NVIDIA L4 vs AMD Instinct MI250X

NVIDIA L4
AMD Instinct MI250X
7680 Shaders
24GB GDDR6
2040MHz
14080 Shaders
128GB HBM2e
1700MHz
Peak AI Performance
1 POPS
INT4 Tensor Sparse
Peak AI Performance
382.98 TOPS
INT4 Tensor
FP32
31.33 TFLOPS
FP32
47.87 TFLOPS
FP16
31.33 TFLOPS
FP16
95.74 TFLOPS
Form Factor
PCIe Card
1.0-Slots
Form Factor
OAM Module
-
TDP
72W
TDP
560W
Power Connectors
-
-
-
1x 16-Pin 12VHPWR
-
-
-
-
-
GB6 OpenCL N/A
0%
GB6 Metal N/A
0%
GB6 Metal 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 POPS
INT4 Tensor Sparse
Peak AI Performance
382.98 TOPS
INT4 Tensor
-
-
-
-
-
-
FP8
-
250.68 TFLOPS Tensor (FP16 Accumulate)
501.35 TFLOPS Tensor (FP16 Accumulate) Sparse
250.68 TFLOPS Tensor (FP32 Accumulate)
501.35 TFLOPS Tensor (FP32 Accumulate) Sparse
-
-
-
-
-
-
FP16
31.33 TFLOPS
125.34 TFLOPS Tensor (FP16 Accumulate)
250.68 TFLOPS Tensor (FP16 Accumulate) Sparse
125.34 TFLOPS Tensor (FP32 Accumulate)
250.68 TFLOPS Tensor (FP32 Accumulate) Sparse
FP16
95.74 TFLOPS
382.98 TFLOPS Tensor (FP16 Accumulate)
-
382.98 TFLOPS Tensor (FP32 Accumulate)
-
FP32
31.33 TFLOPS
-
-
FP32
47.87 TFLOPS
95.74 TFLOPS Tensor
-
FP64
490 GFLOPS
-
FP64
47.87 TFLOPS
95.74 TFLOPS Tensor
BF16
31.33 TFLOPS
125.34 TFLOPS Tensor
250.68 TFLOPS Tensor Sparse
BF16
-
382.98 TFLOPS Tensor
-
TF32
62.67 TFLOPS Tensor
125.34 TFLOPS Tensor Sparse
-
-
-
INT4
501.35 TOPS Tensor
1 POPS Tensor Sparse
INT4
382.98 TOPS Tensor
-
INT8
-
250.68 TOPS Tensor
501.35 TOPS Tensor Sparse
INT8
-
382.98 TOPS Tensor
-
INT32
15.67 TOPS
-
-
Ray Tracing
72.4 TOPS
-
-
Pixel Fillrate
163.2 GPixel/s
Pixel Fillrate
-
-
-
-
-
Texture Fillrate
489.6 GTexel/s
Texture Fillrate
1496 GTexel/s
Manufacturer
NVIDIA
Manufacturer
AMD
Chip Designer
NVIDIA
Chip Designer
AMD
Architecture
Ada Lovelace
Architecture
CDNA 2
Family
Server
Family
Instinct
Codename
NV184
AD104
-
-
Codename
Aldebaran
Aldebaran XTX
Variant
Aldebaran XTX
Market Segment
Server
Market Segment
Server
Release Date
3/21/2023
Release Date
11/8/2021
Foundry
TSMC
-
Foundry
TSMC
-
Fabrication Node
4N
-
Fabrication Node
N6
-
Die Size
295 mm²
-
Die Size
2x 724 mm²
-
Transistor Count
35.8 Billion
-
Transistor Count
2x 28 Billion
-
Transistor Density
121.56M/mm²
-
Transistor Density
38.67M/mm²
-
Form
PCIe Card
Form
OAM Module
Shading Units
7680 Shaders
-
Shading Units
14080 Shaders
-
Texture Mapping Units
240 TMUs
Texture Mapping Units
880 TMUs
Render Output Units
80 ROPs
Render Output Units
-
Tensor Cores
240 T-Cores
Tensor Cores
880 T-Cores
Ray-Tracing Cores
60 RT-Cores
-
-
Streaming Multiprocessors
60 SMs
-
-
-
-
Compute Units
220 CUs
-
-
-
-
Graphics Processing Clusters
5 GPCs
-
-
-
-
795MHz Base
2040MHz
-
-
1000MHz Base
1700MHz
-
-
-
-
L1
64KB/SM Tex
128KB/SM
-
-
L1
-
-
16KB/CU
-
L2
48MB Shared
L2
16MB Shared
-
-
-
-
-
-
24GB
GDDR6
-
128GB
HBM2e
ECC
Bus Width
192Bit
Bus Width
8192Bit
Clock
1563MHz
Transfer Rate
12.5GT/s
Bandwidth
300.1GB/s
Clock
1600MHz
Transfer Rate
3.2GT/s
Bandwidth
3276.8GB/s
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
TDP
72W
TDP
560W
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
No Ports
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
No Ports
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
-
-
-
-
Model
2x NVENC 8
Model
2x VCN 2.6
Codec
-
-
-
-
-
-
-
-
AVC (H.264)
HEVC (H.265)
-
AV1
-
-
Codec
-
-
-
-
-
-
-
-
AVC (H.264)
HEVC (H.265)
-
-
-
-
Model
NVDEC 5
No Decoders
Codec
MPEG-1
MPEG-2
MPEG-4
-
VC-1
VP8
VP9
-
AVC (H.264)
HEVC (H.265)
-
AV1
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Direct X
12
Direct 3D
12_3
-
-
-
-
OpenGL
4.6
OpenCL
3.0
Vulkan
1.3
-
-
OpenCL
3.0
-
-
Shader Model
6.7
CUDA
8.9
-
-
PureVideo HD
VP12
VDPAU
Feature Set L
-
-
-
-
GFX
9.4
-
-
-
-
-
-
-
-
-
-
-
-
Power Connectors
-
-
-
-
-
-
1x 16-Pin 12VHPWR
-
-
-
-
-
-
-
-
Slots Required
1.0
PCIe Version
4.0
PCIe Lanes
16
-
-
PCIe Version
4.0
PCIe Lanes
16
-
-
-
-
Multi GPU Support
Supported
Type
Infinity Fabric
Height
56 mm (2.2 in)
Width
169 mm (6.65 in)
Depth
20 mm (0.79 in)
-
-
-
-
-
-
Change Comparison