Gigabyte GeForce RTX 4090 Gaming OC Review 17

Gigabyte GeForce RTX 4090 Gaming OC Review

(17 Comments) »

Introduction

GIGABYTE Logo

In this review we're taking an in-depth look at the Gigabyte GeForce RTX 4090 Gaming OC, which represents the company's value-ended custom-design of NVIDIA's new flagship GPU. The Gaming OC and WindForce brands stood on their own under the Gigabyte marquee, as the company positioned the higher AORUS Gaming range with its various tiers at the top of their product stack. With the RTX 4090, Gigabyte is only designing AORUS Xtreme and AORUS Master tiers (and no lower-end AORUS Elite), which creates room in the company's product stack for the Gaming OC we have here. Don't confuse value for cheap, this is still designed to be a formidable custom-design, with all the bells and whistles enthusiasts and gamers would want. The GeForce RTX 4090 is the absolute best RTX 40-series graphics card you can have (and the only SKU from the series at the time of this review).

At the heart of the RTX 4090 is the new "Ada Lovelace" graphics architecture, which debuts the 3rd generation of NVIDIA RTX, the company's pioneering attempt at adding realism to game visuals, by blending conventional raster 3D graphics with real-time ray traced elements such as lighting, shadows, reflections, illumination, motion-blur, etc., all of which are as close to real-life as possible, thanks to the power of ray tracing. Ada debuts new 3rd generation RT cores that accelerate ray tracing; and 4th generation Tensor cores that do AI; combined with a lot more CUDA cores than the previous generation, running at much higher clock speeds, with much more on-die cache to cushion the memory sub-system.



For Ada, NVIDIA leveraged the 5 nm EUV silicon fabrication process, and achieved three times the transistor-counts over the previous-generation, at comparable GPU die-area and typical board power of 450 W (similar to that of the RTX 3090 Ti). In return the company is promising the kind of performance uplifts we're used to seeing back when Moore's Law still worked for GPUs—about 50 to 100 percent generational performance uplifts. Besides these, the company also introduced a major update to its popular performance enhancement, DLSS 3, which can now generate entire frames in videos or games, without involving the graphics rendering pipeline of the GPU; thereby doubling frame-rates when enabled.

The RTX 4090 is based on the 5 nm "AD102" silicon, which it doesn't quite max-out, but is still spec'd fairly high. It comes with 16,384 CUDA cores (out of 18.432 present on the silicon); 128 RT cores, 512 Tensor cores, 512 TMUs, and 192 ROPs. The memory sub-system is generationally unchanged—24 GB of 21 Gbps GDDR6X memory across a 384-bit wide memory interface; although NVIDIA has significantly increased the on-die caches. The RTX 4090 comes with 72 MB of L2 cache, compared to just 6 MB on the RTX 3090.

The Gigabyte RTX 4090 Gaming OC doesn't look spartan in design despite being the company's value-ended custom-design product. The RGB-illuminated fan intake actually suggests Gigabyte decided in the last minute not to name this an AORUS Elite product, and went with Gaming OC. You get a meaty 4-slot cooling solution with plenty of surface-area for heat-dissipation, and a triple-fan setup. Besides the fan intakes, RGB graces the Gigabyte logo on the card, and you get a sturdy metal backplate. There's also a tiny GPU Boost factory-overclock of 2535 MHz vs. 2525 MHz reference. Handy features such as dual-BIOS, power-status indicators, and a segment-best 4-year product warranty are also present. Gigabyte is pricing this card at $1,700, or a $100 premium over the NVIDIA baseline.

NVIDIA GeForce RTX 4090 Market Segment Analysis
 PriceCoresROPsCore
Clock
Boost
Clock
Memory
Clock
GPUTransistorsMemory
RTX 2080$4002944641515 MHz1710 MHz1750 MHzTU10413600M8 GB, GDDR6, 256-bit
RTX 3060 Ti$4504864801410 MHz1665 MHz1750 MHzGA10417400M8 GB, GDDR6, 256-bit
RX 6700 XT$410
2560642424 MHz2581 MHz2000 MHzNavi 2217200M12 GB, GDDR6, 192-bit
RX 6750 XT$470
2560642495 MHz2600 MHz2250 MHzNavi 2217200M12 GB, GDDR6, 192-bit
RTX 2080 Ti$5504352881350 MHz1545 MHz1750 MHzTU10218600M11 GB, GDDR6, 352-bit
RTX 3070$5305888961500 MHz1725 MHz1750 MHzGA10417400M8 GB, GDDR6, 256-bit
RTX 3070 Ti$6006144961575 MHz1770 MHz1188 MHzGA10417400M8 GB, GDDR6X, 256-bit
RX 6800$5803840961815 MHz2105 MHz2000 MHzNavi 2126800M16 GB, GDDR6, 256-bit
RX 6800 XT$60046081282015 MHz2250 MHz2000 MHzNavi 2126800M16 GB, GDDR6, 256-bit
RTX 3080$6608704961440 MHz1710 MHz1188 MHzGA10228000M10 GB, GDDR6X, 320-bit
RTX 3080 Ti$850102401121365 MHz1665 MHz1188 MHzGA10228000M12 GB, GDDR6X, 384-bit
RX 6900 XT$68051201282015 MHz2250 MHz2000 MHzNavi 2126800M16 GB, GDDR6, 256-bit
RX 6950 XT$95051201282100 MHz2310 MHz2250 MHzNavi 2126800M16 GB, GDDR6, 256-bit
RTX 3090$950104961121395 MHz1695 MHz1219 MHzGA10228000M24 GB, GDDR6X, 384-bit
RTX 3090 Ti$1200107521121560 MHz1950 MHz1313 MHzGA10228000M24 GB, GDDR6X, 384-bit
RTX 4090$1600163841762235 MHz2520 MHz1313 MHzAD10276300M24 GB, GDDR6X, 384-bit
Gigabyte RTX 4090
Gaming OC
$1700163841762235 MHz2535 MHz1313 MHzAD10276300M24 GB, GDDR6X, 384-bit

GeForce Ada Architecture

The Ada graphics architecture heralds the third generation of the NVIDIA RTX technology, an effort toward increasing the realism in game visuals by leveraging real-time ray tracing, without the enormous amount of compute power required to draw purely ray-traced 3D graphics. This is done by blending conventional raster graphics with ray traced elements such as reflections, lighting, and global illumination, to name a few. The 3rd generation of RTX heralds the new higher IPC "Ada" CUDA core, 3rd generation RT core, 4th generation Tensor core, and the new Optical Flow Processor, a component that plays a key role in generating new frames without involving the GPU's main graphics rendering pipeline.


The GeForce Ada graphics architecture driving the RTX 4090 leverages the TSMC 5 nm EUV foundry process to increase transistor counts to a mammoth 76.3 billion transistors, a nearly 3-fold increase over the previous-generation; while the die-size is actually smaller, at 608 mm², compared to 628 mm² of the previous-generation GA102. The GPU features a PCI-Express 4.0 x16 host interface, and a 384-bit wide GDDR6X memory bus, which on the RTX 4090 wires out to 24 GB of memory. The Optical Flow Accelerator (OFA) is an independent top-level component. The chip features two NVENC and one NVDEC units in the GeForce RTX 40-series; while future professional-visualization graphics cards will have all six each of the NVENC and NVDEC components enabled.

The essential component hierarchy is similar to past generations of NVIDIA GPUs. The AD102 silicon features a whopping 12 Graphics Processing Clusters (GPCs), each of these has all the SIMD and graphics rendering machinery, and is a small GPU in its own right. Each GPC shares a raster engine (geometry processing components) and two ROP partitions (each with eight ROP units). The GPC of the AD102 contains six Texture Processing Clusters (TPCs), the main number-crunching machinery. Each of these has two Streaming Multiprocessors (SM), and a Polymorph unit. Each SM contains 128 CUDA cores across four partitions. Half of these CUDA cores are pure-FP32; while the other half is capable of FP32 or INT32. The SM retains concurrent FP32+INT32 math processing capability. The SM also contains a 3rd generation RT core, four 4th generation Tensor cores, some cache memory, and four TMUs. There are 12 SM per GPC, so 1,536 CUDA cores, 48 Tensor cores, and 12 RT cores; per GPC. Twelve GPCs hence add up to 18,432 CUDA cores, 576 Tensor cores, and 144 RT cores. Each GPC contributes 16 ROPs, so there are a mammoth 192 ROPs on the silicon. A 96 MB L2 cache serves as town-square for the various GPCs, memory controllers, and the PCIe host interface, to exchange data. NVIDIA carved the RTX 4090 out of the AD102 by disabling one of the twelve GPCs, and a further two TPCs from two of the other GPCs, as shown in the red-shaded parts of the block diagram above. The RTX 4090 has 72 MB L2 cache enabled (out of 96 MB physically present on the silicon).


The 3rd generation RT core accelerates the most math-intensive aspects of real-time ray tracing, including BVH traversal. Displaced micro-mesh engine is a revolutionary feature introduced with the new 3rd generation RT core, which accelerates the displaced micro-mesh feature. Just as mesh shaders and tessellation have had a profound impact on improving performance with complex raster geometry, allowing game developers to significantly increase geometric complexity; DMMs is a method to reduce the complexity of the bounding-volume hierarchy (BVH) data-structure, which is used to determine where a ray hits geometry. Previously the BVH had to capture even the smallest details to properly determine the intersection point. Ada's ray tracing architecture receives a major performance uplift from Shader Execution Reordering (SER), a software-defined feature that requires awareness from game-engines, to help the GPU reorganize and optimize worker threads associated with ray tracing.


The BVH now needn't have data for every single triangle on an object, but can represent objects with complex geometry as a coarse mesh of base triangles, which greatly simplifies the BVH data structure. A simpler BVH means less memory consumed and helps to greatly reduce ray tracing CPU load, because the CPU only has to generate a smaller structure. With older "Ampere" and "Turing" RT cores, each triangle on an object had to be sampled at high overhead, so the RT core could precisely calculate ray intersection for each triangle. With Ada, the simpler BVH, plus the displacement maps can be sent to the RT core, which is now able to figure out the exact hit point on its own. NVIDIA has seen 11:1 to 28:1 compression in total triangle counts. This reduces BVH compile times by 7.6x to over 15x, in comparison to the older RT core; and reducing its storage footprint by anywhere between 6.5 to 20 times. DMMs could reduce disk- and memory bandwidth utilization, utilization of the PCIe bus, as well as reduce CPU utilization. NVIDIA worked with Simplygon and Adobe to add DMM support for their tool chains.


Opacity Micro Meshes (OMM) is a new feature introduced with Ada to improve rasterization performance, particularly with objects that have alpha (transparency data). Most low-priority objects in a 3D scene, such as leaves on a tree, are essentially rectangles with textures on the leaves where the transparency (alpha) creates the shape of the leaf. RT cores have a hard time intersecting rays with such objects, because they're not really in the shape that they appear (they're really just rectangles with textures that give you the illusion of shape). Previous-generation RT cores had to have multiple interactions with the rendering stage to figure out the shape of a transparent object, because they couldn't test for alpha by themselves.


This has been solved by using OMMs. Just as DMMs simplify geometry by creating meshes of micro-triangles; OMMs create meshes of rectangular textures that align with parts of the texture that aren't alpha, so the RT core has a better understanding of the geometry of the object, and can correctly calculate ray intersections. This has a significant performance impact on shading performance in non-RT applications, too. Practical applications of OMMs aren't just low-priority objects such as vegetation, but also smoke-sprites and localized fog. Traditionally there was a lot of overdraw for such effects, because they layered multiple textures on top of each other, that all had to be fully processed by the shaders. Now only the non-opaque pixels get executed—OMMs provide a 30 percent speedup with graphics buffer fill-rates, and a 10 percent impact on frame-rates.


DLSS 3 introduces a revolutionary new feature that promises a doubling in frame-rate at comparable quality, it's called AI frame-generation. While it has all the features of DLSS 2 and its AI super-resolution (scaling up a lower-resolution frame to native resolution with minimal quality loss); DLSS 3 can generate entire frames simply using AI, without involving the graphics rendering pipeline.


Every alternating frame with DLSS 3 is hence AI-generated, without being a replica of the previous rendered frame. This is possible only on the Ada graphics architecture, because of a hardware component called the optical flow accelerator (OFA), which assists in predicting what the next frame could look like, by creating what NVIDIA calls an optical flow-field. OFA ensures that the DLSS 3 algorithm isn't confused by static objects in a rapidly-changing 3D scene (such as a race sim). The process heavily relies on the performance uplift introduced by the FP8 math format of the 4th generation Tensor core. A third key ingredient of DLSS 3 is Reflex. By reducing the rendering queue to zero, Reflex plays a vital role in ensuring that frame-times with DLSS 3 are at an acceptable level, and a render-queue doesn't confuse the upscaler. A combination of OFA and the 4th Gen Tensor core is why the Ada architecture is required to use DLSS 3, and why it won't work on older architectures.

Packaging

Package Front
Package Back


The Card

Graphics Card Front
Graphics Card Back
Graphics Card Front Angled

For the GeForce 40 Series, Gigabyte has designed a new visual theme for their cards. The main cooler is now black, with a gray metal backplate.


In terms of size, the Gaming OC is slightly bigger than the NVIDIA RTX 4090 Founders Edition.

Graphics Card Dimensions

Dimensions of the card are 34.5 x 15.5 cm, and it weighs 2020 g.

Graphics Card Height
Graphics Card Back Angled

Installation requires four slots in your system.

Monitor Outputs, Display Connectors

Display connectivity includes three standard DisplayPort 1.4a ports and one HDMI 2.1a (same as Ampere).

NVIDIA introduced the concept of dual NVDEC and NVENC Codecs with the Ada architecture. This means there are two independent sets of hardware-accelerators; so you can encode and decode two streams of video in parallel or one stream at double the FPS rate. The new 8th Gen NVENC now accelerates AV1 encoding, besides HEVC. You also get an "optical flow accelerator" unit that is able to calculate intermediate frames for videos, to smooth playback. The same hardware unit is used for frame generation in DLSS 3.

Graphics Card Power Plugs

The card uses the new 12+4 pin ATX 12VHPWR connector, which is rated for up to 600 W of power draw. An adapter cable from 4x PCIe 8-pin is included, you can also run the card with just three 8-pins.


This BIOS switch lets you select between the default OC BIOS and a silent BIOS, which runs the fans on a slightly more relaxed fan curve.

Teardown

Graphics Card Cooler Front
Graphics Card Cooler Back

Gigabyte's thermal solution uses a vapor-chamber and eight heatpipes. The main cooler also provides cooling for the VRM circuitry and memory chips—there's a lot of thermal pads here.


The backplate is made of metal and protects the card against damage during installation and handling.

High-resolution PCB Pictures

These pictures are for the convenience of volt modders and people who would like to see all the finer details on the PCB. Feel free to link back to us and use these in your articles, videos or forum posts.

Graphics Card Teardown PCB Front
Graphics Card Teardown PCB Back

High-resolution versions are also available (front, back).

Our Patreon Silver Supporters can read articles in single-page format.
Discuss(17 Comments)
May 18th, 2024 13:21 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts