Intel 0x12B Microcode Update Gimped / Wrecked my RL Overclock

[stahlhart]

Intel i7-13700K
Asus Z690 ROG Strix Gaming-E Wifi
BIOS 3901, dated 09/27/24
P-Cores 56-56-55-55-54-54-53-53, E-Cores 42, Ring 45, +1 eTVB

So last night I decided to take the plunge and apply this supposedly final microcode / BIOS update to the i7 build, and I ran into some bad luck... I had a stable overclock going with x53 all core and single cores at x56 boosting to x57 with TVB, core voltages peaking at about 1.385V give or take a little, IA_VR limited to 1400mV, CEP disabled, AC_LL 0.15, DC_LL 1.05, PL1/2 253W. After updating to 0x12B and returning to the same overclock settings, I noticed that my P-core clocks were now topping at x55, even though I had x56 configured for the first two cores along with +1 TVB, core temperatures were a few degrees higher, and a y-cruncher routine that always used to work now failed with one of those "coefficient too large" errors.

I rolled back to 3802 (0x129), and that's where I'm staying. 0x125 and 0x129 both took the same overclock settings without issue.

 

[ShrimpBrime]

I thought you couldn't roll back firmware once updated. How did you get the firmware to roll back??

 

[phanbuey]

Interesing... had the opposite experience - I was undervolted @5.3 locked needed 1.3v for that - these new patches I can now run 5.4 at 1.258v, 5.5 at 1.3v.

I turned off TVB a while ago since in benching I had more stable FPS, and any game that pushes the CPU you just sit at all core clocks anyways. I think the new microcode just shuts off TVB and TVB voltage optimizations.

 

[stahlhart]

I thought you couldn't roll back firmware once updated. How did you get the firmware to roll back??

Asus has a BIOS flash back feature on this motherboard, and I am assuming others as well. You put the earlier version of the BIOS .CAP on a USB stick and place it in a specifically marked port in back, after you have powered down, and press a small button next to clear CMOS. An LED will flash for a few minutes, and when it goes out completely, you're done.

If I remember correctly, the EZ Flash tool in UEFI only updates to a newer BIOS version.

 

[ShrimpBrime]

Asus has a BIOS flash back feature on this motherboard, and I am assuming others as well. You put the earlier version of the BIOS .CAP on a USB stick and place it in a specifically marked port in back, after you have powered down, and press a small button next to clear CMOS. An LED will flash for a few minutes, and when it goes out completely, you're done.

If I remember correctly, the EZ Flash tool in UEFI only updates to a newer BIOS version.

This doesn't rollback ME firmware that I'm aware of.

Can you double check that? I don't mean micro code or bios version. I'm talking about the ME firmware.

 

[stahlhart]

Interesing... had the opposite experience - I was undervolted @5.3 locked needed 1.3v for that - these new patches I can now run 5.4 at 1.258v, 5.5 at 1.3v.

I turned off TVB a while ago since in benching I had more stable FPS, and any game that pushes the CPU you just sit at all core clocks anyways. I think the new microcode just shuts off TVB and TVB voltage optimizations.

That makes sense. The TVB probably isn't buying me all that much, apart from seeing it bump the P-cores up in HWiNFO. At some point I should probably do an A/B with it on and off to see if it's really adding anything measurable.

But it was troubling that I had x56 configured for the first two P-cores and it was downclocking one / two bins from that. I did not have that issue with either 0x125 or 0x129.

 

[chrcoluk]

Curious how you had TVB working on a chip that doesnt support it. It is a i9 only feature.

For reference a 100mhz bump from 5.3 to 5.4 is only 2% and thats only if its a 2 core or less workload.

 

[stahlhart]

This doesn't rollback ME firmware that I'm aware of.

Can you double check that? I don't mean micro code or bios version. I'm talking about the ME firmware.

OH -- I missed that, sorry for the Captain Obvious moment. You're right, I believe it retains the ME update. I just checked and it's 16.1.30.2307.

Curious how you had TVB working on a chip that doesnt support it. It is a i9 only feature.

For reference a 100mhz bump from 5.3 to 5.4 is only 2% and thats only if its a 2 core or less workload.

The BIOS is allowing me to set it, and if I check HWiNFO I'm hitting 5,700 when my highest P-core is set to 56.

 

[Dr. Dro]

This doesn't rollback ME firmware that I'm aware of.

Can you double check that? I don't mean micro code or bios version. I'm talking about the ME firmware.

It doesn't indeed. ME only updates, never rolls back. It is permanent.

That makes sense. The TVB probably isn't buying me all that much, apart from seeing it bump the P-cores up in HWiNFO. At some point I should probably do an A/B with it on and off to see if it's really adding anything measurable.

But it was troubling that I had x56 configured for the first two P-cores and it was downclocking one / two bins from that. I did not have that issue with either 0x125 or 0x129.

TVB is not supported on Core i7 SKU, nor does it work. Your processor does not support this, so it's completely irrelevant. The option appears, but it does not apply.

 

[ShrimpBrime]

OH -- I missed that, sorry for the Captain Obvious moment. You're right, I believe it retains the ME update. I just checked and it's 16.1.30.2307.



The BIOS is allowing me to set it, and if I check HWiNFO I'm hitting 5,700 when my highest P-core is set to 56.

thank you kindly.

I'm trying to decern if the ME is the culprit or if it's the micro-code or just how they wrote the bios.

 

[Dr. Dro]

thank you kindly.

I'm trying to decern if the ME is the culprit or if it's the micro-code or just how they wrote the bios.

It's likely the microcode. Intel's basically reining in the CPUs which were running wild with volts and current haphazardly all this time. Some performance regression is honestly to be expected on Core i9's, and people pushing their i7's will also feel a hit in achievable clocks unless they go out of their way to disable some or all safeguards.

I haven't had the time to flash the new bios on the Apex Encore yet, so I can't speak for myself - this week was rough here, but I'll keep it in mind and let you guys know if I observe anything different with the 13900KS

 

[phanbuey]

It doesn't indeed. ME only updates, never rolls back. It is permanent.



TVB is not supported on Core i7 SKU, nor does it work. Your processor does not support this, so it's completely irrelevant. The option appears, but it does not apply.

Even though it's not supported, turning on TVB voltage optimization on the mobo did change VID behavior, at least it did for me up until the last 2 bioses.

I havent played with it in a while though, need to check it out again.

 

[stahlhart]

TVB is not supported on Core i7 SKU, nor does it work. Your processor does not support this, so it's completely irrelevant. The option appears, but it does not apply.

Curious how you had TVB working on a chip that doesnt support it. It is a i9 only feature.

I appreciate the correction, thank you -- to be honest, I never checked the processor specifications to confirm this. But any suggestion as to why HWiNFO64 still shows a bin getting added to my maximums for P-core? That's what had me fooled here.

 

[ShrimpBrime]

It's likely the microcode. Intel's basically reining in the CPUs which were running wild with volts and current haphazardly all this time. Some performance regression is honestly to be expected on Core i9's, and people pushing their i7's will also feel a hit in achievable clocks unless they go out of their way to disable some or all safeguards.

I haven't had the time to flash the new bios on the Apex Encore yet, so I can't speak for myself - this week was rough here, but I'll keep it in mind and let you guys know if I observe anything different with the 13900KS

I have this new TUF board and am using the original release bios for 14th gen and I refuse to update the bios beyond that point. I don't see (while at stock) v-core spikes into the 1.50v range and up. But then again, I'm rarely running stock clocks. Anyhow, my reason is because the performance is just outstanding and I don't have a reason to update.

 

[Dr. Dro]

Even though it's not supported, turning on TVB voltage optimization on the mobo did change VID behavior, at least it did for me up until the last 2 bioses.

I havent played with it in a while though, need to check it out again.

Must be some motherboard-level tweak that still applies. Strictly speaking, TVB's exclusive to Core i9, what it does is basically once you've hit the maximum multiplier, it keeps going to an extra multiplier until a certain temperature is reached (thermal velocity).

All core "boost" on 13900KS for example is 5.4. It'll hit up to 5.8 in TBT3.0 (which is likely the tier you're seeing with the i7), but if thermal conditions allow, it'll then push further to target 6.0 - this is limited to the 2 "prime cores" that every CPU has. On Core i7, this last push doesn't occur, which means you'll only be seeing TBT3.0 clocks, even if TVB is enabled on the motherboard options

 

[chrcoluk]

I appreciate the correction, thank you -- to be honest, I never checked the processor specifications to confirm this. But any suggestion as to why HWiNFO64 still shows a bin getting added to my maximums for P-core? That's what had me fooled here.

There is still turbo boost 3 which up clocks the 2 fastest cores. Configurable in bios as well.

 

[ShrimpBrime]

Must be some motherboard-level tweak that still applies. Strictly speaking, TVB's exclusive to Core i9, what it does is basically once you've hit the maximum multiplier, it keeps going to an extra multiplier until a certain temperature is reached (thermal velocity).

All core "boost" on 13900KS for example is 5.4. It'll hit up to 5.8 in TBT3.0 (which is likely the tier you're seeing with the i7), but if thermal conditions allow, it'll then push further to target 6.0 - this is limited to the 2 "prime cores" that every CPU has. On Core i7, this last push doesn't occur, which means you'll only be seeing TBT3.0 clocks, even if TVB is enabled on the motherboard options

Could it be Asus Performance enhancement is enabled??

 

[stahlhart]

There is still turbo boost 3 which up clocks the 2 fastest cores. Configurable in bios as well.

I will look at the settings to confirm this, and will just disable it if it's irrelevant for the processor.

It's still troubling to me that 3901 (0x12B) pulled my P-cores back 200MHz, essentially overriding my prior settings, which were stable, then rendered them unstable on top of that. I'm back to stable now, so it doesn't appear that the ME firmware is at fault.

Just so you can see why I was confused

I just took Asus at their word, but apparently this must be something that isn't TVB being called TVB. It appears to add a bin to the P-cores based on some sort of temperature-based decision (I searched online for "Auto Optimized Temperature Control" with respect to both Asus and Intel, but came up empty). Asus Multicore Enhancement, which I leave disabled, has no effect on these settings.

So if an i9 is in this motherboard, these settings actually implement TVB because it is supported?

 

[Solaris17]

So if an i9 is in this motherboard, these settings actually implement TVB because it is supported?

If an I9 is in the system those settings will actually do something.

They are not greyed out in unsupported configurations. Atleast on this board. They enable it by default. Which to you the user looks like it’s doing something. But only i9s accept and do something with those bit masks.

 

[stahlhart]

Correct. So, what exactly is the difference between TVB and "Auto Optimized Temperature Control"? I can't find any explanation for or definition of what the latter is. But in both cases, it appears that the board will boost the P-core clock -- in the case of an i7, by one or two bins -- if temperature conditions are permitting. I know now that's not TVB, but why is Asus still calling it that?

Look at the HWiNFO statistics -- the P-cores are recording a maximum of 5,700. The highest P-core multiplier in BIOS is 56, which I have set for one or two cores. I have the +1 boost profile applied. If it isn't enabling anything for an i7, then why is HWiNFO adding 100MHz? Are the cores actually running at 5,600?

It's kind of splitting hairs at this point, but I'm just trying to explain why I misinterpreted this. I appreciate all of the feedback I am receiving from the forum.

 

[ShrimpBrime]

Correct. So, what exactly is the difference between TVB and "Auto Optimized Temperature Control"? I can't find any explanation for or definition of what the latter is. But in both cases, it appears that the board will boost the P-core clock -- in the case of an i7, by one or two bins -- if temperature conditions are permitting. I know now that's not TVB, but why is Asus still calling it that?

Look at the HWiNFO statistics -- the P-cores are recording a maximum of 5,700. The highest P-core multiplier in BIOS is 56, which I have set for one or two cores. I have the +1 boost profile applied. If it isn't enabling anything for an i7, then why is HWiNFO adding 100MHz? Are the cores actually running at 5,600?

It's kind of splitting hairs at this point, but I'm just trying to explain why I misinterpreted this. I appreciate all of the feedback I am receiving from the forum.

You enabled +1 TVB overclock profile. So it's boosting 5.7ghz as requested. I don't see anything wrong with it??

The test is in the middle of a stress test or all core benchmark what all core frequencies are at. 5.7ghz single/dual core boosts aren't what you are really after.

Setting 56-56-55-55-54-54-53-53 P-cores, during a full load, you will only run the max speed of the lowest set cores, 5.3ghz all core loads.

You'd be better off aiming for 5.4ghz all core instead. LLC lvl 4 honestly.

 

[eidairaman1]

Asus has a BIOS flash back feature on this motherboard, and I am assuming others as well. You put the earlier version of the BIOS .CAP on a USB stick and place it in a specifically marked port in back, after you have powered down, and press a small button next to clear CMOS. An LED will flash for a few minutes, and when it goes out completely, you're done.

If I remember correctly, the EZ Flash tool in UEFI only updates to a newer BIOS version.

Its known as Crashfree Bios 3 or something

 

[Shrek]

I rolled back to 3802 (0x129), and that's where I'm staying. 0x125 and 0x129 both took the same overclock settings without issue.

The microcode is to protect your CPU, sure you want to go back?

 

[:D:D]

They are not greyed out in unsupported configurations. Atleast on this board.

You'd think there would be a flag to do that, or at least I would. Maybe someone can check the mailbox for fun?

Anyone use an i7-14701E? According to the ark it has TVB but it seems to be a locked processor at 5.4GHz?

Edit: Seems likely an Intel error.

 

[ShrimpBrime]

Its known as Crashfree Bios 3 or something

Easy Flash 3 is built in bios utility. Crash Free Bios 3 is a bootable recovery method. I think it actually comes on the disk that comes with the board.