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

[itsakjt]

Hey guys nice work, it does make me wonder if Intel could have made this work officially.

Just a little comment to bobbyboygaming

As I noticed you set a very high SA and IO voltages, on my own 9900k system (z370) I discovered silent corruption on some SATA/NVME devices which was eventually discovered to be because of the VCCSA voltage too high. So I do suggest you monitor the SMART stats of your IO devices.

In my case, I was getting instability with high VCCSA with OCCT, reducing it increased stability until I found the sweet spot of 1.1V for the clocks I am using. My i5 9600KF needed 1.15V for VCCSA for 5.1 GHz with 4.8 GHz cache. In the 9700K, I was not able to make 4.8 GHz cache stable no matter what voltage I used. Might have something to do with the higher cache memory (12 MB vs 9 MB) on the chips.

 

[rio]

Hey guys nice work, it does make me wonder if Intel could have made this work officially.

Just a little comment to bobbyboygaming

As I noticed you set a very high SA and IO voltages, on my own 9900k system (z370) I discovered silent corruption on some SATA/NVME devices which was eventually discovered to be because of the VCCSA voltage too high. So I do suggest you monitor the SMART stats of your IO devices.

oh they 100% couldve, they just chose not to to sell boards, theres a fair amount of z370 boards that use identical pcbs to z270 equivalents

 

[BobbyBoyGaming]

Just a brief update. I recently reinstalled windows 11 on my 9900K-z170 modded system, and all I did was to change the VCCIO and VCCSA to 1.250 for both (my BIOS does not allow me to set the VCCIO to anything higher than 1.250). And the Vcore, I set to an offset of +40mV

The CPU-Z is showing that my pc is achieving 4.98 (5Ghz) without any issue.

I mention this because Previously, I had been manually changing the clock speed to "All cores = 4.9Ghz, with a cache ratio of 4.6Ghz) but I was having some occasional (infrequent stability issues). However without these manual settings it just seems to be achiving 5GHz, and seems more stable too.. so.. IDK. I guess its even better!

 

[itsakjt]

Just a brief update. I recently reinstalled windows 11 on my 9900K-z170 modded system, and all I did was to change the VCCIO and VCCSA to 1.250 for both (my BIOS does not allow me to set the VCCIO to anything higher than 1.250). And the Vcore, I set to an offset of +40mV

The CPU-Z is showing that my pc is achieving 4.98 (5Ghz) without any issue.

I mention this because Previously, I had been manually changing the clock speed to "All cores = 4.9Ghz, with a cache ratio of 4.6Ghz) but I was having some occasional (infrequent stability issues). However without these manual settings it just seems to be achiving 5GHz, and seems more stable too.. so.. IDK. I guess its even better!

You need to run proper stress test tools like OCCT, Prime95 to ensure 100% stability of the overclock. Typically, for me, if a system is stable in 20 mins of OCCT under extreme mode, it can be considered stable. Like I said, there are a lot of factors at play. For 5.1 GHz core clock, I cannot set 4.6 GHz cache without running into BSODs or lockups or WHEA errors. I could run it but it needs a lot of VCCSA (1.175V). However, I have 4.7 GHz cache rock stable at just 1.1V VCCSA along with 5.1 GHz core clock. There are things that just does not make sense and the only way to find out for sure is repeated stress tests done properly. That's how I like to do it anyway.
For RAM OC, I use MemTest 64 (available on TPU, a bit patience needed for it to work sometimes but well worth it) and AIDA64 Stability test by just selecting memory.

 

[BobbyBoyGaming]

You need to run proper stress test tools like OCCT, Prime95 to ensure 100% stability of the overclock. Typically, for me, if a system is stable in 20 mins of OCCT under extreme mode, it can be considered stable. Like I said, there are a lot of factors at play. For 5.1 GHz core clock, I cannot set 4.6 GHz cache without running into BSODs or lockups or WHEA errors. I could run it but it needs a lot of VCCSA (1.175V). However, I have 4.7 GHz cache rock stable at just 1.1V VCCSA along with 5.1 GHz core clock. There are things that just does not make sense and the only way to find out for sure is repeated stress tests done properly. That's how I like to do it anyway.
For RAM OC, I use MemTest 64 (available on TPU, a bit patience needed for it to work sometimes but well worth it) and AIDA64 Stability test by just selecting memory.

what value are you using for vccio?
does your motherboard allow more that 1.250 vccio?

 

[itsakjt]

what value are you using for vccio?
does your motherboard allow more that 1.250 vccio?

My VCCIO is at 1.1375V. Motherboard allows setting this up to 1.8V.
VCCIO is the voltage for the memory controller (IMC) of the CPU. How much voltage you will need will depend on your memory frequency, memory rank (single will require less voltage, dual rank will require more voltage) and the number of slots you have populated. I have all 4 slots populated with 4*8 GB dual rank memory running at 3100 MHz CL14. I think I got really lucky because this chip seems to be great overall.

 

[itsakjt]

Small update: For the first time ever, I managed to not only boot my RAMs populated in all 4 slots to 3200 MHz on the Maximus VIII Ranger, but got them rock stable. RAM voltage is at 1.4V. I have been using these RAMs since 2017 and this motherboard as well (had an i5 7600K back then) and I was never ever able to even make POST successful with RAMs at 3200 MHz. I knew they would be stable after I got them working at 3266 MHz on my AMD system as on specs (with the same timings you see below) and always thought it had something to do with the board or with the IMC of the CPU.
Later on, when I failed with my i5 9600KF as well, I was almost convinced it was the board because all i5 9th gen can easily do 3200 MHz or even more. After all, all RAMs of 3200 MHz on the QVL of my board were listed as 2 DIMMs and not 4.
Today, I don't know what happened. I just tried it and it just worked! And I got it stable without any struggle or anything.

Only two things have been changed since I wrote this post:
1. CPU - From i5 9600KF to i7 9700K, both R0 stepping.
2. AIO - From a Cooler Master Seidon 120V to a Silverstone PF240-ARGB

I have no idea what made it possible. VCCIO and VCCSA are still at 1.1375V and 1.1000V respectively.
I read somewhere that some AIO pumps interfere with the memory signals but never believed it really. Could it be that? I have no idea. But I am super happy today!

Update: It seems CL14 is stable as well. Tested with Memtest64 and AIDA64 memory test for 1 hour each.

 

[BobbyBoyGaming]

For me it worked at 3600MHz, look at a strange setting in the bios called
DLL (Digital Locked Loop). For me this setting is what allowed my ram to run at its XMP Speed. My ram is very new though so it is more energy efficient I guess. but it is Dual Rank 2x 32Gig Sticks (I wanted to test the 128GB RAM bios mod [and it worked!]), it is a mini ITX board so it has 64 GB of ram in two slots which means the mod worked.

The model of ram that worked for me is this one:

TEAMGROUP T-Create Expert overclocking 10L DDR4 64GB Kit (2 x 32GB) 3600MHz (PC4 28800) CL18 Desktop Memory Module Ram - TTCED464G3600HC18JDC01​

by default this DLL setting (under the RAM menu all the way at the bottom) was set to zero, but if I change it to 3 it works

 

[itsakjt]

For me it worked at 3600MHz, look at a strange setting in the bios called
DLL (Digital Locked Loop). For me this setting is what allowed my ram to run at its XMP Speed. My ram is very new though so it is more energy efficient I guess. but it is Dual Rank 2x 32Gig Sticks (I wanted to test the 128GB RAM bios mod [and it worked!]), it is a mini ITX board so it has 64 GB of ram in two slots which means the mod worked.

The model of ram that worked for me is this one:

TEAMGROUP T-Create Expert overclocking 10L DDR4 64GB Kit (2 x 32GB) 3600MHz (PC4 28800) CL18 Desktop Memory Module Ram - TTCED464G3600HC18JDC01​

View attachment 241214
by default this DLL setting (under the RAM menu all the way at the bottom) was set to zero, but if I change it to 3 it works

That's awesome. I don't have use for 64 GB RAM but I need 32 GB. My RAMs are from 2015 and rated at 2666 MHz CL15. It is a kit of 4 modules with SK hynix chips. For what it does still today, I can't really complain. I have overclocked and have tightened the timings and even tightened secondary and tertiary timings.
However, the motherboard I have is real picky about memory. I tried with my other Fury Beast 2x16 GB RAM kit from my primary system back when I had the i5 9600KF and I was not able to make them POST at 3200 MHz XMP profile. Though now, I am not entirely sure how they would do. I do have that DLL setting as well under DRAM Timing configuration on my board. Looks like I will play with that some day.
One thing is for certain, this CPU I got is absolutely awesome. I still cannot believe I am running dual rank 32 GB 3200 MHz RAMs with all slots populated with VCCIO voltage of just 1.1375V and VCCSA at 1.1V and rock stable. I don't know any Coffee Lake user who has been able to do the same even while keeping such a high overclock.
My next upgrade is going to be the GPU. Prices are going down here in India and is almost close to MSRP now. I am using the 2060 Super on this machine which will run out of warranty on August this year. Performance is still amazing though since I game at 1080p.

 

[BobbyBoyGaming]

I was not so careful with VCCIO and VCCSA so far. I might try to tweak each one in smaller decrements but I may have to dedicate several hours to that. This is not my main computer, and it is actually off at the moment so it is safe.

I did notice that when I install the latest Intel DCH Video driver for the UHD 630 (via the web-browser-based "Intel Driver Assistant") it causes a freeze/crash.

I have tried multiple times with different configurations and this driver causes a system crash during installation no matter what. The one that windows automatically installs works fine though

I think it was this one

For now, it does not matter as windows is not force-feeding me this update yet. but maybe in the future if windows begins to force this update, it may become a problem.

Has anybody had issues with the latest iGPU Intel drivers for the UHD 630 from a 9600K, 9700K or 9900K?

These drivers dont make any difference in performance so it has no point in updating but I was just wondering why this is an issue.

 

[BobbyBoyGaming]

@itsakjt
Have you had any luck with the mod for the resizable BAR?
Sounds very complicated, I wouldn't be surprised if it is not possible

--UPDATE 1--
Going back to the iGPU Driver Update issue I was having. I have turned on "Developer Mode" on Windows 11 and I will try to upgrade the video driver and see if this works.
--
Has anybody else had success updating the iGPU driver with this mod?
Intel driver assistant says that there is graphics driver available. Screenshot below shows that the mod works (z170 with 9900K). And the current video driver version is highlighted (from 2021)

The current version is from 2022, and it is shown below

However, when the driver updated is offered it gives a warning about how "updating might remove customizations".

Since the coffeeTime mod does involve a vBIOS change, I was wondering if that is somehow related to this warning or if it is related to why my driver update installation was failing in the past.

I am just curious. In case anybody has any clue as to what might be going on, I would be interested in any ideas. (even though the iGPU driver update isn't really important). Of course the most important thing is that the coffee lake runs on z170, but I am just curious.


---UPDATE 2---
After I enabled "Developer Mode" and tried to upgrade the video driver again, My computer no longer crashed. It gave me an error this time. Shown Below. And it also produced a error LOG file (attached).

So I think that this is indeed probably related to the customized vBIOS and maybe it is simply not possible to upgrade the video Driver. Hopefully this answers the question to anybody who might be curious about this in the future.

--

 

[itsakjt]

@itsakjt
Have you had any luck with the mod for the resizable BAR?
Sounds very complicated, I wouldn't be surprised if it is not possible

--UPDATE 1--
Going back to the iGPU Driver Update issue I was having. I have turned on "Developer Mode" on Windows 11 and I will try to upgrade the video driver and see if this works.
--
Has anybody else had success updating the iGPU driver with this mod?
Intel driver assistant says that there is graphics driver available. Screenshot below shows that the mod works (z170 with 9900K). And the current video driver version is highlighted (from 2021)
View attachment 243516

The current version is from 2022, and it is shown below
View attachment 243517

However, when the driver updated is offered it gives a warning about how "updating might remove customizations".
View attachment 243514

Since the coffeeTime mod does involve a vBIOS change, I was wondering if that is somehow related to this warning or if it is related to why my driver update installation was failing in the past.
View attachment 243518
I am just curious. In case anybody has any clue as to what might be going on, I would be interested in any ideas. (even though the iGPU driver update isn't really important). Of course the most important thing is that the coffee lake runs on z170, but I am just curious.


---UPDATE 2---
After I enabled "Developer Mode" and tried to upgrade the video driver again, My computer no longer crashed. It gave me an error this time. Shown Below. And it also produced a error LOG file (attached).

So I think that this is indeed probably related to the customized vBIOS and maybe it is simply not possible to upgrade the video Driver. Hopefully this answers the question to anybody who might be curious about this in the future.

View attachment 243521
--

Hi there!
I tried porting Maximus X Hero BIOS to my Maximus VIII Ranger and was almost successful. I was able to get it to pass CPU and even memory training. But it was then stuck on debug code 3b (PCH init) which makes sense since it is Z170 vs Z370. It is definitely possible by changing the PCH init binaries and DXE files but turns out there are a lot of them. I don't have a spare 128 Mb BIOS chip to try and experiment with and hence stopped my experiment there, scared that I am probably going to kill the BIOS chip at some point.
As for your iGPU issue, I will try it once on my i7 9700K. I believe this is more of a software/OS issue you are having rather than the VBIOS at fault. I have a spare SSD to play with and will try the iGPU out once I get some time.

 

[Sweex]

@itsakjt

How did you manage to solder this? My 1st attempt failed cause the solder didnt hold on the resistor.

 

[itsakjt]

@itsakjt

How did you manage to solder this? My 1st attempt failed cause the solder didnt hold on the resistor.

Just basic soldering. I applied some flux to the places and captured a tiny amount of solder on the tip of the soldering iron and soldered the wire.

 

[itsakjt]

Addendum to the original post:

Another ASUS Z270 motherboard, another Coffee Lake mod. I had an i7 9700KF laying around and someone offered me a good deal on an used mint condition ASUS Z270-K motherboard. Modded the BIOS to add Coffee Lake support and unlocked some of the hidden BIOS menu like Boot performance mode etc. Flashed it by connecting my programmer directly to the SPI flash interface on the motherboard. The SuperIO controller here is a Nuvoton NCT5539D and socket occupied sense is implemented on the board. In no time, I checked the datasheet and pin 48 is the SKT_OCC pin. A trace is there which goes to a SMD capacitor and a SMD resistor. The other end of that capacitor is ground. I soldered a tiny wire over that capacitor and we have a winner. Modded the VRM cooling a bit as well. Replaced the thermal pad with a better one of the same thickness. Dropped the temps by 10 degree C (75 degree C now vs 85 degree C before). Also added another small heatsink on one of the top phases which is connected to the IA cores. I left the HD graphics phases as is as I have no use for it, this being a KF chip.
The 9700KF is overclocked to 5 GHz with an AVX offset of 1. As good as it can get with a 120mm cooler and a motherboard of this tier. VRM temps with Prime 95 small FFT stress is 75 degree C with a 120mm fan blowing over the heatsinks.
Everything works. All RAM slots, m.2, SATA ports, PCIe, you name it. At this point, I am confident I will be able to mod any 100 or 200 series board for Coffee Lake support.
This board has a few advantages over the Z170 lineup:
1. 2 x m.2 slots
2. Inbuilt TPM support due to presence of Intel PTT (Platform Trust Technology)
3. Better memory compatibility

Flashing BIOS by connecting CH341A to SPI header

Socket Occupied mod (notice the tiny wire on the capacitor)

First boot

After cleaning the board to perfection

CPU installed

Custom heatsink on one of the power phases belonging to the IA cores. There is a thermal pad between the MOSFETs and the heatsink and the heatsink itself is glued with epoxy based adhesive to the choke below it, cured for about 24 hrs, kept under pressure with a clip and then powered on. The fan above on the chassis helps to keep the entire VRM cool.

CPU-Z

AIDA cache and memory benchmark

The RAMs here are mismatched and one of the sticks does not even have a heatsink hence the timings are not the best. But good to see 3200 MHz CL 17 achievable at 1.35V.
Important note: On this board, isolation of the two pins as noted on the photo on the original post should be done. I was lazy and did not do it and turns out, those pins got very hot and turned blue just like @BobbyBoyGaming also faced. Luckily, the CPU pads were unaffected. Since these are reserved pins, I chopped them off with a needle nose plier and I am happy to say nothing is affected. After doing that, I have checked each and every RAM slot, PCIe slot and the list goes and everything works. The board also works with an i5 7600K meaning existing support is not compromised as well by chopping of those pins. If you are not comfortable chopping of the pins, please isolate the pads in your CPU as instructed

 

[z270f]

I have an Asus z270f gaming and want to know if this guide can get the board to use a 8400 (6 cores no HT)

Bios version: 1301
ME version: 11.7.0.1040

I know the board does not have flashback and only has crashfree3 (not sure if it’s useful or not)

Does anyone know if pin isolation is needed for this board? I plan to do find the pin 102 and ground it for the superIO.

Is there a way to flash the modded bios safely without bios programmer? If I screw up, only way to get it back is via bios programmer? I heard most of the programmer out there can’t do the needed 3.3v, 5v will kill the bios chip??? Any cheaper programmer with clip that works with 3.3v?

Is AMI Firmware Update Utility v5.X.X.X (AfuEfix64.efi) going to able to flash the modded bios???
I found this want want to see if following it is safe approach: https://rog.asus.com/forum/showthre...-in-DOS-with-USB-tutorial-Intel-AMD-roll-back

is it better to use the port-bios? [ASUS_Z270F_PORT_Z370F_V2001] or mod the latest original from asus?

Thanks in advance.

 

[itsakjt]

I have an Asus z270f gaming and want to know if this guide can get the board to use a 8400 (6 cores no HT)

Bios version: 1301
ME version: 11.7.0.1040

I know the board does not have flashback and only has crashfree3 (not sure if it’s useful or not)

Does anyone know if pin isolation is needed for this board? I plan to do find the pin 102 and ground it for the superIO.

Is there a way to flash the modded bios safely without bios programmer? If I screw up, only way to get it back is via bios programmer? I heard most of the programmer out there can’t do the needed 3.3v, 5v will kill the bios chip??? Any cheaper programmer with clip that works with 3.3v?

Is AMI Firmware Update Utility v5.X.X.X (AfuEfix64.efi) going to able to flash the modded bios???
I found this want want to see if following it is safe approach: https://rog.asus.com/forum/showthre...-in-DOS-with-USB-tutorial-Intel-AMD-roll-back

is it better to use the port-bios? [ASUS_Z270F_PORT_Z370F_V2001] or mod the latest original from asus?

Thanks in advance.

1. It is possible to run Coffee Lake CPUs on any 100/200 series board.
2. Pin isolation is recommended for any Z270 series board from ASUS to prevent damage to contact pads of CPU or pins of socket.
3. It is recommended to use external programmer to flash BIOS as ASUS boards have some regions locked and cannot be flashed. No software can bypass those locks as well. The only other way is a hard mod which involves momentarily grounding one of the pins of the audio chip.
4. I have the CH341A unmodded and so far, I have flashed numerous BIOS chips e with it and all of them work perfectly fine.

 

[z270f]

1. It is possible to run Coffee Lake CPUs on any 100/200 series board.
2. Pin isolation is recommended for any Z270 series board from ASUS to prevent damage to contact pads of CPU or pins of socket.
3. It is recommended to use external programmer to flash BIOS as ASUS boards have some regions locked and cannot be flashed. No software can bypass those locks as well. The only other way is a hard mod which involves momentarily grounding one of the pins of the audio chip.
4. I have the CH341A unmodded and so far, I have flashed numerous BIOS chips e with it and all of them work perfectly fine.

If I'm going to get the CH341A anyway, I can first try out AMI Firmware Update Utility to flash first? If that fails, I can still use the programmer to load the modded or original bios to recover? I don't see the realtek chipset on z270f gaming...

I found it odd that coffetime doesn't let me patch the 128gb ram limit on 1501 but available to 1301 officially. Does that mean Asus already enable the 128gb ram in 1501?

I guess it's still better to use coffeetime to mod the official bios other than using port from z370f?

Thanks

 

[itsakjt]

If I'm going to get the CH341A anyway, I can first try out AMI Firmware Update Utility to flash first? If that fails, I can still use the programmer to load the modded or original bios to recover? I don't see the realtek chipset on z270f gaming...

I found it odd that coffetime doesn't let me patch the 128gb ram limit on 1501 but available to 1301 officially. Does that mean Asus already enable the 128gb ram in 1501?

I guess it's still better to use coffeetime to mod the official bios other than using port from z370f?

Thanks

Using the port from Z370-F if available is the best option since you will have support for newer technologies like resizable bar and per core overclocking of up to 8 cores (depending on CPU). From where are you getting the port? Can you provide the link please?
Also, you can try flashing with AMI firmware update utility but pretty sure it won't work.

 

[z270f]

This is where I found the z370f:

[Guide] How to flash a modded AMI UEFI BIOS

Thank you for great explanation. Regarding the Asus EZ Flash method, I was actually talking about the Windows version of it in AI Suite 3. It seems some body trick the program with the original bios and replace it with mod bios just before clicking the flash button and let the system restart to...

winraid.level1techs.com

This post Rez is following the AFUDOS guide able to flash the z370 ported mod bios to z270f:

[Guide] How to flash a modded AMI UEFI BIOS

@Lost_N_BIOS I tried to fix Error 368 following step 1-5 but somehow end up with a empty Section_PE32_image_Setup IFR.txt (only headers, no detail on BIOS lock). One thing I realize is that on Step 3, the Protocol showed Green EFI, not Green UEFI. I must have done something wrong there. In the...

winraid.level1techs.com

So I'm really curious if loading the bios with AFUDOS is possible. So If I make a backup with AMI utility with the serial#/mac address etc I can just buy the CH341A flasher to recover if things go wrong with AFUDOS?

Thanks.

 

[itsakjt]

This is where I found the z370f:

[Guide] How to flash a modded AMI UEFI BIOS

Thank you for great explanation. Regarding the Asus EZ Flash method, I was actually talking about the Windows version of it in AI Suite 3. It seems some body trick the program with the original bios and replace it with mod bios just before clicking the flash button and let the system restart to...

winraid.level1techs.com

This post Rez is following the AFUDOS guide able to flash the z370 ported mod bios to z270f:

[Guide] How to flash a modded AMI UEFI BIOS

@Lost_N_BIOS I tried to fix Error 368 following step 1-5 but somehow end up with a empty Section_PE32_image_Setup IFR.txt (only headers, no detail on BIOS lock). One thing I realize is that on Step 3, the Protocol showed Green EFI, not Green UEFI. I must have done something wrong there. In the...

winraid.level1techs.com

So I'm really curious if loading the bios with AFUDOS is possible. So If I make a backup with AMI utility with the serial#/mac address etc I can just buy the CH341A flasher to recover if things go wrong with AFUDOS?

Thanks.

Awesome. Please let us know your results.
Also, update for me, I checked my ASUS Z270-K and it seems the ASUS TUF Z370 Gaming Plus is the closest equivalent to the Z270-K. I downgraded the ME to 11.7.0.1261 with CoffeeTime and added the fixes and flashed it and WHOA, my board has been transformed to an ASUS Z370 TUF Gaming Plus. The only things that do not work are the Asmedia 1142 controller and the AIO Pump header (though I have found a workaround to make it work but without monitoring though). I have flashed back to the original modded BIOS because I want to get the Asmedia 1142 working for USB 3.1 support.

 

[z270f]

I followed your instruction and got the 8400 to work with z270f gaming.

I had to use external CH341a to connect to SPI port next to the bios to program.

There is only 1 issue, the video output from motherboard doesn't work, do you know if it is bios issue? I'm using 1501 official Asus bios modded with coffeetime 0.99

Thanks

 

[itsakjt]

I followed your instruction and got the 8400 to work with z270f gaming.

I had to use external CH341a to connect to SPI port next to the bios to program.

There is only 1 issue, the video output from motherboard doesn't work, do you know if it is bios issue? I'm using 1501 official Asus bios modded with coffeetime 0.99

Thanks

Hi there, congrats on the mod.

Coming to the iGPU, did you update the VBIOS and GOP using Coffee Time?

 

[z270f]

yes ME 11.7.0.1229, VBIOS 1062 GOP 9.0.1080

Should I try newer 11.8 ME? I don't see option for anything newer for VIBIOS and GOP. When I first loaded the bios the igpu still works i5 7600, but 8400 has to use external gpu card otherwise no screen. The motherboard light is green tho. The i5 8400 already tested in original HP machine the igpu works with hdmi out.

Thanks

 

[itsakjt]

yes ME 11.7.0.1229, VBIOS 1062 GOP 9.0.1080

Should I try newer 11.8 ME? I don't see option for anything newer for VIBIOS and GOP. When I first loaded the bios the igpu still works i5 7600, but 8400 has to use external gpu card otherwise no screen. The motherboard light is green tho. The i5 8400 already tested in original HP machine the igpu works with hdmi out.

Thanks

I see. It should work without issues.
I am wondering, did you isolate the pads on the CPU?
If you did the isolation, it might be possible that more pins are isolated which is causing iGPU not to work.
It would be great if you can take out the CPU once and share us a photo here.