A publishing defect on AMD's own page is worth flagging. The Dual Slot page prints 'Peak Half Precision (FP16 Vector) Performance 123 TFLOPs' where the standard W7900 page prints 61.3 for that row on identical silicon. 123 is the FP16 Matrix figure duplicated into the Vector cell; 61.3 is the correct vector rate. The matrix, INT8 and INT4 rows on both pages agree exactly and are what this row stores.

AMD Radeon PRO W7900 Dual Slot 48GB

RDNA 3 PCIe 2024 5nm Spec confidence: Official
FP32
61
TFLOPS
VRAM
48 GB
GDDR6
Bandwidth
864 GB/s
memory
TDP
295 W
207.8 TFLOPS/kW
Download datasheet

Overview

The Radeon PRO W7900 Dual Slot is the same 96 compute unit, 48 GB ECC card as the original W7900, rebuilt into a double-slot cooler so that four of them fit in one workstation. That is the entire point of the product: 192 GB of GDDR6 and 492 TFLOPS of aggregate FP16 matrix in a machine you can put under a desk. Per card the numbers are unchanged at 61.3 TFLOPS of FP32 vector, 123 TFLOPS of FP16 matrix, 864 GB/s of bandwidth and 295 W of board power. AMD announced it at Computex 2024, a year after the triple-slot original.

Performance

Peak theoretical throughput by precision type

PrecisionPeak
FP64
No verified data available
FP32
32-bit floating point
61TFLOPS
TF32
No verified data available
BF16
No verified data available
FP16
16-bit floating point
123TFLOPS
FP8
No verified data available
FP6
No verified data available
FP4
No verified data available
INT8
8-bit integer
123TOPS
INT4
4-bit integer
245TOPS

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 PRO W7900 Dual Slot 48GB 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 PRO W7900 Dual Slot 48GB

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

Specifications

Architecture

RDNA 3

Form Factor

PCIe

Launch Year

2024

Process Node

5nm

Memory

48 GB GDDR6

Bandwidth

864 GB/s

TDP

295 W

Max Power

295 W

Transistors

57.7 billion

Stream Processors

6,144

Spec Confidence

Official

Full Specifications

Compute Engine
Stream Processors 6,144
Tensor Cores 192 (RDNA 3)
Compute Units 96
Chip Design
Transistors 57.7 billion
Process Node 5nm
Memory
VRAM 48 GB
Memory Type GDDR6
Bandwidth 864 GB/s
Interface Width 384-bit
Interconnect & I/O
PCIe Gen 4 x16
Cache
Infinity Cache 96 MB
Power & Thermal
TDP 295 W
Max Board Power 295 W
Power Connector 2x8-Pin
Cooling Active
Enterprise Features
ECC Memory Yes
Sparsity No
Compute APIs ROCm, OpenCL, HIP
Physical & Media
Card Length 280 mm
Width 2-slot
General
Form Factor PCIe
Architecture RDNA 3
Process Node 5nm
Launch Year 2024

Datasheet & Resources

Data Provenance

Every figure traced to a source

Primary Source

Publisher
AMD
Published
2024-06-02

Data Quality

Spec confidence
Official
Clock basis
Boost
Core precisions with figures
3 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 PRO W7900 Dual Slot have?

The AMD Radeon PRO W7900 Dual Slot delivers 61 TFLOPS FP32 and 123 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 PRO W7900 Dual Slot?

The AMD Radeon PRO W7900 Dual Slot has a TDP (Thermal Design Power) rating of 295 watts.

How much memory does the AMD Radeon PRO W7900 Dual Slot have?

The AMD Radeon PRO W7900 Dual Slot is equipped with 48 GB of VRAM with 864 GB/s of memory bandwidth.

What architecture is the AMD Radeon PRO W7900 Dual Slot based on?

The AMD Radeon PRO W7900 Dual Slot is based on the RDNA 3 architecture, launched in 2024.

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