Help: MBP13” 2018 freezes when connected to Gigabyte Gaming Box RX580
- macOS Mojave 10.14.3 (18D42)
- MacBook Pro 13-inch, Mid-2018 (i5-8259U / 16GB RAM / 256GB SSD)
- Gigabyte RX 580 Gaming Box (GV-RX580IXEB-8GD)
- External 1080p monitor (HDMI) + Keyboard + Mouse <-> Gaming Box <-> MacBook Pro
The first incident was blue square-ish pixels all over my external display while watching a video. My mouse cursor started lagging, and it went completely unresponsive after a minute. There was nothing I could do short of powering off the eGPU, or unplug it. So I unplugged the eGPU from my MBP and everything went back to normal. Plugged it back in, and that's where the issues started happening: the same behavior over and over again. Sometimes it works for awhile (<5minutes everytime), but starts crapping out as described. Other times, the MBP freezes the moment it's plugged in. In some cases there was also a kernel panic right after I unplugged the eGPU after it freezes. I've also noticed the fans on my MBP whirling up (making a woosh sound) right when this happens.
This happens with both internal and external displays. Although there weren't any random bunch of pixels on the internal display, and it tends to work longer when I don't connect it to an external display. As far as I can tell, this always happens a few minutes after/when the eGPU comes under load.
There are a bunch of relevant console log messages, e.g.:
fault 08:36:46.463184 +0800 kernel void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
^ seems to occur the most. There are a few other interesting ones:
default 08:36:53.047064 +0800 kernel [196:0:0] ** AMDRadeonX4000_AMDEllesmereGraphicsAccelerator Device in slot: SLOT--1 ** default 08:36:53.048326 +0800 kernel Trying to restart GPU (Radeon RX 580)... default 08:36:53.050704 +0800 kernel AMD Error: default 08:36:53.050708 +0800 kernel CAILPerEngineReset failed default 08:36:53.051301 +0800 kernel AMD Cail: CAIL detects hungBlocks: 0x00028007 default 08:36:53.051707 +0800 kernel AMD Cail: soft reset done - blockResetResult=0x00008000 fault 08:37:20.582377 +0800 kernel virtual void IOAccelEventMachineFast2::checkGPUProgress() - Signaling hardware error on channel 0.. fault 08:37:20.582398 +0800 kernel void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartSignaled stampIdx=0 type=2 prevType=0 numStamps=20 fault 08:37:20.582399 +0800 kernel void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartEnqueued stampIdx=0 type=2 fault 08:37:20.582419 +0800 kernel virtual void IOAccelEventMachineFast2::checkGPUProgress() - Signaling hardware error on channel 5.. fault 08:37:20.582422 +0800 kernel void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartSignaled stampIdx=5 type=2 prevType=0 numStamps=20 fault 08:37:20.582423 +0800 kernel void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartEnqueued stampIdx=5 type=2 fault 08:37:20.582424 +0800 kernel virtual void IOAccelEventMachineFast2::checkGPUProgress() - Signaling hardware error on channel 6.. fault 08:37:20.582426 +0800 kernel void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartSignaled stampIdx=6 type=2 prevType=0 numStamps=20 fault 08:37:20.582428 +0800 kernel void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartEnqueued stampIdx=6 type=2 fault 08:37:20.582433 +0800 kernel void IOAccelEventMachine2::hardwareErrorEvent(): setting restart type to 2 (channel 0) fault 08:37:20.582434 +0800 kernel void IOAccelEventMachine2::hardwareErrorEvent(): GPURestartDequeued stampIdx=0 type=2
- Relevant console logs: https://pastebin.com/raw/288Luht5
- Kernel panic: https://pastebin.com/raw/qYM6L6ep (note: truncated snapshot data/etc. - they were too big for pastebin)
- GPU Reset: https://pastebin.com/raw/qE0TKFrM
I've tried resetting the SMC/PRAM, using different TB3 ports on my MBP, rebooting with eGPU connected (instead of hotplugging it). I've yet to try a new TB3 cable, or reinstalling macOS - I plan to do both soon, but from the looks of it, it sounds like a hardware issue to me.
I would really appreciate it if anyone could chime in on this, thanks!
Update: I've reinstalled macOS, and managed to get a brand new TB3 cable. Neither works - same behaviour of freezing and subsequent kernel panic when disconnecting. Does anyone have any ideas?
My MacBook Pro 2017 + Razer Core v2 + RX 580.
Exactly same issue here while playing Aerofly FS 2 with Metal backend used. And I have communicated with the game developer for a month. we suspect it's AMD driver's or hardware issue and no way to fix it. I am considering to buy a GTX 1080 card instead.
I’m experiencing this now (months later) - did this ever resolve for you? I can find this error reported in a few places online but I haven’t been able to find someone who returned to their post with a follow up. Any luck for you?
No, seems even worse in Catalina, Os won’t boot up with RX580 eGPU plugged in, and eGPU keep restarting while playing heavy games. Apple just keep saying they can’t reproduce it with 570x what... but I have reproduced it on 17, 19 MBPs. 🙁
Similar problem here: Sometimes (I'm suspecting any sort of video playback) makes macOS freeze in a certain pattern: 5 seconds freeze, then 1 second unfreeze and so on and this repeats for 5 minutes or so until the MacBook shuts off. The mouse stays unfrozen the whole time but any other rendering is frozen. For example I can type something on my keyboard and don't see anything first, and then see it appear 5 seconds later.
I'm suspecting early driver problems in macOS Catalina 10.15.1
The only workaround for me so far: As soon as this problem starts happening I pull out the TB3 cable from the MacBook which will give me the "egpu incorrectly ejected" error but everything recovers after a minute or so and I then reconnect the TB3 cable and everything works as expected.