AMD has announced that it has leap-frogged rival Nvidia in the high-performance computing (HPC) accelerator market once again, launching a card with an impressive 5.07 teraflops of single-precision floating-point performance.
The HPC market might be a small one, but high profit margins mean it's big business for those who can get their feet in the door. Some of the world's fastest computers get their performance from GPU-based accelerator cards, the majority of which are from Nvidia's Tesla family. AMD's latest launch aims to change that, with what it claims is the world's most powerful GPU accelerator board for HPC use.
'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,' claimed David Cummings, senior director and general manager of professional graphics at AMD. 'Designed for large scale multi-GPU support and unmatched compute performance, the AMD FirePro S9150 ushers in a new era of supercomputing. Its 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.'
Designed to win market share from Nvidia's top-end Tesla K40, the FirePro S9150 boasts 5.07 teraflops of compute performance in single-precision mode and 2.53 teraflops in double-precision. Compared to the Tesla K40's 4.29 teraflops and 1.43 teraflops respectively, those are impressive figures - and doubly so when you learn the card has the same thermal design profile (TDP) of 235W, despite an extra 4GB of GDDR5 memory for 16GB total. Its performance comes courtesy of 44 Graphics Core Next (GCN) compute units, totalling 2,816 stream processors.
The FirePro S9150 HPC accelerator is to be joined by a server variant dubbed the FirePro S9050, dropping to 12GB of GDDR5 and 28 GCN compute units for 1,792 stream processors total and a 225W TDP. AMD has yet to confirm pricing for either card.
The HPC market might be a small one, but high profit margins mean it's big business for those who can get their feet in the door. Some of the world's fastest computers get their performance from GPU-based accelerator cards, the majority of which are from Nvidia's Tesla family. AMD's latest launch aims to change that, with what it claims is the world's most powerful GPU accelerator board for HPC use.
'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,' claimed David Cummings, senior director and general manager of professional graphics at AMD. 'Designed for large scale multi-GPU support and unmatched compute performance, the AMD FirePro S9150 ushers in a new era of supercomputing. Its 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.'
Designed to win market share from Nvidia's top-end Tesla K40, the FirePro S9150 boasts 5.07 teraflops of compute performance in single-precision mode and 2.53 teraflops in double-precision. Compared to the Tesla K40's 4.29 teraflops and 1.43 teraflops respectively, those are impressive figures - and doubly so when you learn the card has the same thermal design profile (TDP) of 235W, despite an extra 4GB of GDDR5 memory for 16GB total. Its performance comes courtesy of 44 Graphics Core Next (GCN) compute units, totalling 2,816 stream processors.
The FirePro S9150 HPC accelerator is to be joined by a server variant dubbed the FirePro S9050, dropping to 12GB of GDDR5 and 28 GCN compute units for 1,792 stream processors total and a 225W TDP. AMD has yet to confirm pricing for either card.
MSI MPG Velox 100R Chassis Review
October 14 2021 | 15:04
Want to comment? Please log in.