[Bug 70613] Register spilling is slow

bugzilla-daemon at freedesktop.org bugzilla-daemon at freedesktop.org
Wed Aug 31 12:10:35 UTC 2016


https://bugs.freedesktop.org/show_bug.cgi?id=70613

Eero Tamminen <eero.t.tamminen at intel.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           See Also|                            |https://bugs.freedesktop.or
                   |                            |g/show_bug.cgi?id=91857,
                   |                            |https://bugs.freedesktop.or
                   |                            |g/show_bug.cgi?id=97035

--- Comment #8 from Eero Tamminen <eero.t.tamminen at intel.com> ---
Ran the 30s shader with frag_runner on SKL GT2.  Startup / compile took 2.2s,
so this seems much improved.

Perf profile looks now like this:
16.33%  i965_dri.so  fs_visitor::virtual_grf_interferes
13.45%  i965_dri.so  ra_allocate                   
11.03%  i965_dri.so  ra_add_node_adjacency
 8.76%  i965_dri.so  fs_visitor::assign_regs                                
 6.71%  i965_dri.so  (anon)::ir_copy_propagation_visitor::visit
 5.30%  i965_dri.so  (anon)::ir_copy_propagation_elements_visitor::kill
 4.45%  i965_dri.so  (anon)::ir_copy_propagation_visitor::kill
 3.85%  i965_dri.so 
(anon)::ir_copy_propagation_elements_visitor::handle_rvalue
 3.60%  i965_dri.so  decrement_q.isra.2
 2.78%  i965_dri.so  instruction_scheduler::add_dep
 2.59%  i965_dri.so  ra_add_node_interference
 2.48%  libc-2.23.so _int_malloc
 1.90%  libc-2.23.so _int_free      
 1.08%  libc-2.23.so __libc_calloc      
 1.05%  libc-2.23.so realloc

Profile is similar to ones in bug 91857 and bug 97035, but differs from
profiles for bug 93681 and bug 87103.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-3d-bugs/attachments/20160831/fe139faf/attachment.html>


More information about the intel-3d-bugs mailing list