From mboxrd@z Thu Jan 1 00:00:00 1970 From: bugzilla-daemon-CC+yJ3UmIYqDUpFQwHEjaQ@public.gmane.org Subject: [Bug 90887] PhiMovesPass in register allocator broken Date: Sun, 07 Jun 2015 20:33:35 +0000 Message-ID: References: Mime-Version: 1.0 Content-Type: multipart/mixed; boundary="===============0895008007==" Return-path: In-Reply-To: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: nouveau-bounces-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org Sender: "Nouveau" To: nouveau-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org List-Id: nouveau.vger.kernel.org --===============0895008007== Content-Type: multipart/alternative; boundary="1433709215.cfF4bd1.8354"; charset="UTF-8" --1433709215.cfF4bd1.8354 Date: Sun, 7 Jun 2015 20:33:35 +0000 MIME-Version: 1.0 Content-Type: text/plain; charset="UTF-8" https://bugs.freedesktop.org/show_bug.cgi?id=90887 --- Comment #3 from Ilia Mirkin --- Gah! The inbound edge ordering having to be the same as in phi nodes has bit me before as well. I wonder if it wouldn't be simpler to just fix that, i.e. make phi nodes attach a bb reference to each argument. In any case, I'll be able to take a closer look in a week or so. Great detective work btw. FTR, nva5 should be able to allow manual reclocking on kernel 3.19+ (if you add nouveau.pstate=1), which should help decrease some of the perf difference wrt blob. Also, bug 90347 is completely different... that findFirstUse stuff is pretty dodgy in the first place, and it happens post-RA. (And only on kepler+) -- You are receiving this mail because: You are the QA Contact for the bug. You are the assignee for the bug. --1433709215.cfF4bd1.8354 Date: Sun, 7 Jun 2015 20:33:35 +0000 MIME-Version: 1.0 Content-Type: text/html; charset="UTF-8"

Comment # 3 on bug 90887 from 
Gah! The inbound edge ordering having to be the same as in phi nodes has bit me
before as well. I wonder if it wouldn't be simpler to just fix that, i.e. make
phi nodes attach a bb reference to each argument.

In any case, I'll be able to take a closer look in a week or so. Great
detective work btw.

FTR, nva5 should be able to allow manual reclocking on kernel 3.19+ (if you add
nouveau.pstate=1), which should help decrease some of the perf difference wrt
blob.

Also, bug 90347 is completely different... that findFirstUse stuff is pretty
dodgy in the first place, and it happens post-RA. (And only on kepler+)

You are receiving this mail because:
  • You are the QA Contact for the bug.
  • You are the assignee for the bug.
--1433709215.cfF4bd1.8354-- --===============0895008007== Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: base64 Content-Disposition: inline X19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX18KTm91dmVhdSBt YWlsaW5nIGxpc3QKTm91dmVhdUBsaXN0cy5mcmVlZGVza3RvcC5vcmcKaHR0cDovL2xpc3RzLmZy ZWVkZXNrdG9wLm9yZy9tYWlsbWFuL2xpc3RpbmZvL25vdXZlYXUK --===============0895008007==--