[SOLVED] To Update BIOS, or No?

Mar 2, 2021
3
0
10
Troubleshooting BIOS beep codes and attempting an RMA.

  • Processor: Ryzen 5 2600
  • Motherboard: ASRock B450M-HDV R4.0
  • BIOS: R4.0 P3.20
Storyline
  1. System Build: completely stock. No overclocking. Been working fine for just over a year with no concerns.
  2. L2 Cache Faulty: Computer still boots and runs, but gives me 11 steady beeps (Faulty L2 Cache)
  3. AMD RMA shenanigans: contacted AMD to RMA my processor. (Literally 2 weeks outside the merchant RMA period!) They wanted me to swap out the processor with another one to test. I don't have a spare processor to swap in.
  4. AMD instructs to update BIOS: still giving me the run around on RMA
  5. Motherboard BIOS recommends against updating: ASRock's site says that for Pinnacle processors (Like my Ryzen 5 2600), it recommends against BIOS update.
So:
1.) Should I update the BIOS, anyways?
2.) Any other recommendations for this particular problem to make sure I get my due on replacing this processor?
 
Solution
5. Partially accurate, but not fully.

ASRock, for most of their products, tend to run on the cheaper side. Still pretty good, just cheaper. This meant using the smaller sized bios chip in most of their lineup.

The problems with bios arose with the changes to the zen/zen+/Zen2 architectures, all those different cpus per bios. And the additional requirements of all the ram data tables, Agesa etc.

From the factory, it gets an implanted full bios, everything you do is an update, not a full clean wipe and reinstall. So the updates must also include all the instructions as well as the changes. But there's not much room to add much more on the smaller bios chips.

So.

ASRock split it up. It used 1 change per code, so only 1 thing gets...

Lutfij

Titan
Moderator
Welcome to the forums, newcomer!

Have you taken your current processor over to a donor board to test out that the processor is the one that's at fault? It could be that the board might be faulty and is not recognizing the processor.

Might want to share the rest of your specs like so:
CPU:
Motherboard:
Ram:
SSD.HDD:
GPU:
PSU:
Chassis:

Breadboard the system and try one stick of ram only.
 
  • Like
Reactions: hardwarejunkie9
Mar 2, 2021
3
0
10
Have you taken your current processor over to a donor board to test out that the processor is the one that's at fault? It could be that the board might be faulty and is not recognizing the processor.

Breadboard the system and try one stick of ram only.
Thanks for the warm welcome, Lutfij.

For more details:
  • CPU: Ryzen 5 2600
  • Motherboard: ASRock B450M-HDV R4.0
  • BIOS: UEFI R4.0 P3.20
  • RAM: 16 GB 2133MHz, DIMM
  • AMD StoreMI Hybrid virtual drive:
    • SDD: WD BLue 250 GB
    • HDD: Bog standard SeaGate 1 TB
  • GPU: GeForce GTX 750 Ti
  • PSU: 530 W Rosewill Eco
  • Chassis: Some generic full tower deal off of Newegg Circa 2008.
Not sure where I'd find a donor board, if you've any advice on being able to run one down. (Having a bit of fun being on the literal other side of the world for a couple years, TX -> AUS, so I'm a bit out of my community)
Looking up how to actually breadboard on here, but not 100% confident that I'm less likely to cause more damage trying based on the advice here by shorting pins.

Worth noting that the system boots and runs. Only real issue is the BIOS beeps and it possibly running a bit slower.
 
Troubleshooting BIOS beep codes and attempting an RMA.
...

Tell them that you're on the latest full-support BIOS for Pinnacle Ridge CPU's and the motherboard mfr says do not update beyond that.

Also, if your CPU is faulty attempting to update BIOS could easily brick your motherboard so I wouldn't update under any circumstance unless with another known good CPU. Respond with that and ask them if they'll cover replacement of the motherboard if it does.

Have you reset CMOS? Do so before doing anything else and run your tests with all settings in full-default, no changes.

You're not overclocking now..but when it was 'running fine' were you overclocking? Had you ever experimented with overclocking? How high, what voltages?

How old is your PSU? Do you have another to swap it out? I don't know much about PSU's but being 'Rosewill' it sounds like one bundled with that Newegg case. That makes it a candidate for being a serious POS you should change ASAP anyway and would not be surprised if it fixes your problem.

You don't have a CPU to test the mobo, but do you have another motherboard to test the CPU?

Definitely reduce the variables: one stick of memory and swap them around in different DIMM sockets.

Get a program called MemTest and run it. You actually should run like 12 copies, one for each CPU thread. You tell each copy how many MB of memory to use so be sure to leave enough for the OS to run (about 3 GB) or it will start testing virtual memory. Testing memory like this is also testing the cache; if any of the caches are faulty it should error out pretty quickly but let it run at least 200% for each copy/thread.

BTW: that motherboard does not look good for overclocking even a 2600, 6 core/12 thread cpu. No heatsinking on the one FET per CPU phase I can see; I hope the other one is on bottom-side. While not many AM4 CPU's really need it, the lack of an 8 pin CPU power connector (it's only 4 pins) shows you even Asrock doesn't think it capable of much.

I'm not sure it's wise completely disassembling the system to bench test (breadboard) it. That's really only for systems that won't boot, especially 'newly built' ones that never booted. Even if you did I'd not remove the CPU cooler, and definitely not the CPU, when doing so.
 
Last edited:
Mar 2, 2021
3
0
10
Tell them that you're on the latest full-support BIOS for Pinnacle Ridge CPU's and the motherboard mfr says do not update beyond that.

Also, if your CPU is faulty attempting to update BIOS could easily brick your motherboard so I wouldn't update under any circumstance unless with another known good CPU. Respond with that and ask them if they'll cover replacement of the motherboard if it does.

Have you reset CMOS? Do so before doing anything else and run your tests with all settings in full-default, no changes.

You're not overclocking now..but when it was 'running fine' were you overclocking? Had you ever experimented with overclocking? How high, what voltages?

How old is your PSU? Do you have another to swap it out? I don't know much about PSU's but being 'Rosewill' it sounds like one bundled with that Newegg case. That makes it a candidate for being a serious POS you should change ASAP anyway and would not be surprised if it fixes your problem.

You don't have a CPU to test the mobo, but do you have another motherboard to test the CPU?

Definitely reduce the variables: one stick of memory and swap them around in different DIMM sockets.

Get a program called MemTest and run it. You actually should run like 12 copies, one for each CPU thread. You tell each copy how many MB of memory to use so be sure to leave enough for the OS to run (about 3 GB) or it will start testing virtual memory. Testing memory like this is also testing the cache; if any of the caches are faulty it should error out pretty quickly but let it run at least 200% for each copy/thread.

BTW: that motherboard does not look good for overclocking even a 2600, 6 core/12 thread cpu. No heatsinking on the one FET per CPU phase I can see; I hope the other one is on bottom-side. While not many AM4 CPU's really need it, the lack of an 8 pin CPU power connector (it's only 4 pins) shows you even Asrock doesn't think it capable of much.

I'm not sure it's wise completely disassembling the system to bench test (breadboard) it. That's really only for systems that won't boot, especially 'newly built' ones that never booted. Even if you did I'd not remove the CPU cooler, and definitely not the CPU, when doing so.

Thanks, Drea, You make a lot of good points, and thanks for clarifying a few things. (I had a gut feeling that rushing to BIOS update was too hazardous for the problem).
I'll be sure to try the following and update after:
  1. reset CMOS
  2. Memtest and
  3. look at swapping that power supply out. It's been well-behaved for years, but it's definitely old and a risky part, now; PSU's aren't spendy any ways.
  4. Test swapping my one 16 GB stick around to the other slot.
As for other questions:
  • PSU's probably 6-7 years old. I swapped out the one that came in the original Newegg special, but I can't find the order details for exactly when. It was 2 upgrades ago.
  • Overclocking: not even once. Didn't play with it. Didn't push any limits. Pure stock from start to finish with the stock cooler. Wanted this thing to be reliable budget workhorse. It's been running completely in the clear, full stock, for over a year.
  • No, I don't have a spare mobo. Previous one was an old 2nd gen i5. Been asking around at work to see if anyone has a spare I could borrow.
 
....
  • PSU's probably 6-7 years old. I swapped out the one that came in the original Newegg special, but I can't find the order details for exactly when. It was 2 upgrades ago.
...
Good strategy...

PSU's should be more 'spendy' than you might be thinking. By that, I mean you can get them really cheap but then they're throw-away. Or worse - when a bad one 'poofs' it can take a lot of other hardware along with it. While a good PSU can be considered an investment because it can last through a several upgrade cycles.

But also...if the Rosewill you're using now isn't the one that came with the case it might not be the stinky pile that suggests. Rosewill does make some pretty good ones...I think in their Capstone line for instance. Check some of the PSU tier lists to see if it's one of the good ones.
 

Karadjgne

Titan
Ambassador
5. Partially accurate, but not fully.

ASRock, for most of their products, tend to run on the cheaper side. Still pretty good, just cheaper. This meant using the smaller sized bios chip in most of their lineup.

The problems with bios arose with the changes to the zen/zen+/Zen2 architectures, all those different cpus per bios. And the additional requirements of all the ram data tables, Agesa etc.

From the factory, it gets an implanted full bios, everything you do is an update, not a full clean wipe and reinstall. So the updates must also include all the instructions as well as the changes. But there's not much room to add much more on the smaller bios chips.

So.

ASRock split it up. It used 1 change per code, so only 1 thing gets fixed. If the original code was ABC, the newer update will have just the instruction and data to change ABC to XYZ. A later bios update will change XYZ to 123. But thats the kicker. If you skip the first update, nothing gets fixed because there's no instruction to change ABC to 123 because that's an additional instruction to XYZ to 123, which adds too many instructions to fit in the small chip.

So ASRock updates must be done in stages, read the notes and warnings starting from your current bios to newest.

Partially accurate because the size of the bios chips dictates which info is necessary. If using 1000/2000 cpus, you don't want or need all the added stuff for 3000 or 5000 cpus, or they will erase some of the older code to make room for newer. So ASRock draws a line in the sand and tells you to Stop Here! so you don't loose bios integrity. You only should update in stages not simply skip to newest and stop before adding newest which has no relevance to the cpu you are using.

So if you have a 14nm 1600 on an original release cheap B350, buy a 3600 and update the bios to incorporate 3000 series cpus, chances of getting that 1600 to work again later are slim without a total bios reversion. Which won't happen without a 2600 being used as a go-between cpu.

If your bios Is the latest for your motherboard and you've reached the Stop Here!, then yes, you are done can go no further because anything newer isn't fixing your bios, it's just adding 3000 series cpu data, adding zen+ ram compatibility tables etc.

If you think that maybe the bios is corrupted somehow, a reinstall of the current version might fix that, but I'd not update to a newer version unless it's specifically for your cpu series, and not for the 3000+ series.
 
Last edited:
Solution