G400 MAX slower than TNT?

[hella356]

I replaced my CL 16MB standard (not TNT2) TNT with a MAX and in benchmarks it's about 10 slower. I've tried several different drivers, but the results are the same. I know that the MAX is highly CPU dependent, but this much? Any suggestions about how to speed it up other than getting an Athlon/P3?

AMD K6-3 400
Epox MVP3G-M
256 MB RAM
10,000 RPM U2W SCSI

[Nuno]

What benchmarks? G400 drivers version? What OS? 10% slower at 1024x768x32?

[Stringy]

Yea you don't have enough HP to drive a G400Max, the G400's start to shine with a P3-450(or althon) and up... I'm afraid Matrox Offloads too much to the CPU for the K6-3 to handle...

[DuRaNgO]

first off, you need to be absolutely sure all the Tnt drivers have been removed from your computer, then give us a list of the irq arangement from device manager as there is definately something wrong
jim

------------------
PIII-500mhz @ 620 ! with an Abit BE6 mobo
128mb pc-100
Mill G400 (YAHOO!!!) 32mb @ 167/208 with MGATweak-417mhz, (2.5, 2, 2.5), PD 5.41 & bios 1.5-22
Maxtor 14.3 gb Uata66 hdd
SB Live!
Winblows 98se & DX7
32 lbs. of fans, heatsinks and aluminum ductwork
3DMark Result 5954.69 3DMarks
Synthetic CPU 3D Speed 9325.69 CPU 3DMarks
Rasterizer Score 2791.72 3DRasterMarks
Game 1 - Race 65.07 FPS
Game 2 - First Person 54.89 FPS
Fill Rate 268.79 MTexels/s

[bigluker]

im running a p2-400 and the g400 max screams. If i had a p3 i think i would crap myself at the framerates.

My friends TNT2 ULTRA 175/183 (higher clocked) is a tiny little bit faster in some D3D applications (ie 2 fps higher in Viper Racing), but once you switch to 32 bit the matrox can whoop it. He has a celeron 400.

So either your drivers are messed, or matrox doesnt like AMD K6's. The best solution to your problem...
CYRIX!! Oh baby, the 3D gamer's ultimate CPU>

[hella356]

Win98
Matrox PowerDesk Revision 5.50.005
RR-G Matrox Video Tools Version 1.51.024
I've only had the G400 in this install of 98.
On the previous install of 98, the TNT got scores in the range of 3400 3DMarks, 40 FPS for Game1, 30 FPS for Game2 at 800x600x16. The Max got about the same for each install as the scores below.

IRQ 5 Vortex AU8830 PCI Audio
IRQ 5 Adaptec AHA-2940U2/AHA-2940U2W PCI SCSI Controller
IRQ 5 Vortex AU8830 Multifunction PCI Platform
IRQ 5 IRQ Holder for PCI Steering
IRQ 9 3Com Fast EtherLink XL 10/100Mb TX Ethernet NIC (3C905B-TX)
IRQ 9 VIA Tech PCI to USB Universal Host Controller
IRQ 9 IRQ Holder for PCI Steering
IRQ 11 IRQ Holder for PCI Steering
IRQ 11 Matrox Millennium G400 DualHead - English
IRQ 14 Primary Bus Master IDE controller (dual fifo)
IRQ 14 VIA Bus Master PCI IDE Controller (Ultra DMA)
IRQ 15 Secondary Bus Master IDE controller (dual fifo)
IRQ 15 VIA Bus Master PCI IDE Controller (Ultra DMA)

3DMark99
Resolution: 640*480
Color Depth: 16-bit Color
CPU Optimization: AMD 3DNow!(tm)
Z-Buffer: 16-bit
Frame Buffer: Double buffering
Refresh Rate: 85 Hz
3DMark Result : 2,914 3DMarks
Synthetic CPU 3D Speed : 5,833 CPU 3DMarks
Rasterizer Score : 2,507 3DRasterMarks
Game 1 - Race: 35.0 FPS
Game 2 - First Person: 24.9 FPS
____________________________________________

Resolution: 1024*768
Color Depth: 32-bit Color
CPU Optimization: AMD 3DNow!(tm)
Z-Buffer: 16-bit
Frame Buffer: Double buffering
Refresh Rate: 85 Hz
3DMark Result : 2,869 3DMarks
Synthetic CPU 3D Speed : 5,781 CPU 3DMarks
Rasterizer Score : 1,874 3DRasterMarks
Game 1 - Race: 34.2 FPS
Game 2 - First Person: 24.7 FPS

[Nuno]

Those seem pretty regular numbers for a G400 + a K6 cpu. See if you have v-synch enabled, or use triple buffering instead. And much more important: Games? how are them? smooth? no? 3dmark99 never was very K6 friendly, even with 3dnow.

I know that the MAX is highly CPU dependent

- you said it. It´s true.

The K6-x 400 don´t have all the fpu power that the G400 can have. Face it. Wait, I think there still isn´t a single cpu that has that fpu power! AFAIK, the G400 was still scalling with an Athlon 900...

Just to compare, some results from an Athlon 500 and a G400 clocked to Max speeds:

Resolution 640*480
Color Depth 16-bit Color
Frame Buffer Triple buffering
Refresh Rate VSync Off
CPU Optimization AMD 3DNow!(tm)
3DMark Result 5085,30 3DMarks
Synthetic CPU 3D Speed 8959,95 CPU 3DMarks
Rasterizer Score 0,00 3DRasterMarks
Game 1 - Race 58,99 FPS
Game 2 - First Person 44,69 FPS

Resolution 1024*768
Color Depth 32-bit Color
Frame Buffer Triple buffering
Refresh Rate VSync Off
CPU Optimization AMD 3DNow!(tm)
3DMark Result 4531,87 3DMarks
Synthetic CPU 3D Speed 8810,40 CPU 3DMarks
Rasterizer Score 0,00 3DRasterMarks
Game 1 - Race 48,37 FPS
Game 2 - First Person 42,63 FPS

I previously had a Celeron 450 and it scored in 3dmark about 3500. But I am more impressed with something like 73->104 fps in quake2 demo1 (114 fps with 3dnow! patch) and 19->35 fps in that DMZG Nvidia hi-poly bench.

[hella356]

Yeah, I had the Max in an Athlon 700 I built for a friend for a while and it was MUCH faster, I just didn't expect it to be slower than a TNT. Damn!

[Dyre]

Any possiability you could afford a new mb/chip. Maybe a celery 500 if your now on a budget. You wouldnt need new ram or anything and i think you will be happy with the power increase. Or how about a p3-500 or 550. Those are VER affordable right now.

Just to give you an idea of the scalability
3dMark2000
170/226MAX and pIII 450 = about 1900 3dMarks
DefualtMAX and PIII 450 @ 504 = about 2100 3dMarks
170/226MAX and pIII 450 @ 504 = about 2400 3dMarks!!!!

------------------
PIII 450 @ 504
generic BX motherboard
G400 MAX : )
Maxtor 13 gig UDMA 33
Maxtor 13 gig UDMA 66
Creative 36x CD-ROM
HP 2x2x6x CD-RW
64 megs PC100 RAM
128 megs PC133 RAM
NetGear 10/100
Soundblaster 32 AWE
Diamond SupraExpress 56k
Logitech Wireless Desktop (best $50 ive ever spent)
1 92mm Sunon Fan
4 80mm Sunon Fan's

[Thundrchez]

IMHO, 3Dmark 2000 seems to give stats heavily in favor of nVidia cards vs Matrox cards, but in real performance, the Matrox cards do exceptionally well.

For an example, I've been kicking the tires of an nVidia GeForce 256 SDR card. It gave some overwhelming 3dmark score of 4097, where my G400 gave me 3105. However, in real world gaming, it did not seem a bit faster than my G400.

This was verified by using the Unreal Tournament "utbench" demo/benchmark. There is virtually no difference between a GeForce 256 and a G400. The G400 was only one or two fps avg lower. And in a much more important area, the minimum fps was higher with the G400 than the GeForce 256. When things get really busy, the G400 is better.

Not to mention that the video quality of the G400 completely destroys anything that nVidia makes....

[rylan]

Well, I have a K6-III 450 running o/c to 500, and I am quite please with the results on my MAX. I can run games smoothly, and it looks great. UT at 800x600x32bit with all advanced settings tweaked up.. trilinear filtering etc.
Q3 runs at 1024x768x32bit max detail fairly smoothly too. Can't complain.