AMD publishes no double precision figure for this card, and no lithography row, so both are refused. Vega has no matrix cores, so the 21 TFLOPS FP16 is Rapid Packed Math on the vector units at exactly twice FP32. Same 12.5 billion transistor Vega 10 die as the RX Vega 64, cut from 64 to 56 compute units and clocked lower. amd.com has removed the product page, so the source is a dated Wayback capture of AMD's own page.

AMD Radeon RX Vega 56 8GB

Vega PCIe 2017 Spec confidence: Official
FP32
11
TFLOPS
VRAM
8 GB
HBM2
Bandwidth
410 GB/s
memory
TDP
210 W
50.0 TFLOPS/kW
Download datasheet

Overview

The Radeon RX Vega 56 is the cut-down Vega 10 part, 56 compute units against the RX Vega 64's 64, with the same 8 GB of HBM2 on the same 2048-bit interface but clocked lower for 410 GB/s and a much friendlier 210 W board power. It sits in this catalog as a tier-three historical entry: 8 GB and 10.5 TFLOPS of FP32 do not make a modern inference card, but it is the efficiency point of the Vega generation and the cheapest way to see what HBM2 bandwidth does for memory-bound work. Vega has no matrix cores, so the 21 TFLOPS FP16 AMD quotes is Rapid Packed Math on the vector units at twice FP32, and there is no BF16, FP8, INT8 or INT4 rate published.

Performance

Peak theoretical throughput by precision type

PrecisionPeak
FP64
No verified data available
FP32
32-bit floating point
11TFLOPS
TF32
No verified data available
BF16
No verified data available
FP16
16-bit floating point
21TFLOPS
FP8
No verified data available
FP6
No verified data available
FP4
No verified data available
INT8
No verified data available

Every figure here is dense. The vendor documents no structured (2:4) sparsity mode for this part, so there is no second number to quote.

The Radeon RX Vega 56 8GB in the GPU landscape

Peak FP16 TFLOPS (dense) against TDP, single-GPU parts tracked by Flopper

06001,2001,8002,4003,0000 W250 W500 W750 W1000 W1250 W1500 WInstinct MI355XRadeon RX Vega 56 8GB

Higher and further left is better: more half-precision throughput for less power.

Specifications

Architecture

Vega

Form Factor

PCIe

Launch Year

2017

Process Node

No verified data available

Memory

8 GB HBM2

Bandwidth

410 GB/s

TDP

210 W

Max Power

~241 W

Transistors

12.5 billion

Stream Processors

3,584

Spec Confidence

Official

Full Specifications

Compute Engine
Stream Processors 3,584
Compute Units 56
Base Clock 1.16 GHz
Boost Clock 1.47 GHz
Chip Design
Transistors 12.5 billion
Chiplets 1 (Monolithic)
Memory
VRAM 8 GB
Memory Type HBM2
Bandwidth 410 GB/s
Interface Width 2048-bit
HBM Stacks 2
Memory Clock 1.6 GT/s
Power & Thermal
TDP 210 W
Max Board Power The vendor publishes no maximum board power; the Max Power tile above estimates it from TDP
Enterprise Features
Sparsity No
Compute APIs DirectX 12, OpenGL, OpenCL, Vulkan
General
Form Factor PCIe
Architecture Vega
Process Node No verified data available
Launch Year 2017

Datasheet & Resources

Data Provenance

Every figure traced to a source

Primary Source

Publisher
AMD
Published
2018-11-09

Data Quality

Spec confidence
Official
Clock basis
Boost
Core precisions with figures
2 of 9
Normalization
All values in TFLOPS

Compare cloud providers offering on-demand GPU instances for AI training, inference, and HPC workloads.

Browse GPU Cloud Providers

Similar GPUs

Frequently Asked Questions

How many TFLOPS does the AMD Radeon RX Vega 56 have?

The AMD Radeon RX Vega 56 delivers 11 TFLOPS FP32 and 21 TFLOPS FP16 at peak. Flopper does not currently have a verified FP8 throughput figure for it.

What is the power consumption of the AMD Radeon RX Vega 56?

The AMD Radeon RX Vega 56 has a TDP (Thermal Design Power) rating of 210 watts.

How much memory does the AMD Radeon RX Vega 56 have?

The AMD Radeon RX Vega 56 is equipped with 8 GB of VRAM with 410 GB/s of memory bandwidth.

What architecture is the AMD Radeon RX Vega 56 based on?

The AMD Radeon RX Vega 56 is based on the Vega architecture, launched in 2017.

Stay Updated on GPU Releases

Get notified when new GPUs are added or specifications are updated.

Loading verification...

No spam, unsubscribe anytime.

Back to GPUs