[Mesa-dev] Build mesa-dev on Windows with AVX instruction set problem
Roland Scheidegger
sroland at vmware.com
Wed Oct 25 22:55:35 UTC 2017
Am 26.10.2017 um 00:26 schrieb Ilia Mirkin:
> On Wed, Oct 25, 2017 at 6:15 PM, Fabrício Ceolin
> <fabricio.ceolin at miningmath.com <mailto:fabricio.ceolin at miningmath.com>>
> wrote:
>
> Hi,
>
> Thanks. I recompiled everything (on Windows) using this real machine:
>
> #under msys2
> $ cat /proc/cpuinfo
> processor : 0
> vendor_id : GenuineIntel
> cpu family : 6
> model : 23
> model name : Genuine Intel(R) CPU U2300 @ 1.20GHz
> stepping : 10
> cpu MHz : 1197.000
> cache size : 1024 KB
> physical id : 0
> siblings : 2
> core id : 0
> cpu cores : 2
> apicid : 0
> initial apicid : 0
> fpu : yes
> fpu_exception : yes
> cpuid level : 13
> wp : yes
> flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr
> pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm
> pbe pni dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm xsave
> osxsave lahf_lm dtherm
> clflush size : 64
> cache_alignment : 64
> address sizes : 36 bits physical, 48 bits virtual
> power management:
>
> but I got the same error:
>
> 00007FF85FBF874F 48 83 7A 18 10 cmp qword ptr
> [rdx+18h],10h
> 00007FF85FBF8754 48 8B 6A 10 mov rbp,qword ptr
> [rdx+10h]
> 00007FF85FBF8758 72 03 jb
> std::basic_string<char,std::char_traits<char>,std::allocator<char>
> >::basic_string<char,std::char_traits<char>,std::allocator<char>
> >+2Dh (07FF85FBF875Dh)
> 00007FF85FBF875A 48 8B 32 mov rsi,qword ptr [rdx]
> 00007FF85FBF875D 48 83 FD 10 cmp rbp,10h
> 00007FF85FBF8761 73 27 jae
> std::basic_string<char,std::char_traits<char>,std::allocator<char>
> >::basic_string<char,std::char_traits<char>,std::allocator<char>
> >+5Ah (07FF85FBF878Ah)
> *00007FF85FBF8763 C5 F8 10 06 vmovups xmm0,xmmword ptr
> [rsi] *
> 00007FF85FBF8767 C5 F8 11 01 vmovups xmmword ptr
> [rcx],xmm0
> 00007FF85FBF876B 48 89 69 10 mov qword ptr
> [rcx+10h],rbp
>
> Are there any parameters that I can use for avoid to use AVX (or
> better) instructions?
>
>
> LP_NATIVE_VECTOR_WIDTH=128 should force you to a non-avx path. There's
> also a LP_FORCE_SSE2=1 which will also avoid sse3/4 usage. However all
> this stuff should be getting detected, so it's odd that it's messing up.
> Perhaps run with GALLIUM_DUMP_CPU=1 to see what's being detected?
>
Also, a backtrace when it crashes would be nice. I am not convinced this
is actually happening inside a code-generated shader...
Roland
More information about the mesa-dev
mailing list