATI guys...except for the fact that ATi sponsored the Doom3 theater for the last year, running the game exclusively on ATi hardware. ATi has also used Doom3 extensively for launch events. Yes, clearly a conspiracy.
If by "assembled" you mean that Nvidia performed the sneaky underhanded tactic of recording a timedemo that was never used, you're right. News flash: this is a game that is under development. Nvidia arranged to have their new part tested against a random build from the current codebase, and Nvidia paid for logistics-related costs in getting that done. The horror.
Nvidia clearly arranged (and thus had prior knowledge of) the test as well as had access to the code drop prior to showing up at id (as evidenced by their bringing a pre-recorded timedemo with them). But ATi has had access to Doom3 code for quite some time as well, although the test itself may have been a surprise (as evidenced by the bug preventing the use of Catalyst 3.4 on those tests - something they'd have never allowed to get out had they realized they were going to be tested in such an embarrasingly public manner).
But frankly I'm interested in seeing what happens when pre-release game code containing valid 3D calls is secretly tested against released drivers (even if this one unfairly blindsided ATi and not Nvidia); it helps one gauge the general maturity of those drivers, an area which ATi has made tremendous strides in but arguably has a bit farther to go yet.
For those who apparently pay no attention whatsoever to the gaming industry, Doom3 was leaked to the Internet late last year. Tighter security is in place to try to prevent such a thing from happening again. As a side note, it appears that the poster doesn't understand that "demo" in the reviews refers to a timedemo of a developer playing through parts of the game, not a "demo" as in the game itself (the game engine has been finished, or practically so, for quite some time while the company is busy working on content). Early levels released to the public as a teaser for the full game (if it happens) will occur when the game is released for sale, as with previous id titles. The tests were conducted by running a demo (meaning timedemo) on a current development build. Nothing more.
More words are strung together that give the impression of an argument, without actually containing logic. High-order cluelessness pervades.
Summary:
This was a performance test cum PR stunt by NVidia. Its long-term meaning is significant only in showing that NVidia will have a part that performs well. It's probably not a good indication of final ATi performance (especially on the 256-MB 9800 part), but not because, as implied, ATi has "never seen the code". Conspiracy theorists are wasting their time (and ours) by posting such drivel as the above quote.
Ati, however, clearly did not test Catalyst 3.4 on a 256-MB 9800 part against anything resembling the current codebase prior to releasing Catalyst 3.4, as evidenced by the fact that 3.2 could run the game fine while 3.4 could not (thus preventing the use of the full 256-MB the new 9800 card contained). It isn't a released game, and looking at it was obviously not a priority for them. But they ended up with some egg on their face as a result.
it may be the GFX being so powerful till the CPU limits the performance..........that's how it's the same
no, i am not a fanboy
my last 3 cards were:
ATi Radeon 9700np
GeForce 2 Ti
Voodoo Banshee+Voodoo 2 combo
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter