Initial MAX O/C Results Poor w/ MGA Tweak

[SCompRacer]

SCompRacer wants to run with the big dogs, but can't get off the porch much. Have a P III 500 @ 560 via 112 FSB, G400 MAX and AGP forced at X 2, 75MHz X 2 = 150MHz. Ran tests a couple of times, similar results. Going back to AGP X 1 I get .2% (2 tenths of a percent) improvemment.

No O/C Results
P III @ 560, 112 FSB, MAX @ Stock Clock

3DMark Project:
3DMark Build Version: 200
Date: 9/16/99
Time: 2:31:52 PM
Project Name:My Project
Comments: P III @ 560, 112 FSB, MAX @ Stock Clock, AGP X 2

Project Settings (Template):
Rendering Platform: Internal (Matrox Millennium G400 Dual Head - English)
Resolution: 800*600
Color Depth: 16-bit Color
CPU Optimization: Intel(r) Pentium(r) III
Z-Buffer: 16-bit
Frame Buffer: Triple buffering
Refresh Rate: 84 Hz
Looping: Disabled
Texture Format: 16-bit, 4444 RGBA
Run Tests: Once
Title Screen: Shown

Test Results:
3DMark Result : 5,469 3DMarks
Synthetic CPU 3D Speed : 8,555 CPU 3DMarks
Rasterizer Score : 2,580 3DRasterMarks
Game 1 - Race: 59.1 FPS
Game 2 - First Person: 50.9 FPS
Fill Rate : 247.9 MTexels/s
Fill Rate With Multi-Texturing :251.0 MTexels/s
2MB Texture Rendering Speed: 362.5 FPS
4MB Texture Rendering Speed: 289.9 FPS
8MB Texture Rendering Speed: 199.8 FPS
16MB Texture Rendering Speed: 148.9 FPS
32MB Texture Rendering Speed: 96.5 FPS
Bump Mapping Emboss, 3-pass: 112.6 FPS
Bump Mapping Emboss, 2-pass: 146.9 FPS
Bump Mapping Emboss, 1-pass: 263.0 FPS
Point Sample Texture Filtering Speed: 111.3 %
Bilinear Texture Filtering Speed: 100.0 %
Trilinear Texture Filtering Speed: 66.6 %
Anisotropic Texture Filtering Speed: Not Supported
6 Pixel/individual: 2,082.7 KPolygons/s
6 Pixel/strips: 3,622.1 KPolygons/s
25 Pixel/individual: 2,047.7 KPolygons/s
25 Pixel/strips: 2,260.4 KPolygons/s
50 Pixel/individual: 1,372.9 KPolygons/s
50 Pixel/strips: 1,473.2 KPolygons/s
250 Pixel/individual: 514.6 KPolygons/s
250 Pixel/strips: 554.9 KPolygons/s
1000 Pixel/individual: 173.9 KPolygons/s
1000 Pixel/strips: 191.3 KPolygons/s

O/C w/MGA Tweak
P III @ 560, MAX O/C Fo=351, 2.5/2.0/2.5, AGP X 2


3DMark Project:
3DMark Build Version: 200
Date: 9/16/99
Time: 5:20:37 PM
Project Name:My Project
Comments: P III @ 560, 112 FSB, MAX @ 351 2.5/2.0/2.5, AGP X 2

Project Settings (Template):
Rendering Platform: Internal (Matrox Millennium G400 Dual Head - English)
Resolution: 800*600
Color Depth: 16-bit Color
CPU Optimization: Intel(r) Pentium(r) III
Z-Buffer: 16-bit
Frame Buffer: Triple buffering
Refresh Rate: 84 Hz
Looping: Disabled
Texture Format: 16-bit, 4444 RGBA
Run Tests: Once
Title Screen: Shown

Test Results:
3DMark Result : 5,478 3DMarks
Synthetic CPU 3D Speed : 8,637 CPU 3DMarks
Rasterizer Score : 2,361 3DRasterMarks
Game 1 - Race: 58.8 FPS
Game 2 - First Person: 51.3 FPS
Fill Rate : 227.4 MTexels/s
Fill Rate With Multi-Texturing : 231.5 MTexels/s
2MB Texture Rendering Speed: 321.5 FPS
4MB Texture Rendering Speed: 258.2 FPS
8MB Texture Rendering Speed: 178.7 FPS
16MB Texture Rendering Speed: 134.7 FPS
32MB Texture Rendering Speed: 90.1 FPS
Bump Mapping Emboss, 3-pass: 99.8 FPS
Bump Mapping Emboss, 2-pass: 131.7 FPS
Bump Mapping Emboss, 1-pass: 241.7 FPS
Point Sample Texture Filtering Speed: 111.6 %
Bilinear Texture Filtering Speed: 100.0 %
Trilinear Texture Filtering Speed: 65.9 %
Anisotropic Texture Filtering Speed: Not Supported
6 Pixel/individual: 2,143.1 KPolygons/s
6 Pixel/strips: 3,367.9 KPolygons/s
25 Pixel/individual: 1,889.8 KPolygons/s
25 Pixel/strips: 1,981.3 KPolygons/s
50 Pixel/individual: 1,213.4 KPolygons/s
50 Pixel/strips: 1,305.5 KPolygons/s
250 Pixel/individual: 450.9 KPolygons/s
250 Pixel/strips: 487.7 KPolygons/s
1000 Pixel/individual: 153.8 KPolygons/s
1000 Pixel/strips: 169.0 KPolygons/s

Initially tried Fo @ 450, and when that bombed I started out a little more conservative. With results not getting better, time to stop and get help from you wizards.

BTW, I have 8ns memory, will 7ns produce better results or do I need to look elsewhere?

One other thing, I disabled ECC when performing these tests and have the latest 5.25 drivers.

SCompRacer


------------------
P III 500 @ 560 via 112 FSB, ASUS P2B-LS, 1010 BIOS, w/256 MB 8ns ECC RAM, ECC Enabled since 5.13 driver release, current 5.25 drivers, dual 9 GB Cheetahs, 10 GB IDE DiamondMax, Plextor Ultraplex/Plexwriter, Hitachi IDE DVD CD ROM, SCSI ZIP insider, G400 MAX, SB Live, Sigma Hollywood + DVD decoder, US Robotics 56K Sportster, Sony GDM400PS, UMAX 2400S SCSI scanner, Soundworks 5.1, HP 6P Laser & 890C, LAN & USB enabled. SB Live shares IRQ w/MAX, bus mastering enabled, Win98 SE and NT 4.0 SP4.

[Star F.]

I suspect u get those poor values when o/c'ing, because u changed the clock dividers. Should leave them alone, the standard core-memory-ratio is the best. When going above 300 Mhz pll clock u shuld also turn off automatic duty cycle correction. See the MGA Tweak manual.

[Maggi]

Try again with Matrox's OCer and see if that makes a difference.

------------------
Cheerio,
Maggi
________________________
Asus P2B-DS @ 103MHz FSB
2x P3-450 @ 464MHz
512MB CAS2 SDRAM
Millenium G400 32MB DH

[Scytale]

Yep, try normal dividers. Also, the dawg would better run his beasty benchmarks at 1280x1024x32... 800x600x16 doesn't make much sense for a MAX, you'll be testing your CPU at that resolution.