Quadro FX 2000 performance *is* good

[Uttar]

Hey,

Just to make you know that Amazon International (http://www.amazoninternational.com/html/news/news_current.asp) posted some preliminary benchies of the Quadro FX 2000.

They don't give figures in % or against the ATI Fire GL X1, so I've decided to do that myself based on their Fire GL X1 benchmarks ( and, lucky me, they were done on the exact same system )

So here's the PNY Quadro FX 2000 performance numbers, compared to the PNY Quadro 980 XGL and ATI Fire GL X1:

SPEC View Perf 7, Dual Xeon 2.8Ghz
23.46 vs 18.95 vs 18.13 (
121.5 vs 93.21 vs 89.68
132.6 vs 132.3 vs 105.2
30.9 vs 24.83 vs 26.52
38.55 vs 23.52 vs 22.34
37.04 vs 28.95 vs 21.26

ATI Fire GL X1 & PNY Quadro FX 2000 compared to the PNY Quadro 980 XGL
+04.52% & +29.39%
+03.93% & +35.48%
+25.67% & +26.05%
-06.37% & +16.51%
+05.28% & +72.56%
+36.17% & +74.22%
Average
+11.53% & +42.37%

PNY Quadro FX 2000 compared to the ATI Fire GL X1
+23.80%
+30.35%
+00.23%
+24.45%
+63.90%
+27.94%
Average
+28.44%


As you can see, the PNY Quadro FX 2000 is nearly 30% faster than the ATI FireGL X1 and over 40% faster than the PNY Quadro 980 XGL
The increase in performance over the ATI FireGL X1 is higher than the ATI FireGL X1 increase in performance over the Quadro 980 XGL!

The only problem for the Quadro FX 2000 is that it only got 128MB of RAM ( compared to the ATI FireGL X1 maximum of 256MB ) and that it costs £1240 , compared to the ATI FireGL 128MB price of £500 and 256MB price of £640.
The PNY Quadro 980 XGL now costs £490, so it became very competitive. The Quadro FX 1000 cost is at £780, which is once again very expensive for a card that should theorically be of the same speed as the ATI FireGL 128MB at £500

The real advantage of the Quadro FX thus lies in its features:
1. 12-bit sub-pixel precision, compared to the Wildcat4 7210 10-bit sub-pixel precision ( ATI Fire GL X1 sub-pixel precision unknown )
2. Beyond CineFX support ( 2048 PS instructions vs GFFX 1024 instructions and Fire GL X1 "mere" 96 instructions )
3. Full 128-bit floating-point precision pipeline - ATI FireGL X1 only supports 96 bits in the Pixel Shader, thus offering slightly lower quality ( not like it's really noticable, but it might matter to same people )
4. Optimized for Cg. Seems useless? Well, even if games aren't using Cg all that much, a lot of workstation applications ( discreet, for example ) already got plug-ins in them which allow using Cg in them. And, want it or not, it's quite popular there.
5. Quadro FX 2000 Silent Running when not active ( Flow FX *is* silent when the GPU is not used, but more noisy than other coolers when it's in intensive use - or anyway, that's what nVidia seems to claim )


Here, can't see much more about that good ole Quadro FX without more benchies. Hope you liked this information.


Uttar

[nutball]

What clock-speed does the Fire GL X1 run at? The Quadro FX 2000 is clocked at 400/400 isn't it?

Wondering if we can deduce any NV30 to R300 performance ratios from your numbers plus the relative clock-speeds...

[Uttar]

Originally posted by nutball
What clock-speed does the Fire GL X1 run at? The Quadro FX 2000 is clocked at 400/400 isn't it?

Wondering if we can deduce any NV30 to R300 performance ratios from your numbers plus the relative clock-speeds...

I doubt we can, because maybe ATI & nVidia got different for-workstation optimizations in hardware and in their drivers.
Anyway, the Fire GL X1 runs at 324/290, pretty much like the Radeon 9700 Pro
And the Quadro FX 2000 is indeed clocked at 400/400.

So, assuming both nVidia & ATI got the same workstation optimizations in their cards ( which is certainly wrong ) , the GFFX at 500/500 would be 35.55% faster than the Radeon 9700 Pro


Uttar

[Bigus Dickus]

nVidia has always done a much better job than ATi at workstation optimizations/drivers for their products. This is clearly seen in the Quadro 980 XGL vs. Fire GL X1 performance comparison, which is anything but representative of the gaming performance of the relative chipsets.

[Uttar]

Originally posted by Bigus Dickus
nVidia has always done a much better job than ATi at workstation optimizations/drivers for their products. This is clearly seen in the Quadro 980 XGL vs. Fire GL X1 performance comparison, which is anything but representative of the gaming performance of the relative chipsets.

Indeed, nVidia has generally done a better job with workstation optimizations.
But saying the Quadro 980 XGL vs Fire GL X1 performance got "nothing" to do with GF4 Ti4600 vs Radeon 9700 Pro is not quite true.
Remember the R300 main advantage is with FSAA & Aniso. Without either, the performance advantage is around 40% ( and up to 60% )
And all of those Quadro 980 XGL benchies are done without that.
As Fire GL X1 drivers mature, it's quite likely ATI gets around a 25% advantage over the Quadro 980 XGL

So, saying it got "nothing" to do with R300 performance isn't quite true. But it does not have much to do with it.


Uttar

[nutball]

Hmmm. Well the stuff I want to do with the card is "workstation stuff" (scientific visualisation). The real question is do I splash out now for the Quadro, or wait a few months and get five of the retail cards! Hmmm. :confused:

[fantomas]

nutball, that is ALWAYS the question with computer hardware :)