ASUS Releases Microcode Update to Address Intel 14th Gen and 13th Gen Stability Issues

[btarunr]

ASUS today became the first motherboard vendor to release UEFI firmware updates that encapsulate the latest 126 microcode update by Intel. This microcode update is Intel's response to the stability issues plaguing 14th Gen and 13th Gen Core desktop processors based on the "Raptor Lake" or "Raptor Lake Refresh" silicon, which was caused by an improper implementation of the on-die power management, which caused high voltages to run through the silicon, causing their physical degradation over time. If you are already experiencing stability issues, you should be able to claim an RMA or service under warranty, since your processor has already degraded. If, however, your processor is new, and is stable with all kinds of workloads, including games and compute-heavy productivity, then this UEFI firmware update is crucial in preventing its degradation, as it corrects the issue. Safedisk, a professional overclocker associated with ASUS, posted a list of ASUS ROG and ProArt motherboard models based on the Intel Z790 chipset, with links to their firmware update files. It stands to reason that the company is releasing these updates across its other product lines, such as TUF Gaming and Prime.

You can find the list here.

If you don't find your motherboard model in that list, keep checking the Support section of your motherboard's product page on the ASUS website, you'll either find it there, or one of these days the company will put it up. This aligns with Intel's timeline of a mid-August release for the microcode update.



View at TechPowerUp Main Site

 

[chrcoluk]

Ok its out in the wild then, awaiting info on how the voltages change.

 

[Dr. Dro]

Ok its out in the wild then, awaiting info on how the voltages change.

Already flashed it on my Apex Encore. So far so good. I'll let you know once I used it more, from a first glance, it seems to have a completely redone voltage curve. Volts def seem lower with my 13KS. I need to head out for uni now though, so no time to play around with it.

 

[Calmmo]

Already flashed it on my Apex Encore. So far so good. I'll let you know once I used it more, from a first glance, it seems to have a completely redone voltage curve. Volts def seem lower with my 13KS. I need to head out for uni now though, so no time to play around with it.

Hmm kinda on a wait and see mode here on the old apex, I had 0 problems till about 4months ago after updating bios.. not sure what changed back then but prior to that I could run 5.7 at like 1.32 - 1.34 and then suddenly that started all sorts of random crashes. My regular more extreme oc worked just fine tho.

 

[Dr. Dro]

Hmm kinda on a wait and see mode here on the old apex, I had 0 problems till about 4months ago after updating bios.. not sure what changed back then but prior to that I could run 5.7 at like 1.32 - 1.34 and then suddenly that started all sorts of random crashes. My regular (more extreme oc worked just fine tho)

New code is 0x129 up from 0x125 on the BIOS that's up on the public site right now. Saving the settings profile to USB and reloading it after flash worked, so it's really quick and painless. Not sure if it'll affect any clockability, Intel claimed that it wouldn't in their latest statement about it. Surprised to see the update released so early, I was expecting at least 10 days from now.

Anyway, I'll update if something weird is going on.

 

[piloponth]

Ok its out in the wild then, awaiting info on how the voltages change.

Awaiting re-reviews of the perf.

 

[docnorth]

For now only Z790, but my H770 probably needs a fix too, at least for power limits. PL1 is 253w, PL2 4096w, no option (until now) for adjustment or undervolting. My only possible actions to tame the beast are 90°C thermal limit and restriction of PL2 to 1 sec duration, so practically PL1=PL2. Of course these are needed only for full system scan or benchmarks, otherwise the CPU never exceeds 65°C.

I'm as well awaiting for feedback from Z790 users.

 

[Chaitanya]

For now only Z790, but my H770 probably needs a fix too, at least for power limits. PL1 is 253w, PL2 4096w, no option (until now) for adjustment or undervolting. My only possible actions to tame the beast are 90°C thermal limit and restriction of PL2 to 1 sec duration, so practically PL1=PL2. Of course these are needed only for full system scan or benchmarks, otherwise the CPU never exceeds 65°C.

I'm as well awaiting for feedback from Z790 users.

Also they are beta bioses.

 

[shk021051]

Already flashed it on my Apex Encore. So far so good. I'll let you know once I used it more, from a first glance, it seems to have a completely redone voltage curve. Volts def seem lower with my 13KS. I need to head out for uni now though, so no time to play around with it.

So is Asus profile safe to use again ? or must be on intel default ?

 

[Chrispy_]

I can't wait to see the new benchmarks; How much will the power and performance drop?

One of the negatives in all of the 9700X reviews are that they're hamstrung by the default 65W TDP but honestly that's a bit of an exaggeration since doubling the power draw only adds 5% more performance. IMO it's disingenous to call that "hamstrung".

IMO 65W TDP (~80W peak PPT measured) is a very reasonable default for that will satisfy almost everyone, almost all the time. If you want to chase benchmark records and spend your excess money on overkill cooling, then sure - you have the freedom to do that on anything except a select few A-series boards that don't support PBO+.

My point is that AMD seem to have sacrificed outright performance to sell CPUs that focus on performance/Watt again, and that pleases me. Hopefully Intel are about to (and will continue to) pursue the same thing.

 

[chrcoluk]

I can't wait to see the new benchmarks; How much will the power and performance drop?

One of the negatives in all of the 9700X reviews are that they're hamstrung by the default 65W TDP but honestly that's a bit of an exaggeration since doubling the power draw only adds 5% more performance. IMO it's disingenous to call that "hamstrung".

IMO 65W TDP (~80W peak PPT measured) is a very reasonable default for that will satisfy almost everyone, almost all the time. If you want to chase benchmark records and spend your excess money on overkill cooling, then sure - you have the freedom to do that on anything except a select few A-series boards that don't support PBO+.

My point is that AMD seem to have sacrificed outright performance to sell CPUs that focus on performance/Watt again, and that pleases me. Hopefully Intel are about to (and will continue to) pursue the same thing.

If its a voltage reduction across curve, may well be no loss of performance as it makes it less likely you hit power limit or temp throttle.

No bios for my board yet, so will check tonight if Dr Dro has any more updates.

 

[Dr. Dro]

So is Asus profile safe to use again ? or must be on intel default ?

I do not use ASUS MCE here, so I'm not sure. It seems to have an option to disable limits until the CPU hits 90°C, if you want to try, that seems wise to choose

I can't wait to see the new benchmarks; How much will the power and performance drop?

Little I played I did not notice a performance drop, but I will run a test later today when I am back home and post here

 

[nodata]

Flashed Asus Z790 Hero, core voltage 1,314 V.

 

[Dr. Dro]

There is no performance loss with identical settings. In fact, there is a minuscule increase, well within margin of error in P-core only CB R23. BenchMate validated. Identical BIOS settings.

BIOS 1402, microcode revision 125

BIOS 1503, microcode revision 129

I don't have a data point with all P- and E- enabled from the last BIOS, but I will run it and update this post

Here you go, official Intel extreme power delivery profile (320 W, 400A ICCMax limits). Standard cooling, not increased to 100% or anything (iPPCs sound like jets, spare my poor ears). Looks normal to me, this is around what an air-cooled, stock 13900KS/14900K at default Intel settings should score. Although I question, why is iTSC reporting as "Unreliable" since I re-enabled the E-cores... Eh, i'll figure it out later. I'm probably gonna run this CPU with hyper-threading disabled from here on out. 30°C reduction as far as I can tell, and sustains clocks better. It's really gonna make me buy one of those TEC coolers for it someday

Bonus: V/F points appear to remain identical

 

[Chrispy_]

If its a voltage reduction across curve, may well be no loss of performance as it makes it less likely you hit power limit or temp throttle.

No bios for my board yet, so will check tonight if Dr Dro has any more updates.

I think I get what you mean - No loss of performance if you're on a weak cooler and were throttling beforehand?
How many people who knowingly bought the ultra power-hungry 13th/14th gen are capping their performance with cheap coolers though?

 

[Dr. Dro]

I think I get what you mean - No loss of performance if you're on a weak cooler and were throttling beforehand?
How many people who knowingly bought the ultra power-hungry 13th/14th gen are capping their performance with cheap coolers though?

If you can even call a Noctua NH-D15S tricked out with two iPPC-3000s (running at 2000 RPM or so) and PTM7950 under it "cheap" - these i9s are heavy duty, man.

My cooling straight isn't adequate for these chips under full load.

 

[Super Firm Tofu]

There is no performance loss with identical settings. In fact, there is a minuscule increase, well within margin of error in P-core only CB R23. BenchMate validated. Identical BIOS settings.

BIOS 1402, microcode revision 125

View attachment 358001

BIOS 1503, microcode revision 129

View attachment 358000


I don't have a data point with all P- and E- enabled from the last BIOS, but I will run it and update this post

Here you go, official Intel extreme power delivery profile (320 W, 400A ICCMax limits). Standard cooling, not increased to 100% or anything (iPPCs sound like jets, spare my poor ears). Looks normal to me, this is around what an air-cooled, stock 13900KS/14900K at default Intel settings should score.

View attachment 358006

Bonus: V/F points appear to remain identical

View attachment 358007

Your memory controller voltage is a little higher than I'd be comfortable with @1.403

For 8200, i'm at 1.24.

 

[Dr. Dro]

Your memory controller voltage is a little higher than I'd be comfortable with @1.403

For 8200, i'm at 1.24.

Your kit is probably newer and better than mine, but I will take a good look at it. When I set this up I only set the DRAM voltage to 1.4 and SA to 1.2. Did not change other settings, I presume it just set MC to that value as well.

Will lower it to 1.25 and let it running Karhu while I have lunch thanks

 

[Super Firm Tofu]

Your kit is probably newer and better than mine, but I will take a good look at it. When I set this up I only set the DRAM voltage to 1.4 and SA to 1.2. Did not change other settings, I presume it just set MC to that value as well.

Will lower it to 1.25 and let it running Karhu while I have lunch thanks

Yeah, I think 1.4v was default for the MC after setting XMP.

 

[Dr. Dro]

Yeah, I think 1.4v was default for the MC after setting XMP.

I lowered it to 1.20 and the IVR MC voltage from 1.4 to 1.2 as well. I'll let it running for an hour or so while I have lunch. Don't think it'll error out though. It's already at 500% as we speak

 

[Chrispy_]

If you can even call a Noctua NH-D15S tricked out with two iPPC-3000s (running at 2000 RPM or so) and PTM7950 under it "cheap" - these i9s are heavy duty, man.

My cooling straight isn't adequate for these chips under full load.

Nothing cheap about a boosted NH-D15, but 13900KS can easily drink 400W+ if you let it, right? Even at 320W that's a lot to ask of anything using heatpipes.

 

[Dr. Dro]

Nothing cheap about a boosted NH-D15, but 13900KS can easily drink 400W+ if you let it, right? That's custom-loop territory for sure.

Yup. I appreciate those new Ryzen chips just a little bit more when I think of it. You'll get it all with a 9950X in a regular PC, not the case here

 

[Chrispy_]

Yup. I appreciate those new Ryzen chips just a little bit more when I think of it. You'll get it all with a 9950X in a regular PC, not the case here

The 9000-series dual-CCD models are what I'm looking forward to. The 170W (230W PPT requirements) of our typical 7900X workstations have been marred by the need for higher-end motherboards, which means "gamer" boards mostly that are both more expensive and full of pointless bling/gamer features that are irrelevant and unwelcome to a business workstation.

Being able to drop a 9900X or 9950X into an A620 board and run a modest air cooler is hugely appealing.

 

[LittleBro]

So? Any changes to VID table? Upper limit lowered?

 

[Dr. Dro]

So? Any changes to VID table? Upper limit lowered?

No changes for the VID table itself on my i9-13900KS. However, it looks like it doesn't juice the CPU as aggressively. Temperatures are a bit lower and benchmarks are consistently a bit higher (eTVB).