r9600pro first results & pics

[caater]

i got this powercolor radeon9600pro for testing yesterday.
i was kinda surprised it wasn't using the glued-down hsf like
reference boards did.. but rather it looked like radeon 9700.
in fact - it has the same ram as 9700pro.
but.. 0.13 process is simply amazing.. the core is sooo tiny
and no shim this time
guess they don't give a **** bout us crushing cores on el cheapo cards :P

and to be honest - it's oc results are rather disappointing.
max with def cooler was 465core/330mem.
with h20, 485core was ok, but it at 500core it did
hang right away.
but the card's soo incredibly cool.. i measured max 46C
at the backside when overclocking and running 3dmark..
r9700pro was ~65C w/ def cooler.
when watercooled, temps are like 30-34C. and that's like 10-15C
lower than 9700pro w/cooled.

so, according to temps - this is not a cooling issue. baby needs some voltage.
Anyone knows what def vgpu is for r9600pro?


first testresults with 2.8c proc and 465/330 rad were:
3dm2001se - 13165, 3dm03 - 4022

i'll look for w/cooling max and then bench with full cpu power





and.. ready to roll

[Estide]

----------------------------
but.. 0.13 process is simply amazing.. the core is sooo tiny
----------------------------
*wow* he's very small...thats right
But only 13XXX Points @3dmarks2001?!?
But 4XXX Points @ edmark2003 is like an (bad) 9700Np

Thx for Information

Estide

[caater]

stock r9700pro was 16k points at 2001SE with that cpu.
unfortunately i don't have a 9500 :/

Originally posted by Estide
----------------------------
but.. 0.13 process is simply amazing.. the core is sooo tiny
----------------------------
*wow* he's very small...thats right
But only 13XXX Points @3dmarks2001?!?
But 4XXX Points @ edmark2003 is like an (bad) 9700Np

Estide

[Estide]

The The 0.13µ prozess makes the card cooler (like palomino 0.18µ and t-Bred 0.13µ )

330 Mem with the Samsung 2.86NS chips isn't very high?!?

http://www.xtremesystems.org/forums/...threadid=14039 here is a thread to Mod 9600

Estide

[caater]

this card unfortunately doesnt have a semtech vrm.. but i'll search
teh net tonite

but 330 ram isnt too shabby, i have to admit it.. it artifacted like hell at 345 and eventually locked up..

so i think that reference 9600 boards definately had higher voltages.. and with voltmod.. agp power is the limit
and ofcoz.. as far as i can see, external powerconnector mod
looks also possibility

Originally posted by Estide
The The 0.13µ prozess makes the card cooler (like palomino 0.18µ and t-Bred 0.13µ )

330 Mem with the Samsung 2.86NS chips isn't very high?!?

http://www.xtremesystems.org/forums/...threadid=14039 here is a thread to Mod 9600

Estide

[antipop]

That's not a really good score in 3dmark, what is your benching system?

[caater]

p4 2.8C + aopen ax4c max + 2x256megs corsair @
dual ddr400 2 2 6 2.

as i stated before.. r9700pro w/def clocks 2001SE score was 16000... +- 10 points
and default 3dmark03 was 4800.. if that matters.

Originally posted by antipop
That's not a really good score in 3dmark, what is your benching system?

[jmke]

the high score in 3Dmark03 is due to the R9600 supporting more DX9 features then its older brother
but performance wise it still less then the R9500

anybody using the 3Dmark03 bench is just looking at some pretty pictures, because the endscore means nothing , thanks to Futuremarks greed for $$ and nvidia/ati's greed for market domination.

yippikajee

[Estide]

Originally posted by jmke
the high score in 3Dmark03 is due to the R9600 supporting more DX9 features then its older brother
but performance wise it still less then the R9500

anybody using the 3Dmark03 bench is just looking at some pretty pictures, because the endscore means nothing , thanks to Futuremarks greed for $$ and nvidia/ati's greed for market domination.

yippikajee

Youre right, but i Think that 3dMark is good to compare the different cards. To see what performance have your card is another thing.

estide

[saaya]

the core is smaller because the 9600 only has 4 pipes afaik, less transistors...

about ocing, ati lowered the gpu voltage a lot to reduce the ocability of the cards. they probably want to have some room left to release an updated rv350 in the future that will run faster.

only 13K with that system oced and all? fck thats low... thats slower than a ti4200 i hoped it would perform better : /

[jmke]

Originally posted by Estide
is good to compare the different cards.

if the compared cards use the same driver and are from the same "nest" (ATI/nVidia/SiS/etc.. etc)

that way you don't have to worry about any unfair optimalisations!

[caater]

or those darn OEMs didnt stop whining bout high power
consumption on low end cards..

and bout voltage.. i think they (ati) did the opposite way.
increased voltage of press cards, so they oc high and look good.

Originally posted by saaya

about ocing, ati lowered the gpu voltage a lot to reduce the ocability of the cards. they probably want to have some room left to release an updated rv350 in the future that will run faster.

[saaya]

yeah good possibility... or they only send cards to people like Kyle and the pre samples didnt oc any better

[antipop]

Originally posted by saaya
yeah good possibility... or they only send cards to people like Kyle and the pre samples didnt oc any better

ROFL even with a 30MHz over stock core he would call that a great oc'er. He could help nvidia :smileysex

[saaya]

roflmao but he has an affair with bfg tech it seems

[wierdAl]

em... did you try to up the voltage on the gpu?, becouse i think it could do better... it's 0.13 micron.... it has to go higher...

[caater]

i'll up my cpu just as my water can handle and then benchmark it..
and then.. start lookin for voltmod possibilties :p
i have one 10k pot lying around

Originally posted by wierdAl
em... did you try to up the voltage on the gpu?, becouse i think it could do better... it's 0.13 micron.... it has to go higher...

[caater]

i ran 3dmarks now with ~3.6ghz proc.. not too shabby..
3dmark2001se - 14337
3dmark03 - 4168

i'll measure some voltages.. etc..