Moore Threads MTT S3000 vs NVIDIA RTX PRO 6000 Blackwell Server
We compared two Desktop platform GPUs: 32GB VRAM Moore Threads MTT S3000 and 96GB VRAM RTX PRO 6000 Blackwell Server to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
Moore Threads MTT S3000 's Advantages
Lower TDP (250W vs 600W)
NVIDIA RTX PRO 6000 Blackwell Server 's Advantages
Boost Clock2617MHz
More VRAM (96GB vs 32GB)
Larger VRAM bandwidth (1.79TB/s vs 448.0GB/s)
19968 additional rendering cores
Score
Benchmark
FP32 (float)
Moore Threads MTT S3000
15.56 TFLOPS
RTX PRO 6000 Blackwell Server
+709%
126 TFLOPS
Graphics Card
Unknown
Release Date
Mar 2025
MUSA-Chunxiao
Generation
Server Blackwell
Desktop
Type
Desktop
PCIe 5.0 x16
Bus Interface
PCIe 5.0 x16
Clock Speeds
1900 MHz
Base Clock
1590 MHz
0 MHz
Boost Clock
2617 MHz
1750 MHz
Memory Clock
1750 MHz
Memory
32GB
Memory Size
96GB
GDDR6
Memory Type
GDDR7
256bit
Memory Bus
512bit
448.0GB/s
Bandwidth
1.79TB/s
Render Config
32
Compute Units
-
-
SM Count
188
4096
Shading Units
24064
256
TMUs
752
256
ROPs
192
-
Tensor Cores
752
-
RT Cores
188
-
L1 Cache
128 KB (per SM)
4 MB
L2 Cache
128 MB
-
-
-
Theoretical Performance
486.4 GPixel/s
Pixel Rate
502.5 GPixel/s
486.4 GTexel/s
Texture Rate
1968 GTexel/s
31.13 TFLOPS
FP16 (half)
126.0 TFLOPS
15.56 TFLOPS
FP32 (float)
126.0 TFLOPS
243.2 GFLOPS
FP64 (double)
1.968 TFLOPS
Board Design
250W
TDP
600W
600 W
Suggested PSU
1000 W
2x DisplayPort 1.4a
Outputs
4x DisplayPort 2.1b
1x 8-pin
Power Connectors
1x 16-pin
Graphics Processor
MTT Gen 2
GPU Name
GB202
-
-
-
MUSA-Chunxiao
Architecture
Blackwell 2.0
TSMC
Foundry
TSMC
12 nm
Process Size
5 nm
22 billion
Transistors
92.2 billion
Unknown
Die Size
750 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
1.4
-
CUDA
12.0
6.5
Shader Model
6.8
