March Patch Tuesday updates boost AMD Ryzen’s performance on Windows 10

Madalina Dinita
by Madalina Dinita
Madalina has been a Windows fan ever since she got her hands on her first Windows XP computer. She is interested in all things technology, especially emerging technologies -- AI and DNA computing in... Read more
Posted:
Affiliate Disclosure

Recent reports have revealed that AMD Ryzen’s performance on Windows 10 is not as good as expected due to various scheduler problems. Reportedly, the Windows 10 scheduler is not able to correctly identify Ryzen’s principal core threads from virtual SMT threads. As a result, the OS doesn’t assign the tasks to a principal core thread, scheduling many of them to a virtual SMT thread. This reduces the overall CPU performance.

AMD Ryzen is an excellent processor, much faster than any previous AMD CPU. Ryzen is exceptional when it comes to productivity, but its leaves a lot to be desired when it comes to games.

The best word to describes the whole story created around AMD Ryzen’s performance is “ambiguous”. On the one hand, Microsoft has already officially acknowledged the CPU’s performance issues and promised to roll out a hotfix to solve this problem. On the other hand, AMD says that there’s no Windows 10 scheduling issue affecting Ryzen’s performance.

However, some Windows 10 users recently confirmed that this month’s Patch Tuesday updates have fixed some of the scheduler issues affecting Ryzen’s performance.

AMD Ryzen and the March Patch Tuesday updates

According to a series of benchmark tests performed on a Windows 10 computer running the latest Patch Tuesday updates, the operating system’s scheduler now distributes the tasks in a more even manner, reducing stress on certain threads. In other words, Microsoft appears to have optimized the tool to work better on AMD Ryzen computers.

OP said that it won’t crash running some FMA and AVX stress test and works are distributed to correct physical cores in chess.

Some improvements in stress test which don’t push CPU to 100% load like SuperPI 32M、AIDA64 Ram latency、x264FHD.
But still not completely fixed, some program which take less threads, like chess, when you set 4 cores , works will be distributed across CCX (2+2).

For more information, you can check out this Chinese forum.

RELATED STORIES YOU NEED TO CHECK OUT:

This article covers:Topics: