16" MBP GPU Panics

[gazwas]

There seems to be a lot of ‘disproving’ going on in this thread. I will find out tonight what Apple did to fix my machine but I am guessing it will be the exact same fix as @StolenPhone - a new Logicboard.

I don’t think we’re all disproving, but just doubtful this is hardware failure as all 16 MBP’s are effected. I have a freshly opened BTO 16 MBP that is a March 2020 build with the latest 10.15.3 OS. I ran the GB5 metal test on the Intel GPU and got a score of 2230. The test totally hung on the face detect test and console reported GPU unresponsive, hardware error and GPU restarts as @multidima above. The computer was basically frozen as this stage and very slow to respond. I think the test was finally cancelled and all other tests completed quickly.

The MBP kept running, no restarts, no crashes and no error messages.

Unless all the Intel CPU’s are defective (?) this has to be driver issues....... doesn’t it?

 

[visualinventor]

Well, after testing some tonight on my newly repaired MacBook Pro 16" - I think a little bit of what everyone has been saying is correct.

It's pretty much what @StolenPhone was alluding to.

1. GeekBench 5 pegging the Intel GPU is a way to test GPU kernel panics. Your machine will crash or become unresponsive while the test is running.

2. I think the GeekBanch 5 Face Detection test is too much for the internal Intel 630 GPU resulting in GPU restarts.

3. I think metal isn't currently working correctly under Catalina on the Intel 630 GPU. The OpenCL shouldn't be higher than the Metal score. Maybe the OS doesn't use the Intel GPU along with metal but instead uses OpenCL?

4. I think there is/was some issue with the early built MBP 16". I was shipped mine Nov. 25 2019. I've been a Mac user since 1990 - I've never had a machine go to repair 3 months after purchase.

And the final thoughts from the words of the Apple tech(s) that repaired my machine (thanks Apple). In my case it was a hardware logic board failure and an I/O board. Things are significantly speedy again.

 

[gazwas]

Well, after testing some tonight on my newly repaired MacBook Pro 16" - I think a little bit of what everyone has been saying is correct.

It's pretty much what @StolenPhone was alluding to.

1. GeekBench 5 pegging the Intel GPU is a way to test GPU kernel panics. Your machine will crash or become unresponsive while the test is running.

2. I think the GeekBanch 5 Face Detection test is too much for the internal Intel 630 GPU resulting in GPU restarts.

3. I think metal isn't currently working correctly under Catalina on the Intel 630 GPU. The OpenCL shouldn't be higher than the Metal score. Maybe the OS doesn't use the Intel GPU along with metal but instead uses OpenCL?

4. I think there is/was some issue with the early built MBP 16". I was shipped mine Nov. 25 2019. I've been a Mac user since 1990 - I've never had a machine go to repair 3 months after purchase.

And the final thoughts from the words of the Apple tech(s) that repaired my machine (thanks Apple). In my case it was a hardware logic board failure and an I/O board. Things are significantly speedy again.

View attachment 897715

Indeed, I think the issue of faulty LB’s is in pretty limited cases and all the panic in this thread (well for me just receiving a new 16 MBP) is most likely over software or driver issues.

All in all, the 16” MBP is a top draw machine. The screen is really accurate no matter what you read in these forums and the performance for a portable machine is stunning. My GB5 score for a freshly opened machine is 7060 which is eye popping.😃

 

[visualinventor]

My GB5 score for a freshly opened machine is 7060 which is eye popping.😃

Which GPU combo did you use for that score?

or was that CPU?

 

[gazwas]

Which GPU combo did you use for that score?

or was that CPU?

Sorry, should have said CPU.

I’m not convinced any GPU test on the 16” MBP’s AMD or Intel GPU’s are giving up their full potential until Apple improves the quality of their drivers

 

[Camarillo Brillo]

I almost went down the rabbit hole with this one and I stopped myself. My computer works great. I don’t care what geekbench says, or the errors it threw up in console. I don’t have the freezing and kernel panic issues people are talking about so I guess I’m fine. Actually since disabling the radeon gpu I’ve had the smoothest performance yet in logic which is what I bought the computer for. Very happy with how the iGPU is performing on its own for logic, I don’t know why logic wants to use the dGPU as it’s not a graphics intensive program.

Counting on a software fix for any igpu issues, and if it really is a faulty logic board I hope I can trust Apple to make it right. They’ve never done me wrong before. My plan is to keep an eye on this and if it becomes some defect they acknowledge or if my computer starts doing it then I’ll do something about it. I may look into it more after a couple more Catalina updates.

 

[Norbert Mikołajczyk]

I almost went down the rabbit hole with this one and I stopped myself. My computer works great. I don’t care what geekbench says, or the errors it threw up in console. I don’t have the freezing and kernel panic issues people are talking about so I guess I’m fine. Actually since disabling the radeon gpu I’ve had the smoothest performance yet in logic which is what I bought the computer for. Very happy with how the iGPU is performing on its own for logic, I don’t know why logic wants to use the dGPU as it’s not a graphics intensive program.

Counting on a software fix for any igpu issues, and if it really is a faulty logic board I hope I can trust Apple to make it right. They’ve never done me wrong before. My plan is to keep an eye on this and if it becomes some defect they acknowledge or if my computer starts doing it then I’ll do something about it. I may look into it more after a couple more Catalina updates.

Same here, no kernel panics, I love how this machine perform.

Maybe the app you are using wants to use metal or OpenCL to do some calculations.
I think the gpu compute power is for system wide usage, all apps could tap to dgpu via metal to make some advanced calculations to relieve cpu. I would allow logic to use the dgpu and see if the calculation time would go down.

 

[Camarillo Brillo]

The problem is when logic uses the dgpu it wants to use 20+ watts sometimes for no good reason, other times it’s only 2-4 watts. When radeon uses that much power things heat up really fast and the fans get noisy.
With the dgpu disabled I can run logic for hours and the fans won’t go over 1800 rpm or maybe 2k max. The cpu stays like 20 degrees cooler also. And I’ve noticed zero performance hit on the cpu by letting it use its integrated graphics. It actually seems more stable.

... but unfortunately this is only a solution if you’re not using an external monitor and you don’t need the higher graphic performance of the Radeon. For Logic Pro without an external monitor it’s perfect. The external monitor will always use the Radeon though.

 

[Salomonander]

The problem is when logic uses the dgpu it wants to use 20+ watts sometimes for no good reason, other times it’s only 2-4 watts. When radeon uses that much power things heat up really fast and the fans get noisy.
With the dgpu disabled I can run logic for hours and the fans won’t go over 1800 rpm or maybe 2k max. The cpu stays like 20 degrees cooler also. And I’ve noticed zero performance hit on the cpu by letting it use its integrated graphics. It actually seems more stable.

... but unfortunately this is only a solution if you’re not using an external monitor and you don’t need the higher graphic performance of the Radeon. For Logic Pro without an external monitor it’s perfect. The external monitor will always use the Radeon though.

hey that sounds great regarding logic. May i ask which processor you have?

 

[sdp77]

I have a brand new MBP 16", just picked it up yesterday morning. Last might, half way through playing Sid Meier's Railways, my MBP froze. Only way to change that was holding down the power key. When I restarted the computer, a report was ready to send to Apple. How do I retrieve this report (or the info in it) to see what happened?

 

[StolenPhone]

Yeah that’s a panic. Usually the text is in the dialogue box for sending to Apple. You can copy and paste from that, or of you sent it, you can open the Console app (it shows log files). In the left side of the Console app, select Log Reports. Look for a file called panic-full-2020-xx-xx-xxxxxx.xxxx.ips. The “x” characters are replaced with the month, day, and time the panic occurred. It’s a massive json file. But the panic error is with the panicString field which will tell you most of the panic. The macOSPanicString field has the interesting panic info that you can paste here.

 

[simonmet]

This concerns me, but I think there are several unanswered questions:

1. Have any 16” owners achieved normal Geekbench 5 Metal results (assumed to be above 5000) using the Intel UHD 630 without GPU kernel errors and hangs as described in this thread? If so, what is your CPU, OS version; which Radeon GPU did you get and can you send a link to the result? I suffer the problem, like apparently everyone else.
2. Secondly, I heard a claim in this thread or the one on Apple's Support forums that the current 10.15.4 beta solves the aforementioned issue in Geekbench. If so, can we get a confirmation on this? Isn't that something more people can test? If even one person claims this (as they have) then surely that's a point of evidence in favour of it being a driver issue. Another being that OpenCL isn't causing the same issue. I don't currently have a USB drive handy or access to decent internet, so I'd certainly appreciate it if someone else can install the beta to test.

I've heard of several people getting multiple logic board replacements that don't fix the issue, or give them other issues; so that's not something I want to rush into doing when everything else about my unit appears fine (or at least, within spec).

I've only had one kernel panic since receiving my unit over a month ago, and that was early on when I believe it was still on 10.15.2. It was a wake from sleep (while plugged in) panic where the unit restarted and I got the report to send to Apple (which of course I did). This alone is not enough to cause overwhelming concern, even though the console logs are reporting the same apparent hardware fault as everyone else here is getting.

People here are saying that the problem in Geekbench and its associated logs that indicate GPU issues and restarts is proof of a hardware fault and qualifies for a logic board replacement. If that's the case, it appears that the vast majority of 16" MacBook Pros are faulty and need to be repaired, in which case Apple is f****ed. I'm not ready to jump to that conclusion. Why, for instance, are 15" units with the same CPU apparently not affected? That's why I'm looking for further evidence to the two questions above: because many people seem to be reporting issues in the Geekbench test, with at least some people suggesting or arguing it indicates a hardware flaw. There needs be some further information on this before a bunch of people start trying to get their hardware fixed or replaced.

Finally, I'm aware that some people seem to be experiencing very frequent and constant kernel panics that may be a separate issue, but for now I want to focus on the issue in Geekbench and what that means or implies for existing users. It's obviously a concern that a simple benchmarking software is failing in this way.

 

[gazwas]

I’m confident it’s a software or driver issue highlighted by GB5 as all 16” MBP’s show the same results. The simple fact OpenCL results are not effected and Metal tests in GB4 reports no such errors points to software over hardware.

Additionally, as there seem to be no issues with 15”MBP’s that use the same Intel CPU/GPU I can oly assume it has something to do with the new AMD Navi drivers and the auto switching between GPU’s.

 

[StolenPhone]

Finally, I'm aware that some people seem to be experiencing very frequent and constant kernel panics that may be a separate issue, but for now I want to focus on the issue in Geekbench and what that means or implies for existing users. It's obviously a concern that a simple benchmarking software is failing in this way.

Yes, frequent panics are indeed a separate issue. It affects a batch of early MBPs that were produced in November and December.

The Geekbench issues that produce slowness (but do not panic consistently) is likely software. But only time will tell on the next release or two of Catalina.

 

[gazwas]

The Geekbench issues that produce slowness (but do not panic consistently) is likely software. But only time will tell on the next release or two of Catalina.

Apple have nearly 12 months of tweaking until RDNA 2 (Navi 2) arrives and considering they still don’t have the W5700X available for the Mac Pro which I assume is partly down to quality drivers I think its safe to say things will improve.

The 16” MBP was the first and still the only Mac using AMD’s newest Navi GPU’s so I imagine they are still ironing out the bumps. Apple only offer AMD GPU options and AMD have shown they are fully invested in these cards so Apple has no choice but to fix issues like this.

 

[visualinventor]

My personal assumption was that Metal 2 should be able to run on the Intel 630 GPU under Catalina. It does run on Metal 1 in older os’s. I’m not sure with the shift to metal 2 that the internal GPU can run it now.

 

[gazwas]

My personal assumption was that Metal 2 should be able to run on the Intel 630 GPU under Catalina. It does run on Metal 1 in older os’s. I’m not sure with the shift to metal 2 that the internal GPU can run it now.

But then the same Intel GPU on the 2019 15” MBP would have the problem too. No mentions of this on the most OCD Mac forum I know suggests its works fine (?).

It has to be the AMD (switcheroo) drivers or a bug in Geekbench 5 IMO.

 

[simonmet]

Yes, frequent panics are indeed a separate issue. It affects a batch of early MBPs that were produced in November and December.

The Geekbench issues that produce slowness (but do not panic consistently) is likely software. But only time will tell on the next release or two of Catalina.

My unit has a Manufacture Date (according to Coconut Battery) of 6th Jan, 2020; though I didn’t get the unit until 28 Jan. I got a BTO model but only upgraded the VRAM from the base i9.

I hope I’m clear of manufacturing problems, but like I said, I’ve only had the one kernel panic that I can remember. How late in December do you think there might’ve been batch problems?

Kind of wishing I didn’t even read this thread or download Geekbench. I’d have had no idea there was a problem and lived in ignorant bliss! I wonder if anyone has asked John Poole for his opinion.

I do think those people claiming the hang in Geekbenchis is definitely related to a hardware problem might have misinterpreted the Console logs, but I’m no expert on OS X so I don’t know.

 

[StolenPhone]

I hope I’m clear of manufacturing problems, but like I said, I’ve only had the one kernel panic that I can remember. How late in December do you think there might’ve been batch problems?

Kind of wishing I didn’t even read this thread or download Geekbench. I’d have had no idea there was a problem and lived in ignorant bliss! I wonder if anyone has asked John Poole for his opinion.

First you need to see the panic. The ones with the anomaly are getting “Powerplay failed to resume” or “Powerplay failed to initialize”. It will happen when it wakes from sleep or if you run something that needs to switch to the AMD, like a game, Adobe Illustrator, etc. it will start intermittently and slowly progress to more consistently.

the Gerkbench 5 metal intel GPU test faithfully tickles the bug and if you have it, it will panic with the above panic message.

if you don’t have that issue, then you shouldn’t be concerned.

 

[Camarillo Brillo]

I had one instance of my 16” crashing while asleep overnight, or upon trying to wake it. It was within the first week or two I owned it, and it may have been before I even updated Catalina the first time, I can’t remember if it was on .1 or .2. Didn’t save the logs so I’m not sure what they say, is there an easy way to look up something like that that happened 3 months ago? Sounds like more hassle than I’m willing to put in since everything works fine since then but I’ll keep a close eye on this issue.

It only happened once, and I assumed that one of the Catalina updates fixed it.

Mine was built in December.

 

[sdp77]

Yeah that’s a panic. Usually the text is in the dialogue box for sending to Apple. You can copy and paste from that, or of you sent it, you can open the Console app (it shows log files). In the left side of the Console app, select Log Reports. Look for a file called panic-full-2020-xx-xx-xxxxxx.xxxx.ips. The “x” characters are replaced with the month, day, and time the panic occurred. It’s a massive json file. But the panic error is with the panicString field which will tell you most of the panic. The macOSPanicString field has the interesting panic info that you can paste here.

Here is the macOSPanicString field:
"macOSPanicString" : "panic(cpu 0 caller 0xffffff8012e288b7): \"ATY,Boa::setPowerState(0xffffff86a7f96000 : 0xffffff7f963a1056, 2 -> 1) timed out after 45414 ms\"@\/BuildRoot\/Library\/Caches\/com.apple.xbs\/Sources\/xnu\/xnu-6153.81.5\/iokit\/Kernel\/IOServicePM.cpp:5302\nBacktrace (CPU 0), Frame : Return Address\n0xffffff874a53bb40 : 0xffffff801273bb2b \n0xffffff874a53bb90 : 0xffffff80128734d5 \n0xffffff874a53bbd0 : 0xffffff8012864f4e \n0xffffff874a53bc20 : 0xffffff80126e2a40 \n0xffffff874a53bc40 : 0xffffff801273b217 \n0xffffff874a53bd40 : 0xffffff801273b5fb \n0xffffff874a53bd90 : 0xffffff8012ed2aa9 \n0xffffff874a53be00 : 0xffffff8012e288b7 \n0xffffff874a53be50 : 0xffffff8012e28479 \n0xffffff874a53be60 : 0xffffff8012e3f82e \n0xffffff874a53bea0 : 0xffffff8012e27228 \n0xffffff874a53bec0 : 0xffffff801277d7e5 \n0xffffff874a53bf40 : 0xffffff801277d311 \n0xffffff874a53bfa0 : 0xffffff80126e213e \n\nBSD process name corresponding to current thread: kernel_task\nBoot args: chunklist-security-epoch=0 -chunklist-no-rev2-dev\n\nMac OS version:\n19D76\n\nKernel version:\nDarwin Kernel Version 19.3.0: Thu Jan 9 20:58:23 PST 2020; root:xnu-6153.81.5~1\/RELEASE_X86_64\nKernel UUID: A8DDE75C-CD97-3C37-B35D-1070CC50D2CE\nKernel slide: 0x0000000012400000\nKernel text base: 0xffffff8012600000\n__HIB text base: 0xffffff8012500000\nSystem model name: MacBookPro16,1 (Mac-E1008331FDC96864)\nSystem shutdown begun: NO\n\nSystem uptime in nanoseconds: 17112866573527\nlast loaded kext at 15190092297743: >usb.IOUSBHostHIDDevice\t1.2 (addr 0xffffff7f967c0000, size 45056)\nlast unloaded kext at 16058764908499: >!UAudio\t320.49 (addr 0xffffff7f99ac9000, size 434176)\nloaded kexts:\n@filesystems.afpfs\t11.2\n@nke.asp-tcp\t8.1\n@kext.AMDRadeonX6000\t3.0.5\n@kext.AMDRadeonServiceManager\t3.0.5\n>!AGraphicsDevicePolicy\t4.7.2\n@fileutil\t20.036.15\n@AGDCPluginDisplayMetrics\t4.7.2\n>!AHV\t1\n|IOUserEthernet\t1.0.1\n|IO!BSerialManager\t7.0.3f5\n>AGPM\t111.4.2\n>!APlatformEnabler\t2.7.0d0\n>X86PlatformShim\t1.0.0\n>!AUpstreamUserClient\t3.6.8\n>pmtelemetry\t1\n>AGDCBacklightControl\t4.7.2\n>!A!IKBLGraphics\t14.0.4\n@Dont_Steal_Mac_OS_X\t7.0.0\n>!AThunderboltIP\t3.1.3\n>BridgeAudioCommunication\t6.66\n>!AHIDALSService\t1\n>!ATopCaseHIDEventDriver\t3430.1\n>!AMCCSControl\t1.13\n>!AMuxControl2\t4.7.2\n>!ABridgeAudio!C\t6.66\n>!AGFXHDA\t100.1.424\n>!A!IPCHPMC\t2.0.1\n>!A!ICFLGraphicsFramebuffer\t14.0.4\n>!A!ISlowAdaptiveClocking\t4.0.0\n>!AAVEBridge\t6.1\n@filesystems.autofs\t3.0\n>BCMWLANFirmware4355.Hashstore\t1\n>BCMWLANFirmware4364.Hashstore\t1\n>BCMWLANFirmware4377.Hashstore\t1\n>!ABCMWLANBusInterfacePCIe\t1\n@filesystems.hfs.kext\t522.0.9\n@BootCache\t40\n@!AFSCompression.!AFSCompressionTypeDataless\t1.0.0d1\n@!AFSCompression.!AFSCompressionTypeZlib\t1.0.0\n>!AVirtIO\t1.0\n@filesystems.apfs\t1412.81.1\n@private.KextAudit\t1.0\n>!ASmartBatteryManager\t161.0.0\n>!AACPIButtons\t6.1\n>!ASMBIOS\t2.1\n>!AACPIEC\t6.1\n>!AAPIC\t1.7\n$!AImage4\t1\n@nke.applicationfirewall\t303\n$TMSafetyNet\t8\n@!ASystemPolicy\t2.0.0\n|EndpointSecurity\t1\n$SecureRemotePassword\t1.0\n@kext.AMDRadeonX6100HWLibs\t1.0\n@kext.AMDRadeonX6000HWServices\t3.0.5\n|IOAVB!F\t800.17\n>!ASSE\t1.0\n>!ABacklightExpert\t1.1.0\n@kext.AMDRadeonX6000Framebuffer\t3.0.5\n@!AGPUWrangler\t4.7.2\n>!AActuatorDriver\t3430.1\n>!AHIDKeyboard\t209\n>X86PlatformPlugin\t1.0.0\n>!AHS!BDriver\t3430.1\n>IO!BHIDDriver\t7.0.3f5\n>!AMultitouchDriver\t3430.1\n>!AInputDeviceSupport\t3430.1\n>!ASMBus!C\t1.0.18d1\n@kext.AMDSupport\t3.0.5\n>!AGraphicsControl\t4.7.2\n|IONDRVSupport\t569.4\n|IO!BHost!CUARTTransport\t7.0.3f5\n|IO!BHost!CTransport\t7.0.3f5\n>!A!ILpssUARTv1\t3.0.60\n>!A!ILpssUARTCommon\t3.0.60\n>!AOnboardSerial\t1.0\n>IOPlatformPlugin!F\t6.0.0d8\n@!AGraphicsDeviceControl\t4.7.2\n|IOAccelerator!F2\t438.3.1\n|IOGraphics!F\t569.4\n|IOSlowAdaptiveClocking!F\t1.0.0\n@plugin.IOgPTPPlugin\t810.1\n|IOEthernetAVB!C\t1.1.0\n@kext.triggers\t1.0\n>usb.cdc.ncm\t5.0.0\n>usb.cdc\t5.0.0\n>usb.networking\t5.0.0\n>usb.!UHostCompositeDevice\t1.2\n>!ABCMWLANCore\t1.0.0\n>mDNSOffloadUserClient\t1.0.1b8\n>IOImageLoader\t1.0.0\n|IO80211!FV2\t1200.12.2b1\n>corecapture\t1.0.4\n|IOSkywalk!F\t1\n|IOSurface\t269.6\n@filesystems.hfs.encodings.kext\t1\n|IOAudio!F\t300.2\n@vecLib.kext\t1.2.0\n|IOSerial!F\t11\n>usb.!UVHCIBCE\t1.2\n>usb.!UVHCI\t1.2\n>usb.!UVHCICommonBCE\t1.0\n>usb.!UVHCICommon\t1.0\n>!AEffaceableNOR\t1.0\n|IOBufferCopy!C\t1.1.0\n|IOBufferCopyEngine!F\t1\n|IONVMe!F\t2.1.0\n>!AThunderboltPCIDownAdapter\t2.5.4\n>!AThunderboltDPInAdapter\t6.2.5\n>!AThunderboltDPAdapter!F\t6.2.5\n>!AHPM\t3.4.4\n>!A!ILpssI2C!C\t3.0.60\n>!A!ILpssDmac\t3.0.60\n>!A!ILpssI2C\t3.0.60\n|IOUSB!F\t900.4.2\n>!AThunderboltNHI\t5.8.6\n|IOThunderbolt!F\t7.6.0\n>usb.!UXHCIPCI\t1.2\n>usb.!UXHCI\t1.2\n>!AEFINVRAM\t2.1\n>!AEFIRuntime\t2.1\n>!ASMCRTC\t1.0\n|IOSMBus!F\t1.1\n|IOHID!F\t2.0.0\n$quarantine\t4\n$sandbox\t300.0\n@kext.!AMatch\t1.0.0d1\n>!AKeyStore\t2\n>!UTDM\t489.80.2\n|IOSCSIBlockCommandsDevice\t422.0.2\n>!ACredentialManager\t1.0\n>KernelRelayHost\t1\n>!ASEPManager\t1.0.1\n>IOSlaveProcessor\t1\n>!AFDEKeyStore\t28.30\n>!AEffaceable!S\t1.0\n>!AMobileFileIntegrity\t1.0.5\n@kext.CoreTrust\t1\n|CoreAnalytics!F\t1\n|IOTimeSync!F\t810.1\n|IONetworking!F\t3.4\n>DiskImages\t493.0.0\n|IO!B!F\t7.0.3f5\n|IO!BPacketLogger\t7.0.3f5\n|IOUSBMass!SDriver\t157.40.7\n|IOSCSIArchitectureModel!F\t422.0.2\n|IO!S!F\t2.1\n|IOUSBHost!F\t1.2\n>usb.!UCommon\t1.0\n>!UHostMergeProperties\t1.2\n>!ABusPower!C\t1.0\n|IOReport!F\t47\n>!AACPIPlatform\t6.1\n>!ASMC\t3.1.9\n>watchdog\t1\n|IOPCI!F\t2.9\n|IOACPI!F\t1.4\n@kec.pthread\t1\n@kec.corecrypto\t1.0\n@kec.Libm\t1\n",

I hope you can make something out of it. Please let me know what you think.

 

[simonmet]

@sdp77 and others:

Can I suggest instead of posting long log files in the message field like that above that you save it as a plain text file (or attach the original log file) and upload that instead? Look for the “Attach files” button.

 

[simonmet]

First you need to see the panic. The ones with the anomaly are getting “Powerplay failed to resume” or “Powerplay failed to initialize”. It will happen when it wakes from sleep or if you run something that needs to switch to the AMD, like a game, Adobe Illustrator, etc. it will start intermittently and slowly progress to more consistently.

the Gerkbench 5 metal intel GPU test faithfully tickles the bug and if you have it, it will panic with the above panic message.

if you don’t have that issue, then you shouldn’t be concerned.

I’ve had difficulty locating the panic file, but I’m quite sure that since it otherwise matched the description of “when it wakes from sleep” that it would include the “Powerplay failed to initialise” message.

So are you saying that, because I’ve had one panic message that induced a forced restart in my 6 weeks of ownership; and none in the past month, that my hardware is faulty; that they‘ll eventually increase in frequency (including while using apps) to the point that the computer becomes unusable? Because that seems to be what you’re saying. If not, at what point of kernel panic frequency are you suggesting this outcome will occur?

Also, is there a difference between a panic that crashes the computer and one that doesn’t (apart from the stated distinction)? I get many “fault” log files that include the word “panic” when running the Geekbench Metal test, but you also suggest that the slow Geekbench results don’t necessarily indicate a hardware flaw. In other words, your messaging seems to be (as I see it) contradictory.

There’s a lot of conjecture in this thread, and people making broad or sweeping claims for which they can’t entirely be certain: eg. “if you get this panic or this string in a log your hardware is faulty” and “it will start intermittently and slowly progress to more consistently”. Apple themselves are sending mixed messages too (as they often do with hardware-related issues) by both agreeing to replace logic boards and telling other users it’s not a hardware fault, for seemingly the same problem!

It would be helpful if people can make clear the distinction between what they’ve found or seen as objective facts and what they think or suspect it means including predictions, and why. I work in science and clear communication is so important yet so commonly poor and it really frustrates! This thread is one giant example of that.

 

[StolenPhone]

I’ve had difficulty locating the panic file, but I’m quite sure that since it otherwise matched the description of “when it wakes from sleep” that it would include the “Powerplay failed to initialise” message.

So are you saying that, because I’ve had one panic message that induced a forced restart in my 6 weeks of ownership; and none in the past month, that my hardware is faulty; that they‘ll eventually increase in frequency (including while using apps) to the point that the computer becomes unusable? Because that seems to be what you’re saying. If not, at what point of kernel panic frequency are you suggesting this outcome will occur?

Also, is there a difference between a panic that crashes the computer and one that doesn’t (apart from the stated distinction)? I get many “fault” log files that include the word “panic” when running the Geekbench Metal test, but you also suggest that the slow Geekbench results don’t necessarily indicate a hardware flaw. In other words, your messaging seems to be (as I see it) contradictory.

There’s a lot of conjecture in this thread, and people making broad or sweeping claims for which they can’t entirely be certain: eg. “if you get this panic or this string in a log your hardware is faulty” and “it will start intermittently and slowly progress to more consistently”. Apple themselves are sending mixed messages too (as they often do with hardware-related issues) by both agreeing to replace logic boards and telling other users it’s not a hardware fault, for seemingly the same problem!

It would be helpful if people can make clear the distinction between what they’ve found or seen as objective facts and what they think or suspect it means including predictions, and why. I work in science and clear communication is so important yet so commonly poor and it really frustrates! This thread is one giant example of that.

I have been very clear on the problem, and just about everyone who has encountered the panic as described have ended up requiring a new logic board. Thats not a broad claim and I AM certain as it was confirmed by Apple and in fact it was Apple who stated about the batch of faulty logic boards. That wasn't my conjecture.. You may wish to reread the thread once again to clarify what has been stated.

I cannot state anything different other than the panic messages. Unless you have the message, you cannot guarantee you have or not have the problem. You need the exact message to know.

Getting 1 panic, especially not having the log should not cause you alarm at this stage. My new laptop got a "BAD MAGIC! (flag set in iBoot panic header), no macOS panic log available" after it awoke once. It is not the same problem and it hasn't happened since. That has been known as a T2 chip issue that many people have experienced and its mostly considered a software bug. I'm not concerned at this point.

If you run Geekbench 5 and get the panics as described in this thread faithfully, then you have the anomaly and need a fix. If you do not, then you have nothing to worry about at this time.

As for frequency, the bug starts out once or twice a week, then slowly becomes more frequent, as in mutliplte times a day, and is reproduced most of the time on awaking from sleep, and 100% using Geekbench.
[automerge]1583764573[/automerge]

Here is the macOSPanicString field:
"macOSPanicString" : "panic(cpu 0 caller 0xffffff8012e288b7): \"ATY,Boa::setPowerState(0xffffff86a7f96000 : 0xffffff7f963a1056, 2 -> 1) timed out after 45414 ms\"

I hope you can make something out of it. Please let me know what you think.

I just needed that part

Thats not the same issue as the anomoly. Keep an eye on it to see if it happens more often.
[automerge]1583764848[/automerge]

I had one instance of my 16” crashing while asleep overnight, or upon trying to wake it. It was within the first week or two I owned it, and it may have been before I even updated Catalina the first time, I can’t remember if it was on .1 or .2. Didn’t save the logs so I’m not sure what they say, is there an easy way to look up something like that that happened 3 months ago? Sounds like more hassle than I’m willing to put in since everything works fine since then but I’ll keep a close eye on this issue.

It only happened once, and I assumed that one of the Catalina updates fixed it.

Mine was built in December.

Yeah I wouldn't worry if it happened once. Keep an eye on it if it happens more often and watch for the evil panic message as described in this thread.

 

[gazwas]

If you run Geekbench 5 and get the panics as described in this thread faithfully, then you have the anomaly and need a fix. If you do not, then you have nothing to worry about at this time.

I think here lies the problem with this discussion in that there is no clear message from you as to what the pattern to the 'anomoly' is and people are wrongly assuming GB5 reporting errors is a sign of logic board failures. Or at least that is how I read your reply I've quoted above.

Simply getting hardware errors and restart error messages in console is not an indication of people having your original issue and needing a new logic board. Currently with the 16" MBP and OS 10.15.3 when running GB5's metal compute test on the Intel 630 GPU, console reports these errors on all machines and is probably down to software/drivers.

As I think (?) I understand it, your original 'anomoly' was computer crashes and restarts and it was found GB5's metal test could replicate conditions to encourage this behaviour forcing a crash and total system restart. Upon restart, crash log's reported GPU panic's and Apple advised you had a faulty logic Board.

So.......

• GB5 Metal test of Intel GPU - Slow to complete, face detact causes system hangs, low final score, console reoporting things like hardware errors, restarts, time outs etc. This behaviour is NORMAL (if normal means not working very well).
• GB5 Metal test on Intel GPU - Computer crashes, restarts and crash log reports show 'Panic' errors then you possibly have an issue.

Does this sound correct?