projects
/
libreriscv.git
/ commitdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
| commitdiff |
tree
raw
|
patch
|
inline
| side by side (parent:
655fe27
)
Use svg vector images converted from original jpeg ones
author
R Veera Kumar
<vklr@vkten.in>
Wed, 29 Jun 2022 13:25:55 +0000
(18:55 +0530)
committer
R Veera Kumar
<vklr@vkten.in>
Wed, 29 Jun 2022 13:25:55 +0000
(18:55 +0530)
openpower/sv/SimpleV_rationale.mdwn
patch
|
blob
|
history
diff --git
a/openpower/sv/SimpleV_rationale.mdwn
b/openpower/sv/SimpleV_rationale.mdwn
index fe5f0e9a732af1fb4e90c1aa6aa6faddc327b099..a8096629b18718c165ebb339bdb52d952f6970cd 100644
(file)
--- a/
openpower/sv/SimpleV_rationale.mdwn
+++ b/
openpower/sv/SimpleV_rationale.mdwn
@@
-903,9
+903,7
@@
definitely compelling enough to warrant in-depth investigation.
**Use-case: Matrix and Convolutions**
**Use-case: Matrix and Convolutions**
-DRAFT image:
-
-<img src="https://ftp.libre-soc.org/20220614_122128.jpg" width=600 />
+<img src="/openpower/sv/sv_horizontal_vs_vertical.svg" width=600 />
First, some important definitions, because there are two different
Vectorisation Modes in SVP64:
First, some important definitions, because there are two different
Vectorisation Modes in SVP64:
@@
-931,7
+929,7
@@
in full. Vertical-First may be used to perform a LD/ST within
the loop, covered by `svstep`, but it is still not ideal. This
is where the Snitch and EXTRA-V concepts kick in.
the loop, covered by `svstep`, but it is still not ideal. This
is where the Snitch and EXTRA-V concepts kick in.
-<img src="
https://ftp.libre-soc.org/matrix_svremap.jp
g" width=600 />
+<img src="
/openpower/sv/matrix_svremap.sv
g" width=600 />
Imagine a large Matrix scenario, with several values close to zero that
could be skipped: no need to include zero-multiplications, but a
Imagine a large Matrix scenario, with several values close to zero that
could be skipped: no need to include zero-multiplications, but a