Moore Threads MTT S4000 vs NVIDIA A40 PCIe
We compared two Desktop platform GPUs: 48GB VRAM Moore Threads MTT S4000 and 48GB VRAM A40 PCIe to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
Moore Threads MTT S4000 's Advantages
Larger VRAM bandwidth (768.0GB/s vs 695.8GB/s)
NVIDIA A40 PCIe 's Advantages
Boost Clock1740MHz
2560 additional rendering cores
Lower TDP (300W vs 450W)
Score
Benchmark
FP32 (float)
Moore Threads MTT S4000
24.58 TFLOPS
A40 PCIe
+52%
37.42 TFLOPS
Graphics Card
Unknown
Release Date
Oct 2020
MUSA-Chunxiao
Generation
Tesla
Desktop
Type
Desktop
PCIe 5.0 x16
Bus Interface
PCIe 4.0 x16
Clock Speeds
1500 MHz
Base Clock
1305 MHz
-
Boost Clock
1740 MHz
2000 MHz
Memory Clock
1812 MHz
Memory
48GB
Memory Size
48GB
GDDR6
Memory Type
GDDR6
384bit
Memory Bus
384bit
768.0GB/s
Bandwidth
695.8GB/s
Render Config
64
Compute Units
-
-
SM Count
84
8192
Shading Units
10752
512
TMUs
336
512
ROPs
112
128
Tensor Cores
336
-
RT Cores
84
-
L1 Cache
128 KB (per SM)
4 MB
L2 Cache
6 MB
-
-
-
Theoretical Performance
768.0 GPixel/s
Pixel Rate
194.9 GPixel/s
768.0 GTexel/s
Texture Rate
584.6 GTexel/s
49.15 TFLOPS
FP16 (half)
37.42 TFLOPS
24.58 TFLOPS
FP32 (float)
37.42 TFLOPS
384.0 GFLOPS
FP64 (double)
584.6 GFLOPS
Board Design
450W
TDP
300W
850 W
Suggested PSU
700 W
4x DisplayPort 1.4a
Outputs
3x DisplayPort 1.4a
2x 8-pin
Power Connectors
8-pin EPS
Graphics Processor
MTT Gen 2
GPU Name
GA102
-
-
-
MUSA-Chunxiao
Architecture
Ampere
TSMC
Foundry
Samsung
12 nm
Process Size
8 nm
22 billion
Transistors
28.3 billion
Unknown
Die Size
628 mm²
Graphics Features
12 Ultimate (12_2)
DirectX
12 Ultimate (12_2)
4.6
OpenGL
4.6
3.0
OpenCL
3.0
1.4
Vulkan
1.3
-
CUDA
8.6
6.5
Shader Model
6.6
