G400 Vanilla 32MB DH @200 core!

[dneal]

Yeah, I got my G400 Vanilla to do 200MHz core and 200MHz mem.

Unfortunately since my 6ns RAM won't do higher than 206 I can't see much benefit. It's actually slower than 170/220 (with higher memory latencies.)

Imagine DDR on this thing. Can you say bubye Geforce and everyone else?

I played Unreal Tournament at 200/200! Wow.

[birdy]

core!

[Nuno]

Don´t take me wrong, but I highly doubt it. Even the MAXes that have higher yelds have big trouble with clock speeds higher than 180 (and not all of them are able to achieve speeds even close to that). I find it very hard that a G400 core could do 200 Mhz without any kind of supercooling.

Not that you´re trying to cheat us, but you must be doing something wrong. MGAtweak must have its share of bugs (I assume you´re using MGAtweak, it´s the only way you could have change the dividers) and sometimes some PLL/dividers combination could not work at all. My guess is that you are running at 150/200.

[Maniac]

yeah use g4set to verify...

[dneal]

X

There ya go. I think that means 199MHz core. I am using MGATweak and the only divider I changed was the MClk to 2.5, all dividers are 2.5.

As for super cooling, consider this. I purchased a large YSTech case fan (blows over 100cu/ft air/min.) My computer sits on the windowsill. It is winter outside, about 35 degrees F max. My case is 44 degrees F at times. I've touched the G400 (only the heatsink) and it's only warm when at these high speeds, at all speeds for that matter.

I've seen OBVIOUS performance improvements at 174/218. 200/200 is lesser because of the memory bottleneck.

[Maggi]

ouch ... that's not core ...

that's <FONT SIZE=4> hard-core <FONT SIZE=2>



------------------
Cheers,
Maggi
________________________

Working Rig:
Asus P2B-DS @ 103MHz FSB
Double Pentium III-450 @ 464 MHz
4 x 128MB CAS2 SDRAM
Matrox Millennium G400 32MB DualHead
Eye-Q 777 (22" with 127kHz) on primary VGA
Nokia 445Xi (21") on secondary VGA


Home Rig:
Asus P2B-S Bios 1010 @ 100MHz FSB
Celeron 333A @ 500MHz
2 x 128MB CAS2 SDRAM
Matrox Millennium G400 32MB DualHead @ 150/200MHz
CTX VL710T (17")
and a brand new Pioneer 303S SCSI-DVD

[rickn]

actually, i dont believe it is 199. When you set the system PLL that high, it actually slows the card down. Maybe GBM knows.

[Maggi]

Could you make a cross-check and set the PLL to 300MHz and the dividers to 1.5 ???

Should be the same result ...

[Nuno]

Yeah, and 199 IS NOT 200!!



Now really, I tried those settings with MGAtweak and I got a real bad crash... so it´s doing something, just not sure what it is.

Do as Maggi said, set the PLL to 300Mhz (that´s much more within spec) and the dividers to get the same 200/200/200.

If that´s really doing 200 Mhz, you must be the luckiest man on earth (ok ok, and the luckiest matrox user on earth, do you buy it?)

[CrazyBee]

Can I say jealosy?!

Seems the poeple that bought the Max are a bit upset that they cant clock theres teh same.

Since even within a yield there is a huge variance it is totally possible. I have had experience with this with overclocking CPUs.. I have a 233 MMX that can go up to 294 at BELOW normal voltage with a heatsink/fan on it. Just like how most of teh Celeron 366A's did not overclock well below week 22 but every once in awhile there is a Week 7 or 12 that does it just fine.

So many factors in die fabrication its totally possible (within a small degree).

Maybe he should do a graphics test at normal speed, then, at overclocked speed and calculate the difference and see if it corrosponds with the MHz change.

CB

------------------
Abit BX6 Rev.1
Celeron 366A PPGA @ 566, 2.1v
192 meg RAM, CAS2
13.0 gig Maxtor 4320 HD
6.0 gig Maxtor (removeable drive bay)
HP8110i 4x2x24
Creative DVD 5x
Creative DxR2
SB Live! Retail
USB ZIP 100
G400 32MB DH 5ns RAM
Two KDS 17" Trinitron monitors

[dneal]

Well, I took MGATweak to 1.5x like ya'll said. It didn't like that at all. I could do 166 but 175 crashed hard.

Then I tried 2.0x, guess what, maxxed out at 190Mhz, 195 died, 200 died. Probably would do 192 or something.

Just wish I had some 5ns SGRAM, or 4.3 or 3ns.......heh heh.


3DMark 99 MAX Pro...2x
1024x768x32bit

160/200/160=4986
175/175/175=4687
185/185/185=4821
190/190/190=4971
Fillrate =186.6
Fillrate w/ multi=214.2


My system:

P3-558 L2 Cache Latency 4
192MB RAM, Cas 2, 124MHz
Matrox Millenium G400 32MB Vanilla DH\
Creative SBLive Value
Quantum Fireball Plus KA 13.6
Quantum Fireball ST 4.3
3Com Fast Etherlink XL
Toshiba 6X/32X DVD
Sony CDU928E CDR

[Nuno]

Hmmm... that´s strange. Something is going wrong. If it really does 200 Mhz, it should do it no matter what dividers it uses.

Why don´t you post it in the <a href="http://forums.murc.ws/cgi-bin/forumdisplay.cgi?action=topics&forum=MGA+Tools&num ber=6&DaysPrune=5&LastLogin=">MGA Tools forum </a> ?

[This message has been edited by Nuno (edited 13 December 1999).]

[dneal]

Just a note, I think the reason 190/190/190 is slower than 160/200/160 is due to memory constraints. There is simply not enough memory bandwitdth.

[Nuno]

Try using something like 1280x1024x16 16bit z double buffered. It should be less memory bandwidth intensive and scale better with pure core speed.