1. Calculations conducted by AMD Performance Labs as of Sep 18, 2020 for the AMD Instinct™ MI100 (32GB HBM2 PCIe® card)
accelerator at 1,502 MHz peak boost engine clock resulted in 11.54 TFLOPS peak double precision (FP64), 46.1
TFLOPS peak single precision matrix (FP32), 23.1 TFLOPS peak single precision (FP32), 184.6 TFLOPS peak half precision (FP16) peak
theoretical, floating-point performance. AMD TFLOPS calculations conducted with the following equation
for AMD Instinct MI100 GPUs: FLOPS calculations are performed by taking the peak boost engine clock from the highest DPM state and
multiplying it by xx CUs per GPU. Then, multiplying that number by xx stream processors, which exist in
each CU. Then, that number is multiplied by 2 FLOPS per clock for FP32, 4 FLOPS per clock for FP16, to determine TFLOPS. The FP64
TFLOPS rate for MI100 is calculated using 1/2 rate per clock to calculate the FP64/FMA64 TFLOPS. External
results on the NVidia Ampere A100 (40GB) GPU accelerator resulted in 9.7 TFLOPS peak double precision (FP64). 19.5 TFLOPS peak
single precision (FP32), 78 TFLOPS peak half precision (FP16) theoretical, floating-point performance.
Results found at: https://www.nvidia.com/content/dam/en-zz/Solutions/Data-Center/nvidia-ampere-architecture-whitepaper.pdf
MI100-03. 2. Measurements conducted by AMD Performance Labs as of Sep 18, 2020 on the AMD
Instinct™ MI100 accelerator designed with AMD CDNA 7nm FinFET process technology with 1,502 MHz boost engine clock resulted in
184.57 TFLOPS peak half precision (FP16), 46.14 TFLOPS peak single precision (FP32 Matrix) floating-point
performance. The results calculated for Radeon Instinct™ MI50 GPU designed with Vega 7nm FinFET process technology with 1,725 MHz
engine clock resulted in 26.5 TFLOPS peak half precision (FP16), 13.25 TFLOPS peak single precision
(FP32 Matrix) floating-point performance. MI100-04. 3.Works with PCIe® Gen 4.0 and Gen 3.0 compliant motherboards. Performance may
vary from motherboard to motherboard. Refer to system or motherboard provider for individual
product performance and features. 4. ECC support on AMD Instinct™ GPU cards, based on the “AMD CDNA” technology includes full-chip
ECC including HBM2 memory and internal GPU structures. 5. Expanded RAS (Reliability, availability
and serviceability) attributes have been added to the AMDs Instinct™ “AMD CDNA” 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. 6. The AMD
Instinct™ accelerator products come with a three-year limited warranty. Please visit www.AMD.com/warranty page for warranty
details on the specific graphics products purchased. Toll-free phone service available in the U.S. and Canada
only, email access is global. 7. As of SEP 18th, 2020. AMD Instinct™ MI100 built on AMD CDNA technology accelerators support PCIe®
Gen4 providing up to 64 GB/s peak theoretical transport data bandwidth from CPU to GPU per card. Peak
theoretical transport rate performance is calculated by Baud Rate * width in bytes *# directions = GB/s per card. PCIe Gen4: 16 *
2 * 2 = 64 GB/s. AMD Instinct™ MI100 “CDNA” technology-based accelerators include three Infinity Fabric™ links
providing up to 276 GB/s peak theoretical GPU to GPU or Peer-to-Peer (P2P) transport rate bandwidth performance per GPU card.
Combined with PCIe Gen4 compatibility providing an aggregate GPU card I/O peak bandwidth of up to 340
GB/s. 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. MI100s have three links: 92 GB/s
* 3 links per GPU = 276 GB/s. Four GPU hives provide up to 552 GB/s peak theoretical P2P performance: 276GB/s * 4 GPUs = 1,104
GB/s, divided by 2 = 552 GB/s. Dual 4 GPU hives in a server provide up to 1.1 TB/s total peak theoretical direct
P2P performance per server: 552 GB/s per 4 GPU hive * 2 = 1,104 GB/s. AMD Infinity Fabric link technology not enabled: Four GPU
hives provide up to 256 GB/s peak theoretical P2P performance with PCIe® 4.0: 64GB/s per GPU * 4 GPUs = 256
GB/s. Radeon Instinct™ MI50 “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. Radeon Instinct™ MI50 “Vega 7nm” technologybased 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. MI50 based four GPU hives provide up to 368 GB/s peak theoretical P2P
performance: 184 GB/s * 4 GPUs = 736 GB/s, divided by 2 = 368 GB/s. Dual 4 GPU hives in a server provide up
to 736 GB/s total peak theoretical direct P2P performance per server: 368 GB/s per 4 GPU hive * 2 = 736 GB/s.Performance
guidelines are estimated only and may vary. Refer to server manufacture PCIe Gen4 compatibility and performance
guidelines for potential peak performance of the specified server model numbers. Server manufacturers may vary configuration
offerings yielding different results. https://pcisig.com/ , https://www.chipestimate.com/PCI-Ex- press-Gen-4-
a-Big-Pipe-for-Big-Data/Cadence/Technical-Article/2014/04/15,
https://www.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. MI100-06. 8. Calculations by AMD Performance Labs as of OCT 5th, 2020 for
the AMD Instinct™ MI100 accelerator designed with AMD CDNA 7nm FinFET process technology at 1,200 MHz
peak memory clock resulted in 1.2288 TFLOPS peak theoretical memory bandwidth performance. MI100 memory bus interface is 4,096
bits and memory data rate is 2.40 Gbps. The results calculated for Radeon Instinct™ MI50 GPU designed
with “Vega” 7nm FinFET process technology with 1,000 MHz peak memory clock resulted in 1.024 TFLOPS peak theoretical memory
bandwidth performance. MI50 memory bus interface is 4,096 bits and memory data rate is 2.00 Gbps.
AMD memory bandwidth TFLOPs calculations, (x.xx Gbps memory data rate * xxxx bits memory bus interface memory bus interface)/8.
MI100 memory bandwidth = (2.40 Gbps*4,096 bits)/8 = 1.2288 TB/s. MI50 memory bandwidth = (2.00
Gbps*4,096 bits)/8 = 1.024 TB/s. MI100 1.2288/ MI50 1.024 = 1.2x (or 20% higher). CDNA-04. 9. The AMD Instinct™ MI100 accelerator
has 120 compute units (CUs) and 7,680 stream cores. The Radeon Instinct™ MI50 GPU has 60 CUs and
3,840 stream cores. MI100 120 CUs divided by MI50 60 CUs = 2x CUs CDNA-02. 10. Calculations by AMD Performance Labs as of Oct 5,
2020 using the AMD Instinct™ MI100 accelerator with CDNA 7nm FinFet process technology at 1,502 MHz
peak boost engine clock resulted in 11.535 TFLOPS peak theoretical double precision (FP64) floating-point performance. The results
calculated for Radeon Instinct™ MI50 GPU designed with “Vega” 7nm FinFET process technology with 1,725
MHz peak engine clock resulted in 6.62 TFLOPS peak theoretical double precision (FP64) floating-point performance. CDNA-03
item | value |
Application | Workstation, Laptop, Desktop |
Item Condition | PCI Express 3.0 X16 |
Outputs | DisplayPort |
Cooler Type | Fan |
Products Status | New |
Chipset Manufacturer | AMD |
Interface Type | PCI EXPRESS |
Brand Name | AMD |
Place of Origin | China |
Memory Size | 32GB HBM2 |
Memory Interface | 4,096Bits |
Memory Clock | 1.2 GHz |
Memory Bandwidth | Up to 1.2 TB/s |
Board Form Factor | Full-Height, Dual Slot |
Length | 10.5" Long |
Thermal | Passively Cooled |
Max Power | 300W TDP |
For More Information Visit:
INSTINCTMI100
INSTINCTMI100
