DW Fan!!!!! said:Dident dave say something along the lines, that neither Nvidia or ATI would be the same as the ones which were seen at GDC, we have already seen that Nvidia's clocks are lower.
WaltC said:"ATI Radeon R420 Graphics ATI X800 Series
Manufacturing Process .13µ
Transistor Count(millions) 160
Core Speed(MHz) ~500
Memory Speed(MHz) ~500
Memory Interface 256bit
Rendering Pipelines 16
TMUs Per Pipeline 1
Peak Memory Bandwidth(GB/s) ~32.0
Pixel Fillrate(million pixels/sec) ~6,400
Texel Fillrate(million texels/sec) ~6,400
DX9 Pixel Shader Version 3.0
Vertex Shaders(version) 8 or 6 (3.0)
Memory Type GDDR3
FP / Internal precision 32
MAX Memory Size 512MB
AA Sample 8x
Native PCI Express Support Yes (R423)
Native AGP 8x Support Yes (R420)
DirectX Version Support 9.0c "
Hmmmm....wouldn't 500MHz x16 pixels = ~8 Gigapixel peak pixel fill rate, and ditto for megatexels...? ~6400 would indicate 400MHz clock, right?
Joe DeFuria said:True....I am going with the assumption though, that the R420 shader pipeline is no worse than the R3xx pipeline. I think that's a reasonable assumption...but yes, it is an assumption and could turn out to be wrong.
991060 said:Just a side note, folk at nvnews reported 6800Ultra works with a 200W PSU
http://www.nvnews.net/vbulletin/showthread.php?t=27394
Full rig stats....
Biostar IDEQ 200T
P4 3.2GHz Processor
1GB Corsair RAM
250GB Maxtor SATA Hard Drive
52x Plextor CD/RW
6800 Ultra
__________________
Brian Evans
Editor-in-Chief
3DGPU.com
http://www.3dgpu.com
brian.evans@3dgpu.com
engall said:6800U being 600/1000?
It is definitely impossible
Bad_Boy said:engall said:6800U being 600/1000?
It is definitely impossible
nothing is impossible
check this link...seems they have the same information albatron said in their press release.
you go to the site and look under the albatron card and you see...
"Albatronの「GeFORCE6800UV」。コアクロックは600MHz、メモリクロックは1GHz。ビデオ入力対応。冷却機構は1スロットの独自タイプ。発売は5月上旬の予定"
somebody has to get a translator for that site lol
991060 said:There're already some tests showed NV40 have more pixel shader throughput than R3XX(per pipeline per clock).
DW Fan!!!!! said:Bad_Boy said:engall said:6800U being 600/1000?
It is definitely impossible
nothing is impossible
check this link...seems they have the same information albatron said in their press release.
you go to the site and look under the albatron card and you see...
"Albatronの「GeFORCE6800UV」。コアクロックは600MHz、メモリクロックは1GHz。ビデオ入力対応。冷却機構は1スロットの独自タイプ。発売は5月上旬の予定"
somebody has to get a translator for that site lol
You gotta think though, they could have pulled that information straight from albatron's site so it would be the same, im in no clue either way :?
Bjorn said:And there should be room for some optimizations also. Sure, maybe not as much as with the NV3X core because of it's "twitchiness" but i doubt that it won't get faster then what it is now.
Bad_Boy said:DW Fan!!!!! said:Bad_Boy said:engall said:6800U being 600/1000?
It is definitely impossible
nothing is impossible
check this link...seems they have the same information albatron said in their press release.
you go to the site and look under the albatron card and you see...
"Albatronの「GeFORCE6800UV」。コアクロックは600MHz、メモリクロックは1GHz。ビデオ入力対応。冷却機構は1スロットの独自タイプ。発売は5月上旬の予定"
somebody has to get a translator for that site lol
You gotta think though, they could have pulled that information straight from albatron's site so it would be the same, im in no clue either way :?
yeah but they also could have gotten the information confirmed at whatever party that is.
thats why i say, somebody try to get that site translated or somthing.
Joe DeFuria said:991060 said:But we don't yet know how similar the shader cores are from different designs.
True....I am going with the assumption though, that the R420 shader pipeline is no worse than the R3xx pipeline. I think that's a reasonable assumption...but yes, it is an assumption and could turn out to be wrong.
Joe DeFuria said:Seiko said:Unfortunately I think it looks like a 475mhz speed for the Pro won't be that much faster than the NV40 Non ultra (assuming Nvidia can also push the NV40 with 12 pipes a little higher than 400Mhz)
It is interesting...ATI looks to be increasing their clocks for their "full, 4 quad" board relative to the 3 quad, while you're anticipating nvdia doing just the opposite.
While that may indeed turn out to be the case, nvidia would be more or less shooting themselves in the foot IMO. The closer it is in performance to the Ultra, the less they can charge for the Ultra.
12 pipes at 475 Mhz (X800Pro) would be within about 10% of the 6800 Ultra at 400 Mhz, in terms of pixel shading power, and about 20% slower in AA situations based on lower bandwidth of the Pro.
I think if EITHER the X800Pro or the 6800 non-Ultra is that close in performance to the 6800 ultra (while at the same time requiring less stringent cooling and power), the 6800 Ultra is all the sudden a very, very tough card to buy assuimg a $100-200 price difference.
It also doesn't bode well for the supposedly 600Mhz XT version.
While I agree that a 600 Mhz version sounds unlikely given this info...I've had that opinion from the start. I've always been expecting about 500 Mhz, which I think bodes well for the XT.
1) I think ATI will now be very evenly matched in performance compared with Nvidia.
I think this is the one area will ATI will end up having a decided advantage.
2) ATI will still retain a now very slight IQ advantage.
Agree.
3) ATI will loose in the features department.
Agree.
4) ATI will be ahead in the low power consumption department.
Agree.
5) Overall most sites will hand this round to the NV40 family due to it's features and comparable performance/IQ!
I think it will be about evenly split. Some seing the performance increase as not outweighing the lack of PS 3.0...others seeing it the other way.
All in all, both companies will be so evenly matched neither will be able to declare all out victory.
Agreed!
*Sighs, why oh why won't ATI unleash a genuine monster!
Honestly, I think a 500 Mhz, 16 pipe part is a monster....as long as it's paried up with at least 550Mhz+ memory.
People are making a big mistake thinking that the R420 is no more than a 16 pipe version of the R300TacTisIt said:Joe DeFuria said:991060 said:But we don't yet know how similar the shader cores are from different designs.
True....I am going with the assumption though, that the R420 shader pipeline is no worse than the R3xx pipeline. I think that's a reasonable assumption...but yes, it is an assumption and could turn out to be wrong.
What's the status on ATi's Extreme Pipelines?
Aren't they supposed to be a lot more enhanced than the R3xx generation? Unless they couldn't pull it off well enough in the timeframe, and hence why they went with more pipes....I mean, what I 1st heard was that they were gonna have either 8 or 12 Extreme pipes, which are far more efficient than the previous ones (considering they were very good in the 1st place, that would be quite an intrigueing feat). So is the reason that they now have more pipes due to them not being able to enhance the pipes further...
Or were they meant to be a 16 pipe architecture all the while and the extreme pipes were just BS rumours...heh
Hellbinder said:People are making a big mistake thinking that the R420 is no more than a 16 pipe version of the R300TacTisIt said:Joe DeFuria said:991060 said:But we don't yet know how similar the shader cores are from different designs.
True....I am going with the assumption though, that the R420 shader pipeline is no worse than the R3xx pipeline. I think that's a reasonable assumption...but yes, it is an assumption and could turn out to be wrong.
What's the status on ATi's Extreme Pipelines?
Aren't they supposed to be a lot more enhanced than the R3xx generation? Unless they couldn't pull it off well enough in the timeframe, and hence why they went with more pipes....I mean, what I 1st heard was that they were gonna have either 8 or 12 Extreme pipes, which are far more efficient than the previous ones (considering they were very good in the 1st place, that would be quite an intrigueing feat). So is the reason that they now have more pipes due to them not being able to enhance the pipes further...
Or were they meant to be a 16 pipe architecture all the while and the extreme pipes were just BS rumours...heh
Xtreme Pipelines still applies and is not fud.
Shader processing is enhanced over the current cards. In fact I was sure it was PS3.0 for a long time and FP32. Although recently things have been put out that make me doubt my own info.. but then again... If r420 is not PS3.0 then its mighty close and has some EXTREME PS2.0 processing capabilities.
Hellbinder said:People are making a big mistake thinking that the R420 is no more than a 16 pipe version of the R300TacTisIt said:Joe DeFuria said:991060 said:But we don't yet know how similar the shader cores are from different designs.
True....I am going with the assumption though, that the R420 shader pipeline is no worse than the R3xx pipeline. I think that's a reasonable assumption...but yes, it is an assumption and could turn out to be wrong.
What's the status on ATi's Extreme Pipelines?
Aren't they supposed to be a lot more enhanced than the R3xx generation? Unless they couldn't pull it off well enough in the timeframe, and hence why they went with more pipes....I mean, what I 1st heard was that they were gonna have either 8 or 12 Extreme pipes, which are far more efficient than the previous ones (considering they were very good in the 1st place, that would be quite an intrigueing feat). So is the reason that they now have more pipes due to them not being able to enhance the pipes further...
Or were they meant to be a 16 pipe architecture all the while and the extreme pipes were just BS rumours...heh
Xtreme Pipelines still applies and is not fud.
Shader processing is enhanced over the current cards. In fact I was sure it was PS3.0 for a long time and FP32. Although recently things have been put out that make me doubt my own info.. but then again... If r420 is not PS3.0 then its mighty close and has some EXTREME PS2.0 processing capabilities.