According to the latest news, Pawsey SuperComputing will include AMD’s next-generation Instinct MI200 GPUs and EPYC Milan processors in its Setonix supercomputer. The specifications for AMD’s Instinct MI200 HPC GPU accelerator were revealed last week. The Setnoix supercomputer will be one of the first products to integrate CDNA 2 GPUs with MCM technology.
The supercomputer will include over 200,000 EPYC Milan cores and over 750 new next-generation AMD GPUs (Instinct MI200) based on the CDNA 2 architecture. As we know, the Instinct MI200 or the Aldebaran will feature the CDNA 2 architecture and will be the first GPUs to integrate an MCM design.
Setnoix will feature at least 200,000 EPYC Milan cores, and the total number of processors is expected to be 3,125. Its specifications revealed are Over 548TB of system memory, NVMe Near-node storage, 15PB ClusterStor Luster file system with 2.7PB SSD, 90PB Ceph Storage (60PB with 8 + 3 erase coding), as well as a fully configurable HPC environment by software based on the HPE Cray Sashta Platform. The supercomputer will provide up to 50 PetaFlops of power and cost around $ 70 million.
“We recently received $ 70 million in funding to invest in replacing our IT, network and storage infrastructure. Thanks to recently received funding, we have acquired a brand new HPE Cray supercomputer named after a small marsupial endemic to areas near Perth – the computer is called Setonix – [and] as a separate purchase, we have acquired 60 petabytes of object storage that will soon be integrated into the HPE supercomputer in the future. “
2020 AMD Radeon Instinct Accelerators
|Name of the accelerator||AMD Radeon Instinct MI6||AMD Radeon Instinct MI8||AMD Radeon Instinct MI25||AMD Radeon Instinct MI50||AMD Radeon Instinct MI60||AMD Instinct MI100||AMD Instinct MI200|
|GPU architecture||Polaris 10||Fiji XT||Vega 10||Vega 20||Vega 20||Arcturus (cDNA 1)||Aldebaran (CDNA 2)|
|GPU process node||14nm FinFET||28 nm||14nm FinFET||7nm FinFET||7nm FinFET||7nm FinFET||Advanced process node|
|GPU cores||2304||4096||4096||3840||4096||7680||14 336?|
|GPU clock speed||1237 MHz||1000 MHz||1500 MHz||1725 MHz||1800 MHz||~ 1500 MHz||To be determined|
|FP16 Calculate||5.7 TFLOP||8.2 TFLOP||24.6 TFLOP||26.5 TFLOP||29.5 TFLOP||185 TFLOP||To be determined|
|FP32 calculation||5.7 TFLOP||8.2 TFLOP||12.3 TFLOP||13.3 TFLOP||14.7 TFLOP||23.1 TFLOP||To be determined|
|FP64 calculation||384 GFLOP||512 GFLOP||768 GFLOP||6.6 TFLOP||7.4 TFLOP||11.5 TFLOP||To be determined|
|VRAM||16 GB of GDDR5||4 GB HBM1||16 GB HBM2||16 GB HBM2||32 GB HBM2||32 GB HBM2||64/128 GB HBM2e?|
|Memory clock||1750 MHz||500 MHz||945 MHz||1000 MHz||1000 MHz||1200 MHz||To be determined|
|Memory bus||256-bit bus||4096-bit bus||2048-bit bus||4096-bit bus||4096-bit bus||4096-bit bus||8192 bits|
|Memory bandwidth||224 GB / s||512 GB / s||484 GB / s||1 TB / s||1 TB / s||1.23 TB / s||~ 2 TB / s?|
|Form factor||Single lunge, full length||Double slit, half length||Double slit, full length||Double slit, full length||Double slit, full length||Double slit, full length||Double slit, full length / OAM|
|Cooling||Passive cooling||Passive cooling||Passive cooling||Passive cooling||Passive cooling||Passive cooling||Passive cooling|
|TDP||150W||175W||300W||300W||300W||300W||To be determined|