AMD Instinct MI100 vs Intel Gaudi

AMD Instinct MI100
Intel Gaudi
7680 Shaders
32GB HBM2
1502MHz
2048 Shaders
32GB HBM2
610MHz
Peak AI Performance
184.57 TFLOPS
FP16 Tensor (FP16 Accumulate)
Peak AI Performance
159.91 TFLOPS
BF16 Tensor
FP32
23.07 TFLOPS
-
-
FP16
46.14 TFLOPS
-
-
Form Factor
PCIe Card
2.0-Slots
Form Factor
OAM Module
-
TDP
300W
TDP
350W
Power Connectors
-
2x 8-Pin
-
-
-
-
-
-
-
GB6 OpenCL N/A
0%
GB6 Metal N/A
0%
GB6 Metal N/A
0%
GB6 Vulkan N/A
0%
GB6 Vulkan N/A
0%
GB5 OpenCL N/A
0%
GB5 OpenCL N/A
0%
GB5 CUDA N/A
0%
GB5 CUDA N/A
0%
GB5 Metal N/A
0%
GB5 Metal N/A
0%
GB5 Vulkan N/A
0%
GB5 Vulkan N/A
0%
OCT 2020.1 N/A
0%
OCT 2020.1 N/A
0%
OCT Metal N/A
0%
OCT Metal N/A
0%
Peak AI Performance
184.57 TFLOPS
FP16 Tensor (FP16 Accumulate)
Peak AI Performance
159.91 TFLOPS
BF16 Tensor
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
FP16
46.14 TFLOPS
184.57 TFLOPS Tensor (FP16 Accumulate)
-
184.57 TFLOPS Tensor (FP32 Accumulate)
-
-
-
-
-
-
-
FP32
23.07 TFLOPS
46.14 TFLOPS Tensor
-
-
-
-
-
FP64
11.54 TFLOPS
-
-
-
-
BF16
-
92.28 TFLOPS Tensor
-
BF16
-
159.91 TFLOPS Tensor
-
-
-
-
-
-
-
INT4
92.28 TOPS Tensor
-
-
-
-
INT8
-
92.28 TOPS Tensor
-
-
-
-
-
-
-
-
-
-
-
-
-
Pixel Fillrate
-
Pixel Fillrate
-
-
-
-
-
Texture Fillrate
720.96 GTexel/s
Texture Fillrate
-
Manufacturer
AMD
Manufacturer
Intel
Chip Designer
AMD
Chip Designer
Intel
Architecture
CDNA 1
Architecture
Gaudi
Family
Instinct
Family
Gaudi
Codename
Arcturus
-
Variant
Arcturus XL
Codename
Gaudi
HL-2000
Variant
HL-2000
Market Segment
Server
Market Segment
Server
Release Date
11/16/2020
Release Date
6/17/2019
Foundry
TSMC
-
Foundry
TSMC
-
Fabrication Node
N7
-
Fabrication Node
16FF
-
Die Size
750 mm²
-
-
-
-
Transistor Count
25.6 Billion
-
-
-
Transistor Density
34.13M/mm²
-
-
-
-
Form
PCIe Card
Form
OAM Module
Shading Units
7680 Shaders
-
Shading Units
2048 Shaders
-
Texture Mapping Units
480 TMUs
Texture Mapping Units
-
Render Output Units
-
Render Output Units
-
Tensor Cores
480 T-Cores
Tensor Cores
8 T-Cores
-
-
-
-
-
-
-
-
Compute Units
120 CUs
Compute Units
1 CU
-
-
-
-
-
-
-
-
-
-
1000MHz Base
1502MHz
-
-
-
610MHz
-
-
-
-
L1
-
-
16KB/CU
-
L1
-
-
-
Unknown
L2
8MB Shared
L2
Unknown
-
-
-
-
-
-
32GB
HBM2
ECC
32GB
HBM2
ECC
Bus Width
4096Bit
Bus Width
4096Bit
Clock
1200MHz
Transfer Rate
2.4GT/s
Bandwidth
1228.8GB/s
Clock
980MHz
Transfer Rate
2GT/s
Bandwidth
1003.5GB/s
-
-
-
-
-
-
-
-
-
-
-
-
-
eSRAM
24MB
3200GB/s
-
-
TDP
300W
TDP
350W
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
No Ports
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
No Ports
Max Resolution
Unknown
Max Resolution
Unknown
Max Resolution Refresh Rate
-
Max Resolution Refresh Rate
-
Variable Refresh Rate
-
-
-
Variable Refresh Rate
-
-
-
Display Stream Compression (DSC)
Not Supported
Display Stream Compression (DSC)
Not Supported
Multi Monitor Support
Unknown
Multi Monitor Support
Unknown
-
-
-
-
Model
VCN 2.5
No Encoders
-
Codec
-
-
-
-
-
-
-
-
AVC (H.264)
HEVC (H.265)
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Model
VCN 2.5
No Decoders
Codec
MPEG-1
MPEG-2
MPEG-4
JPEG
VC-1
-
VP9
-
AVC (H.264)
HEVC (H.265)
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
OpenCL
2.1
-
-
-
-
OpenCL
3.0
-
-
-
-
-
-
GFX
9.4
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Not a Card
-
-
-
Power Connectors
-
-
-
2x 8-Pin
-
-
-
-
-
-
-
-
-
-
-
Slots Required
2.0
PCIe Version
4.0
PCIe Lanes
16
-
-
PCIe Version
4.0
PCIe Lanes
16
Multi GPU Support
Supported
Type
Infinity Fabric
Multi GPU Support
Supported
Type
RoCE
Height
111 mm (4.37 in)
Width
267 mm (10.51 in)
Depth
37 mm (1.46 in)
-
-
-
-
-
-
Change Comparison