Introduction

Sapphire Radeon RX 7900 XT Pulse is among the most affordable custom-design RX 7000 RDNA 3 series graphics cards you can buy right now. The Pulse series from Sapphire represents a cost-effective product that's positioned close to the AMD reference design cards, but with superior thermals and more importantly, a quieter board design. The Radeon RX 7900 XT is the second-fastest RX 7000 series SKU, positioned a notch below the flagship RX 7900 XTX. AMD has looked to undercut NVIDIA's RTX 4080 with these cards, and has aggressively priced them in relation to the green team. The RX 7900 XT launched at $900, but its real-world pricing is now considerably below that with cards reaching $800, to match RTX 4070 Ti.



The Radeon RX 7900 XT has pretty much the same use-case as the flagship RX 7900 XTX: 4K Ultra HD gaming with maxed out settings. AMD claims generational improvements for the ray tracing performance of these cards, although you might want to step away from the presets and take control of a few settings to make ray traced games a bit more fluid at higher resolutions. The RX 7900 series debuts the new RDNA 3 graphics architecture that leverages the switch to the new 5 nm EUV foundry process, significantly higher shader counts, up to 17% IPC improvements for the shaders, and high clock speeds, to offer an over 50% performance/Watt gain over the previous generation, repeating the feat of the RX 6000 series RDNA 2, which springboarded AMD back into the high-end PC graphics segment.

The new "Navi 31" GPU that the RX 7900 series is based on, debuts chiplets to GPUs—a method by which AMD can maximize its foundry allocation of the latest node, and save on costs. The company identified specific components of the GPU that don't benefit from the switch to 5 nm as much as the main graphics rendering machinery does—the memory controllers and last-level caches—and spun them off into tiny chiplets called the Memory Cache Dies (MCDs), built on the 6 nm process. There are six of these on Navi 31, each with a 64-bit GDDR6 memory interface, and a 16 MB segment of the GPU's Infinity Cache; which together make up the GPU's 384-bit GDDR6 memory bus, and 96 MB Infinity Cache.

While the flagship RX 7900 XTX maxes out Navi 31, by enabling all 96 RDNA 3 compute units, the RX 7900 XT is carved out of the silicon by enabling 84 out of 96 compute units, which work out to 5,376 stream processors, 80 2nd Gen Ray Accelerators that each offer a 50% ray intersection performance increase over the previous generation; 336 TMUs, and a mammoth 192 ROPs. While the RX 7900 XTX has 24 GB of 20 Gbps GDDR6 across the chip's full 384-bit memory interface, AMD gave the RX 7900 XT 20 GB of 20 Gbps GDDR6 memory across a slightly narrower 320-bit memory bus. At the GPU-level, this means that one of the six MCDs is disabled, so the Infinity Cache memory size is reduced to 80 MB.

The Sapphire Radeon RX 7900 XT Pulse features a close-to-reference custom-design PCB that draws power from just two 8-pin PCIe power connectors, for a maximum power-delivery configuration of 375 W. The highlight of this card is its reasonably heavy triple-slot cooling solution that has an airy metal cooler shroud with plenty of cutouts to let the heatsink underneath breathe; and of course the heatsink itself, with its multiple aluminium fin-stacks skewered with six heatpipes. This is combined with three Angular Velocity fans that have webbed edges for 100% axial airflow, and double ball-bearings for increased durability. The card offers a small factory-overclock of 2075 MHz Game Clock (vs. 2025 MHz reference), and 2450 MHz Boost Clock (vs. 2394 MHz reference). The memory is untouched at 20 Gbps, at which speed the card enjoys 800 GB/s of memory bandwidth. The RX 7900 XT Pulse is currently selling at $860 on Newegg, below the $900 MSRP this SKU launched at.

Packaging

The Card

Sapphire Pulse design for the RX 7900 Series follows the design theme of the previous generation—black is the dominant color with red highlights. On the back, you'll find a metal backplate, which is designed to let some airflow through the cooler.


Dimensions of the card are 31.5 x 13.5 cm, and it weighs 1420 g.


Installation requires three slots in your system.


Display connectivity includes two standard DisplayPort 2.1 ports (RDNA 2 had 1.4a) and two HDMI 2.1a (same as RDNA 2).

AMD has upgraded their encode/decode setup. It now comes with two independent hardware units that can encode and decode two streams of video in parallel, or one stream at double the FPS rate. There's support for VP9, H.264, H.265 and AV1 decode, and encoding is supported for H.264, H.265 and AV1.


The card uses a classic dual 8-pin power input config, rated for 375 W maximum power. NVIDIA on the other hand uses the new 12+4 pin ATX 12VHPWR connector, which is rated for up to 600 W of power draw.

Teardown

Sapphire's thermal solution provides cooling not only for the GPU, but also the memory chips and voltage regulator circuitry. There's five heatpipes on the cooler.


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.


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

Circuit Board (PCB) Analysis

GPU voltage is a 14-phase design, managed by a Monolithic Power Systems MP2857 controller.


Monolithic MP87997 DrMOS components are used for GPU voltage; they are rated for 70 A of current each.


Memory voltage is a three-phase design, managed by a Monolithic Power Systems MP2856 controller.


For memory, Monolithic MP87997 DrMOS with a 70 A rating are used again.


The GDDR6 memory chips are made by Hynix and carry the model number H56G42AS8DX-014. They are specified to run at 2500 MHz (20 Gbps effective).


AMD's Navi 31 graphics processor is the world's first GPU that uses a chiplet architecture. Note the large die in the center, called "GCD," graphics compute die, which houses the compute units, it is surrounded by six smaller "MCD," memory cache dies, that contain one memory controller interface and one slice of cache each. While they look similar, the MCDs are not HBM chips. The MCDs are fabricated on a 6 nm process at TSMC Taiwan with a die size of 36.6 mm² each, the GCD is fabricated using TSMC's 5 nanometer node, with a die size of 300 mm². The combined transistor count of the GPU is 57.7 billion.