AMD Launches 7nm RDNA Architecture Based Radeon Pro W5700 for Workstations

Back at CES when AMD announced its “built specifically for gamers” RDNA architecture, I had the impression that team red will update its gaming portfolio using the latest Navi GPU and retain GCN for compute-oriented workstation graphics. It seems, however, that AMD plans to use RDNA in both product GPUs and workstations across the board. In the shape of the Radeon Pro W5700, the world’s first 7 nm professional graphics card has been revealed today.

Technically speaking, it seems the same as the Radeon RX 5700 packaging the same GDDR6 memory 36CUs and 8 GB coupled with a 256-bit bus. The only difference is the central clock and the TDP. The W5700 has a 1930MHz (the RX 5700 tops at 1725MHz) much higher boost clock, but this extra performance comes at the cost of a higher TDP. Although the RX 5700 has an 180W power rating, the W5700 is rated to 205W. We realize, however, that the TDP load will be much higher than that. I think the W5700 could be close to the 300W mark as the RX 5700 pulls up as much as 230W under load.

Graphics Card NameRadeon Pro WX 7100Radeon Pro WX 8200Radeon RX 5700Radeon Pro W5700
GPUPolaris 10Vega 10Navi 10Navi 10
Process Node14nm14nm7nm7nm
Compute Units36563636
Stream Processors2304358423042304
ROPs32646464
Clock Speed (Peak)1243 MHz1500 MHz1725 MHz1930 MHz
VRAM8 GB GDDR58 GB HBM28GB GDDR68 GB GDDR6
Memory Bus256-bit2048-bit256-bit256-bit
Memory Bandwidth224 Gbps484 Gbps448 Gbps448 Gbps
TDP150W230W180W205W
Price$799$999$349$799

AMD offers an average performance gain of up to 41% (per watt) compared to the older Vega WX 8200. Looking at the product Navi 10 cards ‘ energy profile, this is consistent with what we’ve seen in gaming workloads. Given that the W5700 is clocked by 200MHz higher, it will be important to see how much of a difference it will make in comparison to the average power consumption.

Compared to the Quadro RTX 4000 from NVIDIA, the W5700 works more or less in line with most applications while receiving the same power level. Of course, in applications that use the RTCores for BVH acceleration and the Tensors for AI-related workloads, the Quadro will still have a big advantage, but it’s also more expensive. In addition, the W5700 will perform roughly the same as the Quadro RTX 4000 for traditional workloads that do not take advantage of NVIDIA’s candies while costing a good $200 less.

Recent Articles

Amazon is Spending $1 Billion in Indian SMBs: CEO Jeff Bezos

Amazon CEO Jeff Bezos has confirmed pledge of $1 billion to empower small and medium-sized enterprises (SMBs) in India. The announcement is...

RedmiBook 13 May be Xiaomi’s First Laptop to Launch in India

Xiaomi is set to launch its Sub-brand laptop manufacturing, RedmiBook in India in the near future, suggested by recent trademark listing. Xiaomi...

Intel Announced Its Upcoming Xe Discrete GPU in CES 2020

At the Consumer Electronics Show (CES) 2020, Intel presented its first discrete GPU, dubbed "DG1." For quite a long time now, the...

Apple A14 chip for 2020 iPhone to be built on 5nm process by TSMC

By September this year, Apple is not scheduled to launch its iPhone lineup 2020. The business appears, however, to have begun to...

Qualcomm Unveils New Generation 3D Sonic Max Fingerprint Sensor with 17x Larger Surface Area

In addition, Qualcomm unveiled the flagship chipset Snapdragon 19865 SoC at its annual qualcomm Snap Dragon Tech Summit, the latest fingerprint display...

Related Stories

Leave A Reply

Please enter your comment!
Please enter your name here

Stay on op - Ge the daily news in your inbox