AMD Instinct MI200 Reportedly Features CDNA 2 MCM Architecture & Launches In Late 2022

AMD launched its first CDNA GPU-based accelerator, the Instinct MI100, terminal year and the latest reports suggest that its successor, the MI200, could launch this yr. The AMD Instinct accelerators feature the CDNA architecture that is designed for compute-intensive and HPC workloads and the 2nd gen of these chips is going to bring a lot of new features and technologies.

AMD Instinct MI200 Reportedly Features CDNA ii MCM Compages & Launches In Late 2021

The details were posted by Komachi over at his Twitter feed and from the looks of it, the Instinct MI200 will be replacing the MI100 as the new flagship accelerator for HPC. Based on the naming convention itself, it looks similar MI200 will be twice as fast as Instinct MI100 but to attain this, AMD is expected to become all out with its MCM design.

The Instinct MI200 is expected to exist launching alongside AMD'south EPYC Trento chips. The Trento line is going to exist a slight variation of the Milan fries for the Supercomputer sector. They are not going to be similar to Genoa every bit that features the side by side-gen Zen 4 cores but information technology should offering a range of new capabilities which nosotros will learn close to launch. As for MI200, information technology is expected to launch in 2021 only we should wait a late Q4 arrival since MI100 was introduced in a similar timeframe (Q4 2020).

Other than that, the AMD MI200 is listed as an MCM 'Special FIO Accelerator' for HPE Cray EX. That could explain where the MI200 name comes from. An MCM (Multi-Fleck-Module) blueprint could mean that we could still exist looking at a Vega Compute GPU merely two of them fused over the same PCB & connected by a next-generation Infinity Fabric interconnect.

AMD Radeon Instinct Accelerators 2020

Accelerator Name	AMD Instinct MI300	AMD Instinct MI250X	AMD Instinct MI250	AMD Instinct MI210	AMD Instinct MI100	AMD Radeon Instinct MI60	AMD Radeon Instinct MI50	AMD Radeon Instinct MI25	AMD Radeon Instinct MI8	AMD Radeon Instinct MI6
GPU Architecture	TBA (CDNA 3)	Aldebaran (CDNA 2)	Aldebaran (CDNA 2)	Aldebaran (CDNA 2)	Arcturus (CDNA 1)	Vega twenty	Vega 20	Vega 10	Fiji XT	Polaris x
GPU Process Node	Avant-garde Procedure Node	6nm	6nm	6nm	7nm FinFET	7nm FinFET	7nm FinFET	14nm FinFET	28nm	14nm FinFET
GPU Dies	4 (MCM)?	2 (MCM)	2 (MCM)	ane (MCM)	one (Monolithic)	1 (Monolithic)	ane (Monolithic)	1 (Monolithic)	ane (Monolithic)	1 (Monolithic)
GPU Cores	28,160?	fourteen,080	xiii,312	6656	7680	4096	3840	4096	4096	2304
GPU Clock Speed	TBA	1700 MHz	1700 MHz	~1700 MHz?	~1500 MHz	1800 MHz	1725 MHz	1500 MHz	grand MHz	1237 MHz
FP16 Compute	TBA	383 TOPs	362 TOPs	~176 TOPs	185 TFLOPs	29.five TFLOPs	26.5 TFLOPs	24.half dozen TFLOPs	viii.2 TFLOPs	5.7 TFLOPs
FP32 Compute	TBA	95.vii TFLOPs	xc.5 TFLOPs	~44 TFLOPs	23.i TFLOPs	14.7 TFLOPs	13.3 TFLOPs	12.3 TFLOPs	8.2 TFLOPs	5.7 TFLOPs
FP64 Compute	TBA	47.nine TFLOPs	45.3 TFLOPs	~22 TFLOPs	11.v TFLOPs	7.4 TFLOPs	half-dozen.vi TFLOPs	768 GFLOPs	512 GFLOPs	384 GFLOPs
VRAM	TBA	128 GB HBM2e	128 GB HBM2e	64 GB HBM2e	32 GB HBM2	32 GB HBM2	16 GB HBM2	16 GB HBM2	4 GB HBM1	16 GB GDDR5
Retentiveness Clock	TBA	iii.2 Gbps	3.2 Gbps	3.2 Gbps?	1200 MHz	1000 MHz	1000 MHz	945 MHz	500 MHz	1750 MHz
Memory Bus	TBA	8192-chip	8192-flake	4096-flake	4096-bit double-decker	4096-bit omnibus	4096-flake coach	2048-bit coach	4096-bit double-decker	256-bit bus
Memory Bandwidth	TBA	3.ii TB/s	3.2 TB/s	1.vi TB/due south	ane.23 TB/s	1 TB/south	1 TB/southward	484 GB/s	512 GB/s	224 GB/s
Course Gene	TBA	OAM	OAM	Dual Slot Card	Dual Slot, Full Length	Dual Slot, Total Length	Dual Slot, Total Length	Dual Slot, Full Length	Dual Slot, One-half Length	Single Slot, Full Length
Cooling	TBA	Passive Cooling	Passive Cooling	Passive Cooling	Passive Cooling	Passive Cooling	Passive Cooling	Passive Cooling	Passive Cooling	Passive Cooling
TDP	TBA	560W	500W?	300W?	300W	300W	300W	300W	175W	150W

Practise note that AMD'due south CDNA two GPU will be fabricated on a make new process node & are confirmed to feature a 3rd Generation AMD Infinity architecture that extends to Exascale past allowing up to 8-Mode coherent GPU connectivity.

According to Coelacanth'south Dream (via Videocardz), it is reported that the MI200 accelerator will utilise a new variant of the GFX900 GPU architecture (Vega) known every bit GFX90A which adds support for full-rate FP64 compute capabilities. Once once again, the AMD MI200 accelerator sounds like it volition go ahead with NVIDIA's time to come Lovelace MCM GPUs for HPC market place.

Source: https://wccftech.com/amd-instinct-mi200-cdna-2-gpu-accelerator-mcm-architecture-2021-launch-rumor/

Posted by: cruzwhicess.blogspot.com

Share this post