systemsoftware
/
Programming-Guides
35
1
Fork 0
Code Issues 17 Pull Requests Projects Releases Wiki Activity

Move the endian-sensitive built-in function table

Browse Source 
Move the table so that it won't have a head-scratching location
in the HTML view.  Update it to add the first_match* and first_mismatch*
functions.

Fixed #39.

Signed-off-by: Bill Schmidt <wschmidt@linux.ibm.com>
pull/69/head
Bill Schmidt 5 years ago
parent 0cea5e8dd1
commit d939b73aac
2 changed files with 176 additions and 150 deletions
Show Stats Download Patch File Download Diff File

314
Intrinsics_Reference/ch_biendian.xml
Unescape Escape View File

@ -762,369 +762,391 @@ a[3] = c;</programlisting>
endian-sensitive built-in functions can be found in <xref endian-sensitive built-in functions can be found in <xref
linkend="VIPR.biendian.sensitive" />. linkend="VIPR.biendian.sensitive" />.
</para> </para>
<section> <table frame="all" pgwide="1" xml:id="VIPR.biendian.sensitive">
<title>Extended Data Movement Functions</title> <title>Endian-Sensitive Built-In Functions</title>
<para> <tgroup cols="3">
The built-in functions in <xref
linkend="VIPR.biendian.vmx-mem" /> map to Altivec/VMX load and
store instructions and provide access to the “auto-aligning”
memory instructions of the VMX ISA where low-order address
bits are discarded before performing a memory access. These
instructions load and store data in accordance with the
program's current endian mode, and do not need to be adapted
by the compiler to reflect little-endian operation during code
generation.
</para>
<para>
Before the bi-endian programming model was introduced, the
<code>vec_lvsl</code> and <code>vec_lvsr</code> intrinsics
were supported. These could be used in conjunction with
<code>vec_perm</code> and VMX load and store instructions for
unaligned access. The <code>vec_lvsl</code> and
<code>vec_lvsr</code> interfaces are deprecated in accordance
with the interfaces specified here. For compatibility, the
built-in pseudo sequences published in previous VMX documents
continue to work with little-endian data layout and the
little-endian vector layout described in this document.
However, the use of these sequences in new code is discouraged
and usually results in worse performance. It is recommended
that compilers issue a warning when these functions are used
in little-endian environments.
</para>
<table frame="all" pgwide="1" xml:id="VIPR.biendian.vmx-mem">
<title>VMX Memory Access Built-In Functions</title>
<tgroup cols="3">
<colspec colname="c1" colwidth="15*" align="center" /> <colspec colname="c1" colwidth="15*" align="center" />
<colspec colname="c2" colwidth="35*" align="center" /> <colspec colname="c2" colwidth="15*" align="center" />
<colspec colname="c3" colwidth="50*" /> <colspec colname="c3" colwidth="15*" align="center" />
<thead> <tbody>
<row> <row>
<entry> <entry>
<para> <para>vec_bperm</para>
<emphasis role="bold">Built-in Function</emphasis>
</para>
</entry> </entry>
<entry> <entry>
<para> <para>vec_mergeh</para>
<emphasis role="bold">Corresponding Power
Instructions</emphasis>
</para>
</entry> </entry>
<entry align="center"> <entry>
<para> <para>vec_signedo</para>
<emphasis role="bold">Implementation Notes</emphasis>
</para>
</entry> </entry>
</row> </row>
</thead>
<tbody>
<row> <row>
<entry> <entry>
<para>vec_ld</para> <para>vec_cipher_be</para>
</entry> </entry>
<entry> <entry>
<para>lvx</para> <para>vec_mergel</para>
</entry> </entry>
<entry> <entry>
<para>Hardware works as a function of endian mode.</para> <para>vec_sld</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_lde</para> <para>vec_cipherlast_be</para>
</entry> </entry>
<entry> <entry>
<para>lvebx, lvehx, lvewx</para> <para>vec_mergeo</para>
</entry> </entry>
<entry> <entry>
<para>Hardware works as a function of endian mode.</para> <para>vec_sldw</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_ldl</para> <para>vec_doublee</para>
</entry> </entry>
<entry> <entry>
<para>lvxl</para> <para>vec_mfvscr</para>
</entry> </entry>
<entry> <entry>
<para>Hardware works as a function of endian mode.</para> <para>vec_sll</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_st</para> <para>vec_doubleh</para>
</entry> </entry>
<entry> <entry>
<para>stvx</para> <para>vec_mule</para>
</entry> </entry>
<entry> <entry>
<para>Hardware works as a function of endian mode.</para> <para>vec_slo</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_ste</para> <para>vec_doublel</para>
</entry> </entry>
<entry> <entry>
<para>stvebx, stvehx, stvewx</para> <para>vec_mulo</para>
</entry> </entry>
<entry> <entry>
<para>Hardware works as a function of endian mode.</para> <para>vec_slv</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_stl</para> <para>vec_doubleo</para>
</entry> </entry>
<entry> <entry>
<para>stvxl</para> <para>vec_ncipher_be</para>
</entry> </entry>
<entry> <entry>
<para>Hardware works as a function of endian mode.</para> <para>vec_splat</para>
</entry> </entry>
</row> </row>
</tbody>
</tgroup>
</table>
<para>
Instead, it is recommended that programmers use the
<code>vec_xl</code> and <code>vec_xst</code> vector built-in
functions to access unaligned data streams. See the
descriptions of these instructions in <xref
linkend="VIPR.vec-ref" /> for further description and
implementation details.
</para>
<table frame="all" pgwide="1" xml:id="VIPR.biendian.sensitive">
<title>Endian-Sensitive Built-In Functions</title>
<tgroup cols="3">
<colspec colname="c1" colwidth="15*" align="center" />
<colspec colname="c2" colwidth="15*" align="center" />
<colspec colname="c3" colwidth="15*" align="center" />
<tbody>
<row> <row>
<entry> <entry>
<para>vec_bperm</para> <para>vec_extract</para>
</entry> </entry>
<entry> <entry>
<para>vec_mergeo</para> <para>vec_ncipherlast_be</para>
</entry> </entry>
<entry> <entry>
<para>vec_sld</para> <para>vec_srl</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_cipher_be</para> <para>vec_extract_fp32_from_shorth</para>
</entry> </entry>
<entry> <entry>
<para>vec_mfvscr</para> <para>vec_pack</para>
</entry> </entry>
<entry> <entry>
<para>vec_sldw</para> <para>vec_sro</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_cipherlast_be</para> <para>vec_extract_fp32_from_shortl</para>
</entry> </entry>
<entry> <entry>
<para>vec_mule</para> <para>vec_pack_to_short_fp32</para>
</entry> </entry>
<entry> <entry>
<para>vec_sll</para> <para>vec_srv</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_doublee</para> <para>vec_extract4b</para>
</entry> </entry>
<entry> <entry>
<para>vec_mulo</para> <para>vec_packpx</para>
</entry> </entry>
<entry> <entry>
<para>vec_slo</para> <para>vec_sum2s</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_doubleh</para> <para>vec_first_match_index</para>
</entry> </entry>
<entry> <entry>
<para>vec_ncipher_be</para> <para>vec_packs</para>
</entry> </entry>
<entry> <entry>
<para>vec_slv</para> <para>vec_sums</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_doublel</para> <para>vec_first_match_or_eos_index</para>
</entry> </entry>
<entry> <entry>
<para>vec_ncipherlast_be</para> <para>vec_packsu</para>
</entry> </entry>
<entry> <entry>
<para>vec_splat</para> <para>vec_unpackh</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_doubleo</para> <para>vec_first_mismatch_index</para>
</entry> </entry>
<entry> <entry>
<para>vec_pack</para> <para>vec_perm</para>
</entry> </entry>
<entry> <entry>
<para>vec_srl</para> <para>vec_unpackl</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_extract</para> <para>vec_first_mismatch_or_eos_index</para>
</entry> </entry>
<entry> <entry>
<para>vec_pack_to_short_fp32</para> <para>vec_permxor</para>
</entry> </entry>
<entry> <entry>
<para>vec_sro</para> <para>vec_unsigned2</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_extract_fp32_from_shorth</para> <para>vec_float2</para>
</entry> </entry>
<entry> <entry>
<para>vec_packpx</para> <para>vec_pmsum_be</para>
</entry> </entry>
<entry> <entry>
<para>vec_srv</para> <para>vec_unsignede</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_extract_fp32_from_shortl</para> <para>vec_floate</para>
</entry> </entry>
<entry> <entry>
<para>vec_packs</para> <para>vec_reve</para>
</entry> </entry>
<entry> <entry>
<para>vec_sum2s</para> <para>vec_unsignedo</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_extract_4b</para> <para>vec_floato</para>
</entry> </entry>
<entry> <entry>
<para>vec_packsu</para> <para>vec_sbox_be</para>
</entry> </entry>
<entry> <entry>
<para>vec_sums</para> <para>vec_xl (ISA 2.07 only)</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_float2</para> <para>vec_gb</para>
</entry> </entry>
<entry> <entry>
<para>vec_perm</para> <para>vec_shasigma_be</para>
</entry> </entry>
<entry> <entry>
<para>vec_unpackh</para> <para>vec_xl_be</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_floate</para> <para>vec_insert</para>
</entry> </entry>
<entry> <entry>
<para>vec_permxor</para> <para>vec_signed2</para>
</entry> </entry>
<entry> <entry>
<para>vec_unpackl</para> <para>vec_xst (ISA 2.07 only)</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_floato</para> <para>vec_insert4b</para>
</entry> </entry>
<entry> <entry>
<para>vec_pmsum_be</para> <para>vec_signede</para>
</entry> </entry>
<entry> <entry>
<para>vec_unsigned2</para> <para>vec_xst_be</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_gb</para> <para>vec_mergee</para>
</entry> </entry>
<entry> <entry>
<para>vec_reve</para> <para></para>
</entry> </entry>
<entry> <entry>
<para>vec_unsignede</para> <para></para>
</entry>
</row>
</tbody>
</tgroup>
</table>
<section>
<title>Extended Data Movement Functions</title>
<para>
The built-in functions in <xref
linkend="VIPR.biendian.vmx-mem" /> map to Altivec/VMX load and
store instructions and provide access to the “auto-aligning”
memory instructions of the VMX ISA where low-order address
bits are discarded before performing a memory access. These
instructions load and store data in accordance with the
program's current endian mode, and do not need to be adapted
by the compiler to reflect little-endian operation during code
generation.
</para>
<para>
Before the bi-endian programming model was introduced, the
<code>vec_lvsl</code> and <code>vec_lvsr</code> intrinsics
were supported. These could be used in conjunction with
<code>vec_perm</code> and VMX load and store instructions for
unaligned access. The <code>vec_lvsl</code> and
<code>vec_lvsr</code> interfaces are deprecated in accordance
with the interfaces specified here. For compatibility, the
built-in pseudo sequences published in previous VMX documents
continue to work with little-endian data layout and the
little-endian vector layout described in this document.
However, the use of these sequences in new code is discouraged
and usually results in worse performance. It is recommended
that compilers issue a warning when these functions are used
in little-endian environments.
</para>
<table frame="all" pgwide="1" xml:id="VIPR.biendian.vmx-mem">
<title>VMX Memory Access Built-In Functions</title>
<tgroup cols="3">
<colspec colname="c1" colwidth="15*" align="center" />
<colspec colname="c2" colwidth="35*" align="center" />
<colspec colname="c3" colwidth="50*" />
<thead>
<row>
<entry>
<para>
<emphasis role="bold">Built-in Function</emphasis>
</para>
</entry>
<entry>
<para>
<emphasis role="bold">Corresponding Power
Instructions</emphasis>
</para>
</entry>
<entry align="center">
<para>
<emphasis role="bold">Implementation Notes</emphasis>
</para>
</entry> </entry>
</row> </row>
</thead>
<tbody>
<row> <row>
<entry> <entry>
<para>vec_insert</para> <para>vec_ld</para>
</entry> </entry>
<entry> <entry>
<para>vec_sbox_be</para> <para>lvx</para>
</entry> </entry>
<entry> <entry>
<para>vec_unsignedo</para> <para>Hardware works as a function of endian mode.</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_insert_4b</para> <para>vec_lde</para>
</entry> </entry>
<entry> <entry>
<para>vec_shasigma_be</para> <para>lvebx, lvehx, lvewx</para>
</entry> </entry>
<entry> <entry>
<para>vec_xl (ISA 2.07 only)</para> <para>Hardware works as a function of endian mode.</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_mergee</para> <para>vec_ldl</para>
</entry> </entry>
<entry> <entry>
<para>vec_signed2</para> <para>lvxl</para>
</entry> </entry>
<entry> <entry>
<para>vec_xl_be</para> <para>Hardware works as a function of endian mode.</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_mergeh</para> <para>vec_st</para>
</entry> </entry>
<entry> <entry>
<para>vec_signede</para> <para>stvx</para>
</entry> </entry>
<entry> <entry>
<para>vec_xst (ISA 2.07 only)</para> <para>Hardware works as a function of endian mode.</para>
</entry> </entry>
</row> </row>
<row> <row>
<entry> <entry>
<para>vec_mergel</para> <para>vec_ste</para>
</entry> </entry>
<entry> <entry>
<para>vec_signedo</para> <para>stvebx, stvehx, stvewx</para>
</entry> </entry>
<entry> <entry>
<para>vec_xst_be</para> <para>Hardware works as a function of endian mode.</para>
</entry> </entry>
</row> </row>
</tbody> <row>
</tgroup> <entry>
<para>vec_stl</para>
</entry>
<entry>
<para>stvxl</para>
</entry>
<entry>
<para>Hardware works as a function of endian mode.</para>
</entry>
</row>
</tbody>
</tgroup>
</table> </table>
<para>
Instead, it is recommended that programmers use the
<code>vec_xl</code> and <code>vec_xst</code> vector built-in
functions to access unaligned data streams. See the
descriptions of these instructions in <xref
linkend="VIPR.vec-ref" /> for further description and
implementation details.
</para>
</section> </section>
<section xml:id="VIPR.biendian.BELE"> <section xml:id="VIPR.biendian.BELE">
<title>Big-Endian Vector Layout in Little-Endian Environments <title>Big-Endian Vector Layout in Little-Endian Environments

12
Intrinsics_Reference/ch_vec_reference.xml
Unescape Escape View File

@ -13606,7 +13606,8 @@ xmlns:xlink="http://www.w3.org/1999/xlink" xml:id="VIPR.vec-ref">
natural element order. If no match, returns the number of natural element order. If no match, returns the number of
characters as an element count in the vector argument.</para> characters as an element count in the vector argument.</para>
<para><emphasis role="bold">Endian considerations:</emphasis> <para><emphasis role="bold">Endian considerations:</emphasis>
None. The element numbering within a register is left-to-right for big-endian
targets, and right-to-left for little-endian targets.
</para> </para>
<indexterm> <indexterm>
@ -13880,7 +13881,8 @@ xmlns:xlink="http://www.w3.org/1999/xlink" xml:id="VIPR.vec-ref">
terminator. If no match or terminator, returns the number of terminator. If no match or terminator, returns the number of
characters as an element count in the vector argument.</para> characters as an element count in the vector argument.</para>
<para><emphasis role="bold">Endian considerations:</emphasis> <para><emphasis role="bold">Endian considerations:</emphasis>
None. The element numbering within a register is left-to-right for big-endian
targets, and right-to-left for little-endian targets.
</para> </para>
<indexterm> <indexterm>
@ -14221,7 +14223,8 @@ xmlns:xlink="http://www.w3.org/1999/xlink" xml:id="VIPR.vec-ref">
natural element order. If no mismatch, returns the number of natural element order. If no mismatch, returns the number of
characters as an element count in the vector argument.</para> characters as an element count in the vector argument.</para>
<para><emphasis role="bold">Endian considerations:</emphasis> <para><emphasis role="bold">Endian considerations:</emphasis>
None. The element numbering within a register is left-to-right for big-endian
targets, and right-to-left for little-endian targets.
</para> </para>
<indexterm> <indexterm>
@ -14479,7 +14482,8 @@ xmlns:xlink="http://www.w3.org/1999/xlink" xml:id="VIPR.vec-ref">
terminator. If no mismatch or terminator, returns the number of terminator. If no mismatch or terminator, returns the number of
characters as an element count in the vector argument.</para> characters as an element count in the vector argument.</para>
<para><emphasis role="bold">Endian considerations:</emphasis> <para><emphasis role="bold">Endian considerations:</emphasis>
None. The element numbering within a register is left-to-right for big-endian
targets, and right-to-left for little-endian targets.
</para> </para>
<indexterm> <indexterm>

Write Preview
Loading…
Cancel
Save