Parhelia Sharkmarks

[az]

Hope it's something to really make you happy

AZ

[GT98]

Nice Results! kinda makes you wonder what they could have done with the card if they went the ATI Route with the higher clockspeed/external power supply!

Time to order that cooljag CPU cooler and some RAM Sinks

[Nuno]

Does anyone notice that the parhelia doesn´t seem to be very bandwidth limited? Increasing the core frequency brings better results than memory overclocking. Hmmmm.....

[TdB]

Does anybody have a download link?
I want to try it with my kyro2.

[thop]

we have a o/c tool! since ant has it i'm pretty sure the betaboys have it too. so the extra cooling wasn't for nothing. deny all you want

[xortam]

thop ... everyone probably knew that the BBs would try o/c'ing with their modified boards. That doesn't mean that they weren't honest in their statements that they modded their cards for extra ruggedness. They're going to have to be running at stock speeds during their beta testing. A lot of MURCers seem to be drawn to tinkering with their HW and the BBs are by no means the exception.

[GT98]

Originally posted by thop
we have a o/c tool! since ant has it i'm pretty sure the betaboys have it too. so the extra cooling wasn't for nothing. deny all you want

Well the BBs are the Goverment when it comes to Nuclear Weapons or UFOs...

"We cannot Confirm or Yet Deny that Fact"

[Stringy]

well this is some nice info...
so we now know that the P doesn't have a G400 like clock setup, ie one PLL clock and MEM and Core are a divisor of it...

looks like we have independant clocks for Mem and Core...
Nice Job..
250 would be nice, 10% boost... extra cooling and alittle luck might yield 15%...


thanks for the info Ant...
Craig

[Maniac]

Looks like P. scales almost perfecty in this benchmark with the core/mem increase. Increasing the core/mem roughly 12-13% gives a 12-13% gain in score. Now lets see how high this puppy can be o/c

[Reverend Maynard]

Don't we need some sort of reference numbers to compare to the set you posted? I am not familiar with the program you used or the number it generates. Basically what I am getting at is this, in the first set of numbers you gained about 6 FPS going from 220 core to 230 core...is this good, is it bad, how are we to know?

Basically if someone says "I OC'ed my video card and gained 700 points in 3dMark2K," I am going to have an understanding of the speed gained. But going from 164 FPS to 170 FPS in a benchmark I've never heard of kind of leaves me scratching my head. Any chance we can see a more standardized benchmark? Regardless of various bias' that may exist, it would be nice to have a frame of reference when looking at the overclocked numbers.

[Ant]

If you guys suggest a benchmark I'll run it although it won't really provide a point of reference unless you want it compared to my G400, the point of reference is the default Parhelia speed. Sharkmark is good as it utilises the Parhelia fully, the benchmarks were to show how it scaled, not as a comparison against any other boards that was not the point of the excercise. If one of you wants to send me a GF4 though I'll do the tests

[Sinistral]

- Ant, you can still give us the results of 3DMark2001. Comparing, core clock scalability with that of the Parhelia alone. As 3DMark is well reknowned. People can independantly compare the results to other cards.

What i mean is:

275:220 = 5500 3DMarks
275:230 = 6000 3DMarks
...

280:220 = 5700 3DMarks
300:300 = 12000 3DMarks -

[Ant]

If we can get a consensus of opinion as to the key benchmarks that will give us most information I will give them a good run through and update with new driver releases. I'd prefer that to be one or two benchmarks that I can thoroughly test over a range of speeds.

[DukeP]

GLExcess/XSMark v1.2

This is a really nice looking demo and bench utility, which I always find is important, as one have to sit and wait for endless runs anyhow.

http://www.glexcess.com/xsmark/

This and then 3dbench2001se.

Happy doodling, dont burn any bridges!


~~DukeP~~

[ganyaik]

Hi,
Ant, could you please run 3dmark2k1 for e.g. and send the detailed results? This way we would be able to compare the values to other cards also. I'm particularly interested in triangle throughput and fillrate changes. Also could you please give a go to the newest drivers? Maybe they'll have some positive effect on the performance too.
Thanx in advance,
Chris