NVIDIA ® The Quadro ® RTX ™ 4000
is the world's first ray tracing GPU
The Quadro RTX 4000 combines the NVIDIA Turing GPU architecture with the latest memory and display technologies to provide the best performance and functionality in a single-slot PCI-e form factor. Enjoy greater fluidity with realistic rendering, experience faster performance with AI-enabled applications, and create detailed, realistic VR experiences more cost-effectively in a wider range of workstation chassis configurations.
The Quadro RTX 4000 has 36 RT cores to accelerate ray tracing, 288 Tensor cores to accelerate AI, and 8 GB GDDR6 memory to accommodate large data sets. It is equipped with three DisplayPort 1.4 connectors with HDR support and has the latest VirtualLink connector (the DisplayPort panel can also be driven via the included USB Type-C to DP adapter) to simplify the connection to the next-generation VR HMD. In addition, combined with Quadro Sync II, you can deploy large-scale digital signage solutions at higher density and lower cost.
Incredible application performance
Experience fast, interactive, professional application performance
Latest NVIDIA Turing GPU architecture and ultra-fast graphics memory
New RT core for real-time ray tracing
36 RT cores to accelerate realistic ray tracing rendering
NVIDIA RTX technology brings real-time rendering to professionals
Tensor Core accelerates AI workflow
288 Tensor Cores support AI development and training
· Incredible inference performance is perfect for "edge" deployments
Next-generation VR support
VirtualLink simplifies HMD wiring
Advanced rendering and shadowing capabilities for immersive VR
Performance and availability features
Revolutionary real-time ray tracing acceleration
Turing integrates a new hardware-based ray tracing engine and is the industry's first graphics processor to support real-time ray tracing. A Quadro RTX 4000 board renders complex professional models with physically accurate shadows, reflections, and refraction, giving users instant insight. Quadro RTX 4000-based systems work with applications that leverage APIs such as NVIDIA OptiX, Microsoft DXR, and Vulcan ray tracing to provide powerful power for a truly interactive design workflow to provide instant feedback for unprecedented productivity Level.
RT kernel
For the first time, a new hardware-based dedicated ray tracing technology allows the GPU to render film quality, realistic objects and environments in real time through physically accurate shadows, reflections and refractions. The real-time ray tracing engine works with NVIDIA OptiX, Microsoft DXR, and Vulkan APIs to provide realism far beyond what traditional rendering technologies can achieve. The RT kernel uses a small amount of light projected through pixels to accelerate BVH traversal and ray-casting functions.
Enhanced Tensor Core
The new mixed-precision kernel specially designed for deep learning matrix arithmetic can provide 8 times TFLOPS for training compared to the previous generation. The Quadro RTX 4000 uses 288 tensor cores; each Tensor core performs 64 floating-point fusion multiplication (FMA) operations per clock, and each SM performs a total of 1024 separate floating-point operations per clock. In addition to supporting FP16 / FP32 matrix operations, the new Tensor core also adds INT8 (2048 integer operations per clock) and experimental INT4 and INT1 (binary) precision modes for matrix operations.
Mesh shading
Provides computation-based geometry pipelines to accelerate processing and culling of geometrically complex models and scenes, thereby increasing performance by up to 2x.
Motion adaptive shadow
Provide finer control to allocate GPU capabilities (for example, apply more cycles to detailed areas of the scene and less cycles to less detailed areas) to improve performance and maintain the same image quality, Or reduce the image quality by 50% of the time required to generate shadow pixels.
Variable Rate Shadow (VRS)
Better control of pixel shadow rate; very effective for motion blur and polarized shadow effects. This feature enables shadow and geometry samples to be processed at different rates for more efficient execution.
Texture space shading
Decoupling shadows from screen space improves shadow efficiency and reusability.
Multi-view rendering (MVR)
Single-pass extended stereo renders multiple views with a unique view original position or view direction in a single pass.
High-speed GDDR6 memory
The Quadro RTX 4000's built-in Turing's optimized 8 GB GDDR6 memory subsystem provides the industry's fastest graphics memory (with a peak bandwidth of 416 GB / s) and is an ideal platform for processing latency-sensitive applications for large data sets.
H.264 and HEVC encoding / decoding engines
Transcoding, video editing and other encoding applications faster than real-time performance with dedicated decoding engines dedicated to H.264 and HEVC and dedicated decoding engines dedicated to 3D / computing pipelines.
Turing VR
Ray tracing and variable rate shadowing improve visual quality, while multi-view rendering provides wider field of view and supports next-generation HMDs and displays. The RT core provides accurate acoustic simulation, and VirtualLink requires only a single cable to make VR setup easier than ever.
Picture quality
Full Screen Anti-Aliasing (FSAA)
Up to 64x FSAA (128x in SLI mode) significantly reduces visual aliasing artifacts or "aliasing" for unparalleled image quality and highly realistic scenes.
32K texture and rendering processing
Texture rendering from 32K x 32K surfaces to support applications that require the highest resolution and high-quality image processing.
Display function
NVIDIA Quadro Mosaic ™ Technology
Transparently scale desktops and applications with up to 8 GPUs and 32 displays through a single workstation, while providing complete performance and image quality.
DisplayPort 1.4
Supports up to four 8K (7680 x 4320) displays at 60 Hz, up to four 5K (5120 x 2880) displays at 60 Hz, or four 4K (3840 x 2160 or 4096 x 2160) at up to 120 Hz ). The Quadro RTX 4000 supports HDR on DisplayPort 1.4 (SMPTE 2084 / 2086.BT.2020), with 4K 10b / 12b HEVC decoding at 60 Hz, and 4K 10b HEVC encoding at 60 Hz. Three monitors are connected to the supplied DP connector, while the fourth uses the included USB Type-C to DP adapter.
NVIDIA nView Advanced Desktop Software
Gain unprecedented end-user control over your desktop experience, increasing productivity in a single large or multi-monitor environment.
NVIDIA Quadro Sync II
Synchronize the display and image output of up to 32 displays from 8 GPUs (connected via two Quadro Sync II boards) in a single system, reducing the number of systems required to create an advanced video visualization environment.
OpenGL Quad Buffered Stereo Support
A smooth, immersive 3D stereo experience for professional applications.
Professional 3D stereo sync
Synchronize the 3D stereo hardware directly with the Quadro graphics card via a dedicated connection, enabling powerful control over stereo effects.
Super high resolution desktop support
Use a high-resolution display device with a maximum display desktop size of 32K for more mosaic topology options.
Virtual link
VirtualLink is an industry standard alternative to USB Type-C. It provides HBR3 DisplayPort with 4 display channels, USB 3.1 Gen2 SuperSpeed data, and up to 27 W of power for next-generation VR HMDs (head-mounted displays).
Software support
NVIDIA RTX Real-Time Ray Tracing Platform
A software framework that makes real-time ray tracing possible, portable and displayable.
Provides interoperability between rasterization, ray tracing, computing and AI / deep learning.
New Turing ray tracing acceleration in OptiX, DXR and Vulkan.
NVIDIA MDL (now open source) and supports Pixar's Universal Scene Description (USD) for portability and consistency.
NVIDIA NGX: AI for Vision Applications
Provides a neural graphics framework for image processing.
Provides end-user applications with AI-based capabilities to enhance graphics, photos, images, and video processing.
NGX functionality leverages the Tensor core to maximize performance and efficiency.
Allows NVIDIA to bring advancements in AI capabilities to developers who place them in applications through NVIDIA's pre-trained neutral network.
NVIDIA NGX technology
AI InPainting allows removing existing content from an image and replacing it with a realistic computer-generated alternative.
AI Up-Res uses AI to increase the resolution of an image or video by 2x, 4x, or 8x by interpreting the image and intelligently placing the data in the new image to create new pixels.
DLSS (Deep Learning Super Sampling) removes jagged lines to smooth them, resulting in higher quality images than using other technologies.
AI Slow Motion inserts interpolated frames into the video stream to provide smooth slow motion video.
AI-optimized software
Deep learning frameworks such as Caffe2, MXNet, CNTK, and TensorFlow provide significantly faster training times and higher multi-node training performance. GPU-accelerated libraries such as cuDNN, cuBLAS, and TensorRT provide higher performance for deep learning inference and high-performance computing (HPC) applications.
NVIDIA CUDA ® Parallel Computing Platform
Native execution of standard programming languages like C / C ++ and Fortran, as well as APIs such as CUDA, OpenCL, OpenACC and Direct Compute to accelerate such things as ray tracing, video and image processing, finite element analysis or computational fluid dynamics (Especially) and other technologies.
Unified memory
A single seamless 49-bit virtual address space allows data to be migrated transparently between the full allocation of CPU and GPU memory.
NVIDIA GPUDirect Video
GPUDirect for Video avoids unnecessary copies of system memory and CPU overhead, which speeds up communication between the GPU and video I / O devices.
NVIDIA Enterprise Management Tools
Maximize system uptime, seamlessly manage large-scale deployments, and remotely control graphics and display settings for efficient operations.
3D graphics architecture
Scalable geometry
· Hardware Segmentation Engine
· NVIDIA GigaThread ™ engine with 7 asynchronous replication engines
· Shader Model 5.1 (OpenGL 4.5 and DirectX 12)
Up to 32K x 32K texture and rendering processing
Transparent multisampling and supersampling
16x angle independent anisotropic filtering
32-bit filtering and blending by component floating point texture
64x Full Screen Anti-Aliasing (FSAA) / 128x FSAA in SLI mode
MPEG-2, MPEG-4 Part 2 Advanced Simple Profile, H.264, HEVC, MVC, VC1, DivX (version 3.11 and higher) and Flash (10.1 and higher) decoding acceleration
Dedicated H.264 and HEVC encoders
· Blu-ray dual stream hardware acceleration (supports HD picture-in-picture playback)
· NVIDIA GPU Boost (automatically increases GPU engine throughput to maximize application performance)
NVIDIA CUDA Parallel Processing Architecture
· New RT (ray tracing) core for each SM
Turing SM architecture (streaming multi-processor design, which can provide higher processing efficiency)
· Dynamic parallelism (the GPU dynamically spawns new threads without returning to the CPU)
· Mixed precision (1, 4, 8, 16, 32, and 64-bit) calculations
API support includes: CUDA C, CUDA C ++, DirectCompute 5.0, OpenCL, Java, Python, and Fortran
Configurable up to 96 KB of RAM (shared memory size dedicated to each SM)
Advanced display functions
Supports any combination of four connected displays
· Three DisplayPort 1.4 outputs (supports resolutions such as 3840 x 2160 at 120 Hz, 5120 x 2880 at 60 Hz, and 7680 x 4320 at 60 Hz)
The included UBC Type-C to DP adapter provides the various resolutions and refresh rates mentioned above
DisplayPort to VGA, DisplayPort to DVI (single and dual link) and DisplayPort to HDMI cables (resolution support based on dongle specifications)
· HDR support via DisplayPort 1.4 (SMPTE 2084/2086, BT.2020) (4K 60Hz 10b / 12b HEVC decoding, 4K 60Hz 10b HEVC encoding)
HDCP 2.2 support via DisplayPort and HDMI connectors
12-bit internal display pipeline (hardware support for 12-bit scanning of supported panels, applications and connections)
· NVIDIA 3D Vision ™ technology, 3D DLP, Interleaved and other 3D stereo formats
Full OpenGL quad-buffered stereo support
Underscan / overscan compensation and hardware scaling
· NVIDIA's nView ® multi-screen display technology
Supports large-scale, ultra-high-resolution visualization using the NVIDIA SVS platform, including NVIDIA Mosaic, NVIDIA Sync, and NVIDIA Warp / Blend technologies
Display and HDMI digital audio
Supports the following audio modes: Dolby Digital (AC3), DTS 5.1, Multichannel (7.1) LPCM, Dolby Digital Plus (DD +), and MPEG-2 / MPEG-4 AAC
DisplayPort data rate of 48 KHz
· HDMI digital audio data rates of 44.1 KHz, 48 KHz, 88.2 KHz, 96 KHz, 176 KHz, and 192 KHz
16-, 20-, and 24-bit word lengths
Available accessories
· Connect Quadro RTX 4000 to HDMI display through Yetian DP to HDMI with resolution up to 4K. NVIDIA recommends three included DisplayPort to HDMI adapters, which provide excellent image quality and meet professional standards.
specification
|
CUDA Core |
2034 |
|
RT kernel |
36 |
|
NVIDIA Tensor Core |
288 |
|
GPU memory |
8GB GDDR6 |
|
RTX-OPS |
43 T |
|
Ray cast |
6.0 Gigarays / second |
|
Peak single precision FP32 performance |
7.1 TFLOPS |
|
Peak Half Precision FP16 Performance |
14.2 TFLOPS |
|
Peak half precision INT8 performance |
28.5 minutes |
|
Deep learning TFLOPS 1 |
57.0 TFLOPS |
|
Memory bandwidth |
416 GB / s |
|
System interface |
PCI Express 3.0 x16 |
|
Maximum power consumption |
|
|
energy Star |
Yes |
|
Thermal Solution |
Ultra-quiet active fan |
|
Constituent factors |
4.4 "high x 9.5" long, single slot |
|
Display connector |
DisplayPort 1.4 (3) + Virtual Link |
|
DVI-D single link connector |
Yes, via the included adapter |
|
HDMI support |
Yes, via the included adapter |
|
Number of supported displays |
4 |
|
MAX large DP 1.4 resolution |
|
|
5K display support |
HDR 5120 x 2880 at 60 Hz |
|
4K display support |
HDR 4096 x 2160 or 3840 x 2160 at 120 Hz |
|
MAX large DVI-D DL resolution |
2560 x 1600 at 60 Hz via third-party adapter |
|
DVI-D SL maximum resolution |
1920 x 1200 at 60 Hz via the included adapter |
|
HDCP support |
Yes |
|
Professional 3D support |
Via optional stereo connector bracket |
|
Compatible with Quadro Sync II |
Yes (frame lock and sync lock) |
|
Compatible with NVIDIA GPU Direct |
Yes |
|
Graphics API |
Shader Model 5.1, OpenGL 4.5, DirectX 12.0, Vulkan 1.0 |
|
Calculation API |
CUDA, DirectCompute, OpenCL |
|
NVIEW |
Yes |
|
NVIDIA MOSAIC |
Yes |
|
Guarantee |
3 years |
1. Accumulation of FP16 matrix multiplied by FP16 and FP32
Warranty and support
Unrivalled service and commitment to professional graphics customers: 3-year warranty
support platform
· Microsoft Windows 10 (64-bit)
· Microsoft Windows 8.1 and 8 (64-bit)
· Microsoft Windows 7 (64-bit)
Linux-full OpenGL implementation with NVIDIA and ARB extensions (64-bit)
Minimum system hardware requirements
· Intel Core i5, i7, Xeon or higher
AMD Phenom or Opteron level or higher
· PCIe x16 Gen 3 (preferred) expansion slot
8GB system memory (preferred)
Internet connection (for driver installation)
· DisplayPort (preferred) or DVI-compatible display
DisplayPort (preferred), HDMI or DVI compatible display
Included in the package
NVIDIA Quadro RTX 4000 professional graphics
Quadro RTX Quick Start Guide
· Quadro Support Guide
USB Type-C to DP adapter
DisplayPort to DVI-D SL adapter
DisplayPort to HDMI adapter
· Auxiliary power cord
Resources
· Product manual
· Quadro comparison
Quadro Configurator
· Quadro Power Guidelines
3 year limited warranty
页面执行时间：34.180 毫秒 
