News
 

Bookmark and Share

(7) 

At the SC12 supercomputer conference, Nvidia Corp. introduced its Tesla K20-series of graphics processing units-based accelerators for high-performance computing (HPC) applications. The new products are based on the code-named GK110 processors specifically tailored for the HPC world.

Nvidia K20X with 2688 stream processors and 6GB of GDDR5 memory offers 3.95TFLOPS single-precision and 1.31TFLOPS double-precision peak floating point performance. The new family also includes the Tesla K20 accelerator with 2496 stream processors and 5GB of memory, which provides 3.52TFLOPS of single-precision and 1.17TFLOPS of double-precision peak performance. The Tesla K20X GPU accelerator delivers three times higher energy efficiency than previous-generation GPU accelerators and widens the efficiency advantage compared to CPUs.

Thanks to increased double precision floating point performance, Nvidia Tesla K20-series accelerators are suitable for seismic processing, CFD, CAE, financial computing, computational chemistry and physics, data analytics, satellite imaging and weather modeling.

Tesla K20X and K20 GPU accelerators representing more than 30PFLOPS of performance have already been delivered in the last 30 days to various supercomputer clients around the world. 30PFLOPS is equivalent to the computational performance of last year's 10 fastest supercomputers combined.

Tesla Nvidia K20 family of GPU accelerators is shipping today and available for order from leading server manufacturers, including Appro, Asustek Computer, Cray, Eurotech, Fujitsu, HP, IBM, Quanta Computer, SGI, Supermicro, T-Platforms and Tyan, as well as from Nvidia reseller partners.

Tags: Nvidia, Kepler, Tesla, GPGPU

Discussion

Comments currently: 7
Discussion started: 11/13/12 03:03:58 AM
Latest comment: 11/13/12 11:29:30 PM
Expand all threads | Collapse all threads

[1-5]

1. 
How much power does this card consume? AMD's FirePro 10k offers a bit more peak performance but is a power hog at 375 Watts.
0 0 [Posted by: Nightgawk  | Date: 11/13/12 03:03:58 AM]
Reply
- collapse thread

 
I think it depends what you are running. As beck2448 noted below me, it's the overall package that NV offers that makes their HPC cards so much superior.

However, if you look at performance, the S10000 card does offer class-leading SP and DP performance at the same time. If you run a wide variety of applications and need both SP and DP performance, the S10000 is better than either K10 or K20. So it gives you that extra flexibility, albeit the corporate clients probably know exactly if they need SP or DP.

Single Precision
S10000 = 15.77 Gflops / Watt
K10 = 20.36 Gflops / Watt (Best)
K20 = 15.66 Gflops / Watt
K20X = 16.75 Gflops / Watt

Double Precision
S10000 = 3.94 Gflops / Watt
K10 = 0.85 Gflops / Watt
K20 = 5.22 Gflops / Watt
K20X = 5.58 Gflops / Watt (Best)

This gives us an idea that a 2304-2496 shader GK110 could be possible by late spring/summer of 2013 as yields improve even more and 28nm node matures, allowing for lower voltages.
1 2 [Posted by: BestJinjo  | Date: 11/13/12 03:37:03 PM]
Reply

2. 
Nvidia dominates the pro market at almost 90% because they have hardware, software and programmers like the ones that created the effects for films like AVATAR. NVIDIA ported their PantaRay engine to a CUDA-based GPU driven version that runs 25 times faster, utilizing an NVIDIA Tesla S1070 GPU-based server instead of a CPU based server. This allowed rendering times to be shortened from weeks to hours. K20X will be a massive leap in this area.
2 1 [Posted by: beck2448  | Date: 11/13/12 04:21:30 AM]
Reply

3. 
for us regular joe-consumers this new brick does not help in one bit for we have zero BIG super computer in our home. i wonder if those GPU vendors got any pci-express 2.0 or 3.0 x16 cards that we could pop into our desktop and turn those hundreds of GPU cores into CPU cores and running win 7 or Ubuntu as well. that might flick Intel top-down. if that happen Intel will sell an i7 3770k for a few $ and soon we will see Intel pop out 512 core and 1024 cores better yet one with 8192 cores would dominate those super computers in no time.

0 1 [Posted by: idonotknow  | Date: 11/13/12 06:23:36 AM]
Reply

4. 
I wonder how the pro GPU market will affect the advent of cloud gaming? Having a GPU based server might be a good idea for services liike these in the future.
1 0 [Posted by: veli05  | Date: 11/13/12 07:14:27 AM]
Reply

5. 
Still holding out in hope, Not a lot left for Nvidia to reveal, low end,mid range,high end and now server market taken care of, So Nvidia please get to work on that damn 780GTX, The real high end card for consumers
0 0 [Posted by: ozegamer  | Date: 11/13/12 11:29:30 PM]
Reply

[1-5]

Add your Comment




Related news

Latest News

Thursday, August 28, 2014

12:22 pm | AMD Has No Plans to Reconsider Recommended Prices of Radeon R9 Graphics Cards. AMD Will Not Lower Recommended Prices of Radeon R9 Graphics Solutions

Wednesday, August 27, 2014

9:09 pm | Samsung Begins to Produce 2.13GHz 64GB DDR4 Memory Modules. Samsung Uses TSV DRAMs for 64GB DDR4 RDIMMs

Tuesday, August 26, 2014

6:41 pm | AMD Quietly Reveals Third Iteration of GCN Architecture with Tonga GPU. AMD Unleashes Radeon R9 285 Graphics Cards, Tonga GPU, GCN 1.2 Architecture

Monday, August 25, 2014

6:05 pm | Chinese Inspur to Sell Mission-Critical Servers with AMD Software, Power 8 Processors. IBM to Enter Chinese Big Data Market with the Help from Inspur

Sunday, August 24, 2014

6:12 pm | Former X-Bit Labs Editor Aims to Wed Tabletop Games with Mobile Platforms. Game Master Wants to Become a New World of Warcraft