ASRock Radeon RX 9070 XT Steel Legend Dark

Name: ASRock Radeon RX 9070 XT Steel Legend Dark
Brand: ASRock
SKU: 5688

Manufacturer: ASRock

The ASRock Radeon RX 9070 XT Steel Legend Dark is a graphics card based on AMD's RDNA 4.0 architecture, fabricated on a 4 nm process with 53,900 million transistors. It features RGB lighting and supports up to four displays simultaneously, with a physical footprint of 298 mm × 131 mm and a thermal design power rating of 304W.

On the performance side, the card operates at a base clock of 1660 MHz and reaches a turbo frequency of 2970 MHz, delivering 48.6 TFLOPS of floating-point throughput alongside a pixel rate of 380.2 GPixel/s and a texture rate of 760.3 GTexels/s. Its 16GB of GDDR6 memory runs across a 256-bit bus at an effective speed of 20,000 MHz, providing up to 644.6 GB/s of bandwidth. The card includes 4096 shading units, 256 TMUs, and 128 ROPs, supports DirectX 12 Ultimate, OpenGL 4.6, ray tracing, AMD FSR4, and AMD SAM, and connects via one HDMI 2.1b port and three DisplayPort outputs over a PCIe 5.0 interface.

Pros

The 16GB of GDDR6 VRAM across a 256-bit bus provides substantial memory capacity for demanding rendering and compute workloads
A boost clock of 2970 MHz combined with 4096 shading units enables high throughput across graphics tasks
Maximum memory bandwidth of 644.6 GB/s supports fast data transfer between the GPU and its frame buffer
Ray tracing support is included alongside AMD FSR4, offering hardware-accelerated tracing and an upscaling option
ECC memory support adds a layer of data reliability useful for workloads sensitive to memory errors
RGB lighting is built in, and the card can drive up to four displays simultaneously via three DisplayPort and one HDMI 2.1b output

Cons

A TDP of 304W places meaningful demands on system power delivery and case airflow
The card does not support DLSS or XeSS, limiting upscaling options to AMD FSR4 only
No USB-C output is available, which may limit compatibility with certain monitors or display adapters
At 298 mm in length, the card requires a case with adequate clearance for proper fitment
Air-water hybrid cooling is not included, leaving thermal management entirely dependent on the air cooling solution fitted to the card

Who is this for?

This card is well-suited to users running demanding rendering, gaming, or compute workloads that benefit from 16GB of GDDR6 VRAM and a peak throughput of 48.6 TFLOPS, where memory capacity and raw processing headroom matter. The inclusion of ray tracing, AMD FSR4, and DirectX 12 Ultimate makes it a practical fit for users working in modern real-time graphics pipelines. Those who need to drive up to four displays simultaneously via a mix of DisplayPort and HDMI 2.1b outputs will also find the connectivity configuration well-matched to multi-monitor setups, and ECC memory support extends its relevance to users with data-integrity-sensitive workloads.

Who is this NOT for?

Users working in environments where power consumption and heat output are constrained may find the 304W TDP difficult to accommodate without a capable PSU and well-ventilated case. Those relying on DLSS or XeSS upscaling — rather than FSR4 — will find those options absent, which could be a limiting factor depending on the software ecosystem they work within. The 298 mm card length also makes it a poor fit for compact or small-form-factor builds where physical clearance is limited, and the absence of a USB-C output rules it out for users who depend on that connection type for display or device compatibility.

Performance:

GPU clock speed	1660 MHz
GPU turbo	2970 MHz
pixel rate	380.2 GPixel/s
floating-point performance	48.6 TFLOPS
texture rate	760.3 GTexels/s
GPU memory speed	2518 MHz
shading units	4096
texture mapping units (TMUs)	256
render output units (ROPs)	128
Has Double Precision Floating Point (DPFP)	✓

The card runs at a base clock of 1660 MHz and scales up to a turbo frequency of 2970 MHz, driving a floating-point throughput of 48.6 TFLOPS alongside a pixel rate of 380.2 GPixel/s and a texture rate of 760.3 GTexels/s. Underpinning these figures are 4096 shading units, 256 texture mapping units, and 128 render output units, while GPU memory operates at 2518 MHz. The card also supports Double Precision Floating Point (DPFP), adding capability for workloads that require high-accuracy computation.

Memory:

effective memory speed	20000 MHz
maximum memory bandwidth	644.6 GB/s
VRAM	16GB
GDDR version	GDDR6
memory bus width	256-bit
Supports ECC memory	✓

The card is equipped with 16GB of GDDR6 VRAM running across a 256-bit memory bus, reaching an effective memory speed of 20,000 MHz and delivering a maximum bandwidth of 644.6 GB/s. ECC memory support is also included, which helps detect and correct memory errors for workloads where data integrity is a priority.

Features:

DirectX version	DirectX 12 Ultimate
OpenGL version	4.6
OpenCL version	2.2
Supports multi-display technology	✓
supports ray tracing	✓
Supports 3D	✓
supports DLSS	✕
has FSR4	✓
has XeSS (XMX)	✕
AMD SAM / Intel Resizable BAR	AMD SAM
has LHR	✕
has RGB lighting	✓
supported displays	4

The card supports DirectX 12 Ultimate, OpenGL 4.6, and OpenCL 2.2, covering a broad range of graphics and compute workloads. Ray tracing, stereoscopic 3D, and multi-display output across up to four screens are all supported, and AMD FSR4 is available for upscaling. AMD SAM is included for CPU-GPU memory access optimization, while RGB lighting is present on the card itself. DLSS and XeSS (XMX) are not supported, and the card does not feature LHR.

Ports:

has an HDMI output	✓
HDMI ports	1
HDMI version	HDMI 2.1b
DisplayPort outputs	3
USB-C ports	0
DVI outputs	0
mini DisplayPort outputs	0

The card's output configuration consists of three DisplayPort outputs and one HDMI 2.1b port, totaling four display connections. There are no DVI, mini DisplayPort, or USB-C outputs present on this card.

General info:

GPU architecture	RDNA 4.0
Thermal Design Power (TDP)	304W
PCI Express (PCIe) version	5
semiconductor size	4 nm
number of transistors	53900 million
Has air-water cooling	✕
width	298 mm
height	131 mm

Built on the RDNA 4.0 architecture, the GPU is manufactured using a 4 nm process and integrates 53,900 million transistors, with a thermal design power of 304W. It connects via a PCIe 5.0 interface and measures 298 mm in width and 131 mm in height. Cooling is handled without an air-water hybrid solution, and the card does not include that type of cooling system.

Final Verdict

The ASRock Radeon RX 9070 XT Steel Legend Dark is a graphics card that brings together a capable set of specifications under the RDNA 4.0 architecture, with 16GB of GDDR6 VRAM, a 2970 MHz boost clock, and 48.6 TFLOPS of floating-point throughput forming a solid foundation for demanding rendering and compute tasks. Its support for ray tracing, AMD FSR4, DirectX 12 Ultimate, and multi-display output up to four screens rounds out a feature set aimed at users who need both graphical capability and practical versatility. The 304W TDP and 298 mm card length mean it asks something of the system it goes into, and its upscaling support is limited to FSR4. For users building around AMD's ecosystem with adequate power and airflow headroom, this card represents a well-specified option with few obvious gaps in its intended use case.