NVIDIA A100 vs AMD Instinct MI210
NVIDIA A100
AMD Instinct MI210
|
6912 Shaders
80GB HBM2e
1410MHz
|
6656 Shaders
64GB HBM2e
1700MHz
|
|
Peak AI Performance
2.49 POPS
INT4 Tensor Sparse
|
Peak AI Performance
181.04 TOPS
INT4 Tensor
|
|
FP32
19.49 TFLOPS
|
FP32
22.63 TFLOPS
|
|
FP16
77.97 TFLOPS
|
FP16
45.26 TFLOPS
|
|
Form Factor
PCIe Card
2.0-Slots
|
Form Factor
PCIe Card
2.0-Slots
|
|
TDP
250W
|
TDP
300W
|
|
Power Connectors
-
|
Power Connectors
1x 8-Pin
|
Peak AI Performance
|
|
A100 - 2.49 POPS INT4 Tensor Sparse
x13.75
Instinct MI210 - 181.04 TOPS INT4 Tensor
x1
FP32
|
|
A100 - 19.49 TFLOPS FP32
x1
Instinct MI210 - 22.63 TFLOPS FP32
x1.16
FP16
|
|
A100 - 77.97 TFLOPS FP16
x1.72
Instinct MI210 - 45.26 TFLOPS FP16
x1
|
|
A100 - 1935.4GB/s
x1.18
Instinct MI210 - 1638.4GB/s
x1
|
|
A100 - 250W
x1
Instinct MI210 - 300W
x1.2
|
|
A100 - GB6 OpenCL 200,130
x1.44
Instinct MI210 - GB6 OpenCL 139,440
x1
|
0%
|
0%
|
|
GB6 Vulkan N/A
0%
|
|
0%
|
0%
|
|
0%
|
0%
|
|
0%
|
0%
|
|
0%
|
0%
|
|
0%
|
0%
|
|
0%
|
0%
|
|
Manufacturer
NVIDIA
|
Manufacturer
AMD
|
|
Chip Designer
NVIDIA
|
Chip Designer
AMD
|
|
Architecture
Ampere
|
Architecture
CDNA 2
|
|
Family
Server
|
Family
Instinct
|
|
Codename
NV170
GA100
-
-
|
Codename
Aldebaran
Aldebaran XL
Variant
Aldebaran XL
|
|
Market Segment
Server
|
Market Segment
Server
|
|
Release Date
6/28/2021
|
Release Date
3/22/2022
|
|
Foundry
TSMC
|
Foundry
TSMC
|
|
Fabrication Node
7N
|
Fabrication Node
N6
|
|
Die Size
826 mm²
|
Die Size
724 mm²
|
|
Transistor Count
54.2 Billion
|
Transistor Count
28 Billion
|
|
Transistor Density
65.62M/mm²
|
Transistor Density
38.67M/mm²
|
|
Form
PCIe Card
|
Form
PCIe Card
|
|
Shading Units
6912 Shaders
|
Shading Units
6656 Shaders
|
|
Texture Mapping Units
432 TMUs
|
Texture Mapping Units
416 TMUs
|
|
Render Output Units
160 ROPs
|
Render Output Units
-
|
|
Tensor Cores
432 T-Cores
|
Tensor Cores
416 T-Cores
|
|
Streaming Multiprocessors
108 SMs
|
-
-
|
|
-
-
|
Compute Units
104 CUs
|
|
765MHz Base
1410MHz
|
1000MHz Base
1700MHz
|
|
Peak AI Performance
2.49 POPS
INT4 Tensor Sparse
|
Peak AI Performance
181.04 TOPS
INT4 Tensor
|
|
FP16
77.97 TFLOPS
311.87 TFLOPS Tensor (FP16 Accumulate)
623.74 TFLOPS Tensor (FP16 Accumulate) Sparse
311.87 TFLOPS Tensor (FP32 Accumulate)
623.74 TFLOPS Tensor (FP32 Accumulate) Sparse
|
FP16
45.26 TFLOPS
181.04 TFLOPS Tensor (FP16 Accumulate)
-
181.04 TFLOPS Tensor (FP32 Accumulate)
-
|
|
FP32
19.49 TFLOPS
-
|
FP32
22.63 TFLOPS
45.26 TFLOPS Tensor
|
|
FP64
9.75 TFLOPS
19.49 TFLOPS Tensor
|
FP64
22.63 TFLOPS
45.26 TFLOPS Tensor
|
|
BF16
38.98 TFLOPS
311.87 TFLOPS Tensor
623.74 TFLOPS Tensor Sparse
|
BF16
-
181.04 TFLOPS Tensor
-
|
|
TF32
155.94 TFLOPS Tensor
311.87 TFLOPS Tensor Sparse
|
-
-
-
|
|
INT4
1.25 POPS Tensor
2.49 POPS Tensor Sparse
|
INT4
181.04 TOPS Tensor
-
|
|
INT8
623.74 TOPS Tensor
1.25 POPS Tensor Sparse
|
INT8
181.04 TOPS Tensor
-
|
|
INT32
19.49 TOPS
|
-
-
|
|
Pixel Fillrate
225.6 GPixel/s
|
Pixel Fillrate
-
|
|
Texture Fillrate
609.12 GTexel/s
|
Texture Fillrate
707.2 GTexel/s
|
|
L1
64KB/SM Tex
192KB/SM
-
|
L1
-
-
16KB/CU
|
|
L2
40MB Shared
|
L2
8MB Shared
|
|
80GB
HBM2e
-
|
64GB
HBM2e
ECC
|
|
Bus Width
5120Bit
|
Bus Width
4096Bit
|
|
Clock
1512MHz
Transfer Rate
3GT/s
Bandwidth
1935.4GB/s
|
Clock
1600MHz
Transfer Rate
3.2GT/s
Bandwidth
1638.4GB/s
|
|
TDP
250W
|
TDP
300W
|
|
Max Resolution
Unknown
|
Max Resolution
Unknown
|
|
Max Resolution Refresh Rate
|
Max Resolution Refresh Rate
|
|
Variable Refresh Rate
G-Sync
FreeSync
|
Variable Refresh Rate
-
-
|
|
Display Stream Compression (DSC)
Not Supported
|
Display Stream Compression (DSC)
Not Supported
|
|
Multi Monitor Support
Unknown
|
Multi Monitor Support
Unknown
|
|
No Encoders
-
|
Model
VCN 2.6
|
|
-
-
-
|
Codec
AVC (H.264)
HEVC (H.265)
|
|
Model
5x NVDEC 4
|
No Decoders
|
|
Codec
MPEG-1
MPEG-2
MPEG-4
VC-1
VP8
VP9
AVC (H.264)
HEVC (H.265)
|
-
-
-
-
-
-
-
-
|
|
OpenCL
3.0
Vulkan
1.2
|
OpenCL
3.0
-
-
|
|
CUDA
8.0
-
-
PureVideo HD
VP10
VDPAU
Feature Set J
|
-
-
GFX
9.4
-
-
-
-
|
|
Power Connectors
-
1x 8-Pin EPS
|
Power Connectors
1x 8-Pin
-
|
|
Slots Required
2.0
PCIe Version
4.0
PCIe Lanes
16
|
Slots Required
2.0
PCIe Version
4.0
PCIe Lanes
16
|
|
Multi GPU Support
Supported
Type
NVLink
|
Multi GPU Support
Supported
Type
Infinity Fabric
|
|
Height
111 mm (4.37 in)
Width
267 mm (10.51 in)
Depth
40 mm (1.57 in)
|
Height
111 mm (4.37 in)
Width
267 mm (10.51 in)
Depth
37 mm (1.46 in)
|
Copy Link