A40 PCIe — Full Specifications

nvidia Ampere · FP32: 37.42 TFLOPS · Memory: 48.0 GB GDDR6 · TDP: 300.0W

Last updated:

Calculated from published specifications — a Quanteta index, not measured FPS / benchmark results.

Quanteta Performance Indices

QOV ? Overall Performance Index 15.0 / 100 top 4%
QAI ? Compute (AI) Strength 4.2 / 100 top 14%
QGP ? Graphics Index 31.7 / 100 top 3%
QWP ? Power Efficiency Index 21.0 / 100 top 10%
QME ? Memory Bandwidth Density 0.2 / 100 top 69%

Methodology

Full Specifications

GPU ID
nvidia-a40-pcie
Vendor
nvidia
Manufacturer
NVIDIA
GPU Chip
GA102
Architecture
Ampere
Generation
Server Ampere(Axx)
Process
8.0nm
Transistors
28.3B
Die Size
628.0 mm²
Release
2020-10-05
Bus Interface
PCIe 4.0 x16
Base Clock
1305.0 MHz
Boost Clock
1740.0 MHz
Memory
48.0 GB GDDR6
Memory Bus
384.0-bit
Bandwidth
695.8 GB/s
Shaders
10752.0
TMUs
336.0
ROPs
112.0
Tensor Cores
336.0
RT Cores
84.0
TDP
300.0W
FP32
37.42 TFLOPS
FP16
37.42 TFLOPS
FP64
0.58 TFLOPS
Source
TechPowerUp
Check price on Amazon

Search link to Amazon.co.jp. We do not collect price or stock data — Amazon shows the current price and availability. As an Amazon Associate, Quanteta earns from qualifying purchases.

Related GPUs

GPUs with the closest value on each axis. A descriptive neighbourhood derived from the calculated indices — not a recommendation.

Closest Overall Performance Index

Arc A770M 14.9 QOV 14.9 / 100 A100 SXM4 40 GB 14.9 QOV 14.9 / 100 Radeon Pro W6900X 15.1 QOV 15.1 / 100

Closest Compute (AI) Strength

Radeon RX 6950 XT 4.2 QAI 4.2 / 100 Tesla V100 DGXS 16 GB 4.2 QAI 4.2 / 100 Tesla V100 DGXS 32 GB 4.2 QAI 4.2 / 100

Closest Graphics Index

Radeon RX 9070 GRE 16 GB 31.6 QGP 31.6 / 100 Arc A750 31.5 QGP 31.5 / 100 RTX PRO 4000 Blackwell 31.3 QGP 31.3 / 100

Closest Memory Bandwidth

Tesla P10 694.3 GB/s QME 0.3 / 100 GeForce RTX 4080 716.8 GB/s QME 0.5 / 100 GeForce RTX 4070 Ti SUPER AD102 672.3 GB/s QME 0.5 / 100
Compare · ← Back to all GPUs