From 362736a01d34361ccc36ffdaef4e8755d5f1e202 Mon Sep 17 00:00:00 2001 From: lkcl Date: Sun, 16 Apr 2023 10:30:15 +0100 Subject: [PATCH] --- openpower/sv/remap.mdwn | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/openpower/sv/remap.mdwn b/openpower/sv/remap.mdwn index ea022d008..22b0cc135 100644 --- a/openpower/sv/remap.mdwn +++ b/openpower/sv/remap.mdwn @@ -62,11 +62,17 @@ group, just as it does when REMAP is not active. In its general form, REMAP is quite expensive to set up, and on some implementations may introduce latency, so should realistically be used only where it is worthwhile. Given that even with latency the fact -that up to 127 operations can be requested to be issued (from a single +that up to 127 operations can be Deterministically issued (from a single instruction) it should be clear that REMAP should not be dismissed for *possible* latency alone. Commonly-used patterns such as Matrix Multiply, DCT and FFT have helper instruction options which make REMAP -easier to use. +easier to use + +*Programmer's note: with the exception +of Indexed REMAP all of these helper instructions are static operands +as this helps Hardware by keeping Schedules 100% Deterministic. +Where this is inconvenient the SVSHAPE SPRs 0-3 may be written to directly, +bearing in mind this may have additional latency on some implementations.* There are four types of REMAP: -- 2.30.2