Intel Gaudi 3 vs NVIDIA A2

Intel Gaudi 3
NVIDIA A2
16384 Shaders
128GB HBM2e
438MHz
1280 Shaders
16GB GDDR6
1770MHz
Peak AI Performance
1.84 PFLOPS
FP8 Tensor (FP32 Accumulate)
Peak AI Performance
145 TOPS
INT4 Tensor Sparse
FP32
14.35 TFLOPS
FP32
4.53 TFLOPS
FP16
28.7 TFLOPS
FP16
4.53 TFLOPS
Form Factor
OAM Module
-
Form Factor
PCIe Card
1.0-Slots
TDP
900W
TDP
60W
-
-
-
-
-
-
-
-
-
-
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
145 TOPS
INT4 Tensor Sparse
-
-
-
-
-
-
FP8
-
-
-
1.84 PFLOPS Tensor (FP32 Accumulate)
-
-
-
-
-
-
-
FP16
28.7 TFLOPS
-
-
458.75 TFLOPS Tensor (FP32 Accumulate)
-
FP16
4.53 TFLOPS
18.13 TFLOPS Tensor (FP16 Accumulate)
36.25 TFLOPS Tensor (FP16 Accumulate) Sparse
18.13 TFLOPS Tensor (FP32 Accumulate)
36.25 TFLOPS Tensor (FP32 Accumulate) Sparse
FP32
14.35 TFLOPS
229.38 TFLOPS Tensor
-
FP32
4.53 TFLOPS
-
-
-
-
-
FP64
70 GFLOPS
-
BF16
28.7 TFLOPS
1.84 PFLOPS Tensor
-
BF16
4.53 TFLOPS
18.13 TFLOPS Tensor
36.25 TFLOPS Tensor Sparse
TF32
458.75 TFLOPS Tensor
-
TF32
9.06 TFLOPS Tensor
18.12 TFLOPS Tensor Sparse
-
-
-
INT4
72.5 TOPS Tensor
145 TOPS Tensor Sparse
-
-
-
-
INT8
-
36.25 TOPS Tensor
72.5 TOPS Tensor Sparse
-
-
INT32
2.27 TOPS
-
-
Ray Tracing
8.9 TOPS
Pixel Fillrate
-
Pixel Fillrate
56.64 GPixel/s
-
-
-
-
Texture Fillrate
-
Texture Fillrate
70.8 GTexel/s
Manufacturer
Intel
Manufacturer
NVIDIA
Chip Designer
Intel
Chip Designer
NVIDIA
Architecture
Gaudi 3
Architecture
Ampere
Family
Gaudi
Family
Server
Codename
Gaudi 3
Gaudi 3
-
-
Codename
NV177
GA107
-
-
Market Segment
Server
Market Segment
Server
Release Date
4/9/2024
Release Date
11/10/2021
Foundry
TSMC
-
Foundry
Samsung
-
Fabrication Node
N5
-
Fabrication Node
8N
-
-
-
-
Die Size
200 mm²
-
-
-
Transistor Count
8.7 Billion
-
-
-
-
Transistor Density
43.50M/mm²
-
Form
OAM Module
Form
PCIe Card
Shading Units
16384 Shaders
-
Shading Units
1280 Shaders
-
Texture Mapping Units
-
Texture Mapping Units
40 TMUs
Render Output Units
-
Render Output Units
32 ROPs
Tensor Cores
64 T-Cores
Tensor Cores
40 T-Cores
-
-
Ray-Tracing Cores
10 RT-Cores
-
-
Streaming Multiprocessors
10 SMs
Compute Units
8 CUs
-
-
-
-
-
-
-
-
-
-
-
-
-
438MHz
-
-
1440MHz Base
1770MHz
-
-
-
-
L1
-
-
-
Unknown
L1
64KB/SM Tex
128KB/SM
-
-
L2
Unknown
L2
2MB Shared
-
-
-
-
-
-
128GB
HBM2e
ECC
16GB
GDDR6
-
Bus Width
8192Bit
Bus Width
128Bit
Clock
1808MHz
Transfer Rate
3.6GT/s
Bandwidth
3702.8GB/s
Clock
782MHz
Transfer Rate
6.3GT/s
Bandwidth
100GB/s
-
-
-
-
eSRAM
96MB
12800GB/s
-
-
-
-
-
-
-
-
-
-
-
TDP
900W
TDP
60W
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
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
-
Model
NVENC 7
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Codec
-
-
-
-
-
-
-
-
AVC (H.264)
HEVC (H.265)
-
-
-
-
No Decoders
Model
NVDEC 5
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Codec
MPEG-1
MPEG-2
MPEG-4
-
VC-1
VP8
VP9
-
AVC (H.264)
HEVC (H.265)
-
AV1
-
-
-
-
-
-
Direct X
12
Direct 3D
12_2
-
-
OpenCL
3.0
-
-
OpenGL
4.6
OpenCL
3.0
Vulkan
1.2
-
-
-
-
-
-
-
-
-
-
Shader Model
6.6
CUDA
8.6
-
-
PureVideo HD
VP11
VDPAU
Feature Set K
Not a Card
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
PCIe Version
5.0
PCIe Lanes
16
Slots Required
1.0
PCIe Version
4.0
PCIe Lanes
8
Multi GPU Support
Supported
Type
RoCE
-
-
-
-
-
-
-
-
-
-
Height
69 mm (2.72 in)
Width
168 mm (6.61 in)
Depth
20 mm (0.79 in)
Change Comparison