NVIDIA Maxwell Speculation Thread

[swaaye]

Some GM204 and GK104 cards are not exactly huge. My Zotax 970 is about half the length of my 6970.

 

[iMacmatician]

FYI, the base clock for the laptop GTX 980 is 1064 MHz compared to the desktop GTX 980's 1126 MHz, although that number is a minimum and NVIDIA says that it "varies by model."

 

[Deleted member 13524]

Isn't the normal 980m exactly the same chip but with shaders deactivated and at lower clocks?

I don't understand how these laptop chips can be so small and the desktop cards so unbelievably massive. Especially considering some 980m versions have 8GB of RAM.

These laptops are equally huge, and the card in that pic seems to be larger than the MXM III standard.

 

[swaaye]

Plus the display I/O and cooling aren't shown.

 

[swaaye]

Triple post!? How did I do this.

 

[gongo]

Full Maxwell non-Titan coming soon? ...or 23/24 activated SMM for a 980Ti Ultra?
Or dual 980Ti on a single PCB?

Nvidia Conducts “Secret” Meeting, Previews Upcoming Graphics Card
These types of limited press briefings have notably happened prior to several prominent graphics card launches from Nvidia over the past few years and have become somewhat of a tradition. Thankfully as we did not attend nor have we signed any NDA we can bring you some of the details that we’ve managed to secure so far.

Read more: http://wccftech.com/nvidia-conducted-secret-meeting-press-york-city/#ixzz3muWY537E

 

[Grall]

Pure clickbait.

Nothing concrete whatsoever, just an announcement of an announcement about an event which might have been completely fabricated to begin with. I feel dumb to have even clicked that link, but I didn't see it was Wfftech, or I wouldn't have.

Please, mark appropriately with warnings and stuff next time, alright?

 

[fellix]

Well, to be honest, Nvidia hadn't had dual-GPU board since the GK110-based Titan Z (not counting Tesla SKUs) and with the prospect of dual Fiji coming soon, it is logical to assume there would be an appropriate answer to the challenge.

 

[iMacmatician]

WCCFTech has posted a followup report that claims to have more details related to the claimed secret meeting in New York.

Apparently the upcoming product is a GM200-based dual-GPU card.

The “secret” briefing mentioned above took place last week. Review samples of Nvidia’s upcoming dual GM200 graphics board have been issued since then and are now in circulation. We’ve learned that most of the lucky few who have managed to secure a sample unit have pretty much completed the bulk of their benchmarking and testing of the card. So when we say it’s coming out soon, we do mean soon.

 

[DuckThor Evil]

There definitely was a meeting with press invited. Even Linus mentioned it more than a week ago.

Starting from one minute into the video.

 

[pjbliverpool]

Dual GPU cards are so boring. Give me Pascal already! (Although a full GM200 at overclocked 980Ti speeds would also be very nice).

 

[spworley]

The new GTX GPU is likely a fully enabled dual GM204 card with 8+8=16GB. NVidia has already shown the hardware configuration with the Tesla M60 announcement last month.

 

[Deleted member 2197]

NVIDIA GPUs to Accelerate Microsoft Azure

Deploying the latest version of NVIDIA GRID in its new N-Series virtual machine offering, Azure is the first cloud computing platform to provide NVIDIA GRID 2.0 virtualized graphics for enterprise customers.

For the first time, businesses will have the ability to deploy NVIDIA Quadro-grade professional graphics applications and accelerated computing on-premises, in the cloud through Azure, or via a hybrid of the two using both Windows and Linux virtual machines. Azure will also offer customers supercomputing-class performance, with the addition of the NVIDIA Tesla Accelerated Computing Platform's flagship Tesla K80 GPU accelerators, for the most computationally demanding data center and high performance computing (HPC) applications.

By deploying the Tesla K80 GPU accelerator in its N-Series virtual machines, Azure dramatically expands access to supercomputing-class performance, enabling enterprises worldwide to accelerate their most demanding workloads, without requiring them to invest in, build and maintain dedicated computing resources.

http://www.guru3d.com/news-story/nvidia-gpus-to-accelerate-microsoft-azure.html

 

[lanek]

NVIDIA GPUs to Accelerate Microsoft Azure

http://www.guru3d.com/news-story/nvidia-gpus-to-accelerate-microsoft-azure.html

Way far and under the AMD virtualizuation solution no ?

 

[xpea]

Way far and under the AMD virtualizuation solution no ?

as usual with AMD, the hardware is potent but software /support is lacking. Nvidia GRID is all over the place with big promotion done by VMware, Citrix, Nice and now Microsoft. You find endless resources in their website and you can even try the service for free with VMware: http://www.nvidia.com/object/vmware-trygrid.html
main Nvidia software partners:
http://www.nvidia.com/object/nice.html
http://www.nvidia.com/object/vmware.html
http://www.nvidia.com/object/microsoft.html
http://www.nvidia.com/object/citrix.html
Server and solution partners:
http://www.nvidia.com/object/grid-partners.html
Certified list of servers:
http://www.nvidia.com/object/grid-certified-servers.html
and so on...

now in comparison, AMD: http://www.amd.com/en-us/solutions/servers/virtualization
and that's it, a single page with a whopping total of 1 video, 4 white papers and a link to a Microsoft page that is not even dedicated to virtualization...

Now tell me how they can compete when they don't have any resource to develop/promote the technology ?

 

[fellix]

Looks like AMD's business strategy has been leaning to be more like a vendor for licensed IPs for third parties and not so much as branding/driving force on the market as Nvidia does. The only really strong (consumer) IP left for them is the Radeon line, while the CPU and APU brands have been stuck down the bottom of the bin for years now and the chaotic marketing policies jumping in and out made the damages only more persistent.
Software and platform support is a whole another tragedy -- just compare at how many device driver and middle-ware distributions both AMD and Nvidia have to release each time. One would expect an underdog like AMD to put much more effort and ingenuity in multiple directions, but they still continue to rely on the next "miracle" tech that would buy them some advantage over the competition in a single shot.

 

[Deleted member 2197]

Nvidia adds support for OpenACC parallel programming on x86 architecture processors

NVIDIA today announced availability of its newest PGI Accelerator Fortran, C and C++ compilers (version 15.10) now with support for OpenACC directives-based parallel programming standard on x86 architecture multicore microprocessors. The new compilers allow OpenACC-enabled source code to be compiled for parallel execution on a multicore CPU or a GPU accelerator.

This newest PGI feature compiles OpenACC compute regions for parallel execution across all of the cores in an x86 processor or multi-socket server. The cores are treated in aggregate as a shared-memory accelerator, eliminating all data movement overhead in the resulting OpenACC programs. By default the compiler generates code that uses all the available cores in the system, and several methods exist for programmers to control and fine-tune this behavior.

“We were extremely impressed that we can run OpenACC on a CPU with no code change and get equivalent performance to our OpenMP/MPI implementation, and get 4x faster performance when running on a GPU,” said Wayne Gaudin of the U.K.’s Atomic Weapons Establishment. “From the perspective of performance portability and code future proofing, this is an excellent result.”

“Porting HPC applications from one platform to another is one of the most significant costs in the adoption of breakthrough hardware technologies,” said Buddy Bland, project director at Oak Ridge National Laboratory. “OpenACC for multicore x86 CPUs provides continuity and code portability from existing CPU-only and GPU-enabled applications from machines like Titan to all of DOE’s upcoming major systems as well as portability among those systems.”

While this release targets current x86 machines, NVIDIA also outlined the timing for PGI OpenACC compiler support for IBM POWER, Intel’s Knight’s Landing, and ARM architectures.

http://www.hpcwire.com/2015/10/29/p...add-openacc-support-for-x86-multicore-cpus-2/

 

[iMacmatician]

From HWBattle: "HDMI 2.0 support, which is released GTX 750 variants ?!" (original).

A variant of the GeForce GTX 750 chipset, code names supporting HDMI 2.0 is called GM206-150. (GTX 960/950 family is the GM206.)
GTX 750 series or offering to purchase in mind, HDMI 2.0 graphics for those looking for the cheapest product that seems to be quite merit.
(Said to be soon launched in Japan and Korea.-_- A)

According to this page from Gainward, the GM206-150 has 512 CCs.

 

[homerdog]

Does AMD have any cards that do HDMI 2.0? It's difficult to find this information.

 

[mczak]

Does AMD have any cards that do HDMI 2.0?

Nope.