Command Palette
Search for a command to run...
GPU Leaderboard
Compare GPU performance benchmarks for machine learning and AI workloads
Company
AMDCompany
NVIDIAArchitecture
CDNA4Architecture
NVIDIA Blackwell 2.0Date de publication
2025-06-12 00:00:00
Date de publication
2025-03-18 00:00:00
Mémoire
288 GB HBM3E
Mémoire
96 GB GDDR7
Memory Bandwidth
8.19 TB/sMemory Bandwidth
1.79 TB/sPerformances FP8
Not supported
Performances FP8
Not supported
Performances FP16
24.8 % plus rapide
157.3 TFLOPS
Performances FP16
126 TFLOPS
Performances FP32
24.8 % plus rapide
157.3 TFLOPS
Performances FP32
126 TFLOPS
Performances FP64
3893.9 % plus rapide
78.6 TFLOPS
Performances FP64
1.968 TFLOPS
Cœurs CUDA
1,024
Cœurs CUDA
95.7 % de plus
24,064
Consommation électrique
1400 W
Consommation électrique
133.3% de moins
600 W