Skip to content
HardwareLadder
AI GPU ladder

NVIDIAAI GPU · PCIe

A40

NVIDIA A40 is a Ampere data-center accelerator with 48 GB GDDR6 and 0.696 TB/s of memory bandwidth.

Ladder rank
#22
AI index (H100 SXM = 100)
17.7
Series
Ampere

Specifications

Dense throughput, no sparsity
Architecture Ampere
Form factor PCIe
Process Samsung 8 nm
FP16 / BF16 tensor (dense) 149.7 TFLOPS
FP8 tensor (dense) —
FP4 tensor (dense) —
Memory 48 GB GDDR6
Memory bandwidth 0.696 TB/s
Power 300 W
Launch Oct 2020

AI index

Tensor × bandwidth

100 × √((149.7 ÷ 989.4) × (0.696 ÷ 3.35)) = 17.7

The AI index is the geometric mean of dense 16-bit tensor throughput and memory bandwidth, each relative to the H100 SXM. It is a spec-sheet comparison, not a measured benchmark.

Sources

2 references