The first Geekbench Metal listing of the AMD Radeon Pro 5500M is now available, and it looks like it offers a good upgrade over the Radeon Pro 560X found in the mid-2018 MacBook Pro 15. However, those who already have the end-2018 MacBook Pro 15 with a Radeon Pro Vega 20 might have less of an incentive to upgrade from a GPU standpoint.
https://www.notebookcheck.net/AMD-Radeon-Pro-5500M-Geekbench-Metal-score-8-faster-than-the-Radeon-Pro-Vega-20-and-35-faster-than-the-Radeon-Pro-560X.442747.0.html
Dear Redaktion, you have failed at math.
This new card is actually 9% and 55% faster respectively. Or if you want to stick with the numbers, the title should look like:
"Radeon Pro Vega 20 is 8% slower and the Radeon Pro 560X is 35% slower than the AMD Radeon Pro 5500M Geekbench Metal score"
Quote from: joegalamb on November 17, 2019, 11:49:06
Dear Redaktion, you have failed at math.
This new card is actually 9% and 55% faster respectively. Or if you want to stick with the numbers, the title should look like:
"Radeon Pro Vega 20 is 8% slower and the Radeon Pro 560X is 35% slower than the AMD Radeon Pro 5500M Geekbench Metal score"
You're right. Corrected :)
The Apple store claims their 5500M is 24 compute units (in the Buy screen help text). Hmm.
Apple uses a Pro version with fully enabled Navi 14 silicon.
I'll be interested to see how this shakes out in practice. An artifical benchmark running on the Meal API should be close to the best-case performance scenario for Vega and Polaris, while one of RDNA's big improvements was getting more performance out of the architecture *in practice*. This might be slightly understating the performance improvement in real-world applications.