GPUDesktopReleased Dec 18, 2018

NVIDIA TITAN RTX

24 GB GDDR6 · 384-bit · 4,608 cores

The TITAN RTX is NVIDIA's Turing-generation prosumer card, launched in December 2018 with the fully enabled TU102 GPU and 24 GB of GDDR6 memory. It was aimed at AI researchers, data scientists and content creators rather than gamers, offering up to 130 Tensor TFLOPS for deep-learning training and enough memory for large datasets and 8K video work. Two cards can be joined with an NVLink bridge to pool up to 48 GB of memory. It launched at a price of 2,499 US dollars.

General

Manufacturer: NVIDIA
Model: TITAN RTX
Segment: Desktop
GPU chip: TU102-400
Architecture: Turing
Release date: Dec 18, 2018
Price: Search on Amazon

Memory

Memory size: 24 GB
Memory type: GDDR6
Bus width: 384-bit
Bandwidth: 672 GB/s
Memory clock: 1,750 MHz

Clocks & silicon

GPU clock: 1,350 MHz
Boost clock: 1,770 MHz
Shader units: 4,608 CUDA Cores
Tensor / AI cores: 576
RT cores: 72
Transistors: 18.6 billion
Technology (nm): 12 nm
TSMC 12 nm FFN
TDP / TBP: 280 W

Interface & APIs

Bus interface: PCIe 3.0 x16
SLI / multi-GPU: NVLink (2-way bridge)
DirectX: 12 Ultimate (12_2)
OpenGL: 4.6
Vulkan: 1.4
Other APIs: CUDA 7.5, OpenCL 3.0, DLSS 2

AI support

Precisions: FP8FP16BF16INT8INT4TF32FP64
AI accelerator: 2nd Gen Tensor Cores (576) + 1st Gen RT Cores (72)
Peak AI TOPS: —
Notes: 2nd-gen Tensor Cores. FP16/INT8/INT4 supported; no BF16, TF32 or FP8 on Turing. Full-rate FP16 accumulate for deep learning; FP64 at 1/32 rate.