Author: parsec
Subject: Downgrade BIOS AB350M PRO4 to 2.40?
Posted: 13 Jul 2017 at 10:45am
You can easily change your AB350M Pro4 board's UEFI/BIOS version to any version that is available for your board. Nothing will stop you from doing so, and it will not cause any problem. This is in general true for any ASRock mother board, with a few exceptions.
Examples of the exceptions are the Intel 100 series chipset/Skylake processor boards that can be used with Kaby Lake processors, when paired with the specific Kaby Lake processor compatible UEFI/BIOS versions. If a user of one of these boards with a Kaby Lake processor and Kaby Lake compatible UEFI/BIOS version, later changed to a Skylake only compatible version, the PC would not start. This is also true for Intel X99 boards and Broadwell processors, as well as Intel X79 boards.
You current issue with UEFI version 2.50 with AGESA 1.0.0.6 is strictly a memory compatibility or memory OC issue, and not at all related to your CPU OC. The triple start and stop cycle during POST is the AM4 Advance Boot Training feature attempting to train your memory to work at your attempted memory OC, which is failing with AGESA 1.0.0.6. Any Ryzen user that has over clocked their memory has most likely experienced this with each UEFI/AGESA version, I certainly have. I also use memory that is supposedly specifically designed for Ryzen, G.SKILL FlareX. I have that happen with the first or latest UEFI/BIOS version, if I try a memory OC above 2933. Disable the AM4 Advance Boot Training feature if you don't feel it is helping you, or configure the number of cycle attempts to a smaller number. Both can be configured in the UEFI/BIOS.
The AGESA microcode updates that are supplied solely by AMD, and whose behavior is solely determined by AMD, are updates and fixes for Ryzen's memory support and compatibility. Unfortunately, the terms "support" and "compatibility" are assumed to mean that any random DDR4 memory model will become Ryzen compatible and/or become easier to over clock. While I understand why that is believed to be the case, the reality is it is not the case.
As new AGESA microcode update appear, and are dialing in and tightening up Ryzen's DDR4 memory compatibility (compatibility meaning the memory that is compatible with the Ryzen memory controller, not necessarily making the Ryzen memory controller compatible with more memory models) we are beginning to see what appears to be some DDR4 memory models being left behind from a support aspect. Meaning they no longer work as well with the new AGESA version than they did with the previous AGESA version.
Ryzen users that experience reduced memory compatibility are interpreting this as a problem with the new UEFI itself, or the AGESA microcode. While it is very difficult if not impossible for the user to know exactly what is happening, IMO we are simply seeing Ryzen's memory support being finally sorted out, or starting to be sorted out. That includes reducing the memory OC capability for some models of DDR4 memory, or even causing some models of DDR4 memory to no longer at all with Ryzen. That you have experienced the former yourself with boards from two different manufactures indicates that is the unfortunate, but likely situation.
How you OC your memory with Ryzen is very important. If you are simply applying the XMP profile of your memory, you have not configured it appropriately for a Ryzen system. Kingston HyperX Fury DDR4 was designed for Intel X99 systems, the first use of DDR4 memory in PCs:
https://support.kingston.com/us/hyperx/certifications
You have not supplied any details of how you OC your memory, or have the full model number of your memory, so we cannot comment upon it beyond the few basic things I mentioned.
The Full HD UEFI feature/option has been used for several years now, on many different AMD and Intel platforms. While I agree it would be better to have this option disabled by default, it is actually very rare that a monitor cannot be used with it at its default setting. Personally, I have seven different monitors models (LG, Dell) and a small (40") HD TV that I use with my PCs, all with ASRock boards, and I have never had a problem with the Full HD UEFI feature being enabled. User reports of problems with Full HD UEFI are rare, and mainly associated with using TVs as monitors.
Subject: Downgrade BIOS AB350M PRO4 to 2.40?
Posted: 13 Jul 2017 at 10:45am
You can easily change your AB350M Pro4 board's UEFI/BIOS version to any version that is available for your board. Nothing will stop you from doing so, and it will not cause any problem. This is in general true for any ASRock mother board, with a few exceptions.
Examples of the exceptions are the Intel 100 series chipset/Skylake processor boards that can be used with Kaby Lake processors, when paired with the specific Kaby Lake processor compatible UEFI/BIOS versions. If a user of one of these boards with a Kaby Lake processor and Kaby Lake compatible UEFI/BIOS version, later changed to a Skylake only compatible version, the PC would not start. This is also true for Intel X99 boards and Broadwell processors, as well as Intel X79 boards.
You current issue with UEFI version 2.50 with AGESA 1.0.0.6 is strictly a memory compatibility or memory OC issue, and not at all related to your CPU OC. The triple start and stop cycle during POST is the AM4 Advance Boot Training feature attempting to train your memory to work at your attempted memory OC, which is failing with AGESA 1.0.0.6. Any Ryzen user that has over clocked their memory has most likely experienced this with each UEFI/AGESA version, I certainly have. I also use memory that is supposedly specifically designed for Ryzen, G.SKILL FlareX. I have that happen with the first or latest UEFI/BIOS version, if I try a memory OC above 2933. Disable the AM4 Advance Boot Training feature if you don't feel it is helping you, or configure the number of cycle attempts to a smaller number. Both can be configured in the UEFI/BIOS.
The AGESA microcode updates that are supplied solely by AMD, and whose behavior is solely determined by AMD, are updates and fixes for Ryzen's memory support and compatibility. Unfortunately, the terms "support" and "compatibility" are assumed to mean that any random DDR4 memory model will become Ryzen compatible and/or become easier to over clock. While I understand why that is believed to be the case, the reality is it is not the case.
As new AGESA microcode update appear, and are dialing in and tightening up Ryzen's DDR4 memory compatibility (compatibility meaning the memory that is compatible with the Ryzen memory controller, not necessarily making the Ryzen memory controller compatible with more memory models) we are beginning to see what appears to be some DDR4 memory models being left behind from a support aspect. Meaning they no longer work as well with the new AGESA version than they did with the previous AGESA version.
Ryzen users that experience reduced memory compatibility are interpreting this as a problem with the new UEFI itself, or the AGESA microcode. While it is very difficult if not impossible for the user to know exactly what is happening, IMO we are simply seeing Ryzen's memory support being finally sorted out, or starting to be sorted out. That includes reducing the memory OC capability for some models of DDR4 memory, or even causing some models of DDR4 memory to no longer at all with Ryzen. That you have experienced the former yourself with boards from two different manufactures indicates that is the unfortunate, but likely situation.
How you OC your memory with Ryzen is very important. If you are simply applying the XMP profile of your memory, you have not configured it appropriately for a Ryzen system. Kingston HyperX Fury DDR4 was designed for Intel X99 systems, the first use of DDR4 memory in PCs:
https://support.kingston.com/us/hyperx/certifications
You have not supplied any details of how you OC your memory, or have the full model number of your memory, so we cannot comment upon it beyond the few basic things I mentioned.
The Full HD UEFI feature/option has been used for several years now, on many different AMD and Intel platforms. While I agree it would be better to have this option disabled by default, it is actually very rare that a monitor cannot be used with it at its default setting. Personally, I have seven different monitors models (LG, Dell) and a small (40") HD TV that I use with my PCs, all with ASRock boards, and I have never had a problem with the Full HD UEFI feature being enabled. User reports of problems with Full HD UEFI are rare, and mainly associated with using TVs as monitors.