Nvidia (NASDAQ:NVDA) has launched its latest Tesla CUDA compute card, the Tesla K80 today at Supercomputing 2014 (SC14) in New Orleans.

This follows Nvidia’s announcement last week that it had been awarded a $325 million Department of Energy grant with IBM to help build two 100 Petaflops+ machines for Lawrence Livermore National Laboratories and Oak Ridge National Laboratory. Those supercomputers are expected to be built in 2017 and will use Nvidia’s yet-t0-be-built Volta GPU which comes after Pascal which comes after Maxwell, which is what Nvidia is currently using in their consumer graphics cards, but hasn’t quite started to use in professional just yet. So, until then, people looking for the fastest compute cards are going to want to look at Nvidia’s new Tesla K80.

Nvidia’s new Tesla K80 dual GPU compute card is an interesting product because it once again brings back Nvidia’s dual GPU Tesla products and increases the amount of compute you can squeeze onto a single card. Logically, you would think that the K80 would naturally be two K40’s smacked together into a single card, but that’s not accurate. In order to build the K80, Nvidia actually went with GPUs with similar shader core counts as the Tesla K20, but what’s most important is that they actually did double the onboard memory of the K80 from the K40 to 24 GB of GDDR5. Whenever we talk to anyone looking to do large simulations or scenes, their number one complaint is that they can never have too much VRAM and Nvidia appears to be listening to them by packing a whopping 24 GB of GDDR5 per card or 12GB of GDDR5 per GPU.

According to Nvidia’s specifications for the Tesla K80, it has 4992 shader cores (double that of the K20) which turns out to slightly less than double that of the K40, this is because Nvidia is using two GK-210 GPUs rather than the K40’s GK-110B. However, if you look at Nvidia’s performance claims, they state that the Tesla K80 is capable of 8.74 teraflops single-precision and 2.91 teraflops double-precision. This is more than double that of the K40 GPU which it seeks to replace and almost double that of the K10, Nvidia’s first dual GPU Tesla card. That’s the fantastic thing about Nvidia’s own Tesla cards, the K80 is a Kepler based dual GPU card while the K10 is also a Kepler based dual GPU card and the performance difference is nearly double simply by going from the GK-104 GPU design to the GK-110 (full-blown) Kepler GPU design. The Tesla’s K80 two GK210 GPUs each have 13 SMs that are clocked at 562 Mhz base, 875 Mhz boost thanks to Nvidia’s new GPU boosting features for this Tesla card, a Tesla first and something that came over from the consumer cards. Either way, its a great achievement and this is Nvidia’s next step before introducing a Maxwell-based Tesla compute card.

Nvidia’s Tesla K80 Compute Card – Note, no display out since this is a compute card

Also, if you look at the double-precision performance, it is more than 30 times faster than the original dual Kepler GPU card, the K10 and more than twice as fast as the K40 which means that Nvidia makes up for the lack of shader cores somewhere else. One of those places is in the card’s memory bandwidth which is pumped up to a whopping 480 GB/s which pretty much removes memory bandwidth as a bottleneck in most applications. Unfortunately, even though this card does have double the GPUs and double the memory of the K40, it doesn’t quite have double the memory bandwidth, which once again points to a likelihood of lower memory clock speeds than the K40. If one were to simply double the K40’s memory bandwidth, you would be looking at 566 GB/s, not the current 480 GB/s on the K80, but even so, this pretty much dwarfs anything on the market by over 150 GB/s anyways.

That card would be AMD’s own FirePro W9100, which has been very competitive with Nvidia’s K40 and offers 2,816 stream processors, 16GB GDDR5 memory and 320 GB/s memory bandwidth. This comes out to a peak performance of 5.24 teraflops peak single-precision floating-point performance and 2.62 teraflops peak dual-precision floating-point performance, meaning that now the ball is back in AMD’s court with today’s announcement.

In fact, Nvidia’s Tesla compute cards are also enabling some of the most recent scientific breakthroughs like the ongoing Rosetta mission that the ESA has embarked upon, which most recently landed a probe on a comet. “The Tesla K80 dual-GPU accelerators are up to 10 times faster than CPUs when enabling scientific breakthroughs in some of our key applications, and provide a low energy footprint,” said Wolfgang Nagel, director of the Center for Information Services and HPC at Technische Universität Dresden in Germany. “Our researchers use the available GPU resources on the Taurus supercomputer extensively to enable a more refined cancer therapy, understand cells by watching them live, and study asteroids as part of ESA’s Rosetta mission.”

There is currently no pricing information available for the K80 based on the information we were given by Nvidia, however you can probably ballpark that it will cost upwards of $5000 since the K40 originally sold for that price but has since been discounted down to $3,000 very likely in anticipation of today’s announcement. In terms of availability, Nvidia Tesla K80 dual-GPU compute cards will be available from a variety of server manufacturers, including ASUS, Bull, Cirrascale, Cray, Dell, Gigabyte, HP, Inspur, Penguin, Quanta, Sugon, Supermicro and Tyan, as well as from NVIDIA reseller partners.

The post Nvidia Launches Tesla K80 Dual Kepler Compute Card appeared first on VR World.

Today, the Department of Energy has announced that is has granted $425 million to build two new supercomputers at Oak Ridge National Laboratory and Lawrence Livermore National Laboratories as part of a broader CORAL initiative which is a collaboration of Oak Ridge, Argonne and Lawrence Livermore. $325 million of that will be spent on the actual supercomputer building while an additional $100 million will be used for the FastForward2 program, which is designed to encourage and enable hardware vendors to increase performance and efficiency for the next generation.

The first supercomputer, to be known as Summit, will be installed at Oak Ridge National Laboratory and will replace the currently existing ‘Titan’ supercomputer which is capable of a peak performance of 27 petaflops. Summit will be capable of delivering between 150 and 300 peak petaflops and will be used for ‘open science’. The Sierra supercomputer, designed to replace the existing Sequoia will be used for nuclear security simulations and will be capable of speeds in excess of 100 petaflops as well. Both systems will be faster than the world’s fastest supercomputer right now, Tianhe-2 in China, which currently clocks in at 55 petaflops of peak performance. Argonne’s hardware win is yet to be announced, but will be unveiled at a later date.

In order to achieve this level of performance, the laboratories participating in the CORAL initiative are harnessing the power of IBM’s (NYSE: IBM) Power 9 architecture CPUs and Nvidia’s (NASDAQ:NVDA) yet-t0-be-announced Volta GPUs. This means that since this machine is expected to come online in 2017, that we can very likely expect to see Volta GPUs in 2017. The project will use Mellanox’s interconnect technologies to connect the systems together, but in order to connect the GPU to the CPU, they will be using Nvidia’s own NVLink GPU interconnect. NVLink is Nvidia’s own proprietary interconnect specifically designed to increase the communication speed between GPUs and Nvidia is working with IBM to get this interconnect embedded directly into the IBM Power CPUs that will be powering these different supercomputer designs. Additionally, the Summit supercomputer will also be using IBM’s own IBM Elastic Storage using GPFS technology and will store 120 petabytes of data.

The system as a whole, Summit, will only use 10% more power than Titan but will deliver approximately 5-10 the performance of Titan, illustrating where supercomputer designs are headed and how the Department of Energy is really trying to drive high performance increases while also promoting energy efficiency. The expected performance for Summit has already been stated to be between 150 and 300 petaflops, however, this is thanks to over 3400 compute nodes, each delivering 40 teraflops of performance alone. Each node will consist of IBM Power 9 CPU(s) and Nvidia Volta GPU(s), unfortunately we do not know if each node will be a dual processor node or how many GPUs will fit into each node, but the expectation would be a dual processor node with at least 2 GPUs per node.

a

This hardware win for IBM and Nvidia is a huge one because it illustrates that the Open Power partnership between the two companies is working and that it can enable IBM to ship more CPUs. This is a very big deal for IBM and Nvidia because this is the first supercomputer in the US in a long time that will be built without either Intel or AMD CPUs. It also means that Nvidia will finally make use of NVLink, which they announced will be coming out with the Pascal GPU, the predecessor to Volta. Nvidia has already said we can expect to see Pascal in 2016, which means the transition from Pascal to Volta will be a fairly quick one.

The Department of Energy has stated that the whole purpose of these new supercomputer designs is to enable exascale computing. Both Coral and FastForward2 are supposed to enable hardware manufacturers to help their customers build efficient and powerful suptercomputers capable of over 1 exaflop (or 1000 petaflops). And if they can get the Summit supercomputer to 300 petaflops, that’s going to be a huge step forward to achieving exascale computing and an exaflop supercomputer.

The post IBM and Nvidia to Build 100 Petaflop+ Supercomputers appeared first on VR World.