GPU

NVIDIA GeForce RTX 4090

Memory

24GB GDDR6X, 384bit, 21Gbps, 1313MHz, 1008GB/​s

overclock

-

clock Boost

2520MHz

clock Base

2235MHz

Cooling

3x Axial-fan (100mm) with 0dB-zero-fan-mode

TDP/TGP

450W (NVIDIA)

Power connections

1x 16-Pin PCIe 5.0 (via adapter: 4x 8-Pin PCIe)

Interfaces

1x HDMI 2.1a, 3x DisplayPort 1.4a

Interface

PCIe 4.0 x16

Special features

Backplate

total height

Quad Slot

Dimensions

331x150x70mm (LxWxH)

Type

PCIe card (full height)

PCI slot bracket

full height

Computing power

82.58 TFLOPS (FP16), 82.58 TFLOPS (FP32), 1.29 TFLOPS (FP64)

Chip-data

Chip-Architecture

Ada Lovelace (from 2022/​Q3)

Chip-title

AD102-300-A1 /​ AD102-301-A1

Chip-Configuration

128SM (16384 ALUs /​ 512 TMUs /​ 176 ROPs), 128 RT Cores, 512 Tensor Cores, 96MB L2 cache

Chip-Manufacturing Process

TSMC 5nm "NVIDIA 4N"

Chip size

608mm², monolithisch, 76.3 Mrd. Transistoren

Chip-functions

Raytracing (3rd Gen NVIDIA RTX), NVIDIA Tensor (4th Gen), HDCP 2.3

API-support

DirectX 12 Ultimate (12_2) /​ CUDA 8.9 /​ Vulkan 1.3 /​ OpenCL 3.0 /​ OpenGL 4.6 /​ Shader Model 6.7

Transcoding-Engine

2x NVENC (8th Gen, max. Sessions: 5), 1x NVDEC (5th Gen), Info

encoding

AV1, H.265 (4K YUV 4:2:0 /​ 4K YUV 4:4:4 /​ 4K Lossless /​ 8K /​ HEVC 10bit support /​ HEVC B Frame support), H.264 (YUV 4:2:0 /​ YUV 4:4:4 /​ Lossless)

decoding

AV1 (8bit /​ 10bit), H.265 (8bit 4:2:0 /​ 10bit 4:2:0 /​ 12bit 4:2:0 /​ 8bit 4:4:4 /​ 10bit 4:4:4 /​ 12bit 4:4:4), H.264, VP9 (8bit /​ 10bit /​ 12bit), VP8, VC-1, MPEG-2, MPEG-1