(GPU) AMD Radeon RX 5500 XT

(Profile updated as of 16th March 2020)

Tiny baby Navi! So cute, so adorable.

(click for full images)

(picture 1) Navi 14 graphics processor, surrounded by four GDDR6 memory chips, populating its 128-bit memory interface. Depending on the model, these can be eitehr 1GB or 2GB chips (4 or 8GB total capacity).

(picture 2) Architectural Block Diagram for Radeon RX 5500 XT's Navi 14 silicon. This is not an official diagram, but it does represent the resources within this processor. A single Workgroup Processor has been disabled. (Diagram credit: Anandtech).

(picture 3) Actual silicon die shot of the Navi 14 graphics processor in RX 5500 XT. Note the single disabled Workgroup Processor, containing two Compute Units; each CU with 64 stream processors and 4 texture units.

(GPU) AMD Radeon RX 5500 XT
(GPU) AMD Radeon RX 5500 XT
(GPU) AMD Radeon RX 5500 XT

Graphics Card Information

Graphics Card: AMD Radeon RX 5500 XT

Graphics Card Manufacturer: Advanced Micro Devices

Graphics Card Release Date: December 12, 2019

Graphics Card MSRP: $199 USD (8GB), $169 USD (4GB)

Graphics Processor Codename: "Navi 14"

Graphics Processor Manufacturer: Advanced Micro Devices

Graphics Processor Implementation: Cut die

Graphics Interface: PCI-E 8x Gen4

Architecture: RDNA1

Lithography Process: TSMC 7nm (N7) FinFET

Approximate die size: 158mm²

SashleyCat's GPU die Size Rating: tiny

Approximate Transistor Count: 6,400 Million

Approximate Transistor Density: 40.5 Million / Square Milimetre


GPU Features

Double-speed FP16 Shading: Yes (Rapid Packed Math)

Asynchronous Compute Capability: Full

DirectX Hardware Support: DX12.1 (FL 12_1)

Dedicated DXR Acelleration on chip: No

Variable-rate Shading: No

Adv. Geometry shading: Yes (Primitive Shader Triangle Culling)

Adv. Geometry shading (Programmable/DX12 Mesh Shaders): No

AI/ML Acceleration: No

Advanced Memory Management: No

Integer and Float Shader Co-execution: No

Tile-based Renderer: Unknown (likely yes) *



GPU Computing Resources

GPU Substructures: 1 Shader Engine, 11 Workgroup Processors (12 Full Chip)

Graphics Cores: 22 Compute Units (24Full Chip)

Graphics Cores per Substructure: 1 Shader engine with 11, each Workgroup Processor contains two Compute Units

Total Stream Processors (ALU/Shaders): 1408 (1536 Full Chip)

Stream Processors per Graphics Core: 64

Graphics Core SIMD Structure: 2 x 32

Total Special Execution Units: 44 Scalar Units (48 Full Chip) *

Special Execution Units per Graphics Core: 2 Scalar Units per Compute Unit, 4 per Workgroup Processor *

Total Texturing Units: 88 (96 Full Chip)

Texturing Units per Graphics Core: 4 (Texture filter)

Pixel Pipelines (ROPs): 32 (8 x Render Backend with 4 Pixels per clock)

Level 2 shared on-chip cache: 2048 KB (likely, but unconfirmed)

Geometry/Tessellation Processors: 1 Geometry Engine, 2 Primitive Units (Including Tessellation)

Raster Engines: 2 Primitive Units (Including Rasterisation)


GPU Memory Subsystem

Graphics Memory Type: GDDR6

Graphics Memory Standard Capacity: 8192 MB or 4096 MB

Graphics Memory Composition: 4 x 1024 MB GDDR6 DRAM chips or 4x 2048 MB GDDR6 DRAM chips

Graphics Memory Access Granularity: 32-bit (4 bytes)

Graphics Memory Standard Clock Speed / Data Rate: 1750 MHz / 14000 MHz

Graphics Memory Full Interface Width: 128-bit (16 bytes per clock)

Graphics Memory Peak Memory Bandwidth: 224 GB/s


GPU Frequency and Peak performance

Graphics Engine Clock: 1845 MHz *

GPU Computing Power FP16: 10,391,040 Million operations per second (FMA)

GPU Computing Power FP32: 5,195,520 operations per second (FMA)

GPU Computing Power FP64: 324,720 Million operations per second (FMA)

GPU Texturing Rate INT8: 162,360‬ Million Texels per second

GPU Texturing Rate FP16: 162,360 Million Texels per second

GPU Pixel Rate: 59,040 Million Pixels per second

GPU Primitive Rate: 3,690 Million triangles per second rasterised (out). 7,380 Million triangles per second into the pipeline before Hardware culling (in). *


GPU Thermal and Power

Standard Cooling Solution: N/A. Custom models only.

Typical Board Power: 130 W

Maximum Board Power: 156 W

Standard External Power Connectors: 8 pin

Maximum Allowed Junction Temperature (TJ Max): 105*C




Graphics Card description

Radeon RX 5500 XT is a budget, low/mid-range graphics card built on AMD's new RDNA GPU architecture. You can read further information about the RDNA architecture on the RX 5700 Profile page. Launched on December 12, 2019, the RX 5500 XT brings the RDNA / Navi family of graphics architecture down to a more affordable, budget-friendly price point starting at $169 for the 4GB model and $199 for the 8GB. Unlike the RX 5700 series, RX 5500 XT is based on different silicon, known as 'Navi 14'; the GPU is significantly smaller, measuring in at just under 160mm², with 6.4 billion transistors. This video card is likely intended to replace the incumbent occupier of the sub 200 USD price point for Radeon, RX 590 and RX 580 video cards using the Polaris 30 and Polaris 20 GPUs, respectively. RX 5500 XT interestingly doesn't feature the fully enabled Navi 14 silicon; with a single WGP disabled containing two Compute Units, it loses 128 Stream Processors and 8 Texture Units from the full chip, likely for yield reasons. The GPU is somewhat more complex than Polaris 10/20/30; with 6.4 billion transistors, but significantly denser due to the 7nm process technology it is built on. Process and architectural improvements allow RX 5500 XT to provide similar performance to RX 580, but at almost half the total board power consumption; representing major progress for Radeon graphics in this area.



RX 5500 XT offers more updated feature set, including Rapid Packed Math, superior DX12.1 support and is significantly more efficient, but isn't notably faster than RX 580 or RX 590. RX 5500 XT is somewhat unique at 200 USD price point, featuring a relatively narrow 128-bit memory bus requring only four DRAM chips to populate; something often found on entry-level models such as Nvidia's xx50 series cards. However, with fast 14 Gbps GDDR6 memory, the Navi 14 graphics processor achieves 224 GB/s of raw memory bandwidth; almost the same as the RX 580 and 590, and matching cheaper RX 570 models; even at half the bus width due to the doubled data rate of GDDR6.



It is also interesting to note that the Navi 14 GPU has only 8X PCI-E lanes physically present on the silicon, however as with all Navi-based silicon thus far, they are operating at Gen 4.0 speed - with a compatible platform the 8 lanes provide similar bandwidth to 16 lanes of the previous Gen 3. As also indicated by its narrow 128-bit interface; these are cost-saving features reserved for entry-level processors in price-sensitive segments of the market. RX 5500 XT using Navi 14, is almost certainly a sucessor for the RX 460 with Polaris 11 GPU, but due to architectural advancements, achieves RX 480+ performance.





Graphics Card approximate 3D Performance

Sashleycat gaming performance rating (2020): Good for 1080p Gaming at Normal / High settings with 60 FPS, or 1080p Maximum settings and 40 FPS+



The Radeon RX 5500 XT 8GB provides performance similar to the previous-generation Radeon RX 580. This puts it slightly ahead (~<5%) of the cheaper GTX 1650 Super ($149) but behind the similar priced GTX 1660, although the 5500 XT 8GB has more Video memory. Performance is thus similar to the GeForce GTX 1060 6GB, making the card suitable for gaming at normal / high settings and 1920x1080 resolution and 60 FPS, or if a lower framerate is tolerable, the card can easily cope with Maximum settings and 40 FPS+.





Notes

Tile-based Renderer:

I am currently unsure of the nature of the Draw-Stream Binning Rasteriser in Navi silicon. I am fairly certain it does feature some form of Tile-based renderer, however.



Special Execution Units:

I am currently unsure of the amount of Load/Store Units in the CU for Navi. I think it remains the same as GCN, at 16, but I will not list them here until I have solid confirmation, as the new block diagrams for Navi don't actually state LSUs at all.



Graphics Engine Clock:

As with "Vega" based cards, this represents the hard-limited maximum clock speed for the graphics processor. Actual, in-game clock speed will be slightly lower, depending on many factors such as thermals, power use and I assume, silicon quality. It is important to note that "Navi 14" introduces a new "Game clock" rating that will be more realistic of the speeds you will see in 3D, gaming workloads. That speed is 1717 MHz. In my experience with the RX 5500 XT, actual clock speed is closer to 1845 MHz, even under full load.



GPU Primitive Rate:

This is based on my understanding of Navi's Primitive Shading implementation for fast hardware culling.



Misc.

This bit is for my personal opinion on this Graphics card / Graphics processor



Sashleycat's Awesomeness Rating: Good technology, but could be better value.

©2020 by Sashleycat (Read the Legal stuff)