New BIOS available

[chucky2]

Version 2.54.000.

Get it here: 2.54.000

Chuck

[chucky2]

FYI:

Same BIOS (2.1-35) for my card.

Chuck

[chucky2]

Hmmmm.....

After canceling the BIOS flash when it asked me if I wanted to flash the same version, I shut down my computer.

This morning, when I turned it on, right after the loading bar in the Windows XP Pro boot, the screen became corrupted, went blank (like usuall), and then the desktop came up.

Really, everything booted as normal except that corrupted screen for about 2 seconds.

I'll have to see what happens on next boot.

Just to let everyone know....

Chuck

[omegaRED]

A friend of mine who`s running XP always has a corrupted display on startup, 1 or 2 seconds long ...

[chucky2]

I'm not really worried about it (as long as the computer works like it should, I'm not going to worry), I just wanted to let people know.

Also, the Updater wants to copy a file, UtilNt.something or other into some directory on the hard drive. It sucks that after it exits that it doesn't clean up after itself. Mad Matrox.

Chuck

[K6-III]

Hopefully, they finally fixed the broken EMBM of the G550.

[Randy Simons]

Originally posted by K6-III
Hopefully, they finally fixed the broken EMBM of the G550.

Hmm, I believe that's not possible, the "error" is in the hardware. Or actually, in the heads of the Matrox engineers. Or.. actually, in the heads of Matrox's PR folks. "it must be cheap!" Or sumthin' like that.

Gimme my Trilevel Cache!!!

Yeah they don't make them like that anymore today... trilevel rules Too bad I only have 1MB, and that that stinkin' VIA used one-way direct mapped cache
But, K6-III proves that the K6 core was quite good

[Kruzin]

Originally posted by K6-III
Hopefully, they finally fixed the broken EMBM of the G550.

The EMBM with G550 is as good as it's gonna get. No BIOS update is going to fix this issue. It's at a hardware level, and will never be as good as G400's EMBM was...

[K6-III]

They better fix at least that by next gen.

[Wombat]

I really don't see how a 3rd level cache "proves" that the K6-2 (or later) was a very good core. It had a slow FPU, and no cache is going to help that. Throw more cache at just about <I>any</I> architecture, and it will improve.

[Randy Simons]

Originally posted by Wombat
I really don't see how a 3rd level cache "proves" that the K6-2 (or later) was a very good core. It had a slow FPU, and no cache is going to help that. Throw more cache at just about <I>any</I> architecture, and it will improve.

Hmm, a bit offtopic he

It's not as much the three-level cache that proves it, but more the full-speed on-die level 2 cache of the K6-3. On the K6 and K6-2, the lvl2 cache was at the same speed as the FSB, 66MHz or 100MHz.

The K6-3 runs circles around an equally-clocked P2 or P3, while those processors run circles around a K6-2. Thus it seems when the K6-core got the memorybandwith it wants, the core flies.

But the 3rd cache level does provide a nice few of extra % speed. Especially on WindowsNT or Windows2k, a K6-3 with lvl3 cache is very fast. A 400MHz K6-3 is quite a match for a 500MHz P2 on those OSses.

The FPU wasn't really always that bad. It was a low throughput-low latency FPU, vs. a high throughput- high latency FPU on the Intel cores. There are FPU-intensive applications that run faster on a K6 than on a P2, because of the low latency.

BUT tho get on topic again: what effects could a new BIOS on a vidcard have in Windows? Do the Windows-driver still use the BIOS? (why? I think it would be faster to avoid that extra call..)

[WayneHu]

This is an off-topic speaking, too...

The FPU code optimized for Intel Pentium or PentiumPro processor embeds FXCH command in between two math calculation commands to accelerate the x87 code speed for the x87's stack-register architecture. The FPU unit can perform one FXCH and FPU calculation at the same time. And the FPU unit is pipelined.

K6's x87 FPU suffers from such kind of x87 coding style. Its x87 unit is non-pipelined although it has low latency (4 cycles).

K7 does not care about it. And the x87 FPU unit is fully pipelined and super-scaled. So the FPU code optimized for Pentium processor runs faster in K7 than Pentium3.

P4 doesn't like it, either. The x87 unit on it is even worse than the one on Pentium3. So x87 code's efficiency on P4 is not as good as the one on K7 or P3.


main topic...

I think the VGA BIOS still needs to provide APM or ACPI management codes in both real and protected modes?

[[GDI]Raptor]

But does this BIOS update affect preformance in Windows, and whitch Matrox cards is updated in this release?

Matrox has also released new drivers, but only for the G200 MMS card. Perhaps they will release a new driver for G4xx & G550 soon!??

[Inray]

No WindowsXP speed increase here. This is not possible from a bios update. Anyway i believe Matrox people try to fix some P4/G550 compatibility issues they found.
BTW something changed in their G550 PINS format with this BIOS release. I used to increase standard VGA refresh rate for 60Hz modes but after bios update old values don't work anymore. Interesting !

regards

Sakis

[GehRehmee]

Man... if these cards don't stop getting slower and less featureful with each release, i might have to stop buying them. Hmm. wait a minute, that's not right...