Filed under: Tech

AMD Instinct MI200 ‘Aldebaran’ GPUs With 128 GB Memory &…

The Setonix supercomputer from the Pawsey SuperComputing firm is all established to contain AMD’s up coming-gen Intuition MI200 GPUs and EPYC Milan CPUs. The technical specs for Setonix supercomputer had been confirmed by CTO, Ugo Varetto, in the course of ISC 2021.

Setnoix Supercomputer To Be Powered by AMD’s Future-Gen Intuition MI200 128 GB GPUs & EPYC Milan CPUs

The specs of AMD’s Instinct MI200 HPC GPU accelerator ended up uncovered last week. It seems to be like the Setnoix supercomputer is now all set to be one of the initial items to characteristic the CDNA 2 GPUs with MCM technological know-how. The Setnoix supercomputer will be showcasing over two hundred,000 EPYC Milan cores and far more than 750 of the brand name new AMD MI-Next GPUs (Intuition MI200) based on the CDNA two architecture.

The AMD Intuition MI200 GPUs showcasing the CDNA two architecture are codenamed Aldebaran and will be the 1st GPUs to feature an MCM layout. You can locate far more particulars on that below. The GPU will aspect 128 GB of HBM2e memory across an 8192-little bit bus interface. Given that there are at least 750 GPUs showcased in the Setnoix GPU, there’s a overall of 96 Terabytes of graphics memory obtainable on the program.

In conditions of CPUs, the procedure will feature at minimum 200,000 EPYC Milan cores, and looking at that these will most probably be sixty four core areas, we can assume the overall quantity of CPUs to be 3125. Other technical specs consist of 548+ TB of technique memory, In the vicinity of-node NVMe storage, fifteen PB of ClusterStor Lustre filesystem with 2.7 PB SSDs, ninety PB Ceph Storage (60 PB with 8+3 erasure coding) & a totally program configurable HPC environment centered close to the HPE Cray Sashta system. When all is set jointly, the Setnoix supercomputer will provide up to 50 PetaFlops of horsepower & will cost all over 70 Million pounds.

“We recently gained 70 million bucks in funding to make investments in replacing our computing, community and storage infrastructure,” Varetto stated. “With the recently obtained funding, we acquired a brand new HPE Cray supercomputer named soon after a modest marsupial endemic to regions shut to Perth” – the laptop is named Setonix – “[and] in a individual procurement, we obtained sixty petabytes of object storage which will be quickly be integrated with the HPE supercomputer in the long run.”

The AMD driven Setnoix supercomputer will be employed to explore new galaxies and also aid acquire improved diagnostic checks for COVID-19. Presently, the Pawsey Center in Australia is dependent on its Magnus and Galaxy programs which put together have a complete processing electrical power of just 1.eighty three PetaFlops.

AMD Radeon Intuition Accelerators 2020

Accelerator Identify AMD Radeon Intuition MI6 AMD Radeon Instinct MI8 AMD Radeon Instinct MI25 AMD Radeon Instinct MI50 AMD Radeon Instinct MI60 AMD Intuition MI100 AMD Instinct MI200
GPU Architecture Polaris 10 Fiji XT Vega 10 Vega twenty Vega 20 Arcturus (CDNA 1) Aldebaran (CDNA 2)
GPU Procedure Node 14nm FinFET 28nm 14nm FinFET 7nm FinFET 7nm FinFET 7nm FinFET Sophisticated Procedure Node
GPU Cores 2304 4096 4096 3840 4096 7680 fourteen,336?
GPU Clock Speed 1237 MHz 1000 MHz 1500 MHz 1725 MHz 1800 MHz ~1500 MHz TBA
FP16 Compute 5.7 TFLOPs 8.two TFLOPs 24.6 TFLOPs 26.five TFLOPs 29.5 TFLOPs 185 TFLOPs TBA
FP32 Compute five.7 TFLOPs eight.2 TFLOPs twelve.3 TFLOPs thirteen.three TFLOPs fourteen.seven TFLOPs 23.1 TFLOPs TBA
FP64 Compute 384 GFLOPs 512 GFLOPs 768 GFLOPs six.six TFLOPs 7.4 TFLOPs 11.five TFLOPs TBA
VRAM sixteen GB GDDR5 four GB HBM1 sixteen GB HBM2 16 GB HBM2 32 GB HBM2 32 GB HBM2 64/128 GB HBM2e?
Memory Clock 1750 MHz five hundred MHz 945 MHz 1000 MHz one thousand MHz 1200 MHz TBA
Memory Bus 256-bit bus 4096-little bit bus 2048-little bit bus 4096-bit bus 4096-little bit bus 4096-little bit bus 8192-little bit
Memory Bandwidth 224 GB/s 512 GB/s 484 GB/s 1 TB/s one TB/s 1.23 TB/s ~two TB/s?
Form Factor Solitary Slot, Complete Length Dual Slot, Fifty percent Duration Dual Slot, Whole Length Twin Slot, Total Duration Dual Slot, Full Size Dual Slot, Whole Size Twin Slot, Total Size / OAM
Cooling Passive Cooling Passive Cooling Passive Cooling Passive Cooling Passive Cooling Passive Cooling Passive Cooling
TDP 150W 175W 300W 300W 300W 300W TBA

The submit AMD Intuition MI200 ‘Aldebaran’ GPUs With 128 GB Memory & EPYC Milan CPUs To Electricity Setonix Supercomputer With fifty PetaFlops Horsepower by Hassan Mujtaba appeared 1st on Wccftech.