Adtech Global worked with AMD FirePro to develop a High Speed Cluster solution based on GPGPU technology to support efforts in quantum chromodynamics research for the King Abdualziz City for Science and Technology (KACST) and Frankfurt Institute for Advanced Studies (FIAS).
ACST and FIAS were striving to not only maximize performance, but also to optimize this performance for thermal dissipation and power draw efficiency. The challenge in building a supercomputer is no longer the money you’re willing to spend on processors. The challenge now lies within getting enough power and cooling to the supercomputer itself. The SANAM supercomputer was able to break these barriers. It is currently the world’s No. 52 most powerful supercomputer, and the world’s 2nd most power-efficient supercomputer.
Critical to the power efficiency was the use of (420) AMD FirePro S10000 dual-GPU server graphics cards. The SANAM supercomputer is capable of sustaining 421.2 TFLOPS, providing a system energy efficiency of over 2.3 GFLOPS per watt and performing 2,351 million calculations per second per watt.
The video below gives a really insightful tour of the collaboration between Adtech and AMD, as well as some clear visuals on the actual SANAM supercomputer. You can also read the case study which details the custom computing solution they implemented with the help of Adtech Global.
King Abdualziz City for Science and Technology (KACST) in Riyadh, Saudi Arabia, along with Frankfurt Institute for Advanced Studies (FIAS), needed to develop a High Speed Cluster solution based on GPGPU technology to support efforts in quantum chromodynamics research. ACST and FIAS were striving to not only maximize performance, but also to optimize this performance for thermal dissipation and power draw efficiency. This case study details the custom computing solution they implemented with the help of Adtech Global.
Powered by 420 AMD FirePro S10000 dual-GPU server graphics cards, 4.8 thousand of Intel Xeon E5-2650 (8 cores, 2.0GHz) microprocessors and infinaband connectivity, the SANAM supercomputer is capable of sustaining 421.2 TFLOPS, providing a system energy efficiency of over 2.3 GFLOPS per watt and performing 2,351 million calculations per second per watt. The SANAM supercomputer represents a shift from determining performance through FLOPS to focusing on eco-conscious measurements like performance-per-watt. SANAM is currently the world’s No. 52 most powerful supercomputer, and the world’s 2nd most power-efficient supercomputer.
Newegg TV gives a pretty comprehensive interview and overview of the new AMD FirePro W Series discussing the unique points of each card, as well as common features across all cards.
The video covers the FirePro W600, W5000, W7000, W8000, and W9000.
These Newegg videos are really very good. Reminds me of Zappos shoe reviews - comprehensive, useful and no marketing spin.
AMD today announced the AMD Radeon HD 7970. Why this is particularly interesting to the professional CAD/DCC community is because of the new SIMD-based Graphics Core Next (GCN) Architecture first detailed at the 2011 Fusion Developer Summit. Basically GCN enables the card to act both as a a graphics workhorse and a computing (e.g. OpenCL, C++ AMP, DirectCompute) workhorse for the processing of non-3D workloads such as video rendering, photo editing, code cracking, physics FX and scientific calculations.
The 28nm process means the new cards are faster than their 32 nanometer predecessor equivalents, without consuming significantly more power, or generating more heat (even though the number of stream processors increased from 1536 to 2048 and clock speed increased from 880MHz to 925MHz).
Also of note is Eyefinity 2.0 which adds support for stereo 3D, universal bezel compensation, brand new display configurations and an expanded and more immersive field of view. This allows users with 2560x1600 monitors in an AMD Eyefinity 5x1 landscape configuration to achieve a horizontal resolution of 12,800 pixels, driving more than 20 million pixels of screen real estate.
Take aways from my perspective: GPGPU computing is now mainstream. High-performance with low power consumption is mainstream. 3+ display Eyefinity and stereo 3D are mainstream. And lastly, this makes me excited about the next generations of FirePro graphics cards.
The new AMD FirePro V7800P is specially designed for use in rackmount servers, blade servers and PCIe expansion chassis. The passively cooled, half-length, full height card card requires 10 cubic feet / minute of airflow to keep it cool (servers normally deliver three times that rate in a peripheral slot). It can be plugged into a server proper or can be hosted in an external PCI-Express 2.0 chassis. It draws a maximum of 138 watts.
The card supports OpenCL 1.1 and DirectCompute 11 for massively parallel number crunching, as well as OpenGL 4.1 and DirectX 11 for high performance graphics processing. With all 1,440 cores working simultaneously, the FirePro V7800P can deliver 2 teraflops of single-precision and 400 gigaflops of double-precision floating point performance.
In addition to supporting traditional workstation graphics in a client/server model, AMD FirePro V7800P professional graphics also enable GPU compute, remote graphics and Virtual Desktop Infrastructure (VDI) deployments.
At the Embedded Systems Conference Silicon Valley 2011 last week, AMD announced the new Radeon E6760 embedded discrete graphics processor. The E6760 GPU offers embedded system designers OpenCL support as well as support for six independent displays using Eyefinity technology, HDMI 1.4 stereoscopic video, and DisplayPort 1.2 for higher link speeds.
BSN interviewed John Taylor, Director of Client Product and Software Marketing at AMD, regarding the upcoming Fusion Developer Summit to be held June 13-16, 2011 at the Meydenbauer Center in Bellevue, Washington.
Both Microsoft and Arm will be giving keynotes at the event. ARM is going to reiterate their commitment to OpenCL with their own Mali series GPU, tying nicely to AMD’s Fusion APUs, i.e. Ontario (9W) and Zacate (18W) and upcoming Llano APUs (up to 95W).
Also noteworthy from the interview:
With Fusion architecture, CPU (ok, APU) is getting a driver update each and every month for the first time in history of the silicon. The company insists on a vision that by making a continuous Catalyst driver updates for the Ontario/Zacate/Llano APUs, the CPU and GPU parts will be constantly updated with application profiles and API updates. As a result, users of AMD APUs should experience smoother application performance as the time passes by.
Do note that with the Fusion strategy, AMD is no longer just following computing standards, but rather setting them in silicon with smart adjustments. A good example for that is the upcoming C-60 APU that gives you up to 33% higher performance in typical CPU tasks or a massive 43% performance boost in DirectX / OpenGL / OpenCL - all while staying in the same power envelope (9W).
So how do you move open standards along faster? In a word: education.
The OpenCL University Kit introduced by AMD is an easy tool to enable educators to quickly introduce OpenCL learning into their curriculum.
Included in the University Kit is a 13 lecture series, equipped with instructor and speaker notes, as well as code examples where necessary. An advanced understanding of OpenCL is not needed to understand the course materials; students only require a basic knowledge of C/C++ programming. A C/C++ compiler and an OpenCL implementation (such as the AMD APP SDK) are needed to complete the exercises.
Formally know as the ATI Stream SDK, the latest v2.3 release of the GPU + CPU acceleration software development kit is now known as the AMD Accelerated Parallel Processing (APP) SDK. This new version adds support for the new line of Fusions APUs and the Radeon HD 6900 series. It also adds improved OpenCL 1.1 runtime performance to accelerate application performance using both the CPU and GPU.
The industry-leading Cypress GPUs (in the Radeon 5870 and new FirePro line) have made their way into AMD’s FireStream GPU co-processors. These “compute accelerators” are intended for highly parallel, compute-intensive workloads in scientific, financial and academic arenas. Unlike the discrete FirePro/Radeon graphics cards, the FireStream compute accelerators use a passive heat sink so they can slide into rack mounted HPC servers and expansion systems for x64 systems.
From the press release: The AMD FireStream 9350 delivers 2.0 TFLOPS of single precision performance and 400 GFLOPS of double precision floating point performance in a single-slot, 150W solution with 2GB of GDDR5 memory, enabling breakthrough compute density (almost double the previous generation). The AMD FireStream 9370 delivers up to 2.64 TFLOPS of single precision performance and 528 GFLOPS of double-precision performance, and includes 4GB of high-speed GDDR5 memory, at a maximum board power of 225 watts. In addition, the AMD FireStream 9350 and 9370 both support leading industry standard application interfaces, including OpenCL, DirectX 11 and OpenGL.
The FireStream boards are AMD’s higher performance answer to Nvidia’s Tesla Fermi-based co-processors. The higher floating-point performance aside, with the FireStream, AMD is also focused on delivering a solution based on open standards like OpenCL rather than the proprietary solutions like Cuda.
FireUser.com is a community resource for CAD, visualization, 3D, video and engineering professionals to learn about the latest acceleration and display technologies and news with a focus on the AMD FirePro workstation graphics line.