Intel Gaudi 3 vs NVIDIA B200
Intel Gaudi 3
NVIDIA B200
|
16384 Shaders
128GB HBM2e
438MHz
|
18944 Shaders
192GB HBM3e
1980MHz
|
|
Peak AI Performance
1.84 PFLOPS
FP8 Tensor (FP32 Accumulate)
|
Peak AI Performance
18 PFLOPS
FP4 Tensor Sparse
|
|
FP32
14.35 TFLOPS
|
FP32
75.02 TFLOPS
|
|
FP16
28.7 TFLOPS
|
FP16
150.04 TFLOPS
|
|
Form Factor
OAM Module
-
|
Form Factor
SXM6
-
|
|
TDP
900W
|
TDP
1000W
|
|
-
-
-
-
-
|
-
-
-
-
-
|
|
GB6 OpenCL N/A
0%
|
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
1.84 PFLOPS
FP8 Tensor (FP32 Accumulate)
|
Peak AI
Performance
18 PFLOPS
FP4 Tensor Sparse
|
|
-
-
-
|
FP4
9 PFLOPS
Tensor
18 PFLOPS
Tensor Sparse
|
|
FP8
-
-
-
1.84 PFLOPS
Tensor (FP32 Accumulate)
-
|
FP8
-
4.5 PFLOPS
Tensor (FP16 Accumulate)
9 PFLOPS
Tensor (FP16 Accumulate) Sparse
4.5 PFLOPS
Tensor (FP32 Accumulate)
9 PFLOPS
Tensor (FP32 Accumulate) Sparse
|
|
FP16
28.7 TFLOPS
-
-
458.75 TFLOPS
Tensor (FP32 Accumulate)
-
|
FP16
150.04 TFLOPS
2.25 PFLOPS
Tensor (FP16 Accumulate)
4.5 PFLOPS
Tensor (FP16 Accumulate) Sparse
2.25 PFLOPS
Tensor (FP32 Accumulate)
4.5 PFLOPS
Tensor (FP32 Accumulate) Sparse
|
|
FP32
14.35 TFLOPS
229.38 TFLOPS
Tensor
-
|
FP32
75.02 TFLOPS
-
-
|
|
-
-
-
|
FP64
37.51 TFLOPS
35.16 TFLOPS
Tensor
|
|
BF16
28.7 TFLOPS
1.84 PFLOPS
Tensor
-
|
BF16
150.04 TFLOPS
2.25 PFLOPS
Tensor
4.5 PFLOPS
Tensor Sparse
|
|
TF32
458.75 TFLOPS
Tensor
-
|
TF32
1.13 PFLOPS
Tensor
2.25 PFLOPS
Tensor Sparse
|
|
-
-
-
|
-
-
-
|
|
-
-
-
-
|
INT8
-
4.5 POPS
Tensor
9 POPS
Tensor Sparse
|
|
-
-
|
INT32
75.02 TOPS
|
|
-
-
|
-
-
|
|
Pixel
Fillrate
-
|
Pixel
Fillrate
63.36 GPixel/s
|
|
-
-
|
-
-
|
|
Texture
Fillrate
-
|
Texture
Fillrate
1172.16 GTexel/s
|
|
Manufacturer
Intel
|
Manufacturer
NVIDIA
|
|
Chip Designer
Intel
|
Chip Designer
NVIDIA
|
|
Architecture
Gaudi 3
|
Architecture
Blackwell
|
|
Family
Gaudi
|
Family
Server
|
|
Codename
Gaudi 3
Gaudi 3
-
-
|
Codename
NV190
GB100
Variant
Miranda
|
|
Market Segment
Server
|
Market Segment
Server
|
|
Release Date
4/9/2024
|
Release Date
3/18/2024
|
|
Foundry
TSMC
-
|
Foundry
TSMC
-
|
|
Fabrication Node
N5
-
|
Fabrication Node
4NP
-
|
|
-
-
-
|
Die Size
2x 810 mm²
-
|
|
-
-
|
Transistor Count
2x 104 Billion
-
|
|
-
-
-
|
Transistor Density
128.40M/mm²
-
|
|
Form
OAM Module
|
Form
SXM6
|
|
Shading Units
16384 Shaders
-
|
Shading Units
18944 Shaders
-
|
|
Texture Mapping Units
-
|
Texture Mapping Units
592 TMUs
|
|
Render Output Units
-
|
Render Output Units
32 ROPs
|
|
Tensor Cores
64 T-Cores
|
Tensor Cores
592 T-Cores
|
|
-
-
|
-
-
|
|
-
-
|
Streaming Multiprocessors
148 SMs
|
|
Compute Units
8 CUs
|
-
-
|
|
-
-
|
-
-
|
|
-
-
|
-
-
|
|
-
-
-
438MHz
|
1856MHz Tensor
-
-
1980MHz
|
|
-
-
|
-
-
|
|
L1
-
-
-
Unknown
|
L1
64KB/SM Tex
256KB/SM
-
-
|
|
L2
Unknown
|
L2
64MB Shared
|
|
-
-
-
|
-
-
-
|
|
128GB
HBM2e
ECC
|
192GB
HBM3e
-
|
|
Bus Width
8192Bit
|
Bus Width
8192Bit
|
|
Clock
1808MHz
Transfer Rate
3.6GT/s
Bandwidth
3702.8GB/s
|
Clock
3760MHz
Transfer Rate
7.5GT/s
Bandwidth
7700.5GB/s
|
|
-
-
-
-
eSRAM
96MB
12800GB/s
-
-
|
-
-
-
-
-
-
-
-
-
|
|
TDP
900W
|
TDP
1000W
|
|
-
-
|
-
-
|
|
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
No Ports
|
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
No Ports
|
|
Max Resolution
Unknown
|
Max Resolution
Unknown
|
|
Max Resolution Refresh Rate
-
|
Max Resolution Refresh Rate
-
|
|
Variable Refresh Rate
-
-
-
|
Variable Refresh Rate
G-Sync
FreeSync
-
|
|
Display Stream Compression (DSC)
Not Supported
|
Display Stream Compression (DSC)
Not Supported
|
|
Multi Monitor Support
Unknown
|
Multi Monitor Support
Unknown
|
|
-
-
|
-
-
|
|
No Encoders
-
|
No Encoders
-
|
|
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
|
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
|
|
No Decoders
|
Model
7x NVDEC 6
|
|
-
-
-
-
-
-
-
-
-
-
-
-
-
-
|
Codec
MPEG-1
MPEG-2
MPEG-4
-
VC-1
VP8
VP9
-
AVC (H.264)
HEVC (H.265)
-
AV1
-
-
|
|
-
-
-
-
|
-
-
-
-
|
|
-
-
OpenCL
3.0
-
-
|
-
-
OpenCL
3.0
-
-
|
|
-
-
-
-
-
-
-
-
-
-
|
-
-
CUDA
10.0
-
-
PureVideo HD
VP13
VDPAU
Feature Set M
|
|
Not a Card
-
-
-
|
Not a Card
-
-
-
|
|
-
-
-
-
-
-
-
-
|
-
-
-
-
-
-
-
-
|
|
-
-
PCIe Version
5.0
PCIe Lanes
16
|
-
-
PCIe Version
6.0
PCIe Lanes
16
|
|
Multi GPU Support
Supported
Type
RoCE
|
Multi GPU Support
Supported
Type
NVLink
|
|
-
-
-
-
-
-
|
-
-
-
-
-
-
|
Copy Link