2 min read

AMD releases 2.53 TFLOPS server GPU

Channel StrategyInfrastructure

AMD claims its FirePro S9150 server card is the most powerful graphic processing unit (GPU) built for high performance computing.

Built on the company’s Graphic Core Next architecture for high compute workloads, the S9150 delivers up to 2.53 teraflops of peak double precision floating point performance and up to 5.07 TFLOPs of peak single-precision floating point performance. That’s is up to 18 per cent more computing power than other server cards in the market are capable of, according to AMD.

FLOPS or floating-point operations per second, is a measure of computing performance. A teraflop is one trillion floating-point operations per second. TFLOPS is used to measure the performance of supercomputer, some of which are able to run over 1,000 FLOPS.

However, as supercomputers run faster, the more power they consume so it becomes very important to organizations to cut down power consumption in order to reduce cost.

“Today’s supercomputers feature an increasing mix of GPUs, CPUs and co-processors to achieve great performance, and many of them are being implemented in an environmentally responsible manner to help reduce power and water consumption,” said David Cummings, senior director and general manager, professional graphics, AMD. “The S9150’s memory configuration, compute capabilities and performance per watt are unmatched in its class, and can help take supercomputers to the next level of performance and energy efficiency.”

He said the server card is designed specifically for large scale multi-GPU support and high compute performance while maximizing available power budgets.

The S9150 comes with 16GB of GDDR5 memory (33 per cent more than other GPUs) and a maximum power consumption of 235 watts, according to Cummings.

The AMD FirePro S9150 server GPU features:

• 5.07 TFLOPS of peak single-precision floating point performance — up to 18 per cent more than the competition
• Maximum double-precision floating point performance resulting in 2.53 TFLOPS peak double-precision compute performance — half of its peak single precision performance.
• Industry leading memory configuration — 16GB GDDR5 memory, 512-bit memory interface and up to 320 GB/s memory bandwidth
• 2,816 stream processors (44 GCN compute units)
• Error Correcting Code Memory support (external)
• Ready to support OpenCL(TM) 2.0(4)
• 235W maximum power consumption
• AMD STREAM technology

AMD also introduced the new AMD FirePro S9050 server GPU with 12GB of ultra-fast GDDR5 memory, ECC Memory support, cutting-edge graphics and compute performance. The S9050 delivers 3.23 TFLOPS of peak single precision, plus 806 GFLOPS of double precision floating-point performance for effortless processing of challenging compute workloads.

The S9050 server GPU features:

• 12GB GDDR5 memory, 384-bit memory interface and up to 264 GB/s memory bandwidth
• 1,792 stream processors (28 GCN compute units)
• OpenCL 1.2 support
• Error Correcting Code Memory support (internal and external)
• 225W maximum power consumption

The S9150 and S9050 both feature PCIe x16 dual slot form factors and passive thermal solutions more compatible with server environments.

Both cards support AMD STREAM technology which enables AMD’s FirePro S-Series server GPUs to be used for compute intensive workflows leveraging the parallel processing power of AMD GPUs, and to accelerate many applications beyond just graphics.