AMD AGESA 1.0.0.4

[lawood]

My RAM is hynix cjr and never swapped ram, I do have 4 sticks and tested them all individually, the problems (usb dropouts) were present no matter the configuration.

I'm 100% certain the ram is stable, so if thats the case its very strange.

 

[dgianstefani]

I think it was the B-Die memory that kept everything together

I cant speak for USB problems, but a lot of the memory problems I see are from everything that isn't running Samsung B-Die

But that is probably just crazy talk on my part

It's not crazy. AMD is notoriously bad with memory support.

Even on AM5 AMD struggles above 6000 whereas Intel is pushing 7800 on two DIMM and 7200 on four.

 

[Dr. Dro]

I think it was the B-Die memory that kept everything together

I cant speak for USB problems, but a lot of the memory problems I see are from everything that isn't running Samsung B-Die

But that is probably just crazy talk on my part

I was running B-die too... but admittedly a very IMC-heavy configuration with 4 dual-rank sticks... the funny thing is that once the training problems were resolved, it was actually running the 64 GB at 3600 and a crazy low VSOC (0.98v), too. I like this memory kit, I was considering keeping it, but i'll wait and see. This week I will place an import order for a i9-13900KS, and once I have it (should be next month or so), i'll start looking for motherboard, DDR5, etc. - but I don't expect to finish my new build for some time.

It's.... ok. It appears the slow post issue is here to stay. The latest AGESA (the 1.0.0.4 patch to fix the 1.0.0.4 that stopped dual CCD 7600x's from booting) made my stable-since-launch CO unstable. I finally had enough a couple of weeks ago so no more AM5 for me.

I read that problem was being caused due to a bug expecting core 0 (so the first core in the first CCD) being enabled and available, and apparently some of these CPUs don't physically have it, and remapping is done by the AGESA itself... I really wonder if AMD is putting interns to work on such critical code

 

[ir_cow]

Even on AM5 AMD struggles above 6000 whereas Intel is pushing 7800 on two DIMM and 7200 on four.

I'm stuck at 8200 - 8400 kinda works for me for Z790 Apex lol. But yeah AMD has a history of funky memory support.

 

[tabascosauz]

It's.... ok. It appears the slow post issue is here to stay. The latest AGESA (the 1.0.0.4 patch to fix the 1.0.0.4 that stopped dual CCD 7600x's from booting) made my stable-since-launch CO unstable. I finally had enough a couple of weeks ago so no more AM5 for me.

That's sad to hear. I thought by 1004 there were reports that the POST time had been significantly reduced on ASRock?

One of my friends is considering replacing a 8700K with a 7800X3D. Like I told him, the chips are alluring but most without any AM4 experience have no idea what they're getting into. At best, it's learning curve; at worst, it's an unacceptable lack of certainty and stability.

But hey, I like the vote of confidence in the FC140 handling a 13900K!

 

[ir_cow]

That's sad to hear. I thought by 1004 there were reports that the POST time had been significantly reduced on ASRock?

Post times are like 15-40 seconds depending on the configuration now. Pretty much after the first big BIOS revision. "back in my day" we had dail-up and it took minutes to load a webpage.

 

[lawood]

the chips are alluring but most without any AM4 experience have no idea what they're getting into. At best, it's learning curve; at worst, it's an unacceptable lack of certainty and stability.

Yeah.
I was talking about USB because it was the most critical for me.

But there are a meriad of inconsistencies, on my msi motherboard any time I'd turn off PBO and turn it back on I would never boost past 4.0ghz (5600x). Even if I turned it off I would never boost past 4.0. I could push PBO limits, mess with the curve etc. NOTHING worked and the cpu would NEVER boost past 4.0.

The only thing that worked was FULL BIOS RESET and I couldn't turn off PBO or the same behavior would happen.
The only thing that worked was put pbo on advanced with my own limits and NEVER touch it again, if I at some point turned off pbo to test something the boost would never go past 4.0... and it wouldn't matter if I turned it back on, the pbo setting was just borked at that point.

This behavior was resolved on the latest agesa but it was present for months and months, started with 1.2.0.5 randomly.
Now imagine a not so savy person dealing with this...

And the EDC bug doesnt affect me because I'm on a 5600x but I can understand the frustration of some people.

 

[Deleted member 223428]

That's sad to hear. I thought by 1004 there were reports that the POST time had been significantly reduced on ASRock?

One of my friends is considering replacing a 8700K with a 7800X3D. Like I told him, the chips are alluring but most without any AM4 experience have no idea what they're getting into. At best, it's learning curve; at worst, it's an unacceptable lack of certainty and stability.

But hey, I like the vote of confidence in the FC140 handling a 13900K!

Launch day BIOS in my case had post time at 30 seconds (by stopwatch lol). An update or two later that was down to 27 seconds. I just had the typical 2 sticks of single-rank 16GB G.Skill 6000/30. Not the end of the world I know, but deal with some things that need two or three reboots in a row and you're about ready to throw it out the window. Forget about memory tuning, at least for me, with post times that long.

I suppose if you're the type of person who wants to just throw it together, set EXPO, and move on it would be ok. I could have rolled back to see if CO was stable again, but I just ran out of interest.

Lol. Yeah, that FC140 is holding its own. I power limited to 265w I think (?) and I haven't done any tuning yet other than a small negative offset with adaptive vcore. 82c isn't too bad. What gaming I've done is in the mid to high 50's.

 

[freeagent]

Yeah.
I was talking about USB because it was the most critical for me.

But there are a meriad of inconsistencies, on my msi motherboard any time I'd turn off PBO and turn it back on I would never boost past 4.0ghz (5600x). Even if I turned it off I would never boost past 4.0. I could push PBO limits, mess with the curve etc. NOTHING worked and the cpu would NEVER boost past 4.0.

The only thing that worked was FULL BIOS RESET and I couldn't turn off PBO or the same behavior would happen.
The only thing that worked was put pbo on advanced with my own limits and NEVER touch it again, if I at some point turned off pbo to test something the boost would never go past 4.0... and it wouldn't matter if I turned it back on, the pbo setting was just borked at that point.

This behavior was resolved on the latest agesa but it was present for months and months, started with 1.2.0.5 randomly.
Now imagine a not so savy person dealing with this...

And the EDC bug doesnt affect me because I'm on a 5600x but I can understand the frustration of some people.

That's just MSI being MSI.

Meanwhile over here on Team Asus it has been smooth sailing.

 

[lawood]

I will say that gigabyte has been a much better experience overall, the bios is horrendous, msi layout is much much better, but sfuff just seems to work.
Also much better to memory OC (for testing at least) because it was a gamble on MSI if the system booted or not with unstable settings. The gb board in a few seconds will put the ram back at jedec speeds and boot to windows.

 

[tabascosauz]

Launch day BIOS in my case had post time at 30 seconds (by stopwatch lol). An update or two later that was down to 27 seconds. I just had the typical 2 sticks of single-rank 16GB G.Skill 6000/30. Not the end of the world I know, but deal with some things that need two or three reboots in a row and you're about ready to throw it out the window. Forget about memory tuning, at least for me, with post times that long.

I suppose if you're the type of person who wants to just throw it together, set EXPO, and move on it would be ok. I could have rolled back to see if CO was stable again, but I just ran out of interest.

Lol. Yeah, that FC140 is holding its own. I power limited to 265w I think (?) and I haven't done any tuning yet other than a small negative offset with adaptive vcore. 82c isn't too bad. What gaming I've done is in the mid to high 50's.

Really tempting......pricing out a 13700K + Aorus Ultra ITX right now but can't decide if it's worth keeping my 3600/14 B-die and getting the DDR4 board, or splurging for a 6000CL30 Hynix with DDR5 board.

I think no CO would not be acceptable for me, considering it is a huge part of the temp reductions on Raphael. Back at release I think optimumtech did a -30 at 90W on a 7700X and achieved a 30 degree drop with iso-performance in R23.

I will say that gigabyte has been a much better experience overall, the bios is horrendous, msi layout is much much better, but sfuff just seems to work.
Also much better to memory OC (for testing at least) because it was a gamble on MSI if the system booted or not with unstable settings. The gb board in a few seconds will put the ram back at jedec speeds and boot to windows.

Yes, that is also my experience on 5600G and 5700G with iGPU and B-die OC. GB BIOS is one of those things you have to get used to, but once you do you rather quickly find that Gigabyte's hard product is pretty good. Both Asus Strix ITX and MSI ITX did not make the cut in the end, even after a nightmarish RMA experience with Gigabyte.

 

[Kapone33]

And that's great for you.

But if you step out of that bubble you'd quickly see am4 is a mess.

I also never got problems that agesa updates and downgrades didn't fix but I know I'm just lucky.
People were turned off AMD forever with am4 and some of the problems never got fixed, usb's dropouts on the chipset being one. People will try to say that is solved, try to search for that by date.
I've owned b350, b450 and now b550. Also owned 2600, 3600 and now 5600x.
I now a bit about this platform. Love it, but I know its a mess.

edit:
Went on a bit of a search:

5 months ago:
Was the USB dropout bug ever fixed? : Amd (reddit.com)

Half the people with problems, half claiming they don't.

This one is also enlightening, last year:

Spent the weekend troubleshooting and fixing the notoriously hard to replicate AM4 USB disconnectivity issue. It was hard to find info about and I hope this helps others. : Amd (reddit.com)

____________

I did have USB problems on b450, that I fixed with agesa updates so I can understand the frustration of some people.
Because this isnt hardware, its shitty am4 agesas and they never fixed it for everyone. I considerer myself lucky.

I have built over 50 AM4 PCs and not one of them other than Gigabyte have given me issues. The worst was budget X370 boards but I have used every generation with no stability issues since. I have never had that USB issue. Maybe it's people plugging the case USB thinking that it would work as well without any idea of the quality of the cable the case vendor used. I also update to the latest BIOS before I sell a PC to a customer. Some MBs are also flaky by design so I only but Tuf and above from Asus but am happy to get a budget MSI board for any 65 Watt CPU.

Using reddit is not really that good because I could also pull plenty of posts from TPU that show that a small percentage is effected by that and when boards are this inexpensive for AM4 it is good as you can buy a replacement no problem. It is why the 5800X3D is still the number 1 selling CPU in Germany and other places.

It's not crazy. AMD is notoriously bad with memory support.

Even on AM5 AMD struggles above 6000 whereas Intel is pushing 7800 on two DIMM and 7200 on four.

I bet most of those users used Corsair Ram.

 

[erocker]

Yeah.
I was talking about USB because it was the most critical for me.

But there are a meriad of inconsistencies, on my msi motherboard any time I'd turn off PBO and turn it back on I would never boost past 4.0ghz (5600x). Even if I turned it off I would never boost past 4.0. I could push PBO limits, mess with the curve etc. NOTHING worked and the cpu would NEVER boost past 4.0.

The only thing that worked was FULL BIOS RESET and I couldn't turn off PBO or the same behavior would happen.
The only thing that worked was put pbo on advanced with my own limits and NEVER touch it again, if I at some point turned off pbo to test something the boost would never go past 4.0... and it wouldn't matter if I turned it back on, the pbo setting was just borked at that point.

This behavior was resolved on the latest agesa but it was present for months and months, started with 1.2.0.5 randomly.
Now imagine a not so savy person dealing with this...

And the EDC bug doesnt affect me because I'm on a 5600x but I can understand the frustration of some people.

Since the latest AGESA for my x570s Carbon Wifi board, EDC bug has been fixed. Before that, I got around it by using the overclocking options for MSI instead of under AMD in the bios.

 

[Deleted member 223428]

Really tempting......pricing out a 13700K + Aorus Ultra ITX right now but can't decide if it's worth keeping my 3600/14 B-die and getting the DDR4 board, or splurging for a 6000CL30 Hynix with DDR5 board.

I think no CO would not be acceptable for me, considering it is a huge part of the temp reductions on Raphael. Back at release I think optimumtech did a -30 at 90W on a 7700X and achieved a 30 degree drop with iso-performance in R23.

I spent about a week reading reviews/forums to try to decide DDR4 vs DDR5. Without a definitive answer I yolo'd it and bought a set of 6400CL32 that appears to be Hynix M for $185. Anything faster was a big jump in price or had RGB.

I did still have CO, but what was stable for months no longer was. I dropped 5 or so on each core and appeared to be stable again, but I never did corecycler overnight again. Clock speeds lowered, benchmark scores dropped, and temps went up a degree or two. Not the end of the world but I didn't really like that I was getting less performance as time went on.

 

[ir_cow]

I bet most of those users used Corsair Ram.

Not a fan of Corsair eh?

 

[Veseleil]

it was a gamble on MSI if the system booted or not with unstable settings

True, I had to pull out one of the sticks couple of times just to be able to boot into BIOS! Clearing CMOS or other tricks didn't work, the board get stuck on the RAM led, and that's it. MSI BIOS might be designed better, but it's stupid af.

 

[mama]

Post times are like 15-40 seconds depending on the configuration now. Pretty much after the first big BIOS revision. "back in my day" we had dail-up and it took minutes to load a webpage.

I'm at 10 seconds to post on my Gigabyte board. I have noticed weird instability in programs like Gigabyte Contol Center (and it's components) and Cooler Master's MasterPlus software but that resolved when I removed them (which wasn't easy by itself).

True, I had to pull out one of the sticks couple of times just to be able to boot into BIOS! Clearing CMOS or other tricks didn't work, the board get stuck on the RAM led, and that's it. MSI BIOS might be designed better, but it's stupid af.

Yep, happened to me too.

 

[lawood]

True, I had to pull out one of the sticks couple of times just to be able to boot into BIOS! Clearing CMOS or other tricks didn't work, the board get stuck on the RAM led, and that's it. MSI BIOS might be designed better, but it's stupid af.

I just ordered a 5800x3D for it, hope it fits nicely. The 5600x served me well though, great little chip.

Does anyone know if pbo options are available on gigabyte motherboards?
I know that AMD blocks that but some vendors make that available on the motherboard.
I can always use pbo tuner but meh, i dont like random software always running on the background.

 

[BMfan80]

I just ordered a 5800x3D for it, hope it fits nicely. The 5600x served me well though, great little chip.

Does anyone know if pbo options are available on gigabyte motherboards?
I know that AMD blocks that but some vendors make that available on the motherboard.
I can always use pbo tuner but meh, i dont like random software always running on the background.

You can set TDC, EDC and PPT in the bios but, you would need PBO tuner for curve optimiser.
I have a Gigabyte B550 Pro V2 with F15a bios.

 

[lawood]

That's unfortunate, i was hoping to tune the curve in bios.
But I guess I should be happy to able to use pbo tuner at least, will need to create a task for it.
I don't think ill be messing with tdc, edc and ppt limits, kinda scared to burn all that cache. Especially because AMD seemed all worried, blocking eveyrthing, even the CO.

Tomorrow I should get it.

 

[BMfan80]

I set 105w PPT, 70A TDC and 115 EDC.
When I started up for the 1st time and ran Cinebench, the temp shot up to 90.
It went over 140w and I can't remember what my EDC did but, it was around 130a.

It then stayed at an all core of 4450 during C20.

 

[nomadka670]

Every manufacture released 1.0.0.4 for am5 expect Asus-,-

 

[JohH]

Every manufacture released 1.0.0.4 for am5 expect Asus-,-

You're not alone. ASRock still hasn't released 1.0.0.4 on their website either