Fans not spinning, card overheating, hair being torn out. Please help! :c
So, today I learned a hard lesson. Silly ole' me. I assumed that Thunderbolt 3 is just Thunderbolt 3 and if you buy an enclosure it just works, not that some minor V1 vs V2 revision could literally mean compatible vs not compatible. Crazy right? The fact that you have to dig to even see specific V1 incompatibility on Google also didn't help. Sigh.
Anyway, I encountered a really good deal on a Razer Core V1 + RX580 and I hopped on it to use with my MacBook Pro 2017 (4 TB3/Touchbar). I got it home and hooked it up to be sorely disappointed but fortunately the good people on this forum have made a few solutions to this Ti82 incompatibility issue. So I set up Purge Wrangler and viola, it worked! Sick right? Wrong. Now, despite functioning seemingly natively inside macOS, for some reason the fans on my card stop as soon as I log in and the OS initializes and they wont kick back on no matter what - meaning my card shoots straight to 100 degrees plus immediately and throttles and caps itself at 300MHz while taking a beating. It's consistently going even above 100 into hardware damaging territory.
I don't get it. It's not the card - I have two RX580s now from different manufactures and the behavior is identical on both. The enclosure seems to function perfectly and I don't see anything disconnected inside or anything like that. When I plug it in the fans kick up normally and then turn back off (the GPU fans I mean) almost immediately after it's recognized in macOS. I'm pretty well versed in computer hardware, I've been buried in this for 6 hours now and I haven't made an inch of progress even after trying several similar tools with several clean installs thereof.
I'm running 10.15 Catalina - which has honestly been a buggy mess from the get-go. The only reason I upgraded is I wanted SideCar and it's honestly not even worth it. I don't own a Time Machine unfortunately, an attempted downgrade is probably my next step if no one can help.
You guys are geniuses though so please, help me figure this out!
tl;dr - Core V1 on 2017 13" MBP 10.15, patched with Purge Wrangler, now recognized in macOS but GPU fans aren't spinning and card is locked at 300MHz and overheating
I assume all power cables are connected to the GPU - only bringing it up because in rare cases some people did not plug in all cables :p
I presume the issue then is that fans are not spinning up on high GPU load? In terms of just logging into macOS, IIRC, my MSI RX 480 used to stop spinning as well (silent mode), and so does my 5700 XT which I am currently using.
Things to try (test GPU after each step):
- Reset NVRAM and SMC. Note that resetting NVRAM re-enables SIP - so disable it again since you are using purge-wrangler.
- Reinstall macOS Catalina on top of your existing installation and test. This will disable Ti82 patch - rerun.
- Install macOS Mojave on a new APFS volume (to avoid erasing your existing volume) for testing: https://support.apple.com/en-us/HT208891
- Test the GPUs in different system(s) - someone else's Mac, via Bootcamp, or in a PC.
What are you using to check GPU temperatures?
Usually with macOS, we don't expect issues such as these given the static nature of OS-provided drivers. Hopefully the above troubleshooting steps are helpful.
@user99999 I ran multiple cards in my V1 and never had this issue and dont recall any mention of "design defects" which could cause this.
@sampsun great post though, refreshing to see well written and explained help posts rather than... "hElp! mY eGPu dOesn't wOrk wIth usb3, wIll pAy £5 for hElp!".
Thanks for taking the time to reply my friend. To answer your question, yes indeed, that's what I mean. They don't spin up under load. The cables are definitely all plugged in. The new RX580 I got is a Nitro + and uses 8pin + 6pin. I've tried switching the connectors around, just running the 8pin, etc. My original 580 that I already owned is just single 8pin. I also tried switching the connector on that. The behavior is identical despite any of those changes. I know what you mean about them not spinning up until they need to on newer cards, this is something outside of that. I would be idling well above 70 degrees sometimes, especially after trying to troubleshoot for several hours. In my opinion it seems like it's something driver related? Software related in some sense at least? Sigh, I don't know.
1) I tried both of these last night, twice actually. Once just normally - reset, re-disable SIP, proceed - and another time trying to delete purge-wrangler first and reinstall it after. Each time I deleted purge-wrangler after the first two tries I booted in shell mode and ran the command for the recovery that does a complete clean wipe, and I even started using that kext verification tool that's provided with automate-eGPU EFI in between attempts to really make sure I was installing cleanly each time - which I was.
4) I retried both cards inside of a PC and they both seem to still work perfectly normally even under load. I don't have access to another TB3 Mac (if I got a TB2 adapter I could try my old MacBook I suppose). As for bootcamp, I was really, really hoping this solution would work in macOS as my workflow is much more streamlined on it. I have a really high spec gaming PC that I never use and I was really hoping with this solution I could just do some occasional gaming on my MBP and ideally let go of my high-powered rig. That PC has a 1080ti in it, I was considering seeing how that acts inside the eGPU, but it's inside of a waterblock and I definitely didn't feel like messing with all that last night. On top of that, throughout my research I’ve realized that clearly bootcamp eGPU support is hugely complicated too, which is honestly ridiculous.
2 & 3 ) I haven't tried either of these and that's what I'll do next. Are there any widespread issues like this with Catalina? Anyone else saying it broke their purge-wrangler or anything that I just don't see? As I said before, my experience has been particularly buggy in a bunch of different ways. I fear maybe I'll just need to do a clean install anyway which I'm naturally dreading, especially if I don't know it'll fix the problem.
- I've used several different applications. I noticed something was up when I was getting 30k scores in Geekbench. Using the integrated hardware monitor I can see that utilization is indeed jumping up to 100%, so I used iStat and some other similar program which both showed the same insane temps and clock speeds being stuck at 300MHz - which makes perfect sense if the card isn't cooling itself. Is there a better solution out there for monitoring hardware on macOS?
- Exactly, that makes perfect sense and I came into this thinking the exact same way. Sigh. The plan was to sample the eGPU life with this and then upgrade to a 5700xt and let all my other hardware go. It really sucks because if it wasn't throttling it would be working perfectly and seems to be really seamless and awesome.
@User99999 I've seen some stuff online about the USB/ethernet being easily bandwidth capped, but not much outside of that. As a side note, the ethernet doesn't work for me on my MBP, the USB ports (seemingly) do but at restricted speeds.
@Eightarmedpet Thanks man, I try. I want to illustrate that I already do have some basic troubleshooting knowledge in a general sense but I truly just don't understand what's going on here. In Windows it would seem like a textbook bad driver - simple fix, reinstall drivers - but in macOS?? I don't get it.
Sorry I can't be any more help, the only thing I would suggest (although don't expect it to work) would be a fresh macOS instal as it does sound like its driver related.
great post though, refreshing to see well written and explained help posts rather than... "hElp! mY eGPu dOesn't wOrk wIth usb3, wIll pAy £5 for hElp!"
@eightarmedpet indeed, very refreshing to see.
@sampsun thanks for the detailed troubleshooting steps! So the cards themselves aren’t a problem, leaving most likely macOS as the culprit (which, really we don’t get to say often). Let us know how steps 2 and 3 go. To answer your questions, iStat is a good option to monitor temps, there are no widespread issues with purge-wrangler itself (recently a new patch was needed for 10.15.1 beta for Thunderbolt 1 and 2 is all - unrelated here), and yes, it does seem like a driver issue. I assume you are not using a custom BIOS? Some cards (such as XFX) have issues on macOS even though they are recommended by Apple. Usually an installation over an existing one overwrites all system components and drivers in macOS.
Bootcamp setup has become very complicated (system-dependent) with the recent Catalina firmware and Windows updates.
Well guys, I figured it out, but this is no cause for celebration, hahaha.
Here’s what happened :
I did a clean over-install of macOS - no change. Everything was exactly the same. Next, I created a second volume and did a parallel install - same thing, no change whatsoever.
Not looking good. I go for the full wipe and clean install and grit my teeth every moment of the way. I have all my documents backed up, sure, but all the extensive configurations, cache of my whole web drive and 80GB worth of photos, all of it - gone. Ugh! A few tears later the install finishes. I immediately set up the Mac, reboot and disable protection, reboot again and install pure-wrangler immediately and clean - no change whatsoever. How frustrating.
I’ve decided it has to be Catalina so I start gearing up for a downgrade. The thumb drive is ready to go finally, but before that fateful reboot I decide to try another benchmark, why not right? So I fire up LuxMark, for a few minutes nothing changes, and then bam - fan spin! I look at iStat and my heart drops - 126 degrees!? I panic and figure this is some sort of last-stitched safety feature of the card to prevent irreversible damage. Then though, I notice something. Something I had noticed before but just didn’t connect the dots for whatever reason. iStat is reporting my CPU temps at over 100 degrees also? Impossible. I install power gadget and sure enough I’m nowhere near that.
So wait... it can’t be.. does iStat default to FAHRENHEIT? .... IT DOES? Wow. What a silly thing for a program to do. Okay.. so.. then my GPU is actually functioning.. normally? Those are just temperatures though, what’s with the seemingly locked clocks? The super low GeekBench scores? Hm. Wait, GB5 is pretty new I think... are these score databases all based on GB4 and just don’t make that obvious?
Yes. Yes they are.
That gentlemen is the story of how I spent 12+ hours and installed macOS 3 times to troubleshoot a GPU that was working perfectly the entire time.
Thank you and goodnight.
@sampsun I'm sure you don't think its worth it but I have thoroughly enjoyed reading your posts so maybe take some comfort in that? Good to hear its all working fine now!
So wait... it can’t be.. does iStat default to FAHRENHEIT? .... IT DOES? Wow. What a silly thing for a program to do. Okay..
Shit. I’ll be remembering this for the rest of my life.