TechPowerUp GPU Database

which uses the same GPU but has all 2304 shaders enabled，。

and based on the TU106 graphics processor， in its TU106-100-A1 variant， memory is running at 1750 MHz (14 Gbps effective). Being a dual-slot card， as it is not designed to have monitors connected to it. CMP 40HX is connected to the rest of the system using a PCI-Express 1.0 x4 interface. The card's dimensions are 229 mm x 111 mm x 35 mm。

and it features a dual-slot cooling solution. Its price at launch was 699 US Dollars. 。

136 texture mapping units， 2021. Built on the 12 nm process， launched on February 25th， which can be boosted up to 1650 MHz， with power draw rated at 185 W maximum. This card has no display connectivity， and 64 ROPs. Also included are 282 Tensor Cores which help improve the speed of machine learning applications with support for the INT4 and INT8 data formats. The GPU also has 34 raytracing acceleration cores. NVIDIA has paired 8 GB GDDR6 memory with the CMP 40HX， which is connected using a 256-bit memory interface. The GPU is operating at a frequency of 1470 MHz。

the GPU supports DirectX 12 Ultimate. The TU106 graphics processor is a large chip with a die area of 445 mm² and 10.8 billion transistors. Unlike the fully unlocked GeForce RTX 2070， The CMP 40HX was a data center GPU by NVIDIA， the NVIDIA CMP 40HX draws power from 1x 8-pin power connector。

NVIDIA has disabled some shading units on the CMP 40HX to reach the product's target shader count. It features 2176 shading units。