Question Asus B450-F Rog Strig Gaming ATX BOS update 3103 (Agesa ComboAm4 1.0.0.6 beta?) - FRIED CPU? (FIXED!)

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
Reset UEFI to defaults, seemingly flashed OK, but when I re-setup my 4.2Ghz 1.35V all-core OC, wouldn't boot, ever again.

I get a white debug LED, which is "VGA". I also SOMETIMES get four beeps after a clear CMOS. All LEDs (mobo RGB, RAM RGB, GTX 1660 ti RGB), AIO cooler fans and pump LED come on, but can't boot, nor power off with the power button.

Tried both my RX 5700 (used to be primary) and GTX 1660 ti 'Gaming X' (used to be secondary), both individually in each slot.

Also finally swapped out my Antec EDG Gold 750W modular PSU for it's BNIB backup twin. Still white LED, no boot.

Pretty sure it's the board.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
So, I did some more testing. Swapped out PSU with identical backup, no fix, still getting "VGA LED" (white debug Q-LED).

Pulled both RX 5700 and GTX 1660 ti, put in a refurb GTX 1650 that I had yet to test. Still white LED.

So I pulled the CPU (since it had to be CPU or mobo), and dropped in an older used Ryzen R5 1600 CPU.

Bada-boom, diag Q-LED cycled through the various colors like 3 times, then turned green, then I heard a "beep". Sonofa...

So, I guess my new Ryzen R5 3600 got fried.

Whether that was my settings, coincedence, or BIOS bug (the MOST likely reason, since it had been just peachy stable at 4.2Ghz @ 1.350V), I'm not 100% sure. But I'm leaning towards BIOS bug.

Wasn't there a thread on here or on CPU sub-forum, from someone claiming Asus BIOS ADDED TOGETHER VOLTAGE from two different BIOS pages, or something weird?

I left the primary voltage setting for Vcore on the main page on "Auto", and set the frequency and voltage using the separate "AMD Overclocking" settings page. I set that at 4200 and 1350, which is what I had in the prior BIOS version, that was stable and working.

I also set uncore to not clock down (there's a setting for that, uncore power saving disabled), disabled global c-states, and set under "integrated peripherals" to have the USB ports NOT power things when the PC is shut down (switch USB power source to +5V, rather than use +5V SB. I found that the USB3.0 ports were more stable with that setting.)

Maybe I shouldn't have changed all of those settings AT THE SAME TIME. Maybe that screwed up something. Who knows.

Or maybe the "AMD Overclock" page in BIOS is screwed up with the new AGESA ComboAM4 1.0.0.6. (Some mfg's are labeling that a "beta", although I don't believe that Asus did on my B450-F ROG STRIX mobo page.)
 

Rigg

Senior member
May 6, 2020
434
859
106
Sounds like a it could be a bios bug. Sometimes those advanced AMD settings won't clear when you clear CMOS. I'd flash the new bios again with the 1600 installed and see if you can get it to post again with the 3600. I've had weird stuff like this happen on a couple of different B450 boards and 3000 CPU's. If that fails I'd flash back to the bios you used previously and see if that posts with the 3600.

On a side note....I suspect your OC wows are related to that motherboard. In terms of current capacity that motherboard should handle a 3600 overclock without overheating issues but it's still has s pretty terrible VRM. I imagine efficiency, ripple, transient response, and v droop are god awful on that thing. Does it even have LLC? What is your actual v core voltage on SVI2 in HWI64 under heavy all core load with your OC settings? 4.2 Ghz at 1.35v (actual) is way worse than any of the 5 3600 and 2 3600x CPU''s I've owned. Those were almost all pretty early silicon too. 4 of 7 were launch CPU's. Only my 4.5 at 1.28v 3600 has been a 2020 CPU.

The heaviest AVX load i stability test for is real bench though. Maybe you are using a heavier AVX load which would explain the discrepancy.

That sucks hard if you nuked your CPU. Hopefully you can get an RMA sorted out if you did. I wish you good luck.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
Sounds like a it could be a bios bug. Sometimes those advanced AMD settings won't clear when you clear CMOS.
I did not know that.

I'd flash the new bios again with the 1600 installed and see if you can get it to post again with the 3600. I've had weird stuff like this happen on a couple of different B450 boards and 3000 CPU's. If that fails I'd flash back to the bios you used previously and see if that posts with the 3600.
I had similar thoughts. Also, I ordered an MSI (I think) A320M-A PRO MAX board (it was only $55 + $4 ship), I was going to see if the 3600 that I pulled, POSTed in that board.

On a side note....I suspect your OC wows are related to that motherboard. In terms of current capacity that motherboard should handle a 3600 overclock without overheating issues but it's still has s pretty terrible VRM. I imagine efficiency, ripple, transient response, and v droop are god awful on that thing. Does it even have LLC? What is your actual v core voltage on SVI2 in HWI64 under heavy all core load with your OC settings?
I knew that the VRMs weren't the greatest, but they're not the worst, either. They do have heatsinks on them, and I agree, they should be "enough" for a mild 3600 OC.

4.2 Ghz at 1.35v (actual) is way worse than any of the 5 3600 and 2 3600x CPU''s I've owned. Those were almost all pretty early silicon too. 4 of 7 were launch CPU's. Only my 4.5 at 1.28v 3600 has been a 2020 CPU.

The heaviest AVX load i stability test for is real bench though. Maybe you are using a heavier AVX load which would explain the discrepancy.
I could boot Windows at 1.25V, I could run WCG (mostly integer / FPU DC load) at 1.28125V, but for full "PrimeGrid stability" (ask @DrMrLordX , he tried it out too), I need 1.35V. (It's a HEAVILY-OPTIMIZED AVX2-HEAVY bunch of code. Pulls a lot more power than most things, especially if you're running 12 threads of it.)

That sucks hard if you nuked your CPU. Hopefully you can get an RMA sorted out if you did. I wish you good luck.
Like I said, I hope that it isn't actually fully-nuked, and maybe it will run on an A320 "MAX" board, just to test it.

The thing was, I was getting the white (VGA) LED lit, with any GPU installed. So that indicates that CPU's PCI-E x16 lanes were somehow screwed up. I did, after a few CMOS clears, get four beeps out of it. Which, AFAIK, is controlled by the CPU running the BIOS code, so that would mean that the CPU is actually somewhat functional, as far as the actual processing cores go.

I wonder which boards I could use, for a CPU with burned-out PCI-E x
 

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
Sounds like a it could be a bios bug. Sometimes those advanced AMD settings won't clear when you clear CMOS.
I did not know that.

I'd flash the new bios again with the 1600 installed and see if you can get it to post again with the 3600. I've had weird stuff like this happen on a couple of different B450 boards and 3000 CPU's. If that fails I'd flash back to the bios you used previously and see if that posts with the 3600.
I had similar thoughts. Also, I ordered an MSI (I think) A320M-A PRO MAX board (it was only $55 + $4 ship), I was going to see if the 3600 that I pulled, POSTed in that board.

On a side note....I suspect your OC wows are related to that motherboard. In terms of current capacity that motherboard should handle a 3600 overclock without overheating issues but it's still has s pretty terrible VRM. I imagine efficiency, ripple, transient response, and v droop are god awful on that thing. Does it even have LLC? What is your actual v core voltage on SVI2 in HWI64 under heavy all core load with your OC settings?
I knew that the VRMs weren't the greatest, but they're not the worst, either. They do have heatsinks on them, and I agree, they should be "enough" for a mild 3600 OC.

4.2 Ghz at 1.35v (actual) is way worse than any of the 5 3600 and 2 3600x CPU''s I've owned. Those were almost all pretty early silicon too. 4 of 7 were launch CPU's. Only my 4.5 at 1.28v 3600 has been a 2020 CPU.

The heaviest AVX load i stability test for is real bench though. Maybe you are using a heavier AVX load which would explain the discrepancy.
I could boot Windows at 1.25V, I could run WCG (mostly integer / FPU DC load) at 1.28125V, but for full "PrimeGrid stability" (ask @DrMrLordX , he tried it out too), I need 1.35V. (It's a HEAVILY-OPTIMIZED AVX2-HEAVY bunch of code. Pulls a lot more power than most things, especially if you're running 12 threads of it.)

That sucks hard if you nuked your CPU. Hopefully you can get an RMA sorted out if you did. I wish you good luck.
Like I said, I hope that it isn't actually fully-nuked, and maybe it will run on an A320 "MAX" board, just to test it.

The thing was, I was getting the white (VGA) LED lit, with any GPU installed. So that indicates that CPU's PCI-E x16 lanes were somehow screwed up. I did, after a few CMOS clears, get four beeps out of it. Which, AFAIK, is controlled by the CPU running the BIOS code, so that would mean that the CPU is actually somewhat functional, as far as the actual processing cores go.

I wonder which boards I could use, for a CPU with burned-out PCI-E x16 lanes, but working processor cores? If the chipset PCI-E lanes are still OK, then I could use a board that had a secondary PCI-E x16 (physical) slot, connected to the chipset PCI-E lanes, and plug the GPU into the secondary slot. Or possible, use a GPU riser, and use a chipset x1 slot, and mount the GPU vertically somehow with a brace.

Some possibilities, I guess.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
Why'd you flash the board?
Yeah, yeah, I know... it WAS working "perfectly"... but you know... "New and Shiny AGESA 1.0.0.6" drew me in. Was wondering about the new 4000-series desktop APUs, and Zen3 support. Still, no BIOS settings, short of explicit overvoltage, should theoretically "fry" my CPU.

Edit: The Asus BIOS for my board, is somewhat Schizophrenic, there's the normal BIOS page with the voltages and frequency (for both CPU and DRAM and SoC voltage), and then there are the pages under "AMD Overclocking", that's a separate menu system, and seemingly (in previous BIOS version), "overrode" (or at least, if they were left on "Auto" on the normal pages) the "normal" settings. Also, for some reason, with the prior BIOS, I couldn't get the OC stable with settings on the "Normal" pages (side-effects on those settings?), whereas, using the "AMD Overclocking" settings seemed to get it done.
 
  • Wow
Reactions: Magic Carpet

thesmokingman

Platinum Member
May 6, 2010
2,307
231
106
(the MOST likely reason, since it had been just peachy stable at 4.2Ghz @ 1.350V)

I have to say that that is way too much voltage for any Matisse cpu in regards to max fit voltage in my experience. Did you do the PBO test to confirm that voltage as your maximum safe fit voltage?

As for your mb, it looks like it is working considering the 1600 booted fine. Do you have another board to test the 3600 with?
 

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
Well, I suspected that my 3600 wasn't well and truely fried, that it might work on another board.

So, with both GPUs already installed, and my trusty 1600 installed, I reset UEFI to Optimized Defaults, plugged in a 32GB Team Group flash drive with the prior BIOS, flashed it, rebooted, saved and exited, it rebooted, rebooted, and back to failsafe OS install (on SATA SSD). Then I shut down, pulled the 1600, put the 3600 back in, finally managed to get the CM AIO LC kit pump attached (something was weird with it), and booted back up. It POSTed, which was a good sign, so I went into BIOS, cleared to defaults, rebooted, set XMP and FCLK, and Vsoc, rebooted, and then set NVMe RAID mode, and rebooted, and then selected the default boot drive, and rebooted, and back into Windows 10 2004! It's working!
 
  • Like
Reactions: Rigg

Iron Woode

Elite Member
Super Moderator
Oct 10, 1999
30,854
12,353
136
sounds a bit like you panicked rather than sat down and pondered the issue carefully. Modern boards can be wacky at times.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
sounds a bit like you panicked rather than sat down and pondered the issue carefully. Modern boards can be wacky at times.
I don't know about "panicked", more like, "fed up". I cleared CMOS half a dozen times, and that didn't return it to baseline (wouldn't POST, and only beeped once in a great while, four beeps). So I finally popped in a 1600, and it booted, so I left it like that for a while.
 

Iron Woode

Elite Member
Super Moderator
Oct 10, 1999
30,854
12,353
136
I don't know about "panicked", more like, "fed up". I cleared CMOS half a dozen times, and that didn't return it to baseline (wouldn't POST, and only beeped once in a great while, four beeps). So I finally popped in a 1600, and it booted, so I left it like that for a while.
I had a small issue that was similar with my Prime x570-P. I was checking out some BIOS settings and the keyboard slipped and I hit random buttons. Next thing I knew my 3600 was overclocked to 3.9 on all cores. I could hear my fan ramp up and checked temps in the BIOS showed my system hitting 50C.

So, I tried to undo the settings and saved and rebooted. It was still OC'd. I went into the load defaults and rebooted. Still didn't clear it out properly. So I did the manual method of clearing the BIOS. That fixed my issues. I then updated to the newest BIOS which has been working great ever since.

Don't worry, Larry. We have faith in your abilities.
 

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
Don't worry, Larry. We have faith in your abilities.
LOL. I wish I did!@

Edit: I think I do remember reading someone mention that there's some sort of back-and-forth between the BIOS/UEFI and the Ryzen CPUs (APUs too?), when you FIRST install one. Like maybe some sort of digital-signature stuff back and forth, like an SSL connection, between the CPU and platform security processor, and the UEFI and chipset. Honestly, that wouldn't surprise me. If somehow, the BIOS chip got a tiny bit corrupted, that might screw up that exchange, and you would have to actually SWAP CPUs, to re-set that data in the UEFI/CMOS.
 
  • Haha
Reactions: Iron Woode

Kenmitch

Diamond Member
Oct 10, 1999
8,505
2,248
136
You know what they say " If at 1st you don't succeed try, try again " at your own risk that is. Maybe just a bad flash? I guess it could be a bad uEFI version also. I've run into a couple that didn't even like my super duper b-die ram.
 

Magic Carpet

Diamond Member
Oct 2, 2011
3,477
231
106
Definitely, a bug of some sort, these things are very complex in nature. Usually, I don't touch newer hardware for some time, until most obvious bugs get fixed. Too much troubleshooting is never fun. Glad, now it's sorted, Larry.