[SUCCESS] Gigabyte Designare Z390 (Thunderbolt 3) + i7-9700K + AMD RX 580

[CaseySJ]

Thanks for the help

Please un-quote micro-guide from the earlier post.

Also post front page of Hackintool, but blank out any private info. It shows iGPU configuration.

 

[CaseySJ]

I found RadeonBoost to not work with VEGA 64, even the "special" version @CaseySJ has in the mini guide. However @mattystonnie is working on a new setup with SSDT + kext that does seem to give me a pretty good synthetic boost and keeps my fans running as they should. As for real world increases, I've not had a chance to test out yet so your millage certainly may vary.
Testing Thread

It would be interesting to see Before/After benchmarks. And of course we'll wait for real-world apps until you get a chance to test them.

 

[karambit]

We should all do our due diligence and have before and after scores of most apps, esp the ones that are heavily gpu reliant, such as Premiere, Davinci resolve, After effects and Cinema 4D renders NOT the Stress/Bench test.
I have no problem removing Radeon Boost et al and doing the gambit of the whole Lot before , then adding back one at a time!

 

[CaseySJ]

Thanks Casey. All looks well when playing the video, however, this could be completely non-related (and therefore normal behaviour), but whilst that video was playing in the background via Quicktime, I was browsing the Internet/Tonymac etc, and noticed that every time I clicked a link, GFX AVG actually dipped a fair bit (as you can see on the green line). Is this normal? (screenshot attached)

The same behaviour also happens if I open notes and update my personal changelog for the changes/additions I've done to this Hack etc.

Thank you

Yes that's normal. Video playback is not a particularly good performance test because:

• It's a fixed time test. It doesn't make sense, for example, to finish the test in half the time. It just needs to finish on time.
• It's not particularly compute-intensive because the computation is spread out over time. iGPU only needs to decode a little bit of the file to keep the video playing in real time. IGPU does not need to decode the entire 1.2GB file as fast as possible.
• During playback, the full 1.15GHz or 1.2GHz speed of iGPU is not needed, but it's there in case it's needed.

Therefore an encoding or transcoding job might offer a better test.

 

[iRamon]

That pretty much confirms my suspicion. If the chip is flashed properly -- and assuming no physical or electrical damage -- then tbpatch would have worked. In fact, once chip is flashed properly, we can use tbpatch to restore the original firmware without using an external flasher. We can actually use tbpatch to make any additional modifications to the firmware as we wish.

At this point the options are limited:

• Flash original firmware back to the chip.
• But also look closely at the tiny surface-mount resistors/capacitors in the area. You took some exceptionally clear and close-up photos earlier. Check for any damage to those nearby components.

I was able to re-flash the original firmware back to the chip succesfully. A read/backup after flashing confirmed success with the same checksum. Interestingly, flashing would only work when "hot-plugging" VCC Pin8, so starting the write command with VCC disconnected, and immediately after connecting VCC. I had no PSU connected at all.

Nevertheless, my TB XHC Controller and SSP1 and SSP2 ports are still missing in macOS.
I also tried the tbpatch list command, and got the same "device not found" errors.

Before reassembling my system, I took a close-up photo of the chip and it's surrounding surface-mounted components (see below). No damage or missing resistors or capacitors that I can tell.

My best guess is that the board was damaged electrically somehow, by ignorance on my part. Could having a powered on PSU and a connected VCC Pin8 (with Raspberry Pi powered on) at the same time have caused such damage?

Before (ad)venturing into ordering a new Designare and flashing it, I would like to better understand the Z390 Designare's Winbond flashing procedure:

1. Are flashrom reads and writes only possible with PSU powered on? Or is it also possible to flash the chip with only VCC8 powering the chip?
2. Assuming PSU power is required (where VCC must be disconnected before powering on the PSU):
   Am I correct that VCC is only needed for visual confirmation of a correctly attached SOIC8-clip, by powering up the LEDs and allowing inspection of the color and brightness of those LEDs?
3. With PSU powered off and VCC connected, the Designare's LEDs should be red and slightly dim (without blinking) if the clip is properly attached.
4. If the LEDs' are bright yellow, or if they do not light up, the clip is not properly attached. The clip's attachment can then be adjusted by wiggling the clip (with VCC connected) until the LEDs have the proper color and brightness without blinking (red and slightly dim)?
5. After visual confirmation of proper clip attachment: With disconnected VCC and without PSU power, flashrom should partially recognize the chip (as in Supplemental Procedure) ... this and previous LED confirmation indicate proper SOIC8-clip attachment?
6. Then after testing proper clip attachment, VCC must be disconnected before powering on PSU. When switching PSU power ON, the motherboard leds will light up bright white, correct?
7. Then (and only then) all is set up and ready for reading or writing the chip.

Apologies for all my questions. I'm just very keen to understand and prevent another mishap with the next board.

 

[CaseySJ]

Hey @CaseySJ how are you doing? don't know if you remember me... I was having issues with my hackintosh rebooting... Turns out the problem was the i5 9600k, I already sent to Intel and they are refunding me.
So I'm about to buy a new processor and I'm between the i7 9700k and the i7 9700kf... Is the lack of a Igpu in the 9700kf a problem for the hackintosh compatibility?

On NewEgg at the moment both i7-9700K and KF are the same price. I would choose the i7-9700K without hesitation, although technically the 9700KF is still compatible.

 

[karambit]

@jeanvs,

If you require Hyper-threading, the i7 8700k have come down recently BUT are harder to find.
Since BIOS overclocking doesnt add much in the way of performance in MacOsx ~ Hackintosh, I would chose either Hyper-threading or more physical cores ie) the i9-9900k if it's within your budget?

 

[jeanvs]

On NewEgg at the moment both i7-9700K and KF are the same price. I would choose the i7-9700K without hesitation, although technically the 9700KF is still compatible.

I'm actually from Brazil... things here are weird right now (where isn't right?!)... The dollar is super expensive for us right now so importing is out of question not to mention that by importing I would not have the warranty... And also it's hard to find some versions like the non k, I mostly find the 9700k, 9700kf, 9700f with a huge difference between price tags

 

[karambit]

I'm actually from Brazil... things here are weird right now (where isn't right?!)... The dollar is super expensive for us right now so importing is out of question not to mention that by importing I would not have the warranty... And also it's hard to find some versions like the non k, I mostly find the 9700k, 9700kf, 9700f with a huge difference between price tags

Ahh, I see. So get the cheaper one as Casey Recommended

 

[gandem]

@scottkendall @gandem

Now, after a certain amount of time has elapsed I wanted to ask you again, as Power AIC Users, which FW would you recommend? As far as I can tell NVM43 is definitely a no-go. NVM23 was quite praised and NVM50 was said to give similar results, but I'm under the impression a crossflashed NVM33 does give even better results, am I correct? Or did I just read along the wrong posts? Cheers gentlemen

For me, the cross flashed NVM 33 has given me the best results. I have experienced some bugginess- every once in a while my caldigit TS3 dock is not detected. In this scenario, what I believe has happened is the GC-Titan Ridge has locked up. When this has happened I can disable TB in bios, save, exit, then re-enable TB in bios. Then all returns to normal. That seems to be an issue that is not happening as much as before when I initially flashed. In fact it has been two days since it’s happened. Maybe I’ve finally dialed in the right settings-time will tell. My only known limitation is I’m not able to pass dual monitors over TB. For me, NVM 33 is the only firmware that enables 1 monitor over thunderbolt. My second monitor remains plugged into gpu. Unmodified firmware allows both monitors to work over 1 TB port. Even my old TB1 GoFlex carriage works- just gives me an error after it’s ejected and then I pull the cable. Minor issue since I still use it as a bootable backup, which still works!