MI50 AMD | Alldatasheet
Document overview
- Manufacturer or author: Provided By ALLDATASHEET.COM(FREE DATASHEET DOWNLOAD SITE)
- PDF pages: 2
Technical content
The AMD Radeon Instinct™ MI50 server accelerator designed on the world’s First 7nm FinFET technology process brings customers a full-feature set based on the industry’s newest technologies. The MI50 is AMD’s workhorse accelerator offering that is ideal for large scale deep learning. Delivering up to 26.5 TFLOPS of native half-precision (FP16) or up to 13.3 TFLOPS single-precision (FP32) peak floating point performance and INT8 support and combined with 16GB or 32GB of high-bandwidth HBM2 ECC memory, the AMD Radeon Instinct™ MI50 brings customers finely balanced performance needed for enterprise-class, mid-range compute capable of training complex neural networks for a variety of demanding deep learning applications in a cost-effective design Combine this outstanding performance with the MI50’s advanced I/O capabilities, with support for PCIe® Gen 4 technology and the addition of AMD’s dual Infinity Fabric™ Link technology providing up to 248 GB/s of aggregate GPU data transfer bandwidth1; and the advanced efficiencies achieved with the use of AMD’s 2nd Gen “Vega” 7nm technology; and you truly have one of the markets’ most advanced, working-class server GPUs available for a variety of deep learning, cloud and general purpose HPC. AMD RADEON INSTINCT™ MI50 UNLEASH DISCOVERY Optimized Deep Learning Operations Optimized Deep Learning Operations with flexible mixed- precision FP16, FP32, INT8 & INT4 capabilities bring customers supercharged compute performance to meet today’s demanding system requirements of handling large data efficiently for training complex neural networks and running inference against those networks used in deep learning. AMD Infinity Fabric Link Two Infinity Fabric Links per GPU for high speed Direct-Connect GPU hives delivering up to 184 GB/s GPU theoretical peer-to-peer bandwidth – 4.75x faster than using PCIe 3.0 alone.1 Rapid FP64 Performance for HPC Workloads The Radeon Instinct MI50 delivers great double precision performance with up to 6.6 TFLOPS FP64 performance, enabling scientists and researchers across the globe to more efficiently process HPC parallel codes across several industries including life sciences, energy, finance, automotive and aerospace, academics, government, defense and more. Ultra-Fast HBM2 Memory With 16GB or 32GB HBM2 memory utilizing a four-stack memory configuration, the new AMD Radeon Instinct MI50 delivers ultra-high memory bandwidth with up to 1 TB/s memory bandwidth. AMD Radeon Instinct MI50 Ready Servers Product: Altus XE4218GT Form Factor: 4U/2P/8GPU/PCIe Gen 4 + Switch Specs: 2x 2nd Gen AMD EPYC™ Processors + 8x AMD Radeon Instinct™ MI50 GPU accelerators Product: G292-Z20 Form Factor: 2U/1P/8GPU/PCIe Gen 4 + Switch Specs: 1x 2nd Gen AMD EPYC™ Processor + 8x AMD Radeon Instinct™ MI50 GPU accelerators Product: G292-Z40 Form Factor: 2U/2P/8GPU/PCIe Gen 4 + Switch Specs: 2x 2nd Gen AMD EPYC™ Processors + 8x AMD Radeon Instinct™ MI50 GPU accelerators Product: G482-Z52 Form Factor: 4U/2P/8GPU/PCIe Gen 4 Specs: 2x 2nd Gen AMD EPYC™ Processors + 8x AMD Radeon Instinct™ MI50 GPU accelerators Product: G482-Z51 Form Factor: 4U/2P/8GPU/PCIe Gen 4 Specs: 2x 2nd Gen AMD EPYC™ Processors + 8x AMD Radeon Instinct™ MI50 GPU accelerators
©2019 Advanced Micro Devices, Inc. All rights reserved. AMD, the AMD Arrow logo, Radeon, and combinations thereof are trademarks of Advanced Micro Devices, Inc. Thunderbolt is a trademark of Intel Corporation or its subsidiaries in the U.S. and/or other countries. Other product names used in this publication are for identification purposes only and may be trademarks of their respective companies. 1. As of Jun 19, 2019. Radeon Instinct™ MI50 and MI60 “Vega 7nm” technology-based accelerators support PCIe® Gen 4.0 providing up to 64 GB/s peak theoretical transport data bandwidth from CPU to GPU per card. Previous Gen Radeon Instinct compute GPU cards are based on PCIe Gen 3.0 providing up to 32 GB/s peak theoretical transport rate bandwidth performance. Peak theoretical transport rate performance is calculated by Baud Rate * width in bytes *# directions = GB/s per card. PCIe Gen3: 8 * 2 * 2 = 32 GB/s. PCIe Gen4: 16 * 2 * 2 = 64 GB/s. Radeon Instinct™ MI50 and MI60 “Vega 7nm” technology-based accelerators include dual Infinity Fabric™ Links providing up to 184 GB/s peak theoretical GPU to GPU or Peer-to-Peer (P2P) transport rate bandwidth performance per GPU card. Combined with PCIe Gen 4 compatibility providing an aggregate GPU card I/O peak bandwidth of up to 248 GB/s. Performance guidelines are estimated only and may vary. Previous Gen Radeon Instinct compute GPU cards provide up to 32 GB/s peak PCIe Gen 3.0 bandwidth performance. Infinity Fabric Link technology peak theoretical transport rate performance is calculated by Baud Rate * width in bytes *# directions *# links = GB/s per card. Infinity Fabric Link: 23 * 2 * 2 = 92 GB/s. MI50 |MI60 each have two links: 92 GB/s * 2 links per GPU = 184 GB/s. Refer to server manufacture PCIe Gen 4.0 compatibility and performance guidelines for potential peak performance of the specified server model numbers. tomshardware.com/news/pcie-4.0-power-speed-express,32525.html AMD has not independently tested or verified external/third party results/data and bears no responsibility for any errors or omissions therein. RIV-18 3. ECC support on 2nd Gen Radeon Instinct™ GPU cards, based on the “Vega 7nm” technology has been extended to full-chip ECC including HBM2 memory and internal GPU structures. 4. Expanded RAS (Reliability, availability and serviceability) attributes have been added to AMD’s 2nd Gen Radeon Instinct™ Vega 7nm technology-based GPU cards and their supporting ecosystem including software, firmware and system level features. AMD’s remote manageability capabilities using advanced out-of-band circuitry allow for easier GPU monitoring via I2C, regardless of the GPU state. For full system RAS capabilities, refer to the system manufacturer’s guidelines for specific system models. and Canada only, email access is global. © 2020 Advanced Micro Devices, Inc. All rights reserved. AMD, the AMD Arrow logo, Radeon, Radeon Instinct, Infinity Fabric, and combinations thereof are trademarks of Advanced Micro Devices, Inc. OpenCL is a trademark of Apple Inc. used by permission by Khronos. PCIe and PCI Express are registered trademarks of PCI-SIG Corporation. Other product names used in this publication are for identification purposes only and may be trademarks of their respective companies. ROCm Open Ecosystem The ROCm Open eCosystem, is an open-source HPC/Hyperscale-class platform for GPU computing that’s also programming-language independent. ROCm brings choice, minimalism and modular software development to GPU computing. Deep Learning and HPC Applications Cloud / Hyperscale Financial Services Energy Life Sciences Automotive HPC Open Programing Languages OpenMP , HIP , OpenCL and Python Key Features PERFORMANCE Compute Units Stream Processors Peak INT8 Peak FP16 Peak FP32 Peak FP64 Bus Interface 3,840 Up to 53.6 TOPS Up to 26.5 TFLOPS Up to 13.3 TFLOPS Up to 6.6 TFLOPS PCIe®Gen 3 and Gen 4 Supported2 MEMORY Memory Size Memory Interface Memory Clock Memory Band- width 16GB or 32GB HBM2 4,096-Bits
1 GHz
ECC (Full-chip) RAS Support Yes3 Yes4 SCALABILITY Infinity Fabric Links OS Support ROCm Compatible Linux® 64-bit Yes BOARD DESIGN Board Form Factor Length Thermal Max Power Warranty Full-Height, Dual Slot 10.5” Long Passively Cooled 300W TDP Three Year Limited5 For More Information Visit: AMD.com/INSTINCT , AMD.com/ROCm and AMD.com/HPC