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

Sergei Tachenov

Commendable
Jan 22, 2021
64
64
1,610
5600X, B550 Aorus Pro V2, using the latest 1.2.0.5 BIOS (F14), no issues so far. I don't overclock, though (PBO set to Auto, whatever that means), and running a moderately fast RAM kit (3200 CL16, 64 GB worth of Samsung M-Die if that matters). Don't observe any significant differences in temps as well. In fact, I haven't observed any differences after the upgrade.
 
  • Like
Reactions: phenomiix6

court655

Commendable
Oct 6, 2021
9
4
1,515
Makes so much sense now. My windows has been BSOD a lot lately and I thought it was my overclock being unstable at 1.35v with vdroop on auto. But never realized its the bios I recently updated to. I'm using a X570 Auros Master board

Also noticed a performance drop in Cinebench
 
Last edited:
I've noticed some issues ever since F3c BIOS from Gigabyte for my X570S Aorus Master, which has 1.2.0.4, and was about to install F3 for it which has 1.2.0.5. Now the long wait for a proper, stable BIOS...

And yes, this is my first step away from ASUS since the Socket A days, and this is the first regret I have had.
 
Last edited:

OldWalter____IsMyDog

Commendable
Feb 18, 2022
10
6
1,515
I tried running a Gigabyte X570S Aero G with both a 3900X and 5950X on the F3d BIOS update and it was an unmitigated disaster. WHEA errors every few minutes, reboots with no warning. Just utter garbage. Rolled back to F2 and it's been rock solid ever since.
 

_Shatta_AD_

Reputable
Jan 27, 2020
49
30
4,560
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

Well according to Gigabyte, it is “highly recommended” for all users to upgrade to the F15 (I’m on a B550I Aorus Pro AX) BIOS “ASAP” due to critical vulnerabilities that were fixed in AGESA 1.2.0.5 . Anyway, after the upgrade, first thing I noticed is that they removed the option to set PBO max override so you’re stuck with whatever profile you have saved and restoring profiles doesn’t update memory OC parameters requiring manual changes.
Benchmarks after restoring settings showed a 5-12% drop in performance even at similar clock as before not to mention that it fails to OC to the same clock as before without bumping up all the PBO limits resulting in more heat production.
System would also BSOD randomly with message alluding to a new cpu detected and then proceed to resetting the BIOS erasing all manual settings and even the saved profiles. I’m surprised there’s no communication whatsoever from Gigabyte even though there are numerous reports of similar issues popping up everywhere.
 
  • Like
Reactions: -=Quetzalcoatl=-

-=Quetzalcoatl=-

Commendable
Feb 18, 2022
2
0
1,510
Well according to Gigabyte, it is “highly recommended” for all users to upgrade to the F15 (I’m on a B550I Aorus Pro AX) BIOS “ASAP”


Same here, but B550 AORUS ELITE (rev. 1.0). Gigabytes wording for F14:

"
• Major vulnerabilities updates, customers are strongly encouraged to update to this release at the earliest.
Credits to "Assaf Carlsbad and Itai Liba from SentinelOne"
• Introduce capsule BIOS support starting this version."
 
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

Which is generally true, but with AGESA not updating can leave you with security holes and compatibility issues, especially with new motherboards, and, as (in)famously experienced by Ryzen 3000 series early adopters, with CPUs unable to perform to advertised specifications.

AMD Announces BIOS Fix for Ryzen 3000 Boost Clocks, Update Comes September 10th | Tom's Hardware (tomshardware.com)
 
  • Like
Reactions: Roland Of Gilead

court655

Commendable
Oct 6, 2021
9
4
1,515
I just went back to the 1.2.0.3.. Cinebench up 300 pts... and temps running 14 degrees lower. Never seen anything like this before, same exact settings, literally just a bios change. I'll definitely be saving scores and running bench's whenever I do bios updates from now on.
 
Which is generally true, but with AGESA not updating can leave you with security holes and compatibility issues, especially with new motherboards, and, as (in)famously experienced by Ryzen 3000 series early adopters, with CPUs unable to perform to advertised specifications.

AMD Announces BIOS Fix for Ryzen 3000 Boost Clocks, Update Comes September 10th | Tom's Hardware (tomshardware.com)
That's why the qualifier "if it ain't broke" is there. What you described is something that's broken, therefore, there's a reason to update.

However, the infuriating thing about most BIOS updates is almost nobody tells us what the damn thing is doing other than some generic "improves performance" or "updates to this version." This is system level software, it's obviously going to affect the system. I'd like to know what the hell the thing is going to do before I apply it. If anything, why doesn't AMD supply patch notes for each AGESA version? This is on top of my grievances with a so-called hardware company not releasing datasheets for their parts. Or at least, if there is such documentation, they don't make it openly available.

EDIT: I did find out that AMD does have a library of technical documentation (https://www.amd.com/en/support/tech-docs) but there's a lot of documentation that's missing. Like there's no hardware data sheets for Ryzen and there doesn't appear to be anything for their GPUs.
 
Last edited:

Irisena

Commendable
Oct 1, 2019
94
10
1,565
B450m Mortar Max, 5900X. No issue so far, my clocks are still the same, and 3600mhz cl18 RAM is still running just fine. I noticed literally 0 difference between 1.2.0.2 and 1.2.0.5
 
Feb 19, 2022
2
0
10
B550 Aorus Master with 5900x here. After updating to F14 (1.2.0.5), my memory kit (3600MHz, CL16) got damaged - both modules! I tested it on another MOBO, too- random memory errors. Now I'm waiting for RMA.
 
D

Deleted member 14196

Guest
Which is generally true, but with AGESA not updating can leave you with security holes and compatibility issues, especially with new motherboards, and, as (in)famously experienced by Ryzen 3000 series early adopters, with CPUs unable to perform to advertised specifications.

AMD Announces BIOS Fix for Ryzen 3000 Boost Clocks, Update Comes September 10th | Tom's Hardware (tomshardware.com)
You don’t upgrade the BIOS until you need to. that means is you don’t upgrade it until you’re upgrading your hardware and there’s a reason to. All that crap about security ha ha Ha guess what the government has access to your CPU at all times so basically there is no security

I will forgo the BS by not upgrading constantly and you go ahead and ruin your ram and ruin your performance all you want to in the name of “security”. Don’t believe all the hype that you hear
 
  • Like
Reactions: KyaraM and renz496

jp7189

Distinguished
Feb 21, 2012
531
303
19,260
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.
 
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.
Not all security fixes are about adding more and more layers. There are minor bugs that are security vulnerabilities. Heartbleed for example was caused by a buffer overflow due to an off-by-one error. So the fix has zero impact to performance.

Security vulnerabilities that do require more is an indication there's a flaw in the design.
 

Sergei Tachenov

Commendable
Jan 22, 2021
64
64
1,610
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.
There was a reference somewhere. Something about code running at ring 0 (OS kernel and lowest level drivers) can corrupt SMRAM in a partially controlled manner. I saw a reference to this issue here:
View: https://www.reddit.com/r/Amd/comments/smfwyw/anyone_using_gigabyte_b550_boards_what_is_the/
 
  • Like
Reactions: jp7189

d0x360

Distinguished
Dec 15, 2016
134
60
18,670
Indeed, Asus have the 1..2.0.6b out for my mobo! No sign of 1.2.0.5 anywhere. Glad to miss it.

I was actually going to check and see if there was a new bios for my ROG Crosshair VIII Formula... I'm glad I waited.

I'd be really really angry if my kit of DDR4 4000mhz CL14-14-14-34 turned back into it's original 3200mhz state...or worse.