A100

Values shown for A100 40 GB PCIe (the variant deployed in Galvani's a100-galvani partitions). * denotes "with sparsity" — without sparsity values are half.

Spec	A100 40 GB PCIe	A100 80 GB PCIe	A100 40 GB SXM	A100 80 GB SXM
Architecture	NVIDIA Ampere	NVIDIA Ampere	NVIDIA Ampere	NVIDIA Ampere
FP64	9.7 TFLOPS	9.7 TFLOPS	9.7 TFLOPS	9.7 TFLOPS
FP64 Tensor Core	19.5 TFLOPS	19.5 TFLOPS	19.5 TFLOPS	19.5 TFLOPS
FP32	19.5 TFLOPS	19.5 TFLOPS	19.5 TFLOPS	19.5 TFLOPS
TF32 Tensor Core	156 / 312* TFLOPS	156 / 312* TFLOPS	156 / 312* TFLOPS	156 / 312* TFLOPS
BFLOAT16 Tensor Core	312 / 624* TFLOPS	312 / 624* TFLOPS	312 / 624* TFLOPS	312 / 624* TFLOPS
FP16 Tensor Core	312 / 624* TFLOPS	312 / 624* TFLOPS	312 / 624* TFLOPS	312 / 624* TFLOPS
INT8 Tensor Core	624 / 1 248* TOPS	624 / 1 248* TOPS	624 / 1 248* TOPS	624 / 1 248* TOPS
GPU memory	40 GB HBM2	80 GB HBM2e	40 GB HBM2	80 GB HBM2e
Memory bandwidth	1 555 GB/s	1 935 GB/s	1 555 GB/s	2 039 GB/s
Max TDP	250 W	300 W	400 W	400 W
Multi-Instance GPU	up to 7 MIGs @ 5 GB	up to 7 MIGs @ 10 GB	up to 7 MIGs @ 5 GB	up to 7 MIGs @ 10 GB
Form factor	PCIe	PCIe	SXM	SXM
Interconnect	NVLink Bridge for 2 GPUs (600 GB/s), PCIe Gen4 64 GB/s	NVLink Bridge for 2 GPUs (600 GB/s), PCIe Gen4 64 GB/s	NVLink 600 GB/s, PCIe Gen4 64 GB/s	NVLink 600 GB/s, PCIe Gen4 64 GB/s

Architecture detail (from NVIDIA Ampere whitepaper): 6 912 FP32 CUDA cores, 3 456 FP64 CUDA cores, 432 third-generation Tensor cores per A100 GPU.

---

Last update: July 17, 2026
Created: July 17, 2026