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
EDIT: I forgot the obvious red flag.The guy leaking this infoclaimed that the AMD GPUs are manufactured on GloFo's 20nm process....which means that AMD is presently using 20nm and will transition to 28nm (?!) in the future. BTW the ONLY 20nm process GloFo has is a low power process- which doesn't gel with a large monolithic GPU of 200-300W
www.pcgamer.com/nvidias-surprised-the-titan-outsold-the-year-old-gtx-690-in-just-3-months-were-not/
The fake graphs and the clutching at straws in this thread though is funny, now that is a fact.
People forget how early out we got Maxwell 750 benchmarks. 5 months early and Chiphell coming in 3 months earlier but they provided OC 750 Ti samples results and since those numbers looked favorable people gravitated towards them and turned out to be over estimated. If these cards are being released in a similar time-frame to one another its plausible for them and the Sisoft source to have both. Blasphemy
Time will tell if they are true & accurate or not but it seams to be more of a hate for Chiphell having access to the cards.
We are in for a dozy guys, we might be looking at the business/research version of big Maxwell here, but I'm sure a more costumer friendly card a là 780/780ti will be just around the corner.
As someone who's fully embraced 4K gaming, all I can say is: bring it on AMD and Nvidia! :rockout:
some of us see it as fact.
I find Zhul's only post so far quite weird. By its own logic Intel's i7 for the LGA 1150 does not exist because it has been replaced by LGA 2011's i7s. Because AMD is absent on the high end in processors, just like with his assumption "if AMD goes under...X80 series will be all $1000 parts".
As they say more money than sense.
As for the Nvidia vs AMD hate you guys can carry on. I enjoy the childish, hypocrisy & self righteousness.
I don't tar Chiphell with distain for a single poster anymore than I would show distain for TPU because a single forum member chooses to pick an argument using nonsensical data - namely that the first line in the "leakers" post categorically said that the AMD GPUs were manufactured on GloFo's 20nm Low Power Mobility process - a process they would be totally unsuited for (Globalfoundries has stated as much in both interviewand PPS presentation if the naming wasn't enough of a hint) , and the totally farcical situation where AMD are supposedly building GPUs on 20nm, but are moving to 28nm in 2015. When was the last time a vendor moved production to a larger process than they are already supposedly using? No, what I said was that a range of sites - from rumour mills to respected mainstream sites (along with process tech based sites) dismissed the benchmarks as bogus, and the easiest evidence of that proof came from the "leaker" himself stating that the large die GPUs were built on GloFo's 20LPM process.
Videocardz was by no means the only site that were sceptical.Hilbert at G3D called out the charts as fake on the day they showed up: Feel free to believe (and I see two others upvoted your post so they obviously believe as well) that AMD are building GPUs on an 20nm process only to toss it aside to use 28nm, and the only person on the planet leaking gaming benchmarks for unreleased models conveniently just so happens to have all four of the top cards - along with working drivers....just don't try to convince me they are legit when common sense says otherwise - common sense will trump you every time.
AMD inevitably compete on lower pricing. This creates two troll states of consciousness, NV trolls can see AMD as bargain bin trash (and they are not). NV are seen by AMD trolls as exploiting bastards. Which as an NV buyer, I'd say they are.
Point is though, I buy NV because I like buying the best solution for me. At least my 2nd Classified card was a bargain.
Maybe Fiji will be my next choice. People need open, logical minds to make informed tech decisions.
But for those wishing to be brave, why not state here and now who will have the best card next, AMD's Fiji or Big Maxwell?
I'd suggest Maxwell but at leat if I'm wrong, I'll buy the better option for me. Unless AMD's chip is ruined by a poorly made card.
So, I'd pick Fiji XT as the top GPU at the expense of power consumption and framebuffer if HBM is involved since 4GB is the maximum permissable with first-gen HBM.
i bet you have an iphone too as apple only make quality products as well :rofl:
:SMH: