In more general sense this is a question of gaming vs workstation graphics cards. To answer this question Tom’s Hardware did a benchmark between nvidia GTX 260 and Quadro FX 4800.
Quadro FX 4800
XFX GTX 260
Tom’s Hardware benchmark is quite comprehensive and in first part they are comparing Quadro 4800 card against other Quadro and ATI FireGL set of cards.
Results are bit of the mix, but FX 4800 is very close to the top in almost every benchmark.
Finally on page 10 of the benchmark results of test for GTX 260 are listed against FX4800. The number are quite impressive, some of test were a whole magnitude larger for FX 4800.
Test from Viewperf 10 | GTX 260 | FX 4800 |
3dS Max 04 | 11.53 | 46.23 |
catia-02 | 15.22 | 57.95 |
ensigth-02 | 18.34 | 54.47 |
maya-02 | 35.71 | 221.71 |
proe-04 | 14.72 | 60.59 |
sw-01 | 12.94 | 128.71 |
tcvis-01 | 4.77 | 39.36 |
ugnx-01 | 5.89 | 33.72 |
A Quadro FX 4800 moves up to 10 times faster when running workstation applications than the GeForce GTX 280. This leads swiftly to a clear and inescapable conclusion: there's no good reason to use a GeForce graphics card for workstation applications. It just doesn't pay.The problem is that this test does not includes GIS applications, namely ArcGIS. I would love to see a application testing graphics performance in ArcGIS and ArcGlobe. One of major problems might be the fact that most GIS processing (speaking of graphics only) is 2D not 3D. In that regard I believe the gaming cards are doing a fine job. Again, we need a test tool to prove this.
For 3D performance in GIS i tried using Fraps. Fraps can show current frame rate on the screen or 3D window. In ArcGLobe it was constant 25 fps (if memory serves me well).
If any ArcObjects developers are reading this I would love to test your ArcGlobe benchmark tool… If you are working on one please e-mail me on ZergOne@gmail.com.
No comments:
Post a Comment