Slashdot Mirror
GF FX 5900 Ultra vs. ATi Radeon 9800 Pro
Mack writes "OCAddiction.com has their GF FX 5900 Ultra vs. ATi Radeon 9800 Pro article online detailing which card is more powerful. Running a plethora of benchmarks we were anxious to see which card outperformed the other. Quite simple really. We take nVidia's top offering and pair it up against the current top offering from ATi and let them duke it out till the bitter end. Who will come out on top? Let's take a look."
If you haven't heard about the controversy with MadOnion/Futuremark/3dmark2003, check out This article. Kyle @ HardOCP suggests that if you give Futuremark more $$$, they will 'optimize' their benchmark to help out your video card's score.
Now, in this review, we see that GeForceFX 5900 clearly dominates the hardware side of things: .13 vs .15 micron process, 450/850 vs. 380/340 (GPU/Core), 27.2 GB/sec vs. 21.8 GB/sec memory bandwidth, etc. Yet when we start looking at real-world scores, the 9800 keeps up pretty well & even beats the faster GeForceFX 5900 in most tests.
The big exception is the 3DMark2003 score - the GeForceFX 5900 wins 3477 to 2837!!! (!!!).
This can be attributed to one of three things;
1.Speed isn't everything (e.g., AMD vs. Intel CPU's). But of course, the slower Radeon 9800 *is* faster even though it's slower in all the real-world tests.
2.The GeForceFX used WHQL drivers... But despite these 'superior' drivers, the Radeon 9800 still reigned in all the real world tests!
3.3DMark2003 added unfair optimizations to their program to make the nvidia card seem better than ATi's
OMG,
about the 3DMark2003 you are so wrong unfortunately.
1.Speed isn?t everything (e.g., AMD vs. Intel CPU?s). But of course, the slower Radeon 9800 *is* faster even though it?s slower in all the real-world tests.
Its less about the speed of the GPU nowadays, its more about more successful methods of saving memory bandwidth. But basically you're right here.
2.The GeForceFX used WHQL drivers? But despite these ?superior? drivers, the Radeon 9800 still reigned in all the real world tests!
The difference between the WHQL drivers and the beta drivers are, that the WHQL ones use a different method für uninstalling and that they require a reboot when applying the coolbits hack to pverclock the VGA card. With the beta drivers the coolbits are quite often already enabled.
3.3DMark2003 added unfair optimizations to their program to make the nvidia card seem better than ATi?s
Nope.
The other way round. Guess what teh Patch 330 did with the 3DMark2003? It lowered the score of the ATI card by 3 (!) percent. The nvidia card got slower by... more than 30 (!) percent. Yikes.
The 9800 is still the better purchase, the 5900 has little to no overclocking room and needs a massive heatsink to remain "cool". The manufacture process for the 9800 is more mature on the other hand, and it usually clocks about 60Mhz beyond stock for the GPU and about 20Mhz for the RAM giving 440/370, which makes it comfortably faster than the 5900.
Well the site is crawling by now, had several timeouts already, but managed to get to the conclusions (wasn't really worth the effort tbh):::
Conclusion
Let's break down performance of both cards and see which one comes out on top.
UT2k3 - FX 5900 Ultra - While both cards perform well, the FX 5900 comes out on top
AquaMark - R9800 Pro - The R9800 takes home the gold in this real-world benchmark
Comanche 4 - R9800 Pro - The R9800 also wins out by an edge for this nearly obsolete benchmark
Specviewperf 7.0 - R9800 Pro - This one is really close but the #'s lean to the R9800
Code Creatures - FX 5900 Ultra - The 5900 beats up the R9800 pretty good in this intensive benchmark
Splinter Cell - R9800 Pro - Hands down, the R9800 takes it in this awesome game from UBISoft
ShaderMark - R9800 Pro - While the FX 5900 Ultra makes a good showing, the R9800 wins this one
3DMark 01 SE Build 330 - R9800 Pro - The R9800 takes top honors with this tried and true synthetic benchmark
3DMark 03 Build 320 - FX 5900 Ultra - Should we include this? Possibly not, however the FX 5900 wins with WHQL Det Drivers
3D Visual Quality - R9800 Pro hands down
And the winner is.........The FIC ATi Radeon 9800 Pro 128MB. We compared these cards in every category we could think of and in the end, we saw better performance overall from the ATI Radeon 9800 Pro. Did the FX 5900 fail to impress us? No, not at all. We believe both cards are worthy of any good system but we do have to tip our hats to the excellent performance that the Radeon 9800 Pro has showed us here today.
--------
Daniel
Carpe Diem
After buying a 7500 and tinkering with it for a few days, I decided that I didn't want to try anymore, and then traded it for a GeForce 4. It worked perfectly on the first try. I'm not a huge fan of either company, but yes, I still like to buy Nvidia cards.
9800 has a faster transform engine, is slightly ahead at lower resolutions.
5900 has a higher fill rate, is slightly ahead at high resolutions.
Otherwise there are no real differences between the benchmarks and it all comes down to differences any layperson could understand:
The 5900 takes up 2 slots (WTF?) and the 9800 is $100 cheaper (although $399 for a graphics card is still nuts if you ask me).
BTW, the ATI 9800 won the "shootout".
NVIDIA cards because ATI's Linux drivers are not very good compared to NVIDIA's. I won't be buying an ATI card until ATI supports Linux fully like NVIDIA. I do play Linux native-port games in Linux.
This hasn't been true for quite some time.
I have owned numerous high end nvidia and radeon cards, and have never had anything resembling stability from the nvidia cards using the nvidia binary driver (and yes, I've tried all of the tweaks and suggestions Nvidia and others suggest vis-a-vis AGP settings, etc.). This has been true on numerous machines, both single and dual Intel P3 and Athlon XP/MP boxes, with a variety of motherboards, memory configurations, and Linux kernels.
ATI radeon cards on the other hand have been pretty solid, with excellent support via the xfree DRI drivers for most cards, and adequate, reasonably stable support from ATI via their firegl binary-only drivers for those not yet supported.
NVidia has not been king of the Linux hill for quite sometime, and while I have had my gripes with ATI as well, the notorious instability of the Nvidia binary drivers and lackluster support via the xfree DRI drivers has placed me (and my employer) firmly in the ATI camp.
The Future of Human Evolution: Autonomy
Apparently the Radeon Mobility is not supported under Linux either - this has made my search for a new laptop very difficult.
;) People using more cutting edge distro's have been using it longer.
This statement is false. The Mobility Radeon has been supported since Xfree 4.2.
I have been using this chipset with a IBM Thinkpad X22 for almost a year now, and that's with GNU/Debian Linux.
You wan't a great, cheap, superlight laptop with decent 3d support?
Please visit the IBM eBay Store
Laptops are brand new in the box, full warranty, are almost 50% retail, and you are buying directly from Big Blue.
The catch? They're slightly behind the newest models, but hey, with linux support, that's the best way to buy hardware.
Driver stability has not been an issue since they started using numbers to identify Radeons.
Just a Tuna in the Sea of Life
The ATI 'Pro' cards have DVI-D output, however it's incompatible with many monitors at 1600x1200 and higher. It's generally the monitor mfr's fault for not getting the standards quite right, but that's little consolation when you hook your $2000 Viewsonic VP201m or similar up to a Radeon and just get green snow. :-/
WHQL doesn't mean they're better drives, it just means that they passed some MSFT testing bits. If anything, non-WHQL drivers have potential to have higher performance (think a car engine that doesn't have to worry about passing emissions), since they don't have to worry so much about playing nice with -all- available hardware.
my sig's at the bottom of the page.
Interesting that in 3DMARK, the FX 5900 ran away with it.
The FX 5900 ran away with nothing.
First, the Radeon won in 3DMark01.
Second, observe the origin as well as the scale of the 3DMark03 graphs: Graph 1, Graph 2
The difference is grossly exaggerated by the graph's peculiar origin (5700 and 3800 instead of 0) and large scale.
Third, 3DMark03 has been rendered an useless benchmark since it is riddled with nVidia "optimizations," which have been deemed illegitimate by Futuremark's own accord. Even the author of the article acknowledges the dubious nature of 3DMark03: 3DMark 03 Build 320 - FX 5900 Ultra - Should we include this? Possibly not, however the FX 5900 wins with WHQL Det Drivers
New nVidia optimizations (read: cheats) such as the Anistropic Filtering "optimization," which was exposed by renaming the 3DMark03 executable. Of additional note is that one of the 3DMark03 benchmarks was done using 8x Anisotropic Filtering, deeming it even more illegitimate than the rest of the 3DMark03 benchmarks in the article.
Fourth, the Radeon delivers consistently better image quality throughout, as was acknowledged by the author of the article: 3D Visual Quality - R9800 Pro hands down.
Another thing to be kept in mind is that the Radeon may very well do much better relative to the FX 5900 in all benchmarks since nVidia may very well optimizing for all major timedemos (3DMark03 style) and the reviewed did not record his own demos. Feel free to google if you find you would like more information.
Uttering logically derived and empirically supported truths to the disciples of the orthodox establishment.
Sorry nVidia, but I use Linux on PowerPC... If nVidia supported Linux/PowerPC at all, like ATi does, maybe I'd buy one. But till then I'll stick to ATi, no matter if it's slower than nVidia's Linux/x86 drivers.
nVidia does not support Linux; they don't release any specs. ATI does; they give specs to DRI developers. You are not supporting Linux by releasing closed source software, but perhaps Linux/x86.
The new Radeon Catalyst drivers (2.5) have this very interesting note in the change log: "The 3DMark2003 shader optimizations found in previous CATALYST(TM) releases have been removed" Yet Nvidia gets to keep THEIR optimizations... hrm.
For those Windows / dual boot users looking a little on the lower end of the performance and price curve, I just found this hacked driver page and this thread that basically turns certain Radeon 9500 cards (~$135) into 9700s (~$200) by unlocking 4 pixel pipelines on the chip. It doesn't work on all cards, producing visual artifacts on some (some workarounds exist for some users) but given the right hardware, you might pull a good deal of performance out of a mid-priced piece of hardware.
Bleh!
It seems to be readily apparent in the Radeon screenshot (at least the full quality AA screenshot). I use it all the time, it makes 640x480 gaming look a heck of a lot better (which is good for games that are old and don't support higher resolutions). Ultimately, though, it's just one more feature for the list on the side of the box.
-=-=-=-=-=
I'd rather be flamed than ignored.
There really is no clear winner between these two, and they cost the same. So why wouldn't people buy the FX? I prefer to support NVidia because they brought about all the recent great leaps in graphics technology (programmable vertex and pixel shaders, Cg, etc.) whereas ATI hasn't come up with anything particularily impressive.
NVidia is not 3dfx. Don't expect them to die anytime soon.
(I am a professional game programmer. Just thought I'd mention that.)