Next generation RV770 GPU will show up in the end of the second quarte

[dos1986]

AMD lineup for GPU, CPU and notebook platforms revealed

Latest news
Monica Chen, Taipei; Joseph Tsai, DIGITIMES [Tuesday 11 March 2008]

AMD has recently notified its partners that the company expects its share in the discrete graphics card market to rise after its next generation RV770 GPU shows up in the end of the second quarter, growing from the current 35% to reach 50% by the end of this year, according to sources at graphics card makers.

AMD has completed DVT sampling for the B3 stepping Phenom processors, and related triple-core and quad-core parts are expected to appear in the channel in April, AMD stated at a recent media event in Taipei, adding that its 45nm quad-core processors, Shanghai for servers and Deneb for desktops, are both on schedule to launch in the second half of 2008.

AMD's 45nm processors have already entered EVT testing, according to sources at motherboard makers, who added they expect to receive samples by August or September.

AMD is planning to launch a notebook processor family codenamed Swift in the second half of 2009. The processors will come in two versions – dual-core Black Swift and single-core White Swift – both will incorporate a graphics processor. Swift and its companion chipset will form the Shrike platform and will support DDR3 memory, noted sources at notebook makers.

Nvidia had no comment regarding AMD's confidence in claiming a 50% graphics market share.

http://www.digitimes.com/NewsShow/MailHome.asp?datePublish=2008/3/11&pages=PD&seq=210

 

[spoonboy]

What makes me curious/confused is that if the rv770 is supposed to be 40%-50% faster than rv670, as rumoured although the latest news is 40%:

http://www.nordichardware.com/news,7470.html

...and are supposed to have a very high core clock speed AND twice the shaders and TMUs, then surely that should end up being more than 100% faster? (twice the hardware, large core clock bump). Then surely we are missing something here, either that the rumoured improvement of '40%' is in effect refering to the increase in core clock and not the actual improvement in performance, or that the specs of the chip are way off what we have been lead to believe i.e. not twice the shaders as per the most recent rumours I've read, but more along the lines of the 50% shader increase and 100% TMU increase that was rumoured some weeks ago but apparently 'debunked':

Rumour: http://futuremark.yougamers.com/forum/showthread.php?s=f13510ddcc85ab1dca1c340f0ed08ec7&t=74988

'Debunking': http://forums.vr-zone.com/showthread.php?t=244004 (refering I believe to the 100% / 50% rumour I allusioned to above)

Either way, the 40% or 50% or whatever improvement we've been hearing about doesn't add up, it's either AMD trying to pull the wool over our eyes by saying that the clocks of a much more powerful chip are clocked faster than the rv670 by this amount, keeping us and Nvidia in the dark until late Q2 / early Q3 when the cards make their way onto shelves in quantity, in a form more powerful than was previously (i.e. now) imagined, or, that a newer architecture with more shaders and TMUs on board and higher clocks can only improve on the current 38xx series by 40%, which would be a shame to say the least.

So what's the deal ATI?

 

[jnava121]

20 bucks it still won't run crysis on very high... lol crysis eats silicon for breakfast

 

[spoonboy]

lol, im not giving you 20 bucks! Next thing you'll be wagering me that grass is green and the earth revolves around the sun.

 

[reddozen]

could be an exciting year for graphics...

 

[SpinachEater]

What's surprising is AMD/ATI is claiming they will gain 15% graphics market share by years end. From 35% to 50%. Thats a pretty bold claim.

http://www.tomshardware.com/2008/03/11/amd_lineup_for_gpu_cpu_and_notebook_platforms_revealed/

It is indeed.

 

[T8RR8R]

Yes it should be an exciting year for graphics since Intel is going to really rejoin the GPU market, ATI is making a comeback and Nvidia is starting fights with Intel.

I'm guessing that these 40% increase claims will really end up being about 15% for most games and only 40% for some games that already ramp up with every little tweak and upgrade. Although I have high hopes too

I've seen triple SLI Ultra's and dual 3870X2's get eaten by Crysis, so what it's gonna take?

 

[betunn]

I've seen triple SLI Ultra's and dual 3870X2's get eaten by Crysis, so what it's gonna take?

Is it possible to play it on max settings then?
Anyway what's the benefit of running the game in those settings?

 

[spoonboy]

After reading tweakguide.com and my own testing, there are some settings you could quite happily leave on medium with the rest on high or very high, and the visual quality wouldn't be that affected, and you'd have alot higher fps.
If you can live without motion blur, post processing could stay on medium and never move any higher, and if you wanted the 'god-rays' of the very high setting (not the only thing very high changes btw), then there's one command you can enter into the files to force it without losing very much fps at all.
Also, shadows above medium really hacks into your fps and you can easily live without them above medium.
Then particle effects, which apparently drags on your cpu and gpu, dont exactly look 'horrible' on medium, then volumetric effects, which basically controls how fluffy the clouds are. That can also stay quite happily on medium.

The rest on high, run it under dx9 in vista or better yet, xp, don't even think about ANY AA or AF, hey presto any half-way decent dx10 card can run it.

Point is alot of cards can give you a nice experience all the way through the game at a High/medium mix of settings, so sticking all at high or very high is a waste of gpu power if it means your playing at under 25fps or something. So all on high=bit pointless really.

"Is it possible to play it on max settings then?"

Without AA a single card can do it but not above 1280x1024. I think I read forum posts saying people can get 1280x1024 xp hack all very high with 2xAA on a 8800 gtx or ultra only having to take off aa in the second half of the game where things get a bit 'spectacular'.

 

[mactronix]

Got to agree 100% with you there spoonboy,
Lots of people get a new card and think its crap because it cant play the latest games all on "MAX".
Its pretty ridiculous really you would think more people would spend the time trying different combinations of settings to get a good balance/experience from there hardware.
It doesnt take a lot, anyone can get a basic understanding of what does what in the control panels of the various apps that control the finer points of a graphics card. All that's needed is half an hour on Google or a forum even the old fashioned buy a magazine approach all work well.
As with the whole ATI vs Nvidia thing its going to be down to individual likes and dislikes as to what they can live with loosing as far as image quality goes, So while some of the major settings will be the same for most, the fine tuning will be down to the individual.
I have personally spent some time in the past tweaking the older X1*** cards and its amazing the results you can get with a few minutes tinkering.

mactronix

 

[TheGreatGrapeApe]

What makes me curious/confused is that if the rv770 is supposed to be 40%-50% faster than rv670, as rumoured although the latest news is 40%:

http://www.nordichardware.com/news,7470.html

...and are supposed to have a very high core clock speed AND twice the shaders and TMUs, then surely that should end up being more than 100% faster? (twice the hardware, large core clock bump).

It all comes down to bottlenecks. In some things it may be 150% faster, in others maybe 20% depending on what the limitation is. If the ROPs remain the same, and everything else increases, then the max output on easy stuff remain only slightly changed, however on stuff that used to choke the old architecture you'd see larger performance gains.

Personally I think estimations on performance of such an early product are pulled out of people's sphincters.

'Debunking': http://forums.vr-zone.com/showthread.php?t=244004 (refering I believe to the 100% / 50% rumour I allusioned to above)

The interesting thing about that design is the size of the chip and the lack of a shim/socket. The problem though is that the image is artificially stretched or squashed so guessing the actual size of things is diffiicult.

Either way, the 40% or 50% or whatever improvement we've been hearing about doesn't add up, it's either AMD trying to pull the wool over our eyes by saying that the clocks of a much more powerful chip are clocked faster than the rv670 by this amount, keeping us and Nvidia in the dark until late Q2 / early Q3 when the cards make their way onto shelves in quantity, in a form more powerful than was previously (i.e. now) imagined, or, that a newer architecture with more shaders and TMUs on board and higher clocks can only improve on the current 38xx series by 40%, which would be a shame to say the least.

OR... people are starting to get floppy neck run-around-like-theirheads-cutoff syndrome again.

So what's the deal ATI?

What the deal with intel, nV and S3?

 

[TheGreatGrapeApe]

PS there's a bazillion Crysis threads already, go to them for advice.

 

[mactronix]

Yep
I thought people knew better than to expect twice the performance just because it has twice the shaders/TMU's etc. look at the differance between the ATI and Nvidia offerings, there was actually a thread on here the other day that basically said why dosent a R600 card eat a G80 alive as it has more of this and that.
Its the same thing here, the cards will be differant so why should they scale properly ?
mactronix

 

[TheGreatGrapeApe]

Yeah and the other thing is what are they testing on and with what kind of drivers to get these magical numbers?

Geez the G80 and R600 launch with crap driver support and weaker than potential perfomance and people are already talking about performance number on a not finalized product?

 

[dos1986]

What's surprising is AMD/ATI is claiming they will gain 15% graphics market share by years end. From 35% to 50%. Thats a pretty bold claim.

http://www.tomshardware.com/2008/03/11/amd_lineup_for_gpu_cpu_and_notebook_platforms_revealed/

Nvidia's comeback...

http://www.vr-zone.com/articles/Nvidia_Prepares_55nm_G92_Against_AMD_RV770/5645.html

VR-Zone has learned that Nvidia is preparing to shrink G92 process technology from 65nm to 55nm to better compete against the upcoming RV770. DigiTimes today reported that RV770 will come along by end Q2 which is about the same timeframe as the 55nm G92. Most probably, you will see RV770 based mobile parts first as notebook makers have already received RV770 samples for testing. Right now, we are able to confirm some 55nm G92 mobile parts coming up to pit against the 55nm RV770. Most likely, there are 55nm G92 for desktop parts too as Nvidia in desperate needs to lower cost and improve yield in the next couple of months ahead to stay in competition. In any case the 55nm G92 even with higher clocks can't match up against RV770, Nvidia will have to push ahead the schedule of next generation GT200.

 

[spoonboy]

It all comes down to bottlenecks. In some things it may be 150% faster, in others maybe 20% depending on what the limitation is. If the ROPs remain the same, and everything else increases, then the max output on easy stuff remain only slightly changed, however on stuff that used to choke the old architecture you'd see larger performance gains.

Personally I think estimations on performance of such an early product are pulled out of people's sphincters.

'Debunking': http://forums.vr-zone.com/showthread.php?t=244004 (refering I believe to the 100% / 50% rumour I allusioned to above)

The interesting thing about that design is the size of the chip and the lack of a shim/socket. The problem though is that the image is artificially stretched or squashed so guessing the actual size of things is diffiicult.

Either way, the 40% or 50% or whatever improvement we've been hearing about doesn't add up, it's either AMD trying to pull the wool over our eyes by saying that the clocks of a much more powerful chip are clocked faster than the rv670 by this amount, keeping us and Nvidia in the dark until late Q2 / early Q3 when the cards make their way onto shelves in quantity, in a form more powerful than was previously (i.e. now) imagined, or, that a newer architecture with more shaders and TMUs on board and higher clocks can only improve on the current 38xx series by 40%, which would be a shame to say the least.

OR... people are starting to get floppy neck run-around-like-theirheads-cutoff syndrome again.

So what's the deal ATI?

What the deal with intel, nV and S3?

"OR... people are starting to get floppy neck run-around-like-theirheads-cutoff syndrome again." yeah, ...um, cheers for that.