Gigabyte Starts Distributing New Ryzen 5000 Firmware Updates for B550 and X570

Yeah there's been no major issues with F33 here, good to have the 1.2.0.1 AGESA code fixes as of these last updates but I'll probably wait on updating a bit, perhaps until F33 is out (Meaning no longer in beta.) or depending on how it goes if AGESA 1.2.0.2 and the fixes with that are also bundled in before the finalized bios update is out whichever Gigabyte decides on.
(Would be nice to have but it doesn't change too much I suppose whether it's F33H or later or F34A as such.)

 

I cba. With new bios right now. Sad you have to put all the timings again . Im getting stable 1933 fclk. It doesn't get much better. Can't be to greedy with OC or you have to spend weeks on testing and testing again .

 

https://www.tweaktownforum.com/foru...gigabyte-latest-beta-bios?p=978102#post978102

X570

F33 F/G/H

B550

F13 F/G

So a newer 1.2.0.1 AGESA (Patch A)
Ongoing or the same USB 2 fixes. (USB 3 I think was what 1.2.0.2 will resolve.)
And the usual assorted fixes, stability, compatibility updates.

Yeah I suppose this will be updated for a while yet, good stuff.
Getting there though but it reads like there's more to fix.

Or at least until the final update for the 1.2.0.2 AGESA bios and hopefully having most of the ongoing issues resolved at least from the bios and vendor UEFI code.


EDIT: Well the flash worked fine as usual, now to give it a few days and well I doubt anything will happen but it's still a beta bios.

Getting those USB fixes will be good, AGESA 1.2.0.1 and it's updates a nice little bonus and whatever Patch A actually entails.

I don't find the bios to be that sluggish though it's not instant either although I mostly set everything as I want then reboot wait a while (XMP and that.) then back to the bios check so everything is as expected and that nothing was overlooked.
Haven't had any other specific issues, hopefully AMD and Gigabyte plus the other board vendors can work out the remaining bios and UEFI problems now.

Then the rest will be the chipset drivers and add-in components like the Realtek NIC which the upcoming driver should resolve those disconnect issues for one example with the 2.5 GB on these boards alongside the Intel 1 GB chip.


Perhaps some of these things is also more for Vermeer or the integrated and mobile variants of the Zen3 CPU's too whereas Zen2 such as the 3900X I'm using is fairly tested and has most of the code and updates by this point.
(Short of what AMD resolves through the AGESA for the 500 series motherboard issues that is and then stuff like the PCI Express chipset driver covering both the bios and software bits.)

 

Flashed the MSI 1.2.0.1 PATCH A BIOS on my MSI B550 UNIFY

All of the power reporting and PPT + FLCK info is missing in HWiNFO 7.00-4400 , is this the same for everybody on 1.2.0.1 PATCH A ?

1.2.0.1





1.2.0.1 PATCH A

 

try latest hwinfo beta

 

Same Thing

HWiNFO 7.01-4420 BETA


The HWiNFO beta has a new type layout too , looks like the Telemetry the CPU usually sends is disabled in the 1.2.0.1 PATCH A BETA BIOS

 

I'll have a better look tomorrow, laid on the bedsheet ATM.

 

https://www.tweaktownforum.com/foru...abyte/28656-gigabyte-latest-beta-bios/page937

This fixes the memory multiplier issues.

Also new bios updates for the Z590 motherboards.

EDIT: Issue itself to my knowledge being that the available settings are duplicated, nothing critical but good to have it resolved.

 

I can boot and use 1933 and 2000fclk but can't boot 1900. Bug or working as intended?

 

I think it's "working" as such but it might not be stable, different values can be bootable but might throw errors or have instability issues if my understanding about the infinity fabric and clock speed is correct and how it affects the memory speed either at 1:1 or 1:2 linked.
Above 1800 it still seems to be very mixed just how effective or how stable these higher rates are but it also looks like the newer AGESA code has made it more robust but not entirely easy to maintain these levels even with high quality and high binned hardware.

I'm not that great on how this works though, easiest would be to check the event log for recurring memory errors or use something like GPU-Z or HWInfo I believe which can also show these errors indicating that it may be able to pass the boot process but the OS stability may not be 100% if these are recurring and depending on the amount of these errors getting logged.

EDIT: So it's not a bug or at least I would not see it as a bug it's just that some of these rates can be prone to errors still and might be until AMD has some newer hardware available and better support for higher infinite fabric rates and memory clock speeds whether that's something with Warhol or if it's going to take until AM5 and the next hardware cycle.

Just that it stresses the memory or controller but depending on how it operates it may still allow for booting into the OS itself though you would then have to check to ensure it's actually stable.


I'm sticking to 1800 myself it's been mostly OK for the later AMD AGESA code and this hardware though it's possible the Zen3 processors could be able to take it a bit further than my own 3900X on Zen2 can even if the I/O chip hasn't changed much.

I see AMD's statement in regards to 1900 to 2000 being achievable as something only a small set of the very best binned hardware can do reliably and it otherwise operates at a "up to" rule with their own maximum supported rate remaining at 3200 / 1600 in general and good RAM can clear up to 3600 / 1800 but not always without issue.


Still seems like a small set of hardware or memory kits are hitting issues even before XMP factors in so unless that's from faulty hardware or poor bios memory compatibility maybe AMD could improve and fine tune the compatibility further. Not too sure how the DDR4 memory training and compatibility factors into the AGESA but it could also be separate and up to the various motherboard manufacturers including tweaks or timings either for performance or stability which would affect results and maximum possible speeds before encountering errors.

 

What wonders me , even below 1800 , some clocks are bootable some are not. It is hard bootlooop at 1900, 1933 works,, 1966 hard bootloop, 2000 works , zero WHEA ( unless you are crazy curve optimizer settings).

 

Hmm wonder if it's some issue with the AGESA or if it's something the various motherboard partners are doing.

Looking at the changes and the response from the community overall memory compatibility has improved as has stability with high-speed kits and higher speeds for the infinity fabric but AMD hasn't listed too much with the AGESA itself it's mostly been various bug fixes like the ongoing USB issues.

Curious if some of the related settings for the various voltage values and also behavior like voltage drooping could be one explanation, some of these values go into certain timings and modes for what you would get from leaving these on automatic or default.
I'm not entirely sure though the one thing I am aware about is that Gigabyte tends to leave a fairly high target voltage for the SOC where it can go up to near 1.2v instead of 1.0 or 1.1v but it's usually operating at much lower if you monitor the readout for this.

Board itself can also handle it (Or that's what they claim at least.) but it makes me curious if there's other tweaks going on and what might be used by other vendors for attempting to stabilize higher memory clock speeds and the infinity fabric speed.


Interesting though, curious as to what allows it to stabilize at higher values when it will be unbootable at lower modes unless it is hitting some error when it rounds down to the nearest supported.


EDIT: Maybe I'm thinking about it too much from the perspective of it being the memory, perhaps it's more about the infinity fabric and related chipset settings and how the hardware handles that and not anything with the gear-down mode or memory timings and rates.

EDIT: Ah there is some syncing issues as well and potential problems around that.

From:
https://www.techpowerup.com/forums/threads/ryzen-3000-fclk-issue.264073/page-2

 

Man, GB got me flashing my board twice a week nowadays.

 

What i read about FCLK stablity :

IT is VERY true for me.

 

Oh that's interesting, looks like that also explains it a lot better than what I was looking for and why some combinations just won't work when others especially where using higher settings might run fine again, either right on the edge of this curve or completely in that zone then.

 

It was very funny for me as Intel user for many years i tried linear overclocking (small steps until you hit wall) Quite soon i realized something is wrong when my Zen3 couldn't get over 1800 fclk, any change above that(going linear steps) resulted in bootloop, then by maybe just an stupid idea, why not go higher, i tried 1933.. Boot, even more stable, i was like WTF?

Now i use 2000 fclk with pbo disabled. Best performance all-round with good thermals and wattage.

 

if only those profiles worked across versions!!

 

I've been using GB boards for around a decade now so bios is easily setup, but sure if only, especially those fan profiles.

 

Just a update on HWiNFO , BIOS 1.2.0.2 was still the same with the power reporting info missing.

New Version HWiNFO v7.02-4430 fixes the issue with all of the power reporting and PPT + FLCK info

It also adds a few new things , noted with the green bar.

Frequency Limit Global
DRAM Read Bandwidth
DRAM Write Bandwidth

 

Huh.
https://www.tweaktownforum.com/foru...gigabyte-latest-beta-bios?p=978422#post978422

So 1.2.0.1 with some fixes then 1.2.0.1 Patch A with additional fixes included some USB issues getting resolved.

Then 1.2.0.2 with additional fixes but now AMD recommends skipping it and instead 1.2.0.3 is going to be a thing later this month?

Well I wonder why but I guess they found more stuff to fix so it's more along that and just getting to the latest AGESA with all of these resolved issues and less of AMD broke or changed something in the earlier versions although they have tweaked a couple of things as well.