Parhelia Sharkmarks

[ganyaik]

Yep, I think that exactly that would be the case, but I think we won't know it in the near future, because 300MHz without watercooling(or whatsoever) seems quite impossible to me. Anyway, I hope I'm wrong with this.
Have any of you checked the article at http://www.athlonxp.com/modules.php?..._parhelia.html ? Gf4 does 126fps in Sharkmark and an OCd Parhelia does 186! If Sharkmark doesn't use any Matrox specific features then it definitely shows the strength of the card.

[Voyager3D]

Don't think 300MHz will be a posibility, even with water cooling...
At 220MHz Parhelia used about 30W, AGP max is 40W...
So my gess would be that 275MHz would be maximum, if we can't give Parhelia some more juice...

I think we will see 260MHz with watercooling (maby 270MHz)..

[frankymail]

Wait a sec: (30-4)/220*300+5=40.5 watts… that’s running at 300 MHZ core/350 MHZ mem, without any voltage increase; to be stable at those speeds, you’ll need : better cooling and higher voltages, so to get very high speeds, you’ll definitively need a non-AGP power source (wires from the PSU to the card directly, probably somewhere on the power regulation “sector” of the card). Those kinds of mods have been done in the past for G3/G4/R8500, and is standards on the R9700/V5500, so nothing out of the realm of possibilities…

[ganyaik]

BTW, does power consumption grow linearly with the clockspeed? If it does then we'll have still chance for a good 33% increase.

[pgde]

Re: Re: Parhelia Sharkmarks

Originally posted by pgde


Hi:

Where on the CD is Sharkmark? I can't find it???

Thx

Well, I answered my own question. It is only available to editors. See http://forum.matrox.com/mgaforum/For...ML/001247.html

P.

[zandor]

Who's to say your mainboard will actually max out at 40W though? I'd bet most designs will allow for at least a little extra just to be on the safe side, and some will handle a lot of extra current.

[Greebe]

AGP power spec peters out around 40watts on all non "Pro" spec slots.

[ganyaik]

Well, question is that how many watts the card would draw at 0MHz. I think it wouldn't be 0W. So the equation would look somehow like this for a normal card: (base)+220*(watts_per_MHz)=~30w according to Voyager3D. So I think from AGP spec point of view 33% clock speed increase is quite feasible. From P point of view... Who knows?
BTW, does the OCing of the memorys affect power consumption?

[Voyager3D]

If that is correct then..:

30W/220MHz = 3/22 = 0,13636363... W/MHz

then:

200MHz ~ 27W
210MHz ~ 29W
220MHz ~ 30W
230MHz ~ 31W
240MHz ~ 33W
250MHz ~ 34W
260MHz ~ 35W
270MHz ~ 37W
280MHz ~ 38W
290MHz ~ 40W
300MHz ~ 41W

But it is to good to be true...
290MHz would be nice tho..

[R.Carter]

Hang on a second.

Has Matrox released actual power consumption figures for the current 128MB Parhelia?

The AGP 2.0 specification does indicate that the card can draw up-to

VCC3.3 3.3v @ 6.0A = 19.8W
VCC5 5v @ 2.0A = 10W
VCC12 12v @ 1.0A = 12W

from the AGP connector according to table 4-14. So the card actually has 41.8W to play with.

Of course, Matrox has indicated that you would need to bump the voltage to really overclock it.

[Voyager3D]

Originally posted by R.Carter
Hang on a second.

Has Matrox released actual power consumption figures for the current 128MB Parhelia?

No.. It was Haig that said in a o/c thread at forum.matrox.com that is used ~30W...

from the AGP connector according to table 4-14. So the card actually has 41.8W to play with.

Nice..

[Enak]

Don't forget that your power supply needs to have this power handy...

[borat]

how would you volt mod a parhelia then?

[Enak]

I think a bit of soldering might be required...

And the loss of warranty

[Kruzin]

Forget about voltage modding the card.
Doing so will fry the chip in a matter of weeks.