Radeon Instinct MI300X vs Tesla P4
We compared two Professional market GPUs: 192GB VRAM Radeon Instinct MI300X and 8GB VRAM Tesla P4 to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
Radeon Instinct MI300X 's Advantages
Released 7 years and 3 months late
Boost Clock has increased by 89% (2100MHz vs 1114MHz)
More VRAM (192GB vs 8GB)
Larger VRAM bandwidth (5171GB/s vs 192.3GB/s)
16896 additional rendering cores
Tesla P4 's Advantages
Lower TDP (75W vs 750W)
Score
Benchmark
FP32 (float)
Radeon Instinct MI300X
+1332%
81720
Tesla P4
5704
Graphics Card
Dec 2023
Release Date
Sep 2016
Radeon Instinct
Generation
Tesla
Professional
Type
Professional
PCIe 5.0 x16
Bus Interface
PCIe 3.0 x16
Clock Speeds
1000 MHz
Base Clock
886 MHz
2100 MHz
Boost Clock
1114 MHz
2525 MHz
Memory Clock
1502 MHz
Memory
192GB
Memory Size
8GB
HBM3
Memory Type
GDDR5
8192bit
Memory Bus
256bit
5171GB/s
Bandwidth
192.3GB/s
Render Config
304
Compute Units
-
-
SM Count
20
19456
Shading Units
2560
1216
TMUs
160
0
ROPs
64
1216
Tensor Cores
-
16 KB (per CU)
L1 Cache
48 KB (per SM)
16 MB
L2 Cache
2 MB
256 MB
L3 Cache
-
Theoretical Performance
0 MPixel/s
Pixel Rate
71.30 GPixel/s
2554 GTexel/s
Texture Rate
178.2 GTexel/s
653.7 TFLOPS
FP16 (half)
89.12 GFLOPS
81.72 TFLOPS
FP32 (float)
5.704 TFLOPS
81.72 TFLOPS
FP64 (double)
178.2 GFLOPS
Board Design
750W
TDP
75W
1150 W
Suggested PSU
250 W
No outputs
Outputs
No outputs
None
Power Connectors
None
Graphics Processor
Aqua Vanjaram
GPU Name
GP104
-
GPU Variant
GP104-895-A1
CDNA 3.0
Architecture
Pascal
TSMC
Foundry
TSMC
5 nm
Process Size
16 nm
153 billion
Transistors
7.2 billion
1017 mm²
Die Size
314 mm²
Graphics Features
N/A
DirectX
12 (12_1)
N/A
OpenGL
4.6
3.0
OpenCL
3.0
N/A
Vulkan
1.3
-
CUDA
6.1
N/A
Shader Model
6.4
