Adventure: Running 8/9th gen Coffee Lake CPUs on Z170 motherboard (ASUS Maximus VIII Ranger)

[itsakjt]

I knew about two chips, but did not know that they can be quick disconnected. that's good.
I want to leave backup chip with original bios and install new on first one, for safety. will it work like this?
Thanks for help.

Eureka!
I wrote new bios with AFU
works with i7-6700. now i will buy 9-th gen and test it out.
Thanks again.
bios settings:

It is unlikely AFU will downgrade Intel ME version which is required for the mod. 11.7.0.1229 is tried and tested to work with all chips from 6th to 9th gen.
You can verify ME firmware by opening Device Manager>System Devices>Intel Management Engine (double click)>click on the firmware tab.

 

[aviatori]

It is 11.8.50.3425.
Will it work?

 

[itsakjt]

It is 11.8.50.3425.
Will it work?

No it would not. You need 11.7.0.1229 or lower as in Coffee Time to make it work.

 

[aviatori]

No it would not. You need 11.7.0.1229 or lower as in Coffee Time to make it work.

I need separate driver or need to change settings in Coffee Time?

 

[itsakjt]

I need separate driver or need to change settings in Coffee Time?

You will need to change settings in Coffee Time. There is an option to change the Management Engine version. Use that to downgrade it to 11.7.0.1229.
You can try flashing with AFU but I won't recommend it as it can potentially brick the board since it usually cannot flash the ME region since that region is locked.

 

[aviatori]

You will need to change settings in Coffee Time. There is an option to change the Management Engine version. Use that to downgrade it to 11.7.0.1229.
You can try flashing with AFU but I won't recommend it as it can potentially brick the board since it usually cannot flash the ME region since that region is locked.

I tried AFU and it did not downgrade IME version.

11.6.0.1126 will work?
I have an idea. last bios is with this version, after is 11.8... so, I will update normally in UEFI with 11.6.0.1126 and then using AFU will update bios with macrocodes. ME will stay 11.6.0.1126.
there is also 11.0... something and which will be better 11.0 or 11.6. I want to use for now i5-9600K and than i9-9900K.

 

[itsakjt]

I tried AFU and it did not downgrade IME version.

11.6.0.1126 will work?
I have an idea. last bios is with this version, after is 11.8. so, I will update normally in UEFI with 11.6.0.1126 and then using AFU will update bios with macrocodes. ME will stay 11.6.0.1126.
there is also 11.0... something and which will be better 11.0 or 11.6. I want to use for now i5-9600K and than i9-9900K.

ME cannot be downgraded usually using any motherboard flash tool. You can try it but I am certain the ME version will remain the same. The only way is using a CH341A or a programmer similar to it.
Some other methods include flashing using Flash Programming Tool from Intel CSME tools using FPTw.exe -rewrite -f biosfilename.bin but it is only possible if motherboard allows temporarily disabling ME which is rare. Another method is to short a pin on the audio IC but that is risky as well.
My suggestion would be to use the programmer.

 

[aviatori]

ME cannot be downgraded usually using any motherboard flash tool. You can try it but I am certain the ME version will remain the same. The only way is using a CH341A or a programmer similar to it.
Some other methods include flashing using Flash Programming Tool from Intel CSME tools using FPTw.exe -rewrite -f biosfilename.bin but it is only possible if motherboard allows temporarily disabling ME which is rare. Another method is to short a pin on the audio IC but that is risky as well.
My suggestion would be to use the programmer.

Ok.
Thanks again for help.

 

[BobbyBoyGaming]

@itsakjt I was thinking about upgrading the microcodes in coffee time to the last coffeeLake microcodes, i downloaded them from winraid forum and put them in the coffeetime mcu folder but unfortunately the program does not list the new microcodes. Maybe the program must be recompiled or something. Or a settings file must be changed. IDK.

@aviatori Have you performed the pin modification? That may be the hardest part of your mod. Good luck.

@itsakjt I was thinking about upgrading the microcodes in coffee time to the last coffeeLake microcodes (FA, instead of EA), i downloaded them from winraid forum and put them in the coffeetime mcu folder but unfortunately the program does not list the new microcodes. Maybe the program must be recompiled or something. Or a settings file must be changed. IDK.

@aviatori Have you performed the pin modification? That may be the hardest part of your mod. Good luck.

 

[aviatori]

@aviatori Have you performed the pin modification? That may be the hardest part of your mod. Good luck.

hello.
no, not yet but that's not a problem for me. mechanical part is my strong side

I bought this "thing" 9TH COFFEE LAKE REFRESH QQLS
as I understood, it has modifications needed to work with z170, only need bios update. same as i9 9900k, but cheaper and that's why I'm not afraid to make experiments.

 

[BobbyBoyGaming]

hello.
no, not yet but that's not a problem for me. mechanical part is my strong side

I bought this "thing" 9TH COFFEE LAKE REFRESH QQLS
as I understood, it has modifications needed to work with z170, only need bios update. same as i9 9900k, but cheaper and that's why I'm not afraid to make experiments.

I have one too, they do work but they are not 9900K, they dont support fast memory above 2933MHz (on a good motherboard), and they don't actually perform as the Ali Sellers advertise them.
I ran the CPU-Z comparison of my QQTJ0 2.1GHz ES BGA1440 CPU against the 9900KF as reference (while overclocked at 5065 All-Core Boost clock and with the RAM at 3226 MT/s (With BCLK of 110). And I tightened the timings with Samsung B-die to 13-13-13-28.

Anyway the CPU-Z comparison shows that this QQTJ0 only has a multicore performance of 60% that of a 9900K.

Also the Chinese guy on Ali told me this was a CPU with a base clock of 2.8, but the base clock is only 2.1GHz.
So the moral of the story is that these Chinese sellers lie a lot and even though the CPU does work it is not truly what it is being advertised as.
That being said, it could very well be an upgrade from a 9600K

 

[aviatori]

I have one too, they do work but they are not 9900K, they dont support fast memory above 2933MHz (on a good motherboard), and they don't actually perform as the Ali Sellers advertise them.
I ran the CPU-Z comparison of my QQTJ0 2.1GHz ES BGA1440 CPU against the 9900KF as reference (while overclocked at 5065 All-Core Boost clock and with the RAM at 3226 MT/s (With BCLK of 110). And I tightened the timings with Samsung B-die to 13-13-13-28.

Anyway the CPU-Z comparison shows that this QQTJ0 only has a multicore performance of 60% that of a 9900K.

Also the Chinese guy on Ali told me this was a CPU with a base clock of 2.8, but the base clock is only 2.1GHz.
So the moral of the story is that these Chinese sellers lie a lot and even though the CPU does work it is not truly what it is being advertised as.
That being said, it could very well be an upgrade from a 9600K

Yes, it not same but for price/performance and for z170 it's amazing.
Can't you overclock it? around 4GHz.
Now I have i7-6700 (4GHz) and comparing with this it will be huge boost.
I was planning to replace motherboard (and everything) and go for DD5, but now I reconsider it. That performance will be enough for me, with minimum effort and cost.

 

[BobbyBoyGaming]

Yes, it not same but for price/performance and for z170 it's amazing.
Can't you overclock it? around 4GHz.
Now I have i7-6700 (4GHz) and comparing with this it will be huge boost.
I was planning to replace motherboard (and everything) and go for DD5, but now I reconsider it. That performance will be enough for me, with minimum effort and cost.

Yes, as I had said, it was overclocked to 5.065 GHz when I tested it. Which is really good. It hits 5GHz without a problem. But I really wanted my To reach 3200MHz so I had to set the Base clock (BCLK) to 110.
My base CPU clock is then 4.6GHz (multiplier is set to 46) my base clock increases that value because (4600MHz)(0.1)=460+4600=5065, then the RAM was set to 2933, but with baseclock 110, it turns into 3226MHz

I did this on a z370 which was very easy because the z370 does detect all the appropriate voltages for 9th Gen.
But I want to warn you that the z170 will not detect the correct voltages so it will crash a lot.

Luckily, I do have a genuine system (I have multiple computers) that has a z390 and a 9900K.
So I can send you a picture of the official voltages so that you know where to start.

I used to have a z170 with this 9900K that I now have on my z390. and the z170 was not "rock solid stable' for me. It is very hard to get the voltages perfectly right without this side by side reference.

Here it is for your personal reference when you try to get the z170 stable

 

[aviatori]

Yes, as I had said, it was overclocked to 5.065 GHz when I tested it. Which is really good. It hits 5GHz without a problem. But I really wanted my To reach 3200MHz so I had to set the Base clock (BCLK) to 110.
My base CPU clock is then 4.6GHz (multiplier is set to 46) my base clock increases that value because (4600MHz)(0.1)=460+4600=5065, then the RAM was set to 2933, but with baseclock 110, it turns into 3226MHz

I did this on a z370 which was very easy because the z370 does detect all the appropriate voltages for 9th Gen.
But I want to warn you that the z170 will not detect the correct voltages so it will crash a lot.

Luckily, I do have a genuine system (I have multiple computers) that has a z390 and a 9900K.
So I can send you a picture of the official voltages so that you know where to start.

I used to have a z170 with this 9900K that I now have on my z390. and the z170 was not "rock solid stable' for me. It is very hard to get the voltages perfectly right without this side by side reference.

Here it is for your personal reference when you try to get the z170 stable
View attachment 313398
View attachment 313399
View attachment 313400

Tanks for info. It will be helpful.
I mean overclocking chinese cpu, not 9900KF. after overclocking it will be somewhere near with I9, I think.
I know how to overclock. I had overclocking practice since lga775 + xeon e5450 There is also reviews from people and I will use their experience also, to set correct voltage. after I will test it myself. I did same on my i7-6700 (non K) to overclock it. now I know that it needs 1.2v to be stabile at 4GHz and 1.35 at 4.4GHz.
On last picture, first of all you need to set voltage. Auto for overclocking will not be stabile. also load-line calibration must set to Level1, to get stabile voltage.

 

[itsakjt]

@itsakjt I was thinking about upgrading the microcodes in coffee time to the last coffeeLake microcodes, i downloaded them from winraid forum and put them in the coffeetime mcu folder but unfortunately the program does not list the new microcodes. Maybe the program must be recompiled or something. Or a settings file must be changed. IDK.

@aviatori Have you performed the pin modification? That may be the hardest part of your mod. Good luck.

Hello @BobbyBoyGaming,

Yes there is a way. Download the microcode files from github. Now go inside the CoffeeTime subfolders where there is a mcodes folder where the CoffeeTime provided microcodes exist. Copy your downloaded microcodes there and rename them respectively to the same ones as inside the folder (you will need to delete the original files of course). Be very careful of renaming the correct CPUID files.
Then fire up CoffeeTime and select the microcode which you used to rename. Apply it and you will find it will show the upgraded version.

 

[BobbyBoyGaming]

Hello @BobbyBoyGaming,

Yes there is a way. Download the microcode files from github. Now go inside the CoffeeTime subfolders where there is a mcodes folder where the CoffeeTime provided microcodes exist. Copy your downloaded microcodes there and rename them respectively to the same ones as inside the folder (you will need to delete the original files of course). Be very careful of renaming the correct CPUID files.
Then fire up CoffeeTime and select the microcode which you used to rename. Apply it and you will find it will show the upgraded version.

This is a very good idea, thanks. I would be tricking the program into thinking I am using the same EA microcode, but instead putting in the FA (newer). This would bypass the need to recompile it. Very good thinking.

But, before I go through this trouble, I was wondering if you have an opinion on this. Do you think it is worth the trouble? I am currently on microcode BE because it is the microcode the AliExpress seller included (older than EA). And I was worried about compatibility issues or inserting new problems. Now that I know my system is fully working, I would be less scared to try this out.

As a side note: It turned out my ASRock z370 had 1 bent pin, I think when I was removing the CPU locking mechanism to put in the mutant CPU, I dropped a screw or something into the socket and it damaged just 1 pin. I tried to unbend it but it was twisted very strangely. I ended up bending it upward and it is still very bent but it is now at least making good contact. I don't plan on ever removing this CPU again after that so I think it should be fine as long as I don't change the CPU again. The other issue I was having was that I was using some very old RAM by corsair called Vengeance LPX 3000, and it was crashing the system a lot even at the slower clocks. It is not until I put the Samsung B-die into the system that it worked perfectly.

I do know that some newer microcode slows down the CPUs, however I do consider security important so, it is a bit of a hard decision. However the real concern I had was whether or not a newer microcode might remove compatibility with these ES CPUs. I think I was worrying too much about it. I think as long as the microcode is is for the CPU ID 906ED it should continue to work with my particular Engineering Sample CPU since that is the CPU ID it has.

The other question I have is whether or not the new microcode security update even takes effect at all if Intel ME is off? I still currently have Intel ME off. So I am not sure if this impacts the effectiveness of the microcode's updates or if it means that, since ME is off, I have an even greater reason to use the latest Microcode. IDK.
Do you know anything about this? I am currently leaning towards upgrading the MCU to FA by the method/trick you mention. I was thinking about how RPCS3 TSX instruction mode is only supported with ME off, and I wanted to test that before upgrading the MCU, but then I realized my CPU is in-fact a Comet-Lake ES (10th Gen) and so it does not even include TSX to begin with. So there is literally no reason for me to keep this old microcode at this point. If I upgrade the microcode, successful I will then try enabling Intel ME and seeing if it works as you had previously stated, but I will have to try one change at a time otherwise it introduces too much confusion as to what is going wrong if something does go wrong.

It also seems as though no matter which microcode is in the BIOS Windows will automatically use MCU DE, which is newer than BE but older than EA. lol. I am starting to think this microcode upgrade endeavor is completely pointless.

@aviatori Sorry for confusing you, but I am showing you images of another system. The above images were from a z390, which was made for the 9900K specifically, so it can automatically set the correct voltages. I sent you this picture on purpose for your personal benefit. If you don't want to use this information, it is okay by me. Sorry for the confusion. This is the z370 system with the overclocked 1440 CPU. z370 can also set voltages correctly. However the z170 will not.

To come to think of it, and upon closer inspection. I think this CPU-Z performance comparison is incorrect. The QTJ1 was not being pushed to its max boost clock while the test was being executed, and the BCLK also distorts the results of the test. So I think the real-world performance is closer to the 9900K than I initially thought.

 

[itsakjt]

This is a very good idea, thanks. I would be tricking the program into thinking I am using the same EA microcode, but instead putting in the FA (newer). This would bypass the need to recompile it. Very good thinking.

But, before I go through this trouble, I was wondering if you have an opinion on this. Do you think it is worth the trouble? I am currently on microcode BE because it is the microcode the AliExpress seller included (older than EA). And I was worried about compatibility issues or inserting new problems. Now that I know my system is fully working, I would be less scared to try this out.

As a side note: It turned out my ASRock z370 had 1 bent pin, I think when I was removing the CPU locking mechanism to put in the mutant CPU, I dropped a screw or something into the socket and it damaged just 1 pin. I tried to unbend it but it was twisted very strangely. I ended up bending it upward and it is still very bent but it is now at least making good contact. I don't plan on ever removing this CPU again after that so I think it should be fine as long as I don't change the CPU again. The other issue I was having was that I was using some very old RAM by corsair called Vengeance LPX 3000, and it was crashing the system a lot even at the slower clocks. It is not until I put the Samsung B-die into the system that it worked perfectly.

I do know that some newer microcode slows down the CPUs, however I do consider security important so, it is a bit of a hard decision. However the real concern I had was whether or not a newer microcode might remove compatibility with these ES CPUs. I think I was worrying too much about it. I think as long as the microcode is is for the CPU ID 906ED it should continue to work with my particular Engineering Sample CPU since that is the CPU ID it has.

The other question I have is whether or not the new microcode security update even takes effect at all if Intel ME is off? I still currently have Intel ME off. So I am not sure if this impacts the effectiveness of the microcode's updates or if it means that, since ME is off, I have an even greater reason to use the latest Microcode. IDK.
Do you know anything about this?

@aviatori Sorry for confusing you, but I am showing you images of another system. The above images were from a z390, which was made for the 9900K specifically, so it can automatically set the correct voltages. I sent you this picture on purpose for your personal benefit. If you don't want to use this information, it is okay by me. Sorry for the confusion. This is the z370 system with the overclocked 1440 CPU. z370 can also set voltages correctly. However the z170 will not.
View attachment 313644
View attachment 313646
View attachment 313647
View attachment 313648

To come to think of it, and upon closer inspection. I think this CPU-Z performance comparison is incorrect. The QTJ1 was not being pushed to its max boost clock while the test was being executed, and the BCLK also distorts the results of the test. So I think the real-wold performace is closer to the 9900K than I initially thought.

Yes man. The microcode trick with CoffeeTime is tried and tested and it works like a charm.

As for whether it is worth it, from a security standpoint, absolutely. But yes, newer microcodes = slower performance under some specific workloads. I have updated my Maximus VIII Ranger (Z170) as well as modded the Maximus X Formula (Z370) BIOS to have the latest microcodes and also pushed the latest ME update since it is a Z370 board and I haven't noticed any performance difference with the things I do. Benchmark scores were on the margin of error, games run as good as they did with the older microcode.
And yes, as long as you include the microcode of your mutant CPU's CPUID, it will continue to work.
As for Intel ME, try keeping it enabled. If using Z170/270, use 11.7.0.1229 or 11.7.0.1261 (same trick works with Coffee Time just like the microcode one). If using Z370 or Z390, latest Intel ME should work even with mutant CPU though I haven't tested it.
Another thing I would like to add, I have a GIGABYTE B250M D3H board. Terrible motherboard with terrible VRM and for some reason has 64 Mb BIOS chips (two of them for dual BIOS, LOL) and no matter what I did, I was able to add just 3 CPUID's microcodes. I remember I added 506E3 (since I had an i5 6500T), 906EA and 906EB (for testing out the mod with a Pentium Gold G5400 and i3 8100) and it worked.
Later, a friend gave me his i7 9700 (R0 variant, so 906ED for test). I dropped it on that board. And behold! It POSTed! Without the 906ED microcode. I verified the CPUID using CPUZ and it was indeed 906ED. I have no idea how or why it worked but it did. Just to be on the safe side, I later flashed the BIOS including the 906EA, EB and ED microcodes. Do the microcodes share information across CPUID's? Your guess is as good as mine.

 

[BobbyBoyGaming]

Interesting point, I was also thinking about cramming a bunch of different versions of the same CPUID lol.
I noticed that CPU-Z is very bad at identifying the ES mutant CPU, however HWinfo64 is very good at it. I highly recommend installing both and using both.

If you look at my hardware info screenshot it will say it is Comet Lake-H but if you look at CPU-Z it says Coffee Lake. I think HW info is correct because it also says "Xeon" and I think this is what my CPU actually is. Because when you look at the following thread:

[OFFER] Intel CPU Microcode Archives

@wendell Yes, first post is fixed, could you also please make the second post editable? Could you please have a look a the first post in this thread so I can edit it also? Thanks in advance.

winraid.level1techs.com

You see there are two sections which cover this ES mutant CPU. So the only reason 906ED works with the mutant CPU must be because it also includes support for Xeon E (R0). And when you look at the BGA 1440 MCU it has the same version numbers.

So maybe to answer your question you would have had to look at the 906ED codes but make sure that they have a shared version number.
So for example in my case, instead of installing
CPU906ed_plat22_ver000000FA_2023-02-27_PRD_EA5AED82.bin

It would be safer for me to install the last version that is also seen in the 1440 Section which is F8, not FA.
So I would instead replace the microcode with
CPU906ed_plat22_ver000000F8_2023-02-05_PRD_1B015780.bin

This may all be incorrect speculation, but the moral of the story is that upgrading microcode for a mutant CPU is not as straight forward as it would be for a normal CPU and perhaps it requires to be extra careful and check more sections just to be sure to pick the correct microcode.

 

[aviatori]

This is a very good idea, thanks. I would be tricking the program into thinking I am using the same EA microcode, but instead putting in the FA (newer). This would bypass the need to recompile it. Very good thinking.

But, before I go through this trouble, I was wondering if you have an opinion on this. Do you think it is worth the trouble? I am currently on microcode BE because it is the microcode the AliExpress seller included (older than EA). And I was worried about compatibility issues or inserting new problems. Now that I know my system is fully working, I would be less scared to try this out.

As a side note: It turned out my ASRock z370 had 1 bent pin, I think when I was removing the CPU locking mechanism to put in the mutant CPU, I dropped a screw or something into the socket and it damaged just 1 pin. I tried to unbend it but it was twisted very strangely. I ended up bending it upward and it is still very bent but it is now at least making good contact. I don't plan on ever removing this CPU again after that so I think it should be fine as long as I don't change the CPU again. The other issue I was having was that I was using some very old RAM by corsair called Vengeance LPX 3000, and it was crashing the system a lot even at the slower clocks. It is not until I put the Samsung B-die into the system that it worked perfectly.

I do know that some newer microcode slows down the CPUs, however I do consider security important so, it is a bit of a hard decision. However the real concern I had was whether or not a newer microcode might remove compatibility with these ES CPUs. I think I was worrying too much about it. I think as long as the microcode is is for the CPU ID 906ED it should continue to work with my particular Engineering Sample CPU since that is the CPU ID it has.

The other question I have is whether or not the new microcode security update even takes effect at all if Intel ME is off? I still currently have Intel ME off. So I am not sure if this impacts the effectiveness of the microcode's updates or if it means that, since ME is off, I have an even greater reason to use the latest Microcode. IDK.
Do you know anything about this? I am currently leaning towards upgrading the MCU to FA by the method/trick you mention. I was thinking about how RPCS3 TSX instruction mode is only supported with ME off, and I wanted to test that before upgrading the MCU, but then I realized my CPU is in-fact a Comet-Lake ES (10th Gen) and so it does not even include TSX to begin with. So there is literally no reason for me to keep this old microcode at this point. If I upgrade the microcode, successful I will then try enabling Intel ME and seeing if it works as you had previously stated, but I will have to try one change at a time otherwise it introduces too much confusion as to what is going wrong if something does go wrong.

It also seems as though no matter which microcode is in the BIOS Windows will automatically use MCU DE, which is newer than BE but older than EA. lol. I am starting to think this microcode upgrade endeavor is completely pointless.

@aviatori Sorry for confusing you, but I am showing you images of another system. The above images were from a z390, which was made for the 9900K specifically, so it can automatically set the correct voltages. I sent you this picture on purpose for your personal benefit. If you don't want to use this information, it is okay by me. Sorry for the confusion. This is the z370 system with the overclocked 1440 CPU. z370 can also set voltages correctly. However the z170 will not.
View attachment 313644
View attachment 313646
View attachment 313647
View attachment 313648

To come to think of it, and upon closer inspection. I think this CPU-Z performance comparison is incorrect. The QTJ1 was not being pushed to its max boost clock while the test was being executed, and the BCLK also distorts the results of the test. So I think the real-world performance is closer to the 9900K than I initially thought.

For more clarification, that information it will be good, thanks.

But, why are you using auto voltage setting on Chinese cpu?! set it manually and set manual on load line calibration too. also turn off turbo boost (usually it turning off automatically when overclocking, but still), turn off speed step and you will get maximum from that cpu.

Your cpu performance at 4.6GHz. this is not from Chinese seller, but from user.

 

[BobbyBoyGaming]

For more clarification, that information it will be good, thanks.

But, why are you using auto voltage setting on Chinese cpu?! set it manually and set manual on load line calibration too. also turn off turbo boost (usually it turning off automatically when overclocking, but still), turn off speed step and you will get maximum from that cpu.

The auto setting is perfectly stable at the moment. Unless you have a different value to enter manually, it is pointless to put in the same automatic value. For example, VCCIO under "Auto" is 1.2, and VCCSA under "Auto" is 1.25.

I could manually type in 1.2 and 1.25 respectively, but the Auto value is already setting it to that so it makes no difference. The only reason I would want to enter a voltage manually is to lower or increase the one set automatically. I am already hitting my target clocks with perfect stability so I don't see a reason to change those voltages. The only one i might lower is the DRAM one but I will do that later after I do the microcode upgrade and all that.

One thing about the "Auto" setting is that the word "Auto" makes it sound as though the voltage will be changing in an unreliable manner, but that is not really what "Auto" is. "Auto" just means, "preset". It is a very "dumb" preset that is fixated. And if it happens to be set to the value you need then it is all the same. I might be wrong about this.

Unless you suggest I try a different voltage on each setting it makes no difference. I currently don't plan on altering those voltages within the next few weeks. As this is not my main system and it is more of just a "experiment".

The load line calibration can be a little dangerous when set to Level 1. I wouldn't personally set it to that setting right away. Level 2 is usually as high as I am willing to go since It has a protective effect on the hardware and I would rather stay on the safe side.

Yes I do agree that turning off Speed Shift (and Speed Step too) is a good idea especially for benchmarking purposes since it will allow the Windows to take command over the CPU clocks I will turn it off and run the comparison test and see if it approaches the reference 9900KF.

This is still a "Work-in-Progress" for me. I do still have some issues. Like for example, My computer was turning back on after shut down, and now I fixed that by enabling "Deep sleep on S5". But I discovered just now that it still turns back on after going into Stand-By, and so I enabled "Deep Sleep on S4-S5" but it still hasn't fixed this issue. I think Windows "Stand-By" is "S3", but unfortunately the BIOS does not offer deep sleep on S3 so I cant fix this at the BIOS level. I am sure there is a Windows OS workaround for this. I will look into it later.

This will ultimately be an HTPC so I don't really plan on pushing it super hard all the time, but it has been fun overclocking it to see if this Chinese seller ripped me off or not. That was my main goal, was to see if I got ripped off or not. I was just skeptical about my purchase so I wanted to find out if the claims are true. But since this will be an HTPC I do think that it is important for me to have shut-down and stand-by working flawlessly. Otherwise the computer becomes a waste of power.

I suspect these deep sleep/shut-down issues are directly related to the absence of Intel ME.

As @itsakjt suggests I will try to re-enable that function once I upgrade the microcode and see if this fixes my shut-down and stand-by functionality. I think it is related to Intel ME, because it seems as though Intel ME is a "middle-man" between Windows Kernel and the CPU, and as it turns out Windows 11 does not really ever shut down. Microsoft has gotten rid of "real shut down" and most computer only go into something that more closely resembles a hibernation when they are shut off. Microsoft did this in order to decrease boot up time.

Here are the results,

With Speed Shift and Speed Step Off,
Screenshot After the CPU-Z test shows that my scores increased a little bit

However, during the test, HWinfo64 shows that my CPU is not being pushed to the maximum clocks for some strange reason.

The CPU only reaches the max Boost clocks after the CPU benchmark is over. So perhaps this means that my CPU is undergoing thermal throttling. I will have to try a new cooler and a different benchmark software.

I don't currently have liquid cooling on this machine, I have a "Cooler Master Hyper 212 Black Edition" with two Big fans going at full speed.
Not sure what is going on at the moment. But I will have to look into it in the following week.

There is also another setting that I may have to change that is called "BCLK-Aware Adaptive Voltage". However this one could be dangerous to turn off. I don't want to damage my CPU so I will have to be careful. It is currently set to "On" but I may have to set it to off to beat the previous score.

*UPDATE*
I disabled BCLK Aware Adaptive Voltage, and I ran the test. This test is with Speed Shift OFF, Speed Step OFF, & BCLK Adaptive Voltage OFF. The score is the same. lol.

To be Continued...

Yes man. The microcode trick with CoffeeTime is tried and tested and it works like a charm.

As for whether it is worth it, from a security standpoint, absolutely. But yes, newer microcodes = slower performance under some specific workloads. I have updated my Maximus VIII Ranger (Z170) as well as modded the Maximus X Formula (Z370) BIOS to have the latest microcodes and also pushed the latest ME update since it is a Z370 board and I haven't noticed any performance difference with the things I do. Benchmark scores were on the margin of error, games run as good as they did with the older microcode.
And yes, as long as you include the microcode of your mutant CPU's CPUID, it will continue to work.
As for Intel ME, try keeping it enabled. If using Z170/270, use 11.7.0.1229 or 11.7.0.1261 (same trick works with Coffee Time just like the microcode one). If using Z370 or Z390, latest Intel ME should work even with mutant CPU though I haven't tested it.
Another thing I would like to add, I have a GIGABYTE B250M D3H board. Terrible motherboard with terrible VRM and for some reason has 64 Mb BIOS chips (two of them for dual BIOS, LOL) and no matter what I did, I was able to add just 3 CPUID's microcodes. I remember I added 506E3 (since I had an i5 6500T), 906EA and 906EB (for testing out the mod with a Pentium Gold G5400 and i3 8100) and it worked.
Later, a friend gave me his i7 9700 (R0 variant, so 906ED for test). I dropped it on that board. And behold! It POSTed! Without the 906ED microcode. I verified the CPUID using CPUZ and it was indeed 906ED. I have no idea how or why it worked but it did. Just to be on the safe side, I later flashed the BIOS including the 906EA, EB and ED microcodes. Do the microcodes share information across CPUID's? Your guess is as good as mine.

I was able to use the trick to inject the latest Microcodes! I replaced all the EAs with the F4, FA, F0. I have not tested anything yet.

As for the Intel ME, do you know if there might be a GitHub Repository or an Archive of the Intel ME versions much like the Plato GitHub with the Micro codes? Otherwise, I will have to use the Intel ME that is included in the latest official BIOS. But Enabling ME may disable compatibility of z370 with Skylake 6700K. But since I had that bent pin issue I was talking about, I don't think I will ever be changing the CPU on this thing ever again lol.

 

[BobbyBoyGaming]

I found the Intel ME repository! Just for Reference If anybody needs it:

759.59 MB folder on MEGA

31 files

mega.nz

I was able to replace the Intel ME with the newest version, but I have not flashed it yet.

 

[itsakjt]

I found the Intel ME repository! Just for Reference If anybody needs it:

759.59 MB folder on MEGA

31 files

mega.nz

I was able to replace the Intel ME with the newest version, but I have not flashed it yet.
View attachment 313957

For your sleep problem, are you using a wired LAN connection on the system?

 

[BobbyBoyGaming]

For your sleep problem, are you using a wired LAN connection on the system?

I have tried everything that I can think of for that. I disabled "Wake on LAN", "Power On by USB", "Wake Up by USB".
The only thing that worked for me was "Enable Deep Sleep on S4, S5", but it only fixed the Windows 11 "Shut-Down". It did not fixe the "Stand-By". I think Windows Stand-By is technically equal to "S3". Maybe I have to go into the ACPI settings and look at the "Power Saving" settings for other devices. Or something. I don't know. I will have to check the options later and will report back. I would rather fix this at the BIOS level than to do some Windows work-around.

UPDATE:

I enabled all the "C States" in the CPU configuration page of the BIOS and Stand-by/Sleep is now working flawlessly.
These may not be great settings for overclocking, but the overclocking was just out of curiosity, it is more important for me for the computer to behave as expected.
I did a quick and dirty fix, and enabled all the C states, C3 through C7.
Did I only have to enable one of them for Windows 11 sleep mode to work properly? Your guess is as good as mine. I dont think I have the mental energy to figure that out at the moment.

 

[BobbyBoyGaming]

Sad news lol... I was desoldering the BIOS chip for 3rd time to do my final microcode upgrade, and I a lifted the bios chip off the board i ripped some of its solder pads off. it was not melted all the way.
so now i cant resolder the bios chip back on lol.

 

[Scisca]

Hello! I've tried to follow your guide, as I have a ASUS ROG Maximus VIII Ranger motherboard as well. I tried soldering the 102 pin, but now my PC doesn't want to start at all Can you look at the photo and maybe tell me what I did wrong and how can I fix the mistake? Should I desolder the connection? I would appreciate any help very much!