HyperAI
Command Palette
Search for a command to run...
GPU Comparison
Compare GPU performance benchmarks side by side
Basic Information
Basic Information
Release Date
Mar 2023
Release Date
May 2022
Architecture
NVIDIA Hopper
Architecture
Gaudi 2
Memory
96 GB HBM3
Memory
96 GB HBM2e
Memory Bandwidth
3.35 TB/s
Memory Bandwidth
2451.5 GB/s
Performance Metrics
Performance Metrics
CUDA Cores
175.0% more
16,896
CUDA Cores
6,144
FP8 Performance
3,026 TFLOPS
FP8 Performance
-
FP16 Performance
2157.3% faster
248.3 TFLOPS
FP16 Performance
11 TFLOPS
FP32 Performance
1118.2% faster
67 TFLOPS
FP32 Performance
5.5 TFLOPS
FP64 Performance
34 TFLOPS
FP64 Performance
-
Power Consumption
41.7% less
350 W
Power Consumption
600 W