From: lkcl <lkcl@web>
Date: Sat, 8 Feb 2020 12:57:47 +0000 (+0000)
Subject: (no commit message)
X-Git-Tag: convert-csv-opcode-to-binary~3520
X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=5961e39304815ab586785e931bc4f0ededa61ff5;p=libreriscv.git

---

diff --git a/3d_gpu/architecture/dynamic_simd/shift.mdwn b/3d_gpu/architecture/dynamic_simd/shift.mdwn
index bd97bcbf2..4f8153479 100644
--- a/3d_gpu/architecture/dynamic_simd/shift.mdwn
+++ b/3d_gpu/architecture/dynamic_simd/shift.mdwn
@@ -32,3 +32,12 @@ input and a 4-way partition:
     out0 = p00[7..0]
     out1 = pmask[0] ? p01[7..0] : p00[15..8]
 
+# Static Partitioned Shift
+
+Static shift is pretty straightforward: the input is the entire number
+shifted, however clearly due to partitioning some of the bits need to
+be masked out (to zero, or to 1 if it is an arithmetic shift right).
+This can therefore use the class currently known as "MultiShiftRMerge"
+or similar, which can merge in a 1 into the shifted data.  Luckily
+the amount to be 0'd/1'd is also statically computable: it is just that
+the location *where* is dynamic, based on the partition location(s).