I wanted to share some finding's I've discovered on the whole USB disconnectivity issues that have plagued Ryzen owners for years.
To re-cap, this issue got really popular a few years ago after the Ryzen 5000 launch. USB issues surrounding disconnecting problems became so widespread that AMD officially "patched" the issue through new AGESA revisions.
While thankfully, that fixed Ryzen's USB issues for the vast majority of AMD owners. The issue still affects some users, even on the patched AGESA firmware updates.
I am one of the affected people, but only barely. The only USB connectivity issue i have is my keyboard will randomly disconnect like once a month when my system starts. Obviously its barely a problem at all. Regardless, I still wanted to figure out what's going on, and why AMD's USB "patch" does not work for everyone.
Undervolting:
I discovered that SoC voltage and possibly Infinity Fabric voltages are the culprit. When you enable XMP, virtually all motherboards will automatically overvolt both the SoC and infinity fabric voltage to a pre-determined value I believe is set by AMD's AGESA firmware. In my case, enabling XMP will automatically overvolt the SoC and Infinity Fabric on my CPU by 100mv — going from 1.0v to 1.1v (for both IF and SOC). (I've found this to be the case n both my Asus X370 Crosshair Hero and MSI B450 Pro Carbon AC as well as on my Ryzen 7 5800X3D and Ryzen 5 3600).
This in of itself is perfectly (or should be) acceptable, AMD states the maximum safe SoC voltage is 1.2v and I believe the max safe Infinity Fabric voltage is 1.15v. So we are well below dangerous limits.
Regardless, I discovered that either the higher SoC voltage or both the SoC and IF voltages are interfering with USB connectivity somehow. AMD's default out-of-the-box voltages for the SoC is 1.000v, and 0.900v for the infinity fabric. I manually changed my SoC and IF voltages to those exact values while retaining XMP, and I no longer have keyboard disconnects. Actually to be specific i'm running 20mv higher on both voltages because I always add voltage after stress testing at a lower voltage to insure i'm not riding on the edge of stability. Regardless even at this slightly higher voltage i no longer have USB disconnecting problems whatsoever.
If you have USB connectivity issues on your Ryzen system and are on the "patched" AGESA microcode updates, it might be worth undervolting the SoC and or infintiy fabric to see if that helps. Going back down to default voltage values should help, but remember to do a memory stress test to make sure those voltages are stable.
Logically it makes a lot of sense why this is the case. SoC voltage is what connects all of the system's I/O to the CPU including the USB ports. Also, it wouldn't surprise me if AMD only tested the USB firmware patch on a default memory configuration with JEDEC timings which would leave the IF and SoC voltages at their default spec.
To re-cap, this issue got really popular a few years ago after the Ryzen 5000 launch. USB issues surrounding disconnecting problems became so widespread that AMD officially "patched" the issue through new AGESA revisions.
While thankfully, that fixed Ryzen's USB issues for the vast majority of AMD owners. The issue still affects some users, even on the patched AGESA firmware updates.
I am one of the affected people, but only barely. The only USB connectivity issue i have is my keyboard will randomly disconnect like once a month when my system starts. Obviously its barely a problem at all. Regardless, I still wanted to figure out what's going on, and why AMD's USB "patch" does not work for everyone.
Undervolting:
I discovered that SoC voltage and possibly Infinity Fabric voltages are the culprit. When you enable XMP, virtually all motherboards will automatically overvolt both the SoC and infinity fabric voltage to a pre-determined value I believe is set by AMD's AGESA firmware. In my case, enabling XMP will automatically overvolt the SoC and Infinity Fabric on my CPU by 100mv — going from 1.0v to 1.1v (for both IF and SOC). (I've found this to be the case n both my Asus X370 Crosshair Hero and MSI B450 Pro Carbon AC as well as on my Ryzen 7 5800X3D and Ryzen 5 3600).
This in of itself is perfectly (or should be) acceptable, AMD states the maximum safe SoC voltage is 1.2v and I believe the max safe Infinity Fabric voltage is 1.15v. So we are well below dangerous limits.
Regardless, I discovered that either the higher SoC voltage or both the SoC and IF voltages are interfering with USB connectivity somehow. AMD's default out-of-the-box voltages for the SoC is 1.000v, and 0.900v for the infinity fabric. I manually changed my SoC and IF voltages to those exact values while retaining XMP, and I no longer have keyboard disconnects. Actually to be specific i'm running 20mv higher on both voltages because I always add voltage after stress testing at a lower voltage to insure i'm not riding on the edge of stability. Regardless even at this slightly higher voltage i no longer have USB disconnecting problems whatsoever.
If you have USB connectivity issues on your Ryzen system and are on the "patched" AGESA microcode updates, it might be worth undervolting the SoC and or infintiy fabric to see if that helps. Going back down to default voltage values should help, but remember to do a memory stress test to make sure those voltages are stable.
Logically it makes a lot of sense why this is the case. SoC voltage is what connects all of the system's I/O to the CPU including the USB ports. Also, it wouldn't surprise me if AMD only tested the USB firmware patch on a default memory configuration with JEDEC timings which would leave the IF and SoC voltages at their default spec.
Last edited:
Facebook
Twitter
Instagram