From 522f9205c9ed5e1ed23389e2c2b186de352835df Mon Sep 17 00:00:00 2001 From: Raymond Toy <rtoy@google.com> Date: Tue, 27 Mar 2012 14:22:11 -0700 Subject: [PATCH] Set the generator costs for the complex multiply vops to be the number of instructions. This gives the sse3 complex multiply vop a chance to be used if SSE3 is available. (The user needs to :SSE3 to *FEATURES* for this to work. The core of CMUCL only assumes SSE2 is available.) --- src/compiler/x86/float-sse2.lisp | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/compiler/x86/float-sse2.lisp b/src/compiler/x86/float-sse2.lisp index 979268fe9..f54f0721d 100644 --- a/src/compiler/x86/float-sse2.lisp +++ b/src/compiler/x86/float-sse2.lisp @@ -2149,7 +2149,7 @@ (:policy :fast-safe) (:temporary (:scs (complex-double-reg)) t1 t2) (:guard (backend-featurep :sse3)) - (:generator 1 + (:generator 8 ;; Basic algorithm from the paper "The Microarchitecture of the ;; Intel Pentium 4 Processor on 90nm Technololgy" ;; @@ -2176,7 +2176,7 @@ (:policy :fast-safe) (:temporary (:scs (complex-double-reg)) t0 t1 t2) (:temporary (:scs (unsigned-reg)) tmp) - (:generator 1 + (:generator 13 ;; Basic algorithm from the paper "The Microarchitecture of the ;; Intel Pentium 4 Processor on 90nm Technololgy" @@ -2208,7 +2208,7 @@ (:policy :fast-safe) (:temporary (:scs (complex-single-reg)) t0 t1 t2) (:temporary (:scs (unsigned-reg)) tmp) - (:generator 1 + (:generator 14 ;; Basic algorithm from the paper "The Microarchitecture of the ;; Intel Pentium 4 Processor on 90nm Technololgy" -- GitLab