Introduction

NVIDIA today debuted the GeForce RTX 5070 Ti, its third release from the RTX 50-series Blackwell generation. The RTX 5070 Ti is positioned in a gray area between the performance and enthusiast segments, given its starting price of $750. Much like the RTX 4070 Ti from the previous generation, NVIDIA does not explicitly recommend it for 4K Ultra HD gaming, instead slotting it in the rather broad 1440p class, but there should be plenty of performance on tap for 4K Ultra HD gameplay at its native resolution if you know your way around game settings, or make the NVIDIA App find the best settings for you. Then there's always DLSS, and Blackwell introduces DLSS 4 with Multi Frame Generation and image quality improvements in all performance presets thanks to a new Transformer-based AI model at the helm of upscaling. There is no first-party RTX 5070 Ti Founders Edition card from NVIDIA, the company instead sent us the MSI RTX 5070 Ti Ventus 3X, a card priced at the $750 MSRP.



The new GeForce RTX 5070 Ti has a lot in common with the enthusiast-segment RTX 5080 which we reviewed late last month. For starters, the two cards share the same GB203 silicon, and while the RTX 5080 maxes this out, the RTX 5070 Ti is cut down from it. The GB203 silicon physically has 84 streaming multiprocessors (SM), all of which are enabled on the RTX 5080. The RTX 5070 Ti enables 70 out of these 84. Besides the newer architecture, this is a fairly big increase from the previous generation RTX 4070 Ti, which has 60 SM; even its refresh RTX 4070 Ti Super comes with 66 SM, a wee bit shy of the RTX 5070 Ti. With 70 SM on tap, the RTX 5070 Ti enjoys 8,960 CUDA cores, 280 Tensor cores, 70 RT cores, and 280 TMUs. The card gets 96 out of the 112 ROPs present on the GB203, and 48 MB out of the 64 MB present. The RTX 5070 Ti maxes out the 256-bit GDDR7 memory bus of the GB203, just like the RTX 5080, and gets 16 GB of memory—an increase from the 12 GB and 192-bit GDDR6X of the RTX 4070 Ti. The swanky new GDDR7 memory runs at 28 Gbps compared to 30 Gbps on the RTX 5080. The memory bandwidth is hence 896 GB/s compared to the 960 GB/s of its bigger sibling. The final two differentiators between the two are GPU frequencies and total graphics power, with the RTX 5070 Ti boosting up to 2432 MHz, with a 300 W TGP, compared to the 360 W of the RTX 5080.

Much like the RTX 5080 and RTX 5090 launched before it, the RTX 5070 Ti comes with several new technologies thanks to the new Blackwell graphics architecture it's based on. To begin with, it implements Neural Rendering, a feature that brings the awesome power of generative AI directly to the graphics rendering pipeline. The GPU now has the ability to run a generative AI model in tandem with rendering graphics thanks to a new hardware scheduler on these chips, called the AI Management Processor (AMP). Neural Rendering allows certain objects created by the generative AI to be combined with conventional raster 3D graphics, just like RTX brings real time ray traced objects to it. This should vastly improve realism in games. The new Blackwell generation SM comes with concurrent FP32 and INT32 math capability on all CUDA cores, the previous generation Ada only had INT32 capability on half the cores in an SM. The shader execution reordering mechanism is now aware of neural shaders. The 5th generation Tensor cores come with FP4 data format capability, which should max out throughput by trading in precision. The new RT cores come with even more hardware-based components, and are ready for Mega Geometry, which vastly improves the poly count of ray traced objects using hierarchical techniques resembling Mega Textures.

Then there are DLSS 4 and Multi Frame Generation. NVIDIA updated the AI model at the heart of the DLSS upscaler to one based on more advanced Transformers, instead of an older convoluted neural network (CNN). The new Transformer based model is more accurate, and hence there are image quality uplifts to be expected in all performance presets. NVIDIA introduced a new hardware component with Blackwell called Display Flip Metering, with which Blackwell implements Multi Frame Generation, or the ability for the GPU to generate up to three consecutive frames from a single conventionally rendered one, effectively quadrupling frame rates. While the Transformer models for upscaling and ray-reconstruction are available even for older RTX 40-series and RTX 30-series GPUs, Multi Frame Generation is exclusive to the RTX 50-series.

The MSI RTX 5070 Ti Ventus 3X is a simple yet elegant looking piece of hardware, which meets NVIDIA's SFF-Ready spec thanks to its relatively compact dimensions of 30 cm length, 4.9 cm thickness, and 12 cm height. The card features a silver+black two-tone cooler shroud with a design resembling that of the RTX 20-series Founders Edition cards. Three axial flow fans ventilate an aluminium fin-stack heatsink, which pulls heat from the GPU over a sold metal baseplate (instead of a vapor chamber). The card comes with a minor factory overclock of 2482 MHz, compared to 2452 MHz reference. Although this card is launching at $750, you can expect post-launch real world pricing to be much higher. I hope NVIDIA is paying attention to the fact that Grand Theft Auto 6 is launching on consoles first, the PC release will likely come year(s) later, and overpriced PC hardware isn't going to help the platform.

NVIDIA Blackwell Architecture

NVIDIA does not provide a block diagram for the GB203 GPU (we asked), so we had to quickly hack one out from the GB202 diagram. This is accurate just not as pretty.

The GeForce Blackwell graphics architecture heralds NVIDIA's 4th generation of RTX, the late-2010s re-invention of the modern GPU that sees a fusion of real time ray traced objects with conventional raster 3D graphics. With Blackwell, NVIDIA is helping add another dimension, neural rendering, the ability for the GPU to leverage a generative AI to create portions of a frame. This is different from DLSS, where an AI model is used to reconstruct details in an upscaled frame based on its training date, temporal frames, and motion vectors. Today we are reviewing the RTX 5070 Ti. At the heart of this graphics card is the new 5 nm GB203 silicon, the same chip that powers the RTX 5080. This chip has very similar die size and transistor counts to the previous generation AD103 powering the RTX 4080, because both chips are built on the exact same process—TSMC's "NVIDIA 4N", or 5 nm EUV with NVIDIA-specific characteristics—but is based on the newer Blackwell graphics architecture. The GB203 measures 378 mm² in die-area and rocks 45.6 billion transistors (compared to the 378.6 mm² die-area and 45.9 billion transistors of the AD103). This is where the similarities end.

The GB203 silicon is laid out essentially in the same component hierarchy as past generations of NVIDIA GPUs, but with a few notable changes. The GPU features a PCI-Express 5.0 x16 host interface. PCIe Gen 5 has been around since Intel's 12th Gen Core "Alder Lake" and AMD's Ryzen 7000 "Zen 4," so there is a sizable install-base of systems that can take advantage of it. The GPU is of course compatible with older generations of PCIe. The GB203 also features the new GDDR7 memory interface that's making its debut with this generation. The chip features a 256-bit wide memory bus, which is half the bus width of the GB202 powering the RTX 5090. NVIDIA is using this to drive 16 GB of memory at 30 Gbps speeds, yielding 960 GB/s of memory bandwidth, which is a 34% increase over the RTX 4080 and its 22.5 Gbps GDDR6X.

The GigaThread Engine is the main graphics rendering workload allocation logic on the GB203, but there's a new addition, a dedicated serial processor for managing all AI acceleration resources on the GPU, NVIDIA calls this AMP (AI management processor). Other components at the global level are the Optical Flow Processor, a component involved in older versions of DLSS frame generation and for video encoding; and an updated media acceleration engine consisting of two NVENC encode accelerators, and two NVDEC decode accelerators. On the RTX 5070 Ti, one of these two NVDEC units is disabled. The new 9th Gen NVENC video encode accelerators come with 4:2:2 AV1 and HEVC encoding support. The central region of the GPU has the single largest common component, the 64 MB L2 cache. The RTX 5070 Ti is configured with 48 MB of this cache.


Each graphics processing cluster (GPC) is a subdivision of the GPU with nearly all components needed for graphics rendering. On the GB203, a GPC consists of 12 streaming multiprocessors (SM) across 6 texture processing clusters (TPCs), and a raster engine consisting of 16 ROPs. Each SM contains 128 CUDA cores. Unlike the Ada generation SM that each had 64 FP32+INT32 and 64 purely-FP32 SIMD units, the new Blackwell generation SM features concurrent FP32+INT32 capability on all 128 SIMD units. These 128 CUDA cores are arranged in four slices, each with a register file, a level-0 instruction cache, a warp scheduler, two sets of load-store units, and a special function unit (SFU) handling some special math functions such as trigonometry, exponents, logarithms, reciprocals, and square-root. The four slices share a 128 KB L1 data cache, and four TMUs. The most exotic components of the Blackwell SM are the four 5th Gen Tensor cores, and a 4th Gen RT core. The RTX 5070 Ti is carved out from the GB203 by disabling an entire GPC, plus one TPC from the remaining six GPCs, resulting in 70 SM. You hence end up with 8,960 CUDA cores, 280 Tensor cores, 70 RT cores, 280 TMUs, and 96 ROPs.


Perhaps the biggest change to the way the SM handles work introduced with Blackwell is the concept of neural shaders—treating portions of the graphics rendering workload done by a generative AI model as shaders. Microsoft has laid the groundwork for standardization of neural shaders with its Cooperative Vectors API, in the latest update to DirectX 12. The Tensor cores are now accessible for workloads through neural shaders, and the shader execution reordering (SER) engine of the Blackwell SM is able to more accurately reorder workloads for the CUDA cores and the Tensor core in an SM.


The new 5th Gen Tensor core introduces support for FP4 data format (1/8 precision) to fast moving atomic workloads, providing 32 times the throughput of the very first Tensor core introduced with the Volta architecture. Over the generations, AI models leveraged lesser precision data formats, and sparsity, to improve performance. The AI management processor (AMP) is what enables simultaneous AI and graphics workloads at the highest levels of the GPU, so it could be simultaneously rendering real time graphics for a game, while running an LLM, without either affecting the performance of the other. AMP is a specialized hardware scheduler for all the AI acceleration resources on the silicon. This plays a crucial role for DLSS 4 multi-frame generation to work.


The 4th Gen RT core not just offers a generational increase in ray testing and ray intersection performance, which lowers the performance cost of enabling path tracing and ray traced effects; but also offers a potential generational leap in performance with the introduction of Mega Geometry. This allows for ray traced objects with extremely high polygon counts, increasing their detail. Poly count and ray tracing present linear increases in performance costs, as each triangle has to intersect with a ray, and there should be sufficient rays to intersect with each of them. This is achieved by adopting clusters of triangles in an object as first-class primitives, and cluster-level acceleration structures. The new RT cores introduce a component called a triangle cluster intersection engine, designed specifically for handling mega geometry. The integration of a triangle cluster compression format and a lossless decompression engine allows for more efficient processing of complex geometry.


The GB203 and the rest of the GeForce Blackwell GPU family is built on the exact same TSMC "NVIDIA 4N" foundry node, which is actually 5 nm, as previous-generation Ada, so NVIDIA directed efforts to finding innovative new ways to manage power and thermals. This is done through a re-architected power management engine that relies on clock gating, power gating, and rail gating of the individual GPCs and other top-level components. It also worked on the speed at which the GPU makes power-related decisions.


The quickest way to drop power is by adjusting the GPU clock speed, and with Blackwell, NVIDIA introduced a means for rapid clock adjustments at the SM-level.


NVIDIA updated both the display engine and the media engine of Blackwell over the previous generation Ada, which drew some flack for holding on to older display I/O standards such as DisplayPort 1.4, while AMD and Intel had moved on to DisplayPort 2.1. The good news is that Blackwell supports DP 2.1 with UHBR20, enabling 8K 60 Hz with a single cable. The company also updated NVDEC and NVENC, which now support AV1 UHQ, double the H.264 decode performance, MV-HEVC, and 4:2:2 formats.

Neural Rendering

Neural Rendering promises to be as transformative to modern graphics as programmable shaders itself. 3D Graphics rendering evolved from fixed-function over the turn of the century, to programmable shaders, HLSL, geometry shaders, compute shaders, and ray tracing, over the past couple of decades. In 2025, NVIDIA is writing the next chapter in this journey with Blackwell neural shaders. This allows for a host of neural-driven effects, including neural materials, neural volumes, and even neural radiance fields. Microsoft introduced the new Cooperative Vectors API for DirectX in a recent update, making it possible to access Tensor cores within a graphics API. Combined with a new shading language, Slang, this breakthrough enables developers to integrate neural techniques directly into their workflows, potentially replacing parts of the traditional graphics pipeline. Slang splits large, complex functions into smaller pieces that are easier to handle. Given that this is a DirectX standard API feature, there is nothing that stops AMD and Intel from integrating Neural Rendering (Cooperative Vectors) into their graphics drivers.

RTX Neural Materials works to significantly reduce the memory footprint of materials in 3D scenes. Under conventional rendering, the memory footprint of a material is bloated from complex shader code. Neural materials convert shader code and texture layers into a compressed neural representation. This results in up to a 7:1 compression ratio and enables small neural networks to generate stunning, film-like materials in real-time. For example, silk rendered with traditional shaders might lack the multicolored sheen seen in real life. Neural materials, however, capture intricate details like color variation and reflections, bringing such surfaces to life with unparalleled realism—and at a fraction of the memory cost.


The new Neural Radiance Cache, which dynamically trains a neural network during gameplay using the user's GPU, allowing light transport to be cached spatially, enabling near-infinite light bounces in a scene. This results in realistic indirect lighting and shadows with minimal performance impact. NRC partially traces 1 or 2 rays before storing them in a radiance cache, and infers an infinite amount of rays and bounces for a more accurate representation of indirect lighting in the game scene.

DLSS 4 and Multi Frame Generation

DLSS 4 introduces a major leap in image quality and performance. It isn't just a version bump with the introduction of a new feature, namely Multi Frame Generation, but introduces updates to nearly all DLSS sub-features. DLSS from its very beginning relied on AI to reconstruct details in super resolution, and with DLSS 4, NVIDIA is introducing a new transformer-based AI model to succeed the convolutional neural networks previous used, for double the parameters, four times the compute performance, and significantly improved image quality. Ray Reconstruction, introduced with DLSS 3.5, gets a significant image quality update with the new transformer-based model.


To understand Multi Frame Generation, you need to understand how DLSS Frame Generation, introduced with GeForce Ada, works. An Optical Flow Accelerator component gives the DLSS algorithm data to generate an entire frame using a neural network, using information from a previous rendered frame, effectively doubling frame rate. In Multi Frame Generation, AI takes over the functions of optical flow, to predict up to three frames following a conventionally rendered frame, effectively drawing four frames form the rendering effort of one.


Now, assuming this rendered frame is a product of Super Resolution, with the maximum performance setting generating 4x the pixels from a single rendered pixel, you're looking at a possibility where the rendering effort of 1/4th a frame goes into drawing 4 frames, or 15 in every 16 pixels being generated entirely by DLSS. When generating so many frames, Frame Pacing becomes a problem—irregular frame intervals impact smoothness. DLSS 4 addresses these issues by using a dedicated hardware unit inside Blackwell, which takes care of flip metering, reducing frame display variability by 5-10x. The Display Engine of Blackwell contains the hardware for flip metering.

NVIDIA Reflex 2

The original NVIDIA Reflex brought about a significant improvement to the responsiveness of maxed out graphics in competitive online gameplay, by compacting the rendering queue with the goal of reducing the whole system latency by up to 50%. Reflex is mandatory in DLSS 3 Frame Generation, given the latency cost imposed by the technology. Multi-frame generation calls for an equally savvy piece of technology, so we hence have Reflex 2. NVIDIA claims to have achieved a 75% reduction in latency with Frame Warp, which updates the camera (viewport) positions based on user inputs in real-time, and then uses temporal information to reconstruct the frame to display.

Packaging

The Card

MSI's Ventus uses a plastic cooler shroud and a metal backplate. The color theme is black with various shades of gray and metallic highlights. The backplate has a cutout for air to flow through.


Dimensions of the card are 30.0 x 12.0 cm, and it weighs 1060 g.


Installation requires three slots in your system. We measured the card's width to be 49 mm.


Display connectivity includes three standard DisplayPort 2.1b and one HDMI 2.1b.

Standard for all GeForce RTX 50-series Blackwell cards is a new display engine that supports three DisplayPort 2.1b outputs, each capable of UHBR20; and one HDMI 2.1a. Both interfaces support DSC (display stream compression). With DSC enabled, a single DisplayPort on this card can drive 4K 12-bit HDR at 480 Hz; or 8K 12-bit HDR at up to 165 Hz. The RTX 5070 Ti features an updated media acceleration engine with support for 4:2:2 video formats, AV1 UHQ, and MV-HEVC. There are two independent NVENC and NVDEC units.


The card uses a single 16-pin connector, which allows a maximum power draw of 600 W, but the board power limit is set much lower of course.


There is no RGB lighting on the Ventus 3X.