Follow Up--New GeForce2 MX Drivers' Performance

As posted yesterday, we manually flashed the firmware to 1027F and added the v1.1.1 extensions to our System folder. We saw no performance change on a variety of tests, all being virtually identical. Peter Chin did the same thing but tried Rene Trost's ThroughPut v1.5 and was surprised that his performance had actually DECREASED! We ran ThroughPut 1.5 and here are the results:

ThroughPut v1.5 in MB per second

Test	v1.0.9	v1.1.1
CPU	157	60
FPU	342	203
Altivec	570	203
CopyBits	180	181

The results for v1.0.9 (with firmware v1017f1) were obtained right after we received our Power Mac G4 533 DP. The results for v1.1.1 were obtained with firmware v1027F. ThroughPut measures the speed of data passing through the AGP port by means of the indicated unit. The CPU and Altivec numbers for the more recent drivers and firmware are much lower. Curiously, we do not see any difference in performance using Cinebench 2000, Walker atrium spin, scrolling a 136-page PDF file loaded with graphics, MacBench 5.0 Publishing Graphics or Quake III Arena demo127 (1024x768, 32 bit, high). Here are the results:

Comparison of Old & New NVIDIA Firmware/Drivers

Test	v1.0.9	v1.1.1
Cinebench 2000 (score)
- Shading (Cinema 4D)	5.6	5.5
- Shading (OpenGL)	7.0	7.0
- Raytracing (Dual CPU)	12.0	12.1
Walker atrium spin (min. fps)	30	31
PDF Scroll (sec)	27	28
MB 5 Publ. Graphics (score)	3144	3148
Quake III Arena 127 (avg. fps)	41	41