MIT
PROCESSOR DB
Team
Steering committee
Access database
(beta)
Database
NVIDIA CMP 50HX TU102-100-A1
General Information
Manufacturer:
TSMC
Release Date:
6/23/2021
Process Node(nm):
12
Core Count:
-
Total Transistor Count:
18600
Package size (mm²):
-
Die size (mm²):
754
Number of Dies:
-
Transistor density:
24.67
Voltage (low):
-
Voltage (high):
-
Platform:
Processors
GPU: NVIDIA CMP 50HX
General Information
Variant:
TU102-100-A1
TDP (W):
250
Cores:
-
Architecture Information
Architecture:
Turing
Generation:
Mining GPUs
Process Size (nm):
12
Clock Information
Base Clock (MHz):
1350
Boost Clock (MHz):
1545
Memory Clock (MHz):
1750
Cache Information
L0 Cache (KB):
-
L1 Cache (KB):
64
L2 Cache (KB):
5
L3 Cache (KB):
-
Memory Information
Memory Size (GB):
10240
Memory Type:
GDDR6
Memory Bus (bits):
320
Memory Bandwidth (GB/s):
560
Compute Information
Texture Rate (GTexel/s):
-
FP16 Performance (GFLOPS):
22150
FP32 Performance (GFLOPS):
11070
FP64 Performance (GFLOPS):
346.1
Shader Units (SU):
3584
Texture Mapping Units (TMUs):
192
Render Output Units (ROPs):
80
Graphic API Information
Pixel Rate (GPixel/s):
123.6
Slot:
Dual-slot
DirectX:
12 Ultimate (12_2)
OpenGL:
4.6
OpenCL:
3
Vulkan:
1.3
Pixel Shader:
-
Vertex Shader:
-
Benchmarks
Benchmark Name
Score
Economic Data
Year
Price
Database