Unverified50% confidenceFactExact time
A single TPU v4 chip has a peak compute of approximately 275 TFLOPS (BF16 precision), and a TPU v4-32 pod consists of 32 chips interconnected via ICI high-speed interconnects, with a total compute of approximately 8,800 TFLOPS.
1
Sources
50%
Confidence
Long-term
Relevance
8/6/2026
First Seen
Sources
nanoAlphaZero：单文件JAX实现24小时训练2700分棋力
Related Claims
VerifiedThe NVIDIA H100 single-card FP64 double-precision floating-point performance reaches approximately 34 TFLOPS.79% similarUnverifiedThe T4 GPU has 16GB of VRAM and 8.1 TFLOPS of FP32 compute performance.76% similarUnverifiedThe A100 has 40/80GB of VRAM and 19.5 TFLOPS of FP32 performance.73% similarUnverifiedThe NVIDIA A100 single-card compute performance is approximately 312 TFLOPS (FP16).73% similarVerifiedRTX 4090的理论FP16算力为82.6 TFLOPS，显存带宽为1008 GB/s72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/699043API
curl https://kongchang.com/api/v1/knowledge/claims/699043MCP
get_claim(id=699043)