RV350 Pro benchmarks - Don't wait any further.

[Uttar]

Hey,

Just recieved some RV350 Pro benchies from a reliable source.
Strangely, it seems the core clock is of 375Mhz, instead of the standard 400Mhz. Not sure if ATI modified their specs or what. Or maybe the card is at that frequence by error. No idea...

---

Test Setup:

P4 3G, E7205, 1G Mem,
EV350: 378/303 (core/memeory, powetstrip measured)
Driver ver: 6.14.1.6307

---

RV350 Pro

3dmark03 1024 NoAAAF 3170
GT1 123.8
GT2 19.2
GT3 17.9
GT4 18.4
CPU Score 622
Fill Rate (Single-Texturing) 893 MT/C
Fill Rate (Multi-Texturing) 1440 MT/C
Vertex Shader 9.8
Pixel Shader 2.0 25.9
Ragtroll 12.2

3dmark03 1280 NoAAAF 2285

Aquamark 2.1 1024 NoAAAF 62.1
1280 NoAAAF 39.8
1024 4xAA 8xAF Quality 27.9
1280 4xAA 8xAF Quality 18.4
1024 4xAA 8xAF Performance 34.5
1280 4xAA 8xAF Performance 22.7

Commanche 4 1024 NoAAAF 53.54
Demo 1280 NoAAAF 51.07
1024 4xAA 8xAF Quality 31.93
1280 4xAA 8xAF Quality 23.66
1024 4xAA 8xAF Performance 34.24
1280 4xAA 8xAF Performance 25.11

QuakeIII 1024 NoAAAF 245.2
Demo 1280 NoAAAF 166.3
1024 4xAA 8xAF Quality 130.2
1280 4xAA 8xAF Quality 86.3
1024 4xAA 8xAF Performance 136.5
1280 4xAA 8xAF Performance 88.5

SeriousSam2 1024 NoAAAF 78.7
Extreme Quality 1280 NoAAAF 70.6
The Grand 1024 4xAA 8xAF Quality 43.5
Cathedral 1280 4xAA 8xAF Quality 30.2
1024 4xAA 8xAF Performance 60.8
1280 4xAA 8xAF Performance 43.2

Leaked E3 640 Highquality 81.2
DoomIII Altha 1024 Highquality 40.4
V0.02 1280 Highquality 26.7

3dmark01 1024x768 11408
1280x1024 8850
1280 4XAA+8XAF Quality mode 4495

---

And in case anyone is interested, here is a little comparaison I added by finding other scores around the web.
Compared to a FX5600U & R9500P on a different setup:

Quake 3, 1024x768 + 4xAA & 8x AF ( balanced on the FX, performance on the Radeon)

R9500P: 148.1
R9600P: 136.5
FX5600U: 136.1


Uttar

[ReDeeMeR]

Is R9600P going to be cheaper then R9500P ???

[Dazz]

Well on the plus side C4 is more playable on the Radeon 9600Pro :)

Right from what i can tell w/o FSAA & AF it runs faster then the 5600U, w/ FSAA & AF it's runs pretty much the same in most games very strange indeed. The most noticeable diffrence is on Comanche 4 where it's 10fps faster then the 5600U about 20% @ 1024 & 30% at 1280 i believe.

53 vs 44fps 20% right?
51 vs 34fps 35% right?

Once you add FSAA & AF it's only has 3fps lead.

[sebazve]

Originally posted by ReDeeMeR
Is R9600P going to be cheaper then R9500P ???

i guess so. This chip is not so complicated as r300 also its on .13 process

[nagus]

remember... the final card will be clocked @ 400Mhz core.

[Dazz]

I thought the Radeon 9600Pro was slower in SS SE however i was compairing the wrong map. god when will we see reviews of them. I don't know about you but i think it handled DooM III quite well 40fps @ 1024, what do the Radeons 95/9700 get? 40fps is quite as OpenGL runs alot smoother then D3D so 40fps an't bad think of when the first TNT2U cards came out with Q3 :D

[Dazz]

Originally posted by nagus
remember... the final card will be clocked @ 400Mhz core. ATi said tey wern't sure how high they were going to be clocked thats why the delay.

[Joe DeFuria]

Ati's official stance on ALL their 9200/9600 boards is that clock speeds are not final.

They did say, however, that they expect the 9600 pro core clock to be "around 400 MHz." So I would expect at least 375.

[surfhurleydude]

DoomIII Altha 1024 Highquality 40.4

WTF? I don't even get that high of a score with a 2.66 system clocked at 3.10 ghz with a Radeon 9700 Pro!

[Jandar]

remember, the rv350 and r350 have improved cores.

much much better at rendering stencils versus the r300 cores:afro:

[surfhurleydude]

Ya, but the RV350 only has HALF the pipelines.

[GlowStick]

Hi, id like to 'claim' that i know a bit about the doom3 alpha.


There are two possibilities.

1) Was not the alpha but a highly optmized version (very likely)
2) They made the numbers up (unlikely, if they were fake, they did a bad job)
3) Due to the Alphaness of D3 benchmarking is very inaccurate

Now for point 1 that is very likely that someone with a RV350 would happen to have a recent build of Doom 3. I have yet had Nvidia or ATI come to my door and say "Hey, why dont you try this new card for free, befor anyone else has it."

The scores could be made up. Benches are posted of almost every videa card out there, the 9700p and GFFX cannot get close to that on the alpha. Though they surpass it on the versions ATI + Nviida have.

A third thing about the fps in the doom3 alpha is the compete inconsistancy of FPS. When certin sounds are played on any videa card there is allways a odd fps freeze that lasts about 10 seconds. Reason (its a alpha)

[Hellbinder]

guys the Rv350 has improved Stencil support. It also fully used HyperZ while doing Stencils and other things.

The R300 is going to run the Final Verion of DOOM-III faster than the Rv350. But not as fast as the R350. The R300 drivers do not currently support Doubble-sided Stencil on the Alpha leak of doom-III. Also the current build of doom-III is 100x more compex st this point in some areas. More optomized etc.

The final numbers will be much different trust me. otherwise the Rv350 will be faster than a GFFX in doom-III heh.. now imagine that.

Wait for the final numbers and stop benching and giving any weight to the Alpha leak.

[GlowStick]

Originally posted by Hellbinder
guys the Rv350 has improved Stencil support. It also fully used HyperZ while doing Stencils and other things.

The R300 is going to run the Final Verion of DOOM-III faster than the Rv350. But not as fast as the R350. The R300 drivers do not currently support Doubble-sided Stencil on the Alpha leak of doom-III. Also the current build of doom-III is 100x more compex st this point in some areas. More optomized etc.

The final numbers will be much different trust me. otherwise the Rv350 will be faster than a GFFX in doom-III heh.. now imagine that.

Wait for the final numbers and stop benching and giving any weight to the Alpha leak.

The RV350 beat the gffx in d3. That dose go along with the posts on 3dgameman.com that say the RV350 beats the gffx accross the board.

So in the the resutls from Nvidia (witch have a better build than the alpha) and the results show that hte GfFX beats the R300.

But your post states that.

GFFX<RV350<R300 in d3

but the benches we have show R300<GFFX

basicly to sum it up for the RV350 to beat the GFFX, it has to beat the R300 also.

[The Baron]

Glow, Hell's right to the extent that nobody's really sure how well Doom3 is going to perform. The alpha is from JUNE 2002, and since the game won't be out until October 2003 at the earliest... yeah. Wow. Let's benchmark everything with a version that comes out a year and a half before the final thing and somehow give it merit! GOOD IDEA, CHARLIE! ;)

[Uttar]

If you want to compare some other numbers too...
The 9500P numbers come from the FX5600 hardware.fr's preview ( not a bad preview, really, if you understand french )

9500P
GT2: 21,5
GT3: 20,9
GT4: 23,1

9600P
GT2 19.2
GT3 17.9
GT4 18.4


Uttar

[poursoul]

never been to 3dgameman and don't mean any bad will

I always wait to see many benchmarks from many sites before concluding anything like "The RV350 beat the gffx in d3. That dose go along with the posts on 3dgameman.com that say the RV350 beats the gffx accross the board." Believe me i'm goin' for the r350 core cards all the way here, but untill they post benchies along with everyone else it belongs in the rumor mill.

[GlowStick]

Originally posted by The Baron
Glow, Hell's right to the extent that nobody's really sure how well Doom3 is going to perform. The alpha is from JUNE 2002, and since the game won't be out until October 2003 at the earliest... yeah. Wow. Let's benchmark everything with a version that comes out a year and a half before the final thing and somehow give it merit! GOOD IDEA, CHARLIE! ;)

Riiight, but when he says the RV350 will beat theGFFX in doom 3 is comming from..........

Sorry to burst your bubble, RV350 beat NV30 in Doom 3

No.

[The Baron]

Read the post.

The final numbers will be much different trust me. otherwise the Rv350 will be faster than a GFFX in doom-III heh.. now imagine that.

he didn't say that the RV350 beat the GFFX... heh.

[GlowStick]

Originally posted by The Baron
Read the post.



he didn't say that the RV350 beat the GFFX... heh.

ah, i apologise for that misight.

But then again i just realized another thing

Taken from uttars first post "DoomIII Altha 1024 Highquality 40.4"

The Gf FX gets 49 on recent builds useing silimiar config(file). Then again i have no idea what computer that is being tested on.

[Ady]

that RV350 Pro score for default 3dmark03 bench is very close to on par with GFFX ULTRA.

[StealthHawk]

so when are review sites in English going to publish RV350 reviews?

and when are these cards going to show up in stores?

[Dazz]

Originally posted by StealthHawk
and when are these cards going to show up in stores? Sometime in April.

[Uttar]

Originally posted by Dazz
Sometime in April.

Correct. Probably not in late April, though.

BTW, using those numbers, here is yet another comparaison:
Comanche 4, 1024 4xAA 8xAF

R9600P: 34.24
R9500P: 36.7
FX5600U: 27.3

Numbers other than the R9600P's come from Anandtech.


Uttar

[Dazz]

I think ATi made the right move with the higher fill rate, after all when you think about it the 5600U has close to the same bandwith as the 5800 yet is 50% slower due to the pixel pipelines an slight reduection in core speed. Also the RV350 can be cheaper as it's memoy is slower and can reduce the cost more that and it's cheap & simple 0.13m design.