@ganon Yes, it is the updated uefi. I don't have enough time to change the uefi properties yet. But when I discover what happened to mandrake, I'll adjust that and, maybe, update the RAID/SATA/VBios drivers.

Hello all.

I registered here because I have an ASRock Z77 Fatal1ty Professional board with an i5 3570k. There has been no new BIOS released to fix the spectre exploit.

Posting here in the hopes that someone can update the latest P1.70 BIOS with the new microcode to fix spectre on the 3570k.

I would be most grateful if anyone can help. The ASRock BIOS can be found here: http://www.asrock.com/mb/Intel/Fatal1ty%...t=Download#BIOS

Just so you all know, ASrock officially released BIOS 3.0 Beta today: https://www.asrock.com/MB/Intel/Z77%20Ex.../index.asp#BIOS and it has the updated microcode. So if jcorrea's MOD BIOS scares you, this one is plain vanilla. =)

Awesome, thanks for the heads up mate.

Is there anyway to know if this has all the latest SATA/RAID/VBios modules?

No idea.

@jcorrea: Thank you for your efforts :-)

Have you had any luck with finding the reason for updated drivers to cause the problems? Do the updated drivers add any real value?

If I understand correctly, this is the safest version to use: Specter and Meltdown for ASRock Z77 Extreme 4 (2) ?

I've read a lot about how the Meltdown and Spectre fixes cause a lot of performance loss on older systems. Do you know if this is caused by the new microcodes themselves or by the OS part of the fix?

Best regards,
Crocodil


EDIT: I can see that downloads ad MDCLAB are again available. Maybe this will help you with debugging? Also - has anyone tested the BIOS for Extreme4? I would like to try it but I'm affraid to brake my motherboard...

@Crocodil
Sorry about new upgrades on UEFI. I did not discover what's wrong on that.
About upgrading to UEFI on mdclab, if you are unsure or even, if you are afraid to use that, first of all you have to find a way to recover your MB from a possible brick. A good, cheap and safe way is get a ch341a programmer, get a W25Q64BVAIG memory, flash it with a ROM, and replace your original one. If it don't work, you only have to put your original back and try again.
I spent 3 or 4 bucks on a flash programmer, and 2 on a W25Q64BVAIG.

@jcorrea Thank you very much for the explanation. I've never programmed a flash memory, but I've already ordered the needed parts and I will give it a try :-)

Am I correct that this is also the safest way to try your modded BIOS, the one that has caused problems?

Regards,
Crocodil

EDIT:
@jcorrea Are you sure that the BIOS chip is 25Q64BVAIG? I know very little to nothing about the stuff but I was able to read the chip on the motherboard and it reads: 25Q64FVAIQ 1236 (FVAIQ instead of BVAIG)

@Crocodil
Get the mdclab UEFI, cause it has the newest drivers and modules.

About the chip, don't worry, they are compatible, both BVAIG or FVAIG will work.

@jcorrea Thank you very much for the confirmation - I've already ordered the CH341A programmer and a bunch of BVAIG chips :)

From what I was able to find FVAIQ and FVAIG/BVAIG are not directly compatible because the 'Q' stands for 'Quad Enabled' and means that there's some changes in communication (some additional pins are used while in 'G' version they are not). Fortunately it looks like ASRock has implemented a fallback for the 'non-quad' bios chips to work as well :-)

I was able to contact mdclab and it turns out that the latest UEFI is based on BIOS 3.00 so it does have the latest drivers and modules, but it doesn't have the bootable NVME feature :-(

Either way, I think that after some playing around I will go back to original 2.90P. It looks like the microcode updates together with Meltdown and Spectre fixes cause a substantial performance hit especially for Sandy Bridge and Ivy Bridge Systems.

It's not worth running 2.90P. As can be seen in Linuxland, people are constantly refactoring and updating code vulnerable to Meltdown/Spectre and are *slowly* clawing back performance lost to the patches and updates since early January. Given enough time, compiler tweaks, code updates, driver updates, etc. will get us all back to a level at or before Meltdown/Spectre were a problem.....until the next epic CPU bug is discovered of course! =) Maybe some system benchmark util will show a perf drop on 3.00 BIOS but I haven't noticed much if anything in terms of perf drop.

I would think if you're running regular 3.00b BIOS, MDCLAB MOD BIOS (http://www.mdclab.it/intel-socket-1155-z77-extreme-4.html) with all it's updated firmware and such---and if you're feeling brave---will squeeze out just a bit more performance.

@Arturo: I'm afraid that the optimal performance with such old systems will never be restored. I've seen some tests with really nasty drops after all the fixes were applied (system and microcode updates), especially with SSD speeds. I guess I have to weight between system speed and security.

I'm still waiting for the programmer and the bios chips to arrive. Once they arrive, I'll learn to use them and only then I might do some experimentation ;)

Yes, I know some perf will never come back up to the levels of pre-patch on this old hardware. It's just a fact. But if you do go for the MOD BIOS, would be interesting to see 2.90P, 3.00b and MOD BIOS numbers compared.

Zitat von Arturo im Beitrag #43

---

Yes, I know some perf will never come back up to the levels of pre-patch on this old hardware. It's just a fact. But if you do go for the MOD BIOS, would be interesting to see 2.90P, 3.00b and MOD BIOS numbers compared.

---

Hmm, I'd normally think of SisoftSandra but I don't know if anyone even uses that anymore. What about something simple like CrystalMark (https://crystalmark.info/download/archiv...ark2004R7.zip)? Same guy who makes Crystal Disk Info and Crystal Disk Mark. I use those two apps a lot in my tech life but I never knew about CrystalMark. It's from 2017 so not very up to date but very simple benchmark app that should give you a good general idea about before and after results to get percentages of increase or decrease in performance.