Originally Posted by Skylinestar
Just wanna update on this after I update my UEFI to P7.40 (for Spectre/Meltdown patch).
This is related to the last post of this thread, where Xaltar mention how the cache multiplier auto set to a lower value while SkyOC:
With the new P7.40 UEFI, the cache-multiplier doesn't reduce. It's fixed at the high value (it's x36 for my i5 6500). This makes my i5 6500 unable to overclock. Previously, on P7.20, it's 133x24=3192 (as seen in the UEFI setup screen). Now, on P7.40, it's 133x36 = 4788. This cache-clock is just too high, failed to POST. Debug code 32. And the bad thing is that there's no setting for this cache-multi / cache-clock adjustment if we enable SkyOC.
Damn ASRock. Why did you mess with that precious setting?
Anyone goes back to the older UEFI for the purpose of better oc while sacrificing security?
I hope this info is helpful to someone.
I had the same issue with the latest BIOS (7.30) for my Z170-ITX Fatality board. The cache was too high (or so I thought) and non-configurable.
I modified 7.30 with the same edits from the 1.50 Korean BIOS using AMIBCP - changing the access mode from default to user for several menu items. This brought back the cache multiplier and all the other settings from our beloved Korean mod. I couldn't get the same clocks that I used successfully in 1.50 to boot. In fact, I couldn't get the memory to run above 2133 MHz at all. Take this with a grain of salt though, because unfortunately I never was able to get my memory to run at its rated speed (2400 MHz) even with the 1.50 BIOS, so you might have better luck. I have a 2x16GB C15 Patriot kit which seems like an odd bird but I suspect that my ES chip just doesn't handle memory very well. I certainly didn't win the silicon lottery with my QHQG but I haven't tried a very high vcore (my system is housed in a DAN A4 chassis). Many memory dividers fail to boot with either BIOS. I haven't been able to dig up memory timings for my kit but I did try loosening several timings and even tried VDIMM as high as 1.35V (even though the XMP profile only calls for 1.2V).
As for the Meltdown/Spectre microcode (Rev C2), which ASROCK states is included in 7.30...
I ran the Microsoft Powershell test before and after flashing 7.30 and there was no change for my QHQG, it still stated that I needed a newer vendor firmware. I also ran inSpectre from GRC and it reported the following (re Spectre fixes)
"Intel has stated -- will never be made available for this system's processor. (Protection from the Meltdown vulnerability does not require BIOS or processor updates.)"
My guess is that the new microcode doesn't apply to ES chips, or at least, not to mine.
Also, there is a (somewhat ambiguous) comment on the HWTIPS site that (when translated into English) would seem to indicate that it may be possible to update the microcode for protection against Spectre attacks.
Sorry for the bad news. I'm quite interested to read if anyone else makes any headway regarding Spectre fixes for these ES chips though.