- Joined
- Feb 3, 2014
- Messages
- 6
- Motherboard
- Supermicro
- CPU
- Intel
- Graphics
- Nvidia
Hi
I'm trying to fit in my two E5 v2 2687W processors, in lieu of my E5 v2 2660.
These processors have never been sold by Apple, their TDP is too high.
They work fine on an Asus motherboard, very very fine (giacomoleopardo has done an amazing tutorial); but they don't on a SuperMicro motherboard.
AppleACPICPU panicks, with the message "Should have 17 threads, but only 17 for Die 1".
Apparently, the processor presents itself with offsets OS X is not expecting.
There are lots of threads mentioning the problem :
https://www.tonymacx86.com/threads/...-2680v2-gtx770-2x-ssd-32gb-ram-10-9-1.123543/
https://www.tonymacx86.com/threads/...e-xeon-e5-2687w-v2-gtx-660.127138/#post780291
The real solution looks like DSDT patching. The principle is to remove processor (read "thread") device 0, and also the one that are above the real number of threads.
It's very well explained here on the thread from Giacomoleopardo.
I did try DSDT patching, exactly as described, but to no avail (DSDT_Giacomoleopardo_style.dsl is attached)
I tried other things, such as putting a return (Zero) line at the very start of the Method (_STA, 0, NotSerialized) of the Processors that I wanted to hide.
The kernel panic has stayed exactly the same. I know DSDT patching is operating because I can crash the system much earlier if I use a funky truncated DSDT, and debug.log confirms patching is working ("DSDT found in Clover volume OEM folder: EFI\CLOVER\ACPI\patched\DSDT.aml")
I also tried non DSDT options :
- Changing the cores and hyperthreading options in the bios --> the message varies accordingly.
Eg.if I enable only 6 cores on each of the processor : "Should have 12 threads, but only found 24", if in addition I disable hyper-threading : "Should have 6 threads, but only found 12"
- playing with options numcpus, nullCpuPowerManagement.
I compiled a SSDT with ssdtPRGen (attached) (though I'm not sure it's concerned about the problem, which seems more relevant to DSDT).
I looked into the log, and into Clover Source code.
I found that Clover is aware things are going wrong. In AcpiPatcher.c, it warns :
if (ApicCPUNum > 16) {
DBG("Out of control with CPU numbers\n");
break;
I played around quite a bit with Clover source. I managed to get the message to change (was expecting 16 threads, only found 32 at die 3). But while writing this post, I rolled back to an unmodified clover, the message went back to (expecting 16, found 17).
My questions :
- can somebody suggest a way to fix this with DSDT ?
- can somebody suggest modifications to AcpiPatcher.c to make this work?
cheers!
I'm trying to fit in my two E5 v2 2687W processors, in lieu of my E5 v2 2660.
These processors have never been sold by Apple, their TDP is too high.
They work fine on an Asus motherboard, very very fine (giacomoleopardo has done an amazing tutorial); but they don't on a SuperMicro motherboard.
AppleACPICPU panicks, with the message "Should have 17 threads, but only 17 for Die 1".
Apparently, the processor presents itself with offsets OS X is not expecting.
There are lots of threads mentioning the problem :
https://www.tonymacx86.com/threads/...-2680v2-gtx770-2x-ssd-32gb-ram-10-9-1.123543/
https://www.tonymacx86.com/threads/...e-xeon-e5-2687w-v2-gtx-660.127138/#post780291
The real solution looks like DSDT patching. The principle is to remove processor (read "thread") device 0, and also the one that are above the real number of threads.
It's very well explained here on the thread from Giacomoleopardo.
I did try DSDT patching, exactly as described, but to no avail (DSDT_Giacomoleopardo_style.dsl is attached)
I tried other things, such as putting a return (Zero) line at the very start of the Method (_STA, 0, NotSerialized) of the Processors that I wanted to hide.
The kernel panic has stayed exactly the same. I know DSDT patching is operating because I can crash the system much earlier if I use a funky truncated DSDT, and debug.log confirms patching is working ("DSDT found in Clover volume OEM folder: EFI\CLOVER\ACPI\patched\DSDT.aml")
I also tried non DSDT options :
- Changing the cores and hyperthreading options in the bios --> the message varies accordingly.
Eg.if I enable only 6 cores on each of the processor : "Should have 12 threads, but only found 24", if in addition I disable hyper-threading : "Should have 6 threads, but only found 12"
- playing with options numcpus, nullCpuPowerManagement.
I compiled a SSDT with ssdtPRGen (attached) (though I'm not sure it's concerned about the problem, which seems more relevant to DSDT).
I looked into the log, and into Clover Source code.
I found that Clover is aware things are going wrong. In AcpiPatcher.c, it warns :
if (ApicCPUNum > 16) {
DBG("Out of control with CPU numbers\n");
break;
I played around quite a bit with Clover source. I managed to get the message to change (was expecting 16 threads, only found 32 at die 3). But while writing this post, I rolled back to an unmodified clover, the message went back to (expecting 16, found 17).
My questions :
- can somebody suggest a way to fix this with DSDT ?
- can somebody suggest modifications to AcpiPatcher.c to make this work?
cheers!