Oh he does not need to speculate.
The 55nm GT200b will not have the FP64 ALU's at it will be a significant less complex and smaller die.
Who is "he" and why should I take his word for anything?
Nvidia GT200b rumours and speculation thread
- Thread starter nicolasb
- Start date
Who is "he" and why should I take his word for anything?
Good point. A change in naming convention (removal of the "T") would lend a whole lot more credibility to this rumor.
What change ? The current heatspreader shows no GT200...
You're right, that used to be my signature for awhile, actually, yet all the marketing material still refers to GT200. So which is it?
DegustatoR
Veteran
They'll still use GT200s for Tesla/CUDA, ASPs is much higher there anyway.
It's possible, the question is -- how much transistors/die size are they planning to save this way?
I consider it higly unlikely that after nine or so years of re-marketing consumer-level GPUs with minor board- and BIOS-level alterations and a refitted driver package for the high-end professional segment, Nvidia's supposed to be going separate routes again.
GT200b (GTX280+) launch at NVISION 08
http://forums.vr-zone.com/showthread.php?t=304489
If true that + moniker is really awful, they should instead stick to the numbers .. 290. Also wonder the possibilities of a GX2 board.
http://forums.vr-zone.com/showthread.php?t=304489
If true that + moniker is really awful, they should instead stick to the numbers .. 290. Also wonder the possibilities of a GX2 board.
DegustatoR
Veteran
That sounds like they've just made it up.
GX2 is highly doubtful before 40nm.
Mintmaster
Veteran
Every review out there uses GT200. Tech-report, Anandtech, Tom's...
Unlikely that a mere shrink would get the power draw under 150W per chip, it's own memories and other relevant hardware, unless they do some heavy castration in the process aswell.
Mintmaster
Veteran
We only have to look at how G92b has increased power consumption over G92 with only slightly increased clocks. For the big clock increase that people are hoping GT200b has, I can't see how its possible to have reduced power consumption.
GT200b is around the corner but i think they need a new architecture or at least major optimizations with current projects not only going into increase numbers of Shaders, TMUs, ROPs.
I hope NVIDIAs Nextgen will be based on fresh new and effective architecture. If not ATI is going to crush them with it`s Rv870.
I hope NVIDIAs Nextgen will be based on fresh new and effective architecture. If not ATI is going to crush them with it`s Rv870.
Hardware-info is speculating that GT200b isn't as small as a straight shrink would indicate because Nvidia had to spend transistors to increase shader clocks and that overall TDP would be similiar to GT200. Even at the speculated 648/1620 clocks I suspect it would have a hard time keeping up with R700.
http://www.hardware-infos.com/news.php?news=2244
http://www.hardware-infos.com/news.php?news=2244
How likely is it that Nvidia are going to touch their highly optimized custom logic in the shaders just for a mere 200 MHz ALU clock - for there are already GT200 in the market sold with 1400 and higher shaderclocks.
IMO the current situation is a bit of a surprise to Nvidia and the planned improvements of GT200b were fixed before everyone and their dog knew how incredibly fast AMD would be with their sub-300mm²-die.
IMO the current situation is a bit of a surprise to Nvidia and the planned improvements of GT200b were fixed before everyone and their dog knew how incredibly fast AMD would be with their sub-300mm²-die.
Jawed
Legend
[EDIT: hmm, this is sort of a redundant post] What can NVidia change about GT200b in the interval between RV770's launch and GT200b's launch? That interval is 2 months or so. It's the period of one spin.
If 55nm has been of no great significance to the performance of G92b - which has hardly had an easy birth as it certainly wasn't available when NVidia said it would be:
http://www.hardforum.com/showthread.php?t=1325759
then it doesn't look like GT200b will be an easy GPU to produce either - GT200 was late and problematic enough on 65nm.
So, all I'm really saying is that whatever GT200b will be will be at best slightly more than it was always planned to be.
Jawed
If 55nm has been of no great significance to the performance of G92b - which has hardly had an easy birth as it certainly wasn't available when NVidia said it would be:
http://www.hardforum.com/showthread.php?t=1325759
then it doesn't look like GT200b will be an easy GPU to produce either - GT200 was late and problematic enough on 65nm.
So, all I'm really saying is that whatever GT200b will be will be at best slightly more than it was always planned to be.
Jawed
Agreed. I don't think anybody expects GT200b to be a reaction to RV770. But they could have made changes in there based on earlier dissatisfaction with how GT200 turned out.
