AMD

Radeon VII

The Radeon VII is AMD’s 16GB graphics card with a 295W board power rating, built on the Vega 20 architecture. Below you’ll find its full spec sheet and gaming and productivity benchmarks. No board versions are in stock right now, but the specs and benchmarks below still apply whenever it returns to retailers.

Prices update regularly and may differ at the retailer. This page contains affiliate links.

Gaming performance

Gaming score: geometric mean of in-game FPS, normalized so the RTX 5090 = 100. Render and synthetic scores are in the Benchmarks section below.

Gaming scores are normalised performance indexes derived from Tom’s Hardware’s published GPU benchmarks. Performance per dollar and per watt are our own calculations on top of that data.

Radeon VIITop GPU
33.5
100.0

1080p Gaming

21.2
100.0

1440p Gaming

17.8
100.0

4K Gaming

The Radeon VII scores 33.5 at 1080p, 21.2 at 1440p, and 17.8 at 4K on a scale where 100 = the fastest GPU we test (RTX 5090). These scores apply to all board versions of this chip.

Performance-per-dollar rank among all GPUs with live pricing

Value at 1080p

—

Value at 1440p

—

Value at 4K

—

Benchmarks

Standardized benchmark scores for this chip, each compared to the fastest card we track and linked to its source. They apply to every board version.

3D & gaming

3DMark Fire Strike Graphicstechnical.city ↗
Radeon VII27,449 points
GeForce RTX 4090 (best)75,641 points
3DMark Steel NomadUL 3DMark ↗
Radeon VII2,312 points
GeForce RTX 5090 (best)14,544 points
Radeon VII16,227 G3D Mark
GeForce RTX 5090 (best)38,974 G3D Mark

Render

Radeon VII807 samples/min
GeForce RTX 5090 (best)15,017 samples/min

Radeon VII availability

Live retailer stock for this model.

Currently out of stock

None of the board versions of the Radeon VII are in stock at the retailers we track right now. Prices refresh daily, so check back — or compare in-stock alternatives below.

Browse in-stock graphics cards

Specifications

The reference Radeon VII measures 280 mm long.

Performance

Boost clock
1800 MHz
Base clock
1400 MHz
Shader cores
3,840
Compute units
60
Texture units (TMUs)
240
Render outputs (ROPs)
64
RT cores
0

Memory

VRAM
16GB HBM2
Memory bus
4096-bit
Bandwidth
1024 GB/s
Memory speed
2 Gbps
L2 cache
4 MB

GPU die

Architecture
Vega 20
Process node
TSMC N7
Foundry
TSMC
Die size
331 mm²
Transistors
13.2 billion
Generation
Radeon VII

Power

TDP
295W
Recommended PSU
650W
Power connectors
2 8-pin

Display & outputs

Max resolution
7680 × 4320
Max displays
4
HDCP
2.2
Outputs
1 HDMI 2.0b, 3 Displayport 1.4a
PCIe interface
PCIe 3.0 ×16

APIs & software

DirectX
12 (12_1)
Vulkan
1.3
OpenGL
4.6
OpenCL
2.0
Shader model
6.5

Video encoding & decoding

CodecEncodeDecode
H.264 / AVCSupportedSupported
H.265 / HEVCSupportedSupported
AV1Not supportedNot supported
VP9Not supportedSupported

Encoder: VCE 4.1 · Decoder: UVD 7

Features

Upscaling
FSR 2
Also runs
XeSS
Resizable BAR
No

Physical

Reference length
280 mm
Slot width
2 slots
Released
2019-02-07
MSRP
$699.00

3,840 shader cores feed a 4096-bit memory interface delivering 1024 GB/s of peak bandwidth. Compare the Radeon VII against the full GPU lineup on the full graphics card rankings.

Local AI performance

What the Radeon VII can run on your own machine, and how fast.

Largest model it can run (8K context)

Q4 (the usual choice, near-original quality)
~20B
Q8 (closer to original, half the model size)
~12B
FP16 (unquantised, for training and research)
~6.6B

Parameter counts the card holds with room for an 8K conversation, counting the model weights plus its attention cache. A longer conversation needs more memory, so it lowers these figures.

Speed & value

Speed, 8B model at Q4
141-166 tok/s
Memory bandwidth
1024 GB/s
Price per GB of VRAM
—

Generation speed tracks memory bandwidth, because every word produced reads the whole model from memory. Shown as a range.

Models that fit, at Q4

ModelSizeMemory neededEst. speed
GPT Oss 20B20.9B13.2GB—
Granite Code 20B20.1B13.7GB56-66 tok/s
Starcoder2 15B16.0B9.7GB71-84 tok/s

Popular models small enough for this card at Q4, one per family. Speed is left blank for mixture-of-experts models, where a bandwidth estimate is not reliable enough to publish.

Affiliate disclosure: Some links on this page are affiliate links — MaxMyBuild may earn a commission at no extra cost to you if you buy through them. As an Amazon Associate, MaxMyBuild earns from qualifying purchases. For Amazon, current price and availability are shown on the Amazon product page; product prices and availability are accurate as of the date and time of purchase and are subject to change. Prices shown for other retailers are approximate and may differ at checkout. “Search” links are non-affiliate and provided for convenience only.

Advertisement