Friday, January 2nd 2015
Possible NVIDIA GM200 Specs Surface
Somebody sent our GPU-Z validation database a curious looking entry. Labeled "NVIDIA Quadro M6000" (not to be confused with AMD FirePro M6000), with a device ID of 10DE - 17F0, this card is running on existing Forceware 347.09 drivers, and features a BIOS string that's unlike anything we've seen. Could this be the fabled GM200/GM210 silicon?
The specs certainly look plausible - 3,072 CUDA cores, 50 percent more than those on the GM204; a staggering 96 ROPs, and a 384-bit wide GDDR5 memory interface, holding 12 GB of memory. The memory is clocked at 6.60 GHz (GDDR5-effective), belting out 317 GB/s of bandwidth. The usable bandwidth is higher than that, due to NVIDIA's new lossless texture compression algorithms. The core is running at gigahertz-scraping 988 MHz. The process node and die-size are values we manually program GPU-Z to show, since they're not things the drivers report (to GPU-Z). NVIDIA is planning to hold a presser on the 8th of January, along the sidelines of the 2015 International CES. We're expecting a big announcement (pun intended).
The specs certainly look plausible - 3,072 CUDA cores, 50 percent more than those on the GM204; a staggering 96 ROPs, and a 384-bit wide GDDR5 memory interface, holding 12 GB of memory. The memory is clocked at 6.60 GHz (GDDR5-effective), belting out 317 GB/s of bandwidth. The usable bandwidth is higher than that, due to NVIDIA's new lossless texture compression algorithms. The core is running at gigahertz-scraping 988 MHz. The process node and die-size are values we manually program GPU-Z to show, since they're not things the drivers report (to GPU-Z). NVIDIA is planning to hold a presser on the 8th of January, along the sidelines of the 2015 International CES. We're expecting a big announcement (pun intended).
80 Comments on Possible NVIDIA GM200 Specs Surface
What makes it even more unlikely is that AMD will be unveiling the Carrizo APU with the same GPU architecture using 28nm (very likely at this weeks CES). Wouldn't it make more sense to consolidate APUs and GPUs on the same process when the APUs use the same graphics cores as the discrete GPUs? Graphics cards have a short shelf life. You build on the processes suited for the task and readily available. Unless AMD are really late to the party, I think you're setting yourself up for disappointment. The time frame for Fiji (its test and validation boards being sighted on Zauba some months back) almost certainly seems to point to 28nm IMO. Bermuda hasn't been sighted, so if it's a further development for a later launch then a smaller node is a definite possibility.
No.
My finances would be actually quite relaxed to wait till real 20nm or 16nm products. Short shelf life doesn't equate to short service in consumers' cases.
EDIT: Fiji XT boards began shipping two months ago at least
From tape out to finished silicon takes 8-12 weeks ( Tape out > fabrication > die cutting > chip runtime testing > die packaging > board assembly). Eight weeks prior to 7 November makes it four months - a quarter of a year.
So, when will it be released? :D