PSA: Don't overclock your Ryzen systems Bus the 99/100MHz one

[natr0n]

Note that the said system ran stable for months with bus at 101 till now.

Story...I have multiple disk drives for data/games. I noticed my system was randomly locking up and drives showed 100% usage in task manager.

Swapping multiple drives that i know work and event viewer would show retired blocks/sectors/scary stuff and such. I'm like oh shit are my cables bad now.

Then I said to myself just like the old days of my 939 systems you shouldn't use certain sata ports with an oced Bus.

I applied said knowledge and set bus speed back to auto and my drives are normal minus a corrupted empty folder.

So wanted to share that auto/98/99 bus speed is just fine and dandy now and be carefull to not overlook it.

 

[Deleted member 185158]

Note that the said system ran stable for months with bus at 101 till now.

Story...I have multiple disk drives for data/games. I noticed my system was randomly locking up and drives showed 100% usage in task manager.

Swapping multiple drives that i know work and event viewer would show retired blocks/sectors/scary stuff and such. I'm like oh shit are my cables bad now.

Then I said to myself just like the old days of my 939 systems you shouldn't use certain sata ports with an oced Bus.

I applied said knowledge and set bus speed back to auto and my drives are normal minus a corrupted empty folder.

So wanted to share that auto/98/99 bus speed is just fine and dandy now and be carefull to not overlook it.

A lot of the s939 systems with the VIA chipset where locked bus frequency between PCI and FSB. The NF2 boards didn't have the locked bus, so they overclocked via the bus frequency pretty well.

Aside from that, it's doubtful the 101 caused this issue. More so memory faults on the OC'ed (XMP is OC) where the secondary timings on auto may have underlining issues, the 101 bus may have then contributed to the errors.

A lot of this stuff goes hand in hand. If you tweak hard on your daily, then expect issues.


TLDR -
To get the Bus frequency to stay 100mhz, just disable "Spread Spectrum".

 

[tabascosauz]

Story...I have multiple disk drives for data/games. I noticed my system was randomly locking up and drives showed 100% usage in task manager.

Aside from that, it's doubtful the 101 caused this issue. More so memory faults on the OC'ed (XMP is OC) where the secondary timings on auto may have underlining issues, the 101 bus may have then contributed to the errors.

Definitely, memory much more likely. 3800-4000 is on the higher end of things and certainly not guaranteed to be stable, if this is the 4000 18-24-24 1.4V Vulcan kit I"m guessing probably DJR or Micron.

X570 is the only one that's extremely sensitive to FSB overclocking and should be strictly kept at 100, the other ASMedia chipsets (B450/B550) are perfect for base clock OC. Although, messing with base clock on a daily system is not a great idea.

As an aside, one of my SSDs started acting up recently with the same symptoms (100% usage, super slow even though benches look normal) even after cleaning the drive and moving it to an external enclosure. Probably not the same problem as all your drives are behaving that way, but best keep an eye on things to be safe.

But prolonged periods running even slightly unstable memory will certain rack up enough Windows corruption to do all sorts of wacky things. Make sure you test with HCI/TM5/Karhu.

 

[natr0n]

Definitely, memory much more likely. 3800-4000 is on the higher end of things and certainly not guaranteed to be stable, if this is the 4000 18-24-24 1.4V Vulcan kit I"m guessing probably DJR or Micron.

X570 is the only one that's extremely sensitive to FSB overclocking and should be strictly kept at 100, the other ASMedia chipsets (B450/B550) are perfect for base clock OC.

As an aside, one of my SSDs started acting up recently with the same symptoms (100% usage, super slow even though benches look normal) even after cleaning the drive and moving it to an external enclosure. Probably not the same problem as all your drives are behaving that way, but best keep an eye on things to be safe.

But prolonged periods running even slightly unstable memory will certain rack up enough Windows corruption to do all sorts of things.

This is my normal settings never had issues with these. When i go 3800-4000 things get wonky my imc prob is limiting I cant be bothered with it, so I toned it down.
I'm assured it was the 101 bus as thats the only recent thing messed with. Also no errors in windows logs currently. Drives working as normal too.

 

[Chomiq]

This is my normal settings never had issues with these. When i go 3800-4000 things get wonky my imc prob is limiting I cant be bothered with it, so I toned it down.
I'm assured it was the 101 bus as thats the only recent thing messed with. Also no errors in windows logs currently. Drives working as normal too.

You know what's the weirdest thing? That on my GB board hard reset from bad curve optimizer setting would result in bus speed being set to 103.4 MHz immediately after reboot. It would return to 100 afterwards but still, wtf?

 

[natr0n]

You know what's the weirdest thing? That on my GB board hard reset from bad curve optimizer setting would result in bus speed being set to 103.4 MHz immediately after reboot. It would return to 100 afterwards but still, wtf?

Yeah GB boards have weird quirks about them.

 

[tabascosauz]

This is my normal settings never had issues with these. When i go 3800-4000 things get wonky my imc prob is limiting I cant be bothered with it, so I toned it down.
I'm assured it was the 101 bus as thats the only recent thing messed with. Also no errors in windows logs currently. Drives working as normal too.

Event viewer won't show anything for memory instability, only physically/firmware bad CPU cores (cache hierarchy) and Fabric topping out (bus/interconnect). If you don't test, you don't know.

Unfortunately, memory has a very wide range of instability. It's all considered unstable, but only obscenely unstable setups will be plainly visible (no POST/no boot Windows/IRQL_not_less_or_equal/reboots/Windows dismantles itself).

The slight instabilities are no less capable of ruining your Windows/your data, they just take longer. Which is why it's good practice to run sfc /scannow every now and then - corruption comes one-off from many different sources, but if it keeps on coming back then there aren't many culprits other than memory.

 

[Shou Miko]

Have the fsb/bus speed over 100Mhz been a no-go since the start of using NVME storage since the overclock could have a negative effect?

That's what I heard from Buildzoid and read online.

 

[Zach_01]

NVMe drives have almost zero tolerance on base clock OC. Even 101 is not entirely stable.
Never really tested it without NVMe drive on system. Could be the system structure that doesn’t like base clock OC.

Pretty much base clock OC is dead nowadays. Since there is not any separation between base clock and PCI-E bus. Remember the PCI-E lock on 100MHz?

 

[Assimilator]

Like others have said, WTF are you doing overclocking FSB in 2022?

TLDR -
To get the Bus frequency to stay 100mhz, just disable "Spread Spectrum".

Spread spectrum doesn't cause this sort of instability because it constantly switches bus speed to arbitrary values. OP had their bus locked at 101MHz constantly.

 

[ThrashZone]

Hi,
Even on intel systems over 100 BCLK frequency can be really weird.

 

[Shou Miko]

Hi,
Even on intel systems over 100 BCLK frequency can be really weird.

I don't think I personally been BLCK clocking since 4th gen Intel using a K sku and their unlocked Pentium G3258 Intel made.

 

[ThrashZone]

I don't think I personally been BLCK clocking since 4th gen Intel using a K sku and their unlocked Pentium G3258 Intel made.

Hi,
It's a last ditch effort to get a little more memory frequency but yeah last I did was x99 and 100.3 was all she wrote

 

[natr0n]

Like others have said, WTF are you doing overclocking FSB in 2022?


Spread spectrum doesn't cause this sort of instability because it constantly switches bus speed to arbitrary values. OP had their bus locked at 101MHz constantly.

I had reached the limit of about everything or so I thought. my cpu boosts to 5250 so that 5252 looked nice what can I say I go overboard sometimes.

 

[Deleted member 185158]

Hi,
Even on intel systems over 100 BCLK frequency can be really weird.

Like others have said, WTF are you doing overclocking FSB in 2022?


Spread spectrum doesn't cause this sort of instability because it constantly switches bus speed to arbitrary values. OP had their bus locked at 101MHz constantly.

Disabling Spread spectrum to hold the bus frequency steady goes as far back as I can remember. It's one of the main things to disable for overclockers, so I'm not sure where you gathered that information.

WTF am I doing overclocking the Bus frequency in 2022??

I HAVE NO OTHER CHOICE!!

 

[Canned Noodles]

Similarly, I tried raising the bus on my Phenom II system by 10MHz to see what happens, and it broke the windows installation upon booting. I tried again but only by 1MHz and the same thing happened, but this time I was able to recover Windows instead of re-installing it.

Yeah GB boards have weird quirks about them.

and it's a GB board

 

[Deleted member 185158]

Similarly, I tried raising the bus on my Phenom II system by 10MHz to see what happens, and it broke the windows installation upon booting. I tried again but only by 1MHz and the same thing happened, but this time I was able to recover Windows instead of re-installing it.

and it's a GB board

Gotta be doing something wrong. Those easily did 300 bus clocks. On the same platform, that was the only way to OC Athlon II chips as well as the unlocker AMD AM2/+ Athlon 5000 (no plus) which unlocked to a Phenom FX processor.

Even my 8700K was a 400+mhz OC bus frequency. Only for show and tell, but totally doable.

 

[Aquinus]

Almost every type of I/O you can think of off a modern CPU is tied to the PCIe clock. If the bus speed for the CPU is tied to the PCIe clock, then any deviation from 100Mhz, give or take 1mhz, could cause issues. This is particularly true in the case of communication between the chipset and the CPU. Both AMD and Intel have their own branding for the name of the link, but it's essentially just PCIe. For that reason, anything on the chipset or PCIe could be impacted. However, for motherboard where the CPU bus speed and PCIe are not the same clock domain, then you tend to have far more flexibility. It really depends on the hardware you have and how it's configured. There are also cases like my 3930k where you can't go more than 3Mhz for whatever base clock strap you're using, like 100 or 125.