Intel, AMD, VIA & Freescale CPU Microcode Repositories Discussion

Thanks,

Looks all Ok to me…

-cpu306A9_plat12_ver00000020_2018-04-10_PRD_732E4C38
-cpu206A7_plat12_ver0000002E_2018-04-10_PRD_576F65DE

are inserted to my bios with UBU v1.69.17.4 and is ready to flash now.

Scroll down:

Capture1.JPG

New microcodes for older Xeons:
!New_cpu206D6_plat6D_ver0000061D_2018-05-08_PRD_441342A4.bin
!New_cpu206D7_plat6D_ver00000714_2018-05-08_PRD_5A3BF81F.bin

!New_cpu206D6_plat6D_ver0000061D_2018-05-08_PRD_441342A4.zip (18.2 KB)

!New_cpu206D7_plat6D_ver00000714_2018-05-08_PRD_5A3BF81F.zip (19.2 KB)

@plutomaniac

Having an issue with updating my cpu microcode to 306C3 plat32 ver 25. I was able to almost successfully update the microcode file in the bios and correct the FIT table such that after reflashing, rebooting, and then taking a new FPTW64 dump I verified the microcode file was updated from 19 to ver 25 BUT windows and bios setup utility still says ver 19 is loaded!

There is no more ver 19!!!

Maybe a cmos reset??

Thanks

Edit: Apparently with this board when you overclock the bios selects another microcode ver 19 presumably to allow overclocking on a locked processor I would imagine. This bios has two microcode revisions ver 25 which I was able to get working as long as I wasn’t overclocking the cpu, and ver 99 which must be ver 19 though in a hex editor did not match revision 19 at all. Funny thing after I reset cmos bios reported revision 7 for the microcode which turned into 25 after a reboot. So if I want to overclock I have to settle for a less secure less revision microcode version??

Ucode cpu306A9_plat12_ver00000020_2018-04-10_PRD_732E4C38. No oc.
Prime 95 test:
FATAL ERROR: Rounding was 0.4858398438, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Self-test 512K passed!
Self-test 512K passed!
Self-test 512K passed!

Linux* Processor Microcode Data File Version: 20180703

I am a little bit surprised.

From other forums, I know it, that newcomers are welcomed. It commands the courtesy to respond to questions from a person seeking advice as well as possible.

I have not been welcomed in this forum, nor has anybody answered my simple questions.

Either my posts have not arrived and no one can see them, or they are simply ignored.

Hereby I try on the second attempt to ask my questions:

1.) What do the 3 following microcode files have to say? Are they useful for anything?

cpu206A0_plat12_ver00000029_2009-11-02_PRD_D6930586.bin

cpu206A4_plat12_ver00000022_2010-04-14_PRD_8A6F7F91.bin

cpu20652_plat12_ver00000010_2018-02-04_PRD_AC3A178A.bin

2.) plutomaniac, the 2 microcode files you posted cpu306A9_plat12_ver00000020_2018-04-10_PRD_732E4C38 and cpu206A7_plat12_ver0000002E_2018-04-10_PRD_576F65DE obviously are not included in the new official Intel Processor Microcode Package for Linux. (why not …?!)

Where did they originally come from?

At least Question 2 should be easy to answer…

Thank you very much.

cpu906EC_plat22_ver00000096_2018-05-08_PRD_9EF46607

!New_cpu906EC_plat22_ver00000096_2018-05-08_PRD_9EF46607.rar (95.2 KB)

@Dekal

Welcome!

1)
Maybe simply read the documentation included in the project page?
https://github.com/platomav/CPUMicrocodes

2)
Because Intel did not include them

@Dekal ,

Also read this:

https://www.intel.com/content/dam/www/pu…te-guidance.pdf

As far as are whatever microcodes useful for anything it depends. If your overclocking certain microcodes work better than others. Which one it depends on doing your research reading accounts of others which can be subjective. However Ubutool will also lists the best one for overclocking fwiw.

I have an old CPU: i5-750 (Family-Model-Stepping: 06-1e-05). The current Intel microcode revision guidance (June 25th) notes a production version A for my CPU. Does anyone have any idea how I can get this microcode?

The latest version of the BIOS provided by the motherboard manufacturer (Gigabyte) is from 2010 and contains microcode version 7.

The latest Intel Linux Processor Microcode Data File (microcode-20180703.tgz) also contains version 7.

The latest version in the platomav/CPUMicrocode GitHub repository is version 9, which is mentioned in the previous Intel microcode revision guidance (April 3rd). I was able to download it (thanks, platomav!) and insert it in my BIOS image and flash that without problems. I’d like to do the same with version A of the microcode.

Can anyone point me in the right direction to find this microcode?


When we find it from an OEM or similar, it will be uploaded at the github repo which is updated weekly or here faster.

Thank you both for the friendly response.

I have already read through the documentation on the project page several times.

I already know the microcode update guidance.

My 2 questions remain unanswered.

plutomaniac, is not it allowed to know where the 2 microcode files originally came from?

I do not want to be pushy, I will not ask again.

Thank you for your work.

I bet someone pulled or ripped them from a Dell server board. I have noticed that Intel system building partners especially the big ones like Dell get their hands on them first. Could be someone in the industry who probably doesn’t want to take credit or say where or how.

Everything is a conspiracy theory to you guys. Intel does not release changelogs for microcode updates, at least not publicly. Rare exceptions like 2015 ACM or 2018 Spectre apply. For those who don’t ready carefully the GitHub Readme:

Capture.PNG

…?..?..…Thank you anyway.

New microcode in intel package: https://downloadcenter.intel.com/download/27945?v=t

!New_cpu50665_plat10_ver0E00000A_2018-04-20_PRD_902B7544.zip (18.2 KB)

Wish the repository also had older versions and not just the latest one.

cpu206C2_plat03_ver0000001F_2018-05-08_PRD_77DADA73 updated with 1KB padding required for insertion into bios with utilities.



older files can be accessed via the commit history.

cpu206C2_plat03_ver0000001F_2018-05-08_PRD_77DADA73_fixed.zip (11.2 KB)

@plutomaniac , Big thanks for you from your important work.

I wonder why Intel patched my Baytrail 326 µC in C0 Stepping to 836 lately but don’t list it anymore in the latest paper. The linux script shows it has no Spectre 4 capability but is vulnerable to Spectre 3a!