We all knew it was coming, and today it has officially arrived. Nvidia has announced its Turing-powered Titan RTX. Although the chipmaker markets the graphics card for AI researchers, deep learning developers, data scientists, and content creators, the Titan RTX will undoubtedly find its way into many top-tier gaming systems as well.
The Titan RTX is now also known as the T-Rex because Nvidia claims it's the most powerful desktop graphics card on the planet. The graphics card features 4,608 CUDA cores and comes clocked with a 1,350MHz base clock and a 1,770MHz boost clock. With its 576 Tensor cores and 72 RT cores, the Titan RTX can supply up to 130 teraflops of deep learning performance and 11 GigaRays of ray tracing performance. The graphics card also comes equipped with 24GB of high-performance GDDR6 memory, twice the amount of previous-gen Titan models, to deliver up to 672 GB/s of bandwidth. The memory operates at 1,750MHz (14,000MHz effective) across a 384-bit memory interface.
Header Cell - Column 0 | Titan RTX | Titan V | Quadro RTX 5000 | RTX 2080 Ti FE |
---|---|---|---|---|
Architecture (GPU) | Turing (TU102) | Volta (GV100) | Turing (TU104) | Turing (TU102) |
CUDA Cores | 4608 | 5120 | 3072 | 4352 |
Peak FP32 Compute | 16.3 TFLOPS | 13.8 TFLOPS | 11.2 TFLOPS | 14.2 TFLOPS |
Tensor Cores | 576 | 640 | 384 | 544 |
RT Cores | 72 | N/A | 48 | 68 |
Texture Units | 288 | 320 | 192 | 272 |
Base Clock Rate | 1350 MHz | 1200 MHz | 1620 MHz | 1350 MHz |
GPU Boost Rate | 1770 MHz | 1455 MHz | 1815 MHz | 1635 MHz |
Memory Capacity | 24GB GDDR6 | 12GB HBM2 | 16GB GDDR6 | 11GB GDDR6 |
Memory Clock | 14Gbps | 1.7Gbps | 14Gbps | 14Gbps |
Memory Bus | 384-bit | 3072-bit | 256-bit | 352-bit |
Memory Bandwidth | 672 GB/s | 653 GB/s | 448 GB/s | 616 GB/s |
ROPs | 96 | 96 | 64 | 88 |
L2 Cache | 6MB | 4.5MB | ? | 5.5MB |
TDP | 280W | 250W | 265W | 260W |
Transistor Count | 18.6 billion | 21.1 billion | 13.6 billion | 18.6 billion |
Die Size | 754 mm² | 815 mm² | 545 mm² | 754 mm² |
Price | $2499 | $2999 | $2300 | $1199 |
According to Nvidia, the Titan RTX is the Swiss Army Knife of graphics cards. On the AI side, the Titan RTX's 576 Tensor cores and 24GB of memory help improve training and inference for neural networks. Thanks to the presence of NVLink, AI researchers and Deep Learning developers can pair two of these cards together to work with large neural networks and data sets.
Data scientists will be happy to know that Nvidia's Titan RTX also comes with support for Nvidia's RAPIDS platform. RAPIDS is a collection of open-source software libraries for data analytics and machine learning that help accelerate machine learning activities. On the content creation side, the Titan RTX's 72 RT cores give content creators real-time ray tracing and AI functionalities while the graphics card's 24GB of memory and 672 GB/s of bandwidth open the door for real-time 8K video editing.
The Titan RTX has a 280W TDP (thermal design power) rating and draws power from two 8-pin PCIe power connectors. Display outputs include three DisplayPort outputs, one HDMI port, and one USB Type-C port for Nvidia VirtualLink.
The Titan RTX graphics card carries an eye-watering $2,499 price tag and will be available later this month in the U.S. and Europe.
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Zhiye Liu is a news editor and memory reviewer at Tom’s Hardware. Although he loves everything that’s hardware, he has a soft spot for CPUs, GPUs, and RAM.