News AMD's New AGESA Code 1.2.0.5 Wreaks Havoc on Performance, Stability

[Roland Of Gilead]

It doesn’t pay to constantly be upgrading your BIOS unless you need something specific and addresses that case

If it ain’t broke don’t fix it

I don't disagree, necessarily, but with Ryzen and AGESA, updating the bios is more akin to a driver update. Practically every new bios will bring some enhancement. It can be DRAM compatibility, increase in system stability, increase in system performance and also to fix minor bugs (some of which a lot of folk may not even know exists) like a particular GPU being glitchy - Check RTX cards and MSI Mobos. For those that bring more performance, well, why would you not want that? Or stability.? Sure, you can make the argument, that it's possible to brick the machine, same can be said of a GPU bios change/flash. For the most part the process is 'child proof' on most motherboards. Oh, yeah and obviously the security aspect too!

 

[hotaru.hino]

I don't disagree, necessarily, but with Ryzen and AGESA, updating the bios is more akin to a driver update. Practically every new bios will bring some enhancement. It can be DRAM compatibility, increase in system stability, increase in system performance and also to fix minor bugs (some of which a lot of folk may not even know exists) like a particular GPU being glitchy - Check RTX cards and MSI Mobos. For those that bring more performance, well, why would you not want that? Or stability.? Sure, you can make the argument, that it's possible to brick the machine, same can be said of a GPU bios change/flash. For the most part the process is 'child proof' on most motherboards. Oh, yeah and obviously the security aspect too!

And none of these things actually mean anything when they're not tested, verified, or go into detail into what actually is being stated.

• DRAM compatibility? Unless the UEFI is updating the board to be compatible from say DDR4-3200 to DDR4-3600, I don't see how there's better compatibility. Especially if the RAM sticks I bought work fine at their highest rated spec. In the 17 or so years I've built computers or swapped parts around, I've never seen RAM not work unless I did a BIOS/UEFI update.
• System stability? What does that even mean? The system acts the way it was expected? There's less crashes? If less crashes, what even caused those crashes to happen? And similarly with DRAM, I've never seen a BIOS/UEFI update make a system that crashes all the time suddenly not crash. Usually an OS re-install or replacing a failing component does it.
• Performance improvements? Has anyone actually verified this is is the case? And if by "performance improvement" it's "now the CPU boosts properly," that's not a performance improvement. That's fixing something that was broken.

I make an exception to security, but at the same time, one has to consider how the vulnerability can be exploited. If, for example, it only happens if UEFI was configured to go on the internet (which is a thing apparently), then a simple fix for that is disable that feature. Also don't run things willy nilly as an admin, because most UEFI level operations from the OS require it.

 

[Roland Of Gilead]

I've never seen RAM not work unless I did a BIOS/UEFI update.

No-one has suggested this. Only that compatibility can be improved. Example your new shiny DIMMS may not work at XMP/DOCP and may only work manually configuring them. A bios update can fix that. This is a pretty common issue.

System stability? What does that even mean? The system acts the way it was expected? There's less crashes? If less crashes, what even caused those crashes to happen? And similarly with DRAM, I've never seen a BIOS/UEFI update make a system that crashes all the time suddenly not crash.

Like I said earlier, I don't necessarily disagree. In some instances the notes for the bios' are sorely lacking in detail. As such, I would agree that it's hard to discern what 'actual' fix is in place. That doesn't mean it's not going to help improve stability though.

Performance improvements? Has anyone actually verified this is is the case? And if by "performance improvement" it's "now the CPU boosts properly," that's not a performance improvement

Well, I don't know about you and what you do after you update a bios, but yes, for me (and I guess a lot of folks too) I do actually verify (or at least try to) measure any performance increases. I do that with a combination of CPU/GPU and SSD testing among others. Sometimes the performance increase is minimal. Other times more pronounced.

If the vagueness of the bios enhancements is preventing you from upgrading your bios, then that's your choice.

My main point is that updating a bios is not as 'bad' or 'scary' as it used to be, and will often bring some form of improvement. Certainly with Ryzen, updating a bios is something most Ryzen users are well used to now, and do frequently.

 

[Deleted member 14196]

Nobody said it was bad it’s just unnecessary most of the time for most users and folks around here love to throw out advice for people to constantly upgrade their BIOS when it could create more problems then it solves especially if they don’t know what they’re doing and get the wrong BIOS flashed

For instance I’ve seen advice for people who’ve had working systems for years and it’s suddenly not working to upgrade their BIOS and I think that is the stupidest advice in the world. There’s no reason to upgrade the BIOS if everything was working good and then suddenly doesn’t something else is wrong and it isn’t the BIOS

My personal preference is to upgrade the BIOS only to get some new piece of hardware to work that has been added to it or to fix a bug that I am having and that’s it. And then following my own advice I have never been let down once in all of my years of computing which is it going on 40 years

 

[Roland Of Gilead]

Nobody said it was bad it’s just unnecessary most of the time for most users and folks around here love to throw out advice for people to constantly upgrade their BIOS when it could create more problems then it solves especially if they don’t know what they’re doing and get the wrong BIOS flashed

For those reading threads like this it's clear why users might think it a 'bad idea' or 'scary' to update their bios, and think 'unless you have a problem to fix, you should keep away from a bios update'. It's a straight forward process made very easy by mobo manufacturers.

Another example of reasons why someone might upgrade (not to fix an issue) would be to activate something on already existing hardware, like Resizable Bar on Nvidia GPU's. Can only be activated with the bios updated. That doesn't mean the GPU wasn't working as intended.

I would defo agree with you that sometimes, the advice on bios updates and their use isn't always correct or appropriate, but can be often.

In terms of personal preference, I guess that's what this is all about. Some folks like to keep their systems up to date including bios. Others don't. Different horse for different courses. Are either wrong? Probably not.

 

[King_V]

I know I'm coming to this party a month late, but has anyone been able to confirm that AGESA 1.2.0.6 B resolves the problems brought in by 1.2.0.4 and 1.2.0.5?

 

[luxm4n]

I'm using AMD Ryzen 5 2600 (2nd Gen, Zen+) and Gigabyte B450 Aorus Pro WiFi, the best BIOS version for that is should be F5 with AGESA 1.0.0.6 (Specifictly for Pinnacle Ridge), but when I recently change my memory with Kingston Fury Beast 2x8GB XMP 3200Mhz (Fury Beast is new branding from Kingston, so this is new stuff, and yes it non-QVL), and I get unstable system, I found that DRAM Voltage stuck at 1.2V, whatever I do, it's still 1.2V, it should be 1.35V, so as many people suggested to upgrade the latest BIOS (at this time, F63a is the latest version), so I do it, fortunately it's work, that fix the problem (memory compatibility), DRAM Voltage now can get 1.35V and my system can run stable without problem,

But, I found that previous version F50-F62 have instability issue like they say on the news (that why I'm here), agree with that, my system experiencing random black screen with Motherboard Debug LED lights up alternately at RAM and CPU, indicating something wrong with CPU (Memory Controller) and RAM, it need force reset using litle butten on the case, but that problem still happened randomly, I test annother RAM kit, and it's still the same problem.

After upgrading the BIOS to F63a with AMD AGESA V2 1.2.0.6 B, the problem is solved !, normally I'm always avoid BIOS version with letter because it's alpha/beta not final version, but in this case I have no choice and need to use that because the latest version without letter (F5-F62) have issue.

We all know that BIOS upgrade is typically for CPU support, but we all know the fact that Memory Controller is on the CPU, so better CPU compatibility = better Memory compatibility.

In my case, I need upgrading BIOS just because I change the RAM (Non-listed on Motherboard QVL), not the CPU, LOL

 

[Roland Of Gilead]

I'm using AMD Ryzen 5 2600 (2nd Gen, Zen+) and Gigabyte B450 Aorus Pro WiFi, the best BIOS version for that is should be F5 with AGESA 1.0.0.6 (Specifictly for Pinnacle Ridge), but when I recently change my memory with Kingston Fury Beast 2x8GB XMP 3200Mhz (Fury Beast is new branding from Kingston, so this is new stuff, and yes it non-QVL), and I get unstable system, I found that DRAM Voltage stuck at 1.2V, whatever I do, it's still 1.2V, it should be 1.35V, so as many people suggested to upgrade the latest BIOS (at this time, F63a is the latest version), so I do it, fortunately it's work, that fix the problem (memory compatibility), DRAM Voltage now can get 1.35V and my system can run stable without problem,

But, I found that previous version F50-F62 have instability issue like they say on the news (that why I'm here), agree with that, my system experiencing random black screen with Motherboard Debug LED lights up alternately at RAM and CPU, indicating something wrong with CPU (Memory Controller) and RAM, it need force reset using litle butten on the case, but that problem still happened randomly, I test annother RAM kit, and it's still the same problem.

After upgrading the BIOS to F63a with AMD AGESA V2 1.2.0.6 B, the problem is solved !, normally I'm always avoid BIOS version with letter because it's alpha/beta not final version, but in this case I have no choice and need to use that because the latest version without letter (F5-F62) have issue.

We all know that BIOS upgrade is typically for CPU support, but we all know the fact that Memory Controller is on the CPU, so better CPU compatibility = better Memory compatibility.

In my case, I need upgrading BIOS just because I change the RAM (Non-listed on Motherboard QVL), not the CPU, LOL

Just remember to always reset CMOS after each bios update. This will ensure any remnants of old micro code are gone.

 

[Sergei Tachenov]

Isn’t it usually cleared automatically anyway after an update?

 

[KyaraM]

I'm really curious on what the security fixes are. It's not uncommon for security fixes to reduce performance by adding additional checks or disabling optimizations, but this seems rather extreme. Perhaps we're still in the non-disclosure period while they fix a serious hole while trying to hackers from figuring it out too quickly.

Iirc there was a new Spectre vulnerability lately that affected Intel CPUs, so Intel looked into the AMD fix from back in 2017 if memory serves me right. On AMD suggestion, just to make sure people aren't jumping on the hate train again over this. They then discovered that the AMD fix has a security issue, too, and relayed it back to them.

https://www.pcgamer.com/intel-amd-and-arm-cpus-hit-by-new-spectre-vulnerability/

Maybe it's a fix for this? I remember reading about it on Tom's Hardware a while back.

 

[luxm4n]

Just remember to always reset CMOS after each bios update. This will ensure any remnants of old micro code are gone.

Nevermind, I was wrong the motherboard are broken, RMA'd, hope it can be fixed.