NVIDIA A40 videocard review

A40 videocard released by NVIDIA; release date: 5 Oct 2020. The videocard is based on Ampere microarchitecture codenamed GA102.

Core clock speed - 1305 MHz. Boost clock speed - 1740 MHz. Texture fill rate - 584.6 GTexel/s. Pipelines - 10752. Manufacturing process technology - 8 nm. Transistors count - 28300 million. Power consumption (TDP) - 300 Watt.

Memory type: GDDR6. Maximum RAM amount - 48 GB. Memory bus width - 384 bit. Memory clock speed - 1812 MHz (14.5 Gbps effective). Memory bandwidth - 695.8 GB/s.

Benchmarks

Geekbench
OpenCL

Top 1 GPU

This GPU

346719

193656

PassMark
G2D Mark

Top 1 GPU

This GPU

1401

627

PassMark
G3D Mark

Top 1 GPU

This GPU

38287

14665

Name	Value
Geekbench - OpenCL	193656
PassMark - G2D Mark	627
PassMark - G3D Mark	14665

Specifications (specs)

Essentials
Architecture	Ampere
Code name	GA102
Launch date	5 Oct 2020
Place in performance rating	58
Technical info
Boost clock speed	1740 MHz
Core clock speed	1305 MHz
Manufacturing process technology	8 nm
Peak Double Precision (FP64) Performance	1169 GFLOPS (1:32)
Peak Half Precision (FP16) Performance	37.42 TFLOPS (1:1)
Peak Single Precision (FP32) Performance	37.42 TFLOPS
Pipelines	10752
Pixel fill rate	194.9 GPixel/s
Texture fill rate	584.6 GTexel/s
Thermal Design Power (TDP)	300 Watt
Transistor count	28300 million

Video outputs and ports
Display Connectors	3x DisplayPort
Compatibility, dimensions and requirements
Form factor	Dual-slot
Interface	PCIe 4.0 x16
Length	267 mm (10.5 inches)
Recommended system power (PSU)	700 Watt
Supplementary power connectors	8-pin EPS
Width	112 mm (4.4 inches)
API support
DirectX	12.2
OpenCL	3.0
OpenGL	4.6
Shader Model	6.6
Vulkan
Memory
Maximum RAM amount	48 GB
Memory bandwidth	695.8 GB/s
Memory bus width	384 bit
Memory clock speed	1812 MHz (14.5 Gbps effective)
Memory type	GDDR6