radeonsi: initial WIP SI code

This commit adds initial support for acceleration
on SI chips.  egltri is starting to work.

The SI/R600 llvm backend is currently included in mesa
but that may change in the future.

The plan is to write a single gallium driver and
use gallium to support X acceleration.

This commit contains patches from:
Tom Stellard <[email protected]>
Michel Dänzer <[email protected]>
Alex Deucher <[email protected]>
Vadim Girlin <[email protected]>

Signed-off-by: Alex Deucher <[email protected]>

The following commits were squashed in:

======================================================================

radeonsi: Remove unused winsys pointer

This was removed from r600g in commit:

commit 96d882939d612fcc8332f107befec470ed4359de
Author: Marek Olšák <[email protected]>
Date:   Fri Feb 17 01:49:49 2012 +0100

    gallium: remove unused winsys pointers in pipe_screen and pipe_context

    A winsys is already a private object of a driver.

======================================================================

radeonsi: Copy color clamping CAPs from r600

Not sure if the values of these CAPS are correct for radeonsi, but the
same changed were made to r600g in commit:

commit bc1c8369384b5e16547c5bf9728aa78f8dfd66cc
Author: Marek Olšák <[email protected]>
Date:   Mon Jan 23 03:11:17 2012 +0100

    st/mesa: do vertex and fragment color clamping in shaders

    For ARB_color_buffer_float. Most hardware can't do it and st/mesa is
    the perfect place for a fallback.
    The exceptions are:
    - r500 (vertex clamp only)
    - nv50 (both)
    - nvc0 (both)
    - softpipe (both)

    We also have to take into account that r300 can do CLAMPED vertex colors only,
    while r600 can do UNCLAMPED vertex colors only. The difference can be expressed
    with the two new CAPs.

======================================================================

radeonsi: Remove PIPE_CAP_OUTPUT_READ

This CAP was dropped in commit:

commit 04e324008759282728a95a1394bac2c4c2a1a3f9
Author: Marek Olšák <[email protected]>
Date:   Thu Feb 23 23:44:36 2012 +0100

    gallium: remove PIPE_SHADER_CAP_OUTPUT_READ

    r600g is the only driver which has made use of it. The reason the CAP was
    added was to fix some piglit tests when the GLSL pass lower_output_reads
    didn't exist.

    However, not removing output reads breaks the fallback for glClampColorARB,
    which assumes outputs are not readable. The fix would be non-trivial
    and my personal preference is to remove the CAP, considering that reading
    outputs is uncommon and that we can now use lower_output_reads to fix
    the issue that the CAP was supposed to workaround in the first place.

======================================================================

radeonsi: Add missing parameters to rws->buffer_get_tiling() call

This was changed in commit:

commit c0c979eebc076b95cc8d18a013ce2968fe6311ad
Author: Jerome Glisse <[email protected]>
Date:   Mon Jan 30 17:22:13 2012 -0500

    r600g: add support for common surface allocator for tiling v13

    Tiled surface have all kind of alignment constraint that needs to
    be met. Instead of having all this code duplicated btw ddx and
    mesa use common code in libdrm_radeon this also ensure that both
    ddx and mesa compute those alignment in the same way.

    v2 fix evergreen
    v3 fix compressed texture and workaround cube texture issue by
       disabling 2D array mode for cubemap (need to check if r7xx and
       newer are also affected by the issue)
    v4 fix texture array
    v5 fix evergreen and newer, split surface values computation from
       mipmap tree generation so that we can get them directly from the
       ddx
    v6 final fix to evergreen tile split value
    v7 fix mipmap offset to avoid to use random value, use color view
       depth view to address different layer as hardware is doing some
       magic rotation depending on the layer
    v8 fix COLOR_VIEW on r6xx for linear array mode, use COLOR_VIEW on
       evergreen, align bytes per pixel to a multiple of a dword
    v9 fix handling of stencil on evergreen, half fix for compressed
       texture
    v10 fix evergreen compressed texture proper support for stencil
        tile split. Fix stencil issue when array mode was clear by
        the kernel, always program stencil bo. On evergreen depth
        buffer bo need to be big enough to hold depth buffer + stencil
        buffer as even with stencil disabled things get written there.
    v11 rebase on top of mesa, fix pitch issue with 1d surface on evergreen,
        old ddx overestimate those. Fix linear case when pitch*height < 64.
        Fix r300g.
    v12 Fix linear case when pitch*height < 64 for old path, adapt to
        libdrm API change
    v13 add libdrm check

    Signed-off-by: Jerome Glisse <[email protected]>

======================================================================

radeonsi: Remove PIPE_TRANSFER_MAP_PERMANENTLY

This was removed in commit:

commit 62f44f670bb0162e89fd4786af877f8da9ff607c
Author: Marek Olšák <[email protected]>
Date:   Mon Mar 5 13:45:00 2012 +0100

    Revert "gallium: add flag PIPE_TRANSFER_MAP_PERMANENTLY"

    This reverts commit 0950086376b1c8b7fb89eda81ed7f2f06dee58bc.

    It was decided to refactor the transfer API instead of adding workarounds
    to address the performance issues.

======================================================================

radeonsi: Handle PIPE_VIDEO_CAP_PREFERED_FORMAT.

Reintroduced in commit 9d9afcb5bac2931d4b8e6d1aa571e941c5110c90.

======================================================================

radeonsi: nuke the fallback for vertex and fragment color clamping

Ported from r600g commit c2b800cf38b299c1ab1c53dc0e4ea00c7acef853.

======================================================================

radeonsi: don't expose transform_feedback2 without kernel support

Ported from r600g commit 15146fd1bcbb08e44a1cbb984440ee1a5de63d48.

======================================================================

radeonsi: Handle PIPE_CAP_GLSL_FEATURE_LEVEL.

Ported from r600g part of commit 171be755223d99f8cc5cc1bdaf8bd7b4caa04b4f.

======================================================================

radeonsi: set minimum point size to 1.0 for non-sprite non-aa points.

Ported from r600g commit f183cc9ce3ad1d043bdf8b38fd519e8f437714fc.

======================================================================

radeonsi: rework and consolidate stencilref state setting.

Ported from r600g commit a2361946e782b57f0c63587841ca41c0ea707070.

======================================================================

radeonsi: cleanup setting DB_SHADER_CONTROL.

Ported from r600g commit 3d061caaed13b646ff40754f8ebe73f3d4983c5b.

======================================================================

radeonsi: Get rid of register masks.

Ported from r600g commits
3d061caaed13b646ff40754f8ebe73f3d4983c5b..9344ab382a1765c1a7c2560e771485edf4954fe2.

======================================================================

radeonsi: get rid of r600_context_reg.

Ported from r600g commits
9344ab382a1765c1a7c2560e771485edf4954fe2..bed20f02a771f43e1c5092254705701c228cfa7f.

======================================================================

radeonsi: Fix regression from 'Get rid of register masks'.

======================================================================

radeonsi: optimize r600_resource_va.

Ported from r600g commit 669d8766ff3403938794eb80d7769347b6e52174.

======================================================================

radeonsi: remove u8,u16,u32,u64 types.

Ported from r600g commit 78293b99b23268e6698f1267aaf40647c17d95a5.

======================================================================

radeonsi: merge r600_context with r600_pipe_context.

Ported from r600g commit e4340c1908a6a3b09e1a15d5195f6da7d00494d0.

======================================================================

radeonsi: Miscellaneous context cleanups.

Ported from r600g commits
e4340c1908a6a3b09e1a15d5195f6da7d00494d0..621e0db71c5ddcb379171064a4f720c9cf01e888.

======================================================================

radeonsi: add a new simple API for state emission.

Ported from r600g commits
621e0db71c5ddcb379171064a4f720c9cf01e888..f661405637bba32c2cfbeecf6e2e56e414e9521e.

======================================================================

radeonsi: Also remove sbu_flags member of struct r600_reg.

Requires using sid.h instead of r600d.h for the new CP_COHER_CNTL definitions,
so some code needs to be disabled for now.

======================================================================

radeonsi: Miscellaneous simplifications.

Ported from r600g commits 38bf2763482b4f1b6d95cd51aecec75601d8b90f and
b0337b679ad4c2feae59215104cfa60b58a619d5.

======================================================================

radeonsi: Handle PIPE_CAP_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION.

Ported from commit 8b4f7b0672d663273310fffa9490ad996f5b914a.

======================================================================

radeonsi: Use a fake reloc to sleep for fences.

Ported from r600g commit 8cd03b933cf868ff867e2db4a0937005a02fd0e4.

======================================================================

radeonsi: adapt to get_query_result interface change.

Ported from r600g commit 4445e170bee23a3607ece0e010adef7058ac6a11.

1 files changed, 1022 insertions, 0 deletions

diff --git a/src/gallium/drivers/radeon/AMDILConversions.td b/src/gallium/drivers/radeon/AMDILConversions.td
new file mode 100644
index 00000000000..0db66ae8475
--- /dev/null
+++ b/src/gallium/drivers/radeon/AMDILConversions.td
@@ -0,0 +1,1022 @@
+//===-- AMDILConversions.td - TODO: Add brief description -------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//==-----------------------------------------------------------------------===//
+def actos_i16:Pat < (i16 (anyext GPRI8:$src)),
+(IL_ASSHORT_i32
+ (USHR_i32
+  (SHL_i32
+(IL_ASINT_i8 GPRI8:$src),
+   (LOADCONST_i32 24)),
+  (LOADCONST_i32 24))) >;
+
+
+def uctos_i16:Pat < (i16 (zext GPRI8:$src)),
+(IL_ASSHORT_i32
+ (USHR_i32
+  (SHL_i32
+(IL_ASINT_i8 GPRI8:$src),
+   (LOADCONST_i32 24)),
+  (LOADCONST_i32 24))) >;
+
+
+def sctos_i16:Pat < (i16 (sext GPRI8:$src)),
+(IL_ASSHORT_i32
+ (SHR_i32
+  (SHL_i32
+(IL_ASINT_i8 GPRI8:$src),
+   (LOADCONST_i32 24)),
+  (LOADCONST_i32 24))) >;
+
+
+def actoi_i32:Pat < (i32 (anyext GPRI8:$src)),
+(IL_ASINT_i32
+ (USHR_i32
+  (SHL_i32
+(IL_ASINT_i8 GPRI8:$src),
+   (LOADCONST_i32 24)),
+  (LOADCONST_i32 24))) >;
+
+
+def uctoi_i32:Pat < (i32 (zext GPRI8:$src)),
+(IL_ASINT_i32
+ (USHR_i32
+  (SHL_i32
+(IL_ASINT_i8 GPRI8:$src),
+   (LOADCONST_i32 24)),
+  (LOADCONST_i32 24))) >;
+
+
+def sctoi_i32:Pat < (i32 (sext GPRI8:$src)),
+(IL_ASINT_i32
+ (SHR_i32
+  (SHL_i32
+(IL_ASINT_i8 GPRI8:$src),
+   (LOADCONST_i32 24)),
+  (LOADCONST_i32 24))) >;
+
+
+def actol_i64:Pat < (i64 (anyext GPRI8:$src)),
+(LCREATE
+ (USHR_i32
+  (SHL_i32
+(IL_ASINT_i8 GPRI8:$src),
+   (LOADCONST_i32 24)),
+  (LOADCONST_i32 24)),
+ (LOADCONST_i32 0)) >;
+
+
+def uctol_i64:Pat < (i64 (zext GPRI8:$src)),
+(LCREATE
+ (USHR_i32
+  (SHL_i32
+(IL_ASINT_i8 GPRI8:$src),
+   (LOADCONST_i32 24)),
+  (LOADCONST_i32 24)),
+ (LOADCONST_i32 0)) >;
+
+
+def sctol_i64:Pat < (i64 (sext GPRI8:$src)),
+(LCREATE
+ (SHR_i32
+  (SHL_i32
+(IL_ASINT_i8 GPRI8:$src),
+   (LOADCONST_i32 24)),
+  (LOADCONST_i32 24)),
+ (SHR_i32
+  (SHL_i32
+(IL_ASINT_i8 GPRI8:$src),
+   (LOADCONST_i32 24)),
+  (LOADCONST_i32 31))) >;
+
+
+def astoi_i32:Pat < (i32 (anyext GPRI16:$src)),
+(IL_ASINT_i32
+ (USHR_i32
+  (SHL_i32
+(IL_ASINT_i16 GPRI16:$src),
+   (LOADCONST_i32 16)),
+  (LOADCONST_i32 16))) >;
+
+
+def ustoi_i32:Pat < (i32 (zext GPRI16:$src)),
+(IL_ASINT_i32
+ (USHR_i32
+  (SHL_i32
+(IL_ASINT_i16 GPRI16:$src),
+   (LOADCONST_i32 16)),
+  (LOADCONST_i32 16))) >;
+
+
+def sstoi_i32:Pat < (i32 (sext GPRI16:$src)),
+(IL_ASINT_i32
+ (SHR_i32
+  (SHL_i32
+(IL_ASINT_i16 GPRI16:$src),
+   (LOADCONST_i32 16)),
+  (LOADCONST_i32 16))) >;
+
+
+def astol_i64:Pat < (i64 (anyext GPRI16:$src)),
+(LCREATE
+ (USHR_i32
+  (SHL_i32
+(IL_ASINT_i16 GPRI16:$src),
+   (LOADCONST_i32 16)),
+  (LOADCONST_i32 16)),
+ (LOADCONST_i32 0)) >;
+
+
+def ustol_i64:Pat < (i64 (zext GPRI16:$src)),
+(LCREATE
+ (USHR_i32
+  (SHL_i32
+(IL_ASINT_i16 GPRI16:$src),
+   (LOADCONST_i32 16)),
+  (LOADCONST_i32 16)),
+ (LOADCONST_i32 0)) >;
+
+
+def sstol_i64:Pat < (i64 (sext GPRI16:$src)),
+(LCREATE
+ (SHR_i32
+  (SHL_i32
+(IL_ASINT_i16 GPRI16:$src),
+   (LOADCONST_i32 16)),
+  (LOADCONST_i32 16)),
+ (SHR_i32
+  (SHL_i32
+(IL_ASINT_i16 GPRI16:$src),
+   (LOADCONST_i32 16)),
+  (LOADCONST_i32 31))) >;
+
+
+def aitol_i64:Pat < (i64 (anyext GPRI32:$src)),
+(LCREATE
+ (USHR_i32
+  (SHL_i32
+(IL_ASINT_i32 GPRI32:$src),
+   (LOADCONST_i32 0)),
+  (LOADCONST_i32 0)),
+ (LOADCONST_i32 0)) >;
+
+
+def uitol_i64:Pat < (i64 (zext GPRI32:$src)),
+(LCREATE
+ (USHR_i32
+  (SHL_i32
+(IL_ASINT_i32 GPRI32:$src),
+   (LOADCONST_i32 0)),
+  (LOADCONST_i32 0)),
+ (LOADCONST_i32 0)) >;
+
+
+def sitol_i64:Pat < (i64 (sext GPRI32:$src)),
+(LCREATE
+ (SHR_i32
+  (SHL_i32
+(IL_ASINT_i32 GPRI32:$src),
+   (LOADCONST_i32 0)),
+  (LOADCONST_i32 0)),
+ (SHR_i32
+  (SHL_i32
+(IL_ASINT_i32 GPRI32:$src),
+   (LOADCONST_i32 0)),
+  (LOADCONST_i32 31))) >;
+
+
+
+def sctof_f32:Pat < (f32 (sint_to_fp GPRI8:$src)),
+(f32
+ (ITOF
+  (SHR_i32
+   (SHL_i32
+(IL_ASINT_i8 GPRI8:$src),
+    (LOADCONST_i32 24)),
+   (LOADCONST_i32 24)))) >;
+
+
+def uctof_f32:Pat < (f32 (uint_to_fp GPRI8:$src)),
+(f32
+ (UTOF
+  (USHR_i32
+   (SHL_i32
+(IL_ASINT_i8 GPRI8:$src),
+    (LOADCONST_i32 24)),
+   (LOADCONST_i32 24)))) >;
+
+
+def ftosc_i8:Pat < (i8 (fp_to_sint GPRF32:$src)),
+(i8
+ (IL_ASCHAR_i32
+  (BINARY_AND_i32
+(FTOI GPRF32:$src),
+   (LOADCONST_i32 0x000000FF)))) >;
+
+
+def ftouc_i8:Pat < (i8 (fp_to_uint GPRF32:$src)),
+(i8
+ (IL_ASCHAR_i32
+  (BINARY_AND_i32
+(FTOU GPRF32:$src),
+   (LOADCONST_i32 0x000000FF)))) >;
+
+
+def sctod_f64:Pat < (f64 (sint_to_fp GPRI8:$src)),
+(f64 (FTOD
+      (ITOF
+       (SHR_i32
+  (SHL_i32
+(IL_ASINT_i8 GPRI8:$src),
+   (LOADCONST_i32 24)),
+  (LOADCONST_i32 24))))) >;
+
+
+def uctod_f64:Pat < (f64 (uint_to_fp GPRI8:$src)),
+(f64 (FTOD
+      (UTOF
+       (USHR_i32
+  (SHL_i32
+(IL_ASINT_i8 GPRI8:$src),
+   (LOADCONST_i32 24)),
+  (LOADCONST_i32 24))))) >;
+
+
+def dtosc_i8:Pat < (i8 (fp_to_sint GPRF64:$src)),
+(i8
+ (IL_ASCHAR_i32
+  (BINARY_AND_i32
+(FTOI (DTOF GPRF64:$src)),
+   (LOADCONST_i32 0x000000FF)))) >;
+
+
+def dtouc_i8:Pat < (i8 (fp_to_uint GPRF64:$src)),
+(i8
+ (IL_ASCHAR_i32
+  (BINARY_AND_i32
+(FTOU (DTOF GPRF64:$src)),
+   (LOADCONST_i32 0x000000FF)))) >;
+
+
+def sstof_f32:Pat < (f32 (sint_to_fp GPRI16:$src)),
+(f32
+ (ITOF
+  (SHR_i32
+   (SHL_i32
+(IL_ASINT_i16 GPRI16:$src),
+    (LOADCONST_i32 16)),
+   (LOADCONST_i32 16)))) >;
+
+
+def ustof_f32:Pat < (f32 (uint_to_fp GPRI16:$src)),
+(f32
+ (UTOF
+  (USHR_i32
+   (SHL_i32
+(IL_ASINT_i16 GPRI16:$src),
+    (LOADCONST_i32 16)),
+   (LOADCONST_i32 16)))) >;
+
+
+def ftoss_i16:Pat < (i16 (fp_to_sint GPRF32:$src)),
+(i16
+ (IL_ASSHORT_i32
+  (BINARY_AND_i32
+(FTOI GPRF32:$src),
+   (LOADCONST_i32 0x0000FFFF)))) >;
+
+
+def ftous_i16:Pat < (i16 (fp_to_uint GPRF32:$src)),
+(i16
+ (IL_ASSHORT_i32
+  (BINARY_AND_i32
+(FTOU GPRF32:$src),
+   (LOADCONST_i32 0x0000FFFF)))) >;
+
+
+def sstod_f64:Pat < (f64 (sint_to_fp GPRI16:$src)),
+(f64 (FTOD
+      (ITOF
+       (SHR_i32
+  (SHL_i32
+(IL_ASINT_i16 GPRI16:$src),
+   (LOADCONST_i32 16)),
+  (LOADCONST_i32 16))))) >;
+
+
+def ustod_f64:Pat < (f64 (uint_to_fp GPRI16:$src)),
+(f64 (FTOD
+      (UTOF
+       (USHR_i32
+  (SHL_i32
+(IL_ASINT_i16 GPRI16:$src),
+   (LOADCONST_i32 16)),
+  (LOADCONST_i32 16))))) >;
+
+
+def dtoss_i16:Pat < (i16 (fp_to_sint GPRF64:$src)),
+(i16
+ (IL_ASSHORT_i32
+  (BINARY_AND_i32
+(FTOI (DTOF GPRF64:$src)),
+   (LOADCONST_i32 0x0000FFFF)))) >;
+
+
+def dtous_i16:Pat < (i16 (fp_to_uint GPRF64:$src)),
+(i16
+ (IL_ASSHORT_i32
+  (BINARY_AND_i32
+(FTOU (DTOF GPRF64:$src)),
+   (LOADCONST_i32 0x0000FFFF)))) >;
+
+
+
+
+
+def stoc_i8:Pat < (i8 (trunc GPRI16:$src)),
+(IL_ASCHAR_i32
+   (IL_ASINT_i16
+(BINARY_AND_i16 GPRI16:$src,
+     (LOADCONST_i16 0x000000FF)))
+  ) >;
+
+
+def itoc_i8:Pat < (i8 (trunc GPRI32:$src)),
+(IL_ASCHAR_i32
+   (IL_ASINT_i32
+(BINARY_AND_i32 GPRI32:$src,
+     (LOADCONST_i32 0x000000FF)))
+  ) >;
+
+
+def itos_i16:Pat < (i16 (trunc GPRI32:$src)),
+(IL_ASSHORT_i32
+   (IL_ASINT_i32
+(BINARY_AND_i32 GPRI32:$src,
+     (LOADCONST_i32 0x0000FFFF)))
+  ) >;
+
+
+def ltoc_i8:Pat < (i8 (trunc GPRI64:$src)),
+(IL_ASCHAR_i32
+   (BINARY_AND_i32
+(LLO GPRI64:$src),
+    (LOADCONST_i32 0x000000FF))
+  ) >;
+
+
+def ltos_i16:Pat < (i16 (trunc GPRI64:$src)),
+(IL_ASSHORT_i32
+   (BINARY_AND_i32
+(LLO GPRI64:$src),
+    (LOADCONST_i32 0x0000FFFF))
+  ) >;
+
+
+def ltoi_i32:Pat < (i32 (trunc GPRI64:$src)),
+(IL_ASINT_i32
+   (BINARY_AND_i32
+(LLO GPRI64:$src),
+    (LOADCONST_i32 0xFFFFFFFF))
+  ) >;
+
+
+def actos_v2i16:Pat < (v2i16 (anyext GPRV2I8:$src)),
+(IL_ASV2SHORT_v2i32
+ (USHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 24))),
+  (VCREATE_v2i32 (LOADCONST_i32 24)))) >;
+
+
+def uctos_v2i16:Pat < (v2i16 (zext GPRV2I8:$src)),
+(IL_ASV2SHORT_v2i32
+ (USHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 24))),
+  (VCREATE_v2i32 (LOADCONST_i32 24)))) >;
+
+
+def sctos_v2i16:Pat < (v2i16 (sext GPRV2I8:$src)),
+(IL_ASV2SHORT_v2i32
+ (SHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 24))),
+  (VCREATE_v2i32 (LOADCONST_i32 24)))) >;
+
+
+def actoi_v2i32:Pat < (v2i32 (anyext GPRV2I8:$src)),
+(IL_ASV2INT_v2i32
+ (USHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 24))),
+  (VCREATE_v2i32 (LOADCONST_i32 24)))) >;
+
+
+def uctoi_v2i32:Pat < (v2i32 (zext GPRV2I8:$src)),
+(IL_ASV2INT_v2i32
+ (USHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 24))),
+  (VCREATE_v2i32 (LOADCONST_i32 24)))) >;
+
+
+def sctoi_v2i32:Pat < (v2i32 (sext GPRV2I8:$src)),
+(IL_ASV2INT_v2i32
+ (SHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 24))),
+  (VCREATE_v2i32 (LOADCONST_i32 24)))) >;
+
+
+def actol_v2i64:Pat < (v2i64 (anyext GPRV2I8:$src)),
+(LCREATE_v2i64
+ (USHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 24))),
+  (VCREATE_v2i32 (LOADCONST_i32 24))),
+ (VCREATE_v2i32 (LOADCONST_i32 0))) >;
+
+
+def uctol_v2i64:Pat < (v2i64 (zext GPRV2I8:$src)),
+(LCREATE_v2i64
+ (USHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 24))),
+  (VCREATE_v2i32 (LOADCONST_i32 24))),
+ (VCREATE_v2i32 (LOADCONST_i32 0))) >;
+
+
+def sctol_v2i64:Pat < (v2i64 (sext GPRV2I8:$src)),
+(LCREATE_v2i64
+ (SHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 24))),
+  (VCREATE_v2i32 (LOADCONST_i32 24))),
+ (SHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 24))),
+  (VCREATE_v2i32 (LOADCONST_i32 31)))) >;
+
+
+def astoi_v2i32:Pat < (v2i32 (anyext GPRV2I16:$src)),
+(IL_ASV2INT_v2i32
+ (USHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i16 GPRV2I16:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 16))),
+  (VCREATE_v2i32 (LOADCONST_i32 16)))) >;
+
+
+def ustoi_v2i32:Pat < (v2i32 (zext GPRV2I16:$src)),
+(IL_ASV2INT_v2i32
+ (USHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i16 GPRV2I16:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 16))),
+  (VCREATE_v2i32 (LOADCONST_i32 16)))) >;
+
+
+def sstoi_v2i32:Pat < (v2i32 (sext GPRV2I16:$src)),
+(IL_ASV2INT_v2i32
+ (SHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i16 GPRV2I16:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 16))),
+  (VCREATE_v2i32 (LOADCONST_i32 16)))) >;
+
+
+def astol_v2i64:Pat < (v2i64 (anyext GPRV2I16:$src)),
+(LCREATE_v2i64
+ (USHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i16 GPRV2I16:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 16))),
+  (VCREATE_v2i32 (LOADCONST_i32 16))),
+ (VCREATE_v2i32 (LOADCONST_i32 0))) >;
+
+
+def ustol_v2i64:Pat < (v2i64 (zext GPRV2I16:$src)),
+(LCREATE_v2i64
+ (USHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i16 GPRV2I16:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 16))),
+  (VCREATE_v2i32 (LOADCONST_i32 16))),
+ (VCREATE_v2i32 (LOADCONST_i32 0))) >;
+
+
+def sstol_v2i64:Pat < (v2i64 (sext GPRV2I16:$src)),
+(LCREATE_v2i64
+ (SHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i16 GPRV2I16:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 16))),
+  (VCREATE_v2i32 (LOADCONST_i32 16))),
+ (SHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i16 GPRV2I16:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 16))),
+  (VCREATE_v2i32 (LOADCONST_i32 31)))) >;
+
+
+def aitol_v2i64:Pat < (v2i64 (anyext GPRV2I32:$src)),
+(LCREATE_v2i64
+ (USHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i32 GPRV2I32:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 0))),
+  (VCREATE_v2i32 (LOADCONST_i32 0))),
+ (VCREATE_v2i32 (LOADCONST_i32 0))) >;
+
+
+def uitol_v2i64:Pat < (v2i64 (zext GPRV2I32:$src)),
+(LCREATE_v2i64
+ (USHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i32 GPRV2I32:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 0))),
+  (VCREATE_v2i32 (LOADCONST_i32 0))),
+ (VCREATE_v2i32 (LOADCONST_i32 0))) >;
+
+
+def sitol_v2i64:Pat < (v2i64 (sext GPRV2I32:$src)),
+(LCREATE_v2i64
+ (SHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i32 GPRV2I32:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 0))),
+  (VCREATE_v2i32 (LOADCONST_i32 0))),
+ (SHRVEC_v2i32
+  (SHLVEC_v2i32
+(IL_ASV2INT_v2i32 GPRV2I32:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 0))),
+  (VCREATE_v2i32 (LOADCONST_i32 31)))) >;
+
+
+
+def sctof_v2f32:Pat < (v2f32 (sint_to_fp GPRV2I8:$src)),
+(v2f32
+ (ITOF_v2f32
+  (SHRVEC_v2i32
+   (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+    (VCREATE_v2i32 (LOADCONST_i32 24))),
+   (VCREATE_v2i32 (LOADCONST_i32 24))))) >;
+
+
+def uctof_v2f32:Pat < (v2f32 (uint_to_fp GPRV2I8:$src)),
+(v2f32
+ (UTOF_v2f32
+  (USHRVEC_v2i32
+   (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+    (VCREATE_v2i32 (LOADCONST_i32 24))),
+   (VCREATE_v2i32 (LOADCONST_i32 24))))) >;
+
+
+def ftosc_v2i8:Pat < (v2i8 (fp_to_sint GPRV2F32:$src)),
+(v2i8
+ (IL_ASV2CHAR_v2i32
+  (BINARY_AND_v2i32
+(FTOI_v2i32 GPRV2F32:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 0x000000FF))))) >;
+
+
+def ftouc_v2i8:Pat < (v2i8 (fp_to_uint GPRV2F32:$src)),
+(v2i8
+ (IL_ASV2CHAR_v2i32
+  (BINARY_AND_v2i32
+(FTOU_v2i32 GPRV2F32:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 0x000000FF))))) >;
+
+def sctod_v2f64:Pat < (v2f64 (sint_to_fp GPRV2I8:$src)),
+(v2f64 
+ (VINSERT_v2f64
+ (VCREATE_v2f64 
+ (FTOD
+  (VEXTRACT_v2f32
+  (ITOF_v2f32
+   (SHRVEC_v2i32
+    (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+     (VCREATE_v2i32 (LOADCONST_i32 24))),
+    (VCREATE_v2i32 (LOADCONST_i32 24)))),
+  1)
+ )),
+ (FTOD
+  (VEXTRACT_v2f32
+  (ITOF_v2f32
+   (SHRVEC_v2i32
+    (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+     (VCREATE_v2i32 (LOADCONST_i32 24))),
+    (VCREATE_v2i32 (LOADCONST_i32 24)))),
+  2)
+  ), 1, 256)
+ ) >;
+
+def uctod_v2f64:Pat < (v2f64 (uint_to_fp GPRV2I8:$src)),
+(v2f64 
+ (VINSERT_v2f64
+ (VCREATE_v2f64 
+ (FTOD
+  (VEXTRACT_v2f32
+  (UTOF_v2f32
+   (USHRVEC_v2i32
+    (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+     (VCREATE_v2i32 (LOADCONST_i32 24))),
+    (VCREATE_v2i32 (LOADCONST_i32 24)))),
+  1)
+ )),
+ (FTOD
+  (VEXTRACT_v2f32
+  (UTOF_v2f32
+   (USHRVEC_v2i32
+    (SHLVEC_v2i32
+(IL_ASV2INT_v2i8 GPRV2I8:$src),
+     (VCREATE_v2i32 (LOADCONST_i32 24))),
+    (VCREATE_v2i32 (LOADCONST_i32 24)))),
+  2)
+  ), 1, 256)
+ ) >;
+
+
+def dtosc_v2i8:Pat < (v2i8 (fp_to_sint GPRV2F64:$src)),
+(v2i8
+ (IL_ASV2CHAR_v2i32
+  (BINARY_AND_v2i32
+(FTOI_v2i32 (VINSERT_v2f32 
+             (VCREATE_v2f32 
+              (DTOF (VEXTRACT_v2f64 GPRV2F64:$src, 1))),
+             (DTOF (VEXTRACT_v2f64 GPRV2F64:$src, 2)), 1, 256)),
+   (VCREATE_v2i32 (LOADCONST_i32 0x000000FF))))) >;
+
+
+def dtouc_v2i8:Pat < (v2i8 (fp_to_uint GPRV2F64:$src)),
+(v2i8
+ (IL_ASV2CHAR_v2i32
+  (BINARY_AND_v2i32
+(FTOU_v2i32 (VINSERT_v2f32 
+             (VCREATE_v2f32 
+              (DTOF (VEXTRACT_v2f64 GPRV2F64:$src, 1))),
+             (DTOF (VEXTRACT_v2f64 GPRV2F64:$src, 2)), 1, 256)),
+   (VCREATE_v2i32 (LOADCONST_i32 0x000000FF))))) >;
+
+
+def sstof_v2f32:Pat < (v2f32 (sint_to_fp GPRV2I16:$src)),
+(v2f32
+ (ITOF_v2f32
+  (SHRVEC_v2i32
+   (SHLVEC_v2i32
+(IL_ASV2INT_v2i16 GPRV2I16:$src),
+    (VCREATE_v2i32 (LOADCONST_i32 16))),
+   (VCREATE_v2i32 (LOADCONST_i32 16))))) >;
+
+
+def ustof_v2f32:Pat < (v2f32 (uint_to_fp GPRV2I16:$src)),
+(v2f32
+ (UTOF_v2f32
+  (USHRVEC_v2i32
+   (SHLVEC_v2i32
+(IL_ASV2INT_v2i16 GPRV2I16:$src),
+    (VCREATE_v2i32 (LOADCONST_i32 16))),
+   (VCREATE_v2i32 (LOADCONST_i32 16))))) >;
+
+
+def ftoss_v2i16:Pat < (v2i16 (fp_to_sint GPRV2F32:$src)),
+(v2i16
+ (IL_ASV2SHORT_v2i32
+  (BINARY_AND_v2i32
+(FTOI_v2i32 GPRV2F32:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 0x0000FFFF))))) >;
+
+
+def ftous_v2i16:Pat < (v2i16 (fp_to_uint GPRV2F32:$src)),
+(v2i16
+ (IL_ASV2SHORT_v2i32
+  (BINARY_AND_v2i32
+(FTOU_v2i32 GPRV2F32:$src),
+   (VCREATE_v2i32 (LOADCONST_i32 0x0000FFFF))))) >;
+
+
+def sstod_v2f64:Pat < (v2f64 (sint_to_fp GPRV2I16:$src)),
+(v2f64 
+ (VINSERT_v2f64
+ (VCREATE_v2f64 
+ (FTOD
+  (VEXTRACT_v2f32
+  (ITOF_v2f32
+   (SHRVEC_v2i32
+    (SHLVEC_v2i32
+(IL_ASV2INT_v2i16 GPRV2I16:$src),
+     (VCREATE_v2i32 (LOADCONST_i32 16))),
+    (VCREATE_v2i32 (LOADCONST_i32 16)))),
+  1)
+ )),
+ (FTOD
+  (VEXTRACT_v2f32
+  (ITOF_v2f32
+   (SHRVEC_v2i32
+    (SHLVEC_v2i32
+(IL_ASV2INT_v2i16 GPRV2I16:$src),
+     (VCREATE_v2i32 (LOADCONST_i32 16))),
+    (VCREATE_v2i32 (LOADCONST_i32 16)))),
+  2)
+  ), 1, 256)
+ ) >;
+
+def ustod_v2f64:Pat < (v2f64 (uint_to_fp GPRV2I16:$src)),
+(v2f64 
+ (VINSERT_v2f64
+ (VCREATE_v2f64 
+ (FTOD
+  (VEXTRACT_v2f32
+  (UTOF_v2f32
+   (USHRVEC_v2i32
+    (SHLVEC_v2i32
+(IL_ASV2INT_v2i16 GPRV2I16:$src),
+     (VCREATE_v2i32 (LOADCONST_i32 16))),
+    (VCREATE_v2i32 (LOADCONST_i32 16)))),
+  1)
+ )),
+ (FTOD
+  (VEXTRACT_v2f32
+  (UTOF_v2f32
+   (USHRVEC_v2i32
+    (SHLVEC_v2i32
+(IL_ASV2INT_v2i16 GPRV2I16:$src),
+     (VCREATE_v2i32 (LOADCONST_i32 16))),
+    (VCREATE_v2i32 (LOADCONST_i32 16)))),
+  2)
+  ), 1, 256)
+ ) >;
+
+
+def dtoss_v2i16:Pat < (v2i16 (fp_to_sint GPRV2F64:$src)),
+(v2i16
+ (IL_ASV2SHORT_v2i32
+  (BINARY_AND_v2i32
+(FTOI_v2i32 (VINSERT_v2f32 
+             (VCREATE_v2f32 
+              (DTOF (VEXTRACT_v2f64 GPRV2F64:$src, 1))),
+             (DTOF (VEXTRACT_v2f64 GPRV2F64:$src, 2)), 1, 256)),
+   (VCREATE_v2i32 (LOADCONST_i32 0x0000FFFF))))) >;
+
+
+def dtous_v2i16:Pat < (v2i16 (fp_to_uint GPRV2F64:$src)),
+(v2i16
+ (IL_ASV2SHORT_v2i32
+  (BINARY_AND_v2i32
+(FTOU_v2i32 (VINSERT_v2f32 
+             (VCREATE_v2f32 
+              (DTOF (VEXTRACT_v2f64 GPRV2F64:$src, 1))),
+             (DTOF (VEXTRACT_v2f64 GPRV2F64:$src, 2)), 1, 256)),
+   (VCREATE_v2i32 (LOADCONST_i32 0x0000FFFF))))) >;
+
+def stoc_v2i8:Pat < (v2i8 (trunc GPRV2I16:$src)),
+(IL_ASV2CHAR_v2i32
+   (IL_ASV2INT_v2i16
+(BINARY_AND_v2i16 GPRV2I16:$src,
+     (VCREATE_v2i16 (LOADCONST_i16 0x000000FF))))
+  ) >;
+
+
+def itoc_v2i8:Pat < (v2i8 (trunc GPRV2I32:$src)),
+(IL_ASV2CHAR_v2i32
+   (IL_ASV2INT_v2i32
+(BINARY_AND_v2i32 GPRV2I32:$src,
+     (VCREATE_v2i32 (LOADCONST_i32 0x000000FF))))
+  ) >;
+
+
+def itos_v2i16:Pat < (v2i16 (trunc GPRV2I32:$src)),
+(IL_ASV2SHORT_v2i32
+   (IL_ASV2INT_v2i32
+(BINARY_AND_v2i32 GPRV2I32:$src,
+     (VCREATE_v2i32 (LOADCONST_i32 0x0000FFFF))))
+  ) >;
+
+
+def ltoc_v2i8:Pat < (v2i8 (trunc GPRV2I64:$src)),
+(IL_ASV2CHAR_v2i32
+   (BINARY_AND_v2i32
+(LLO_v2i64 GPRV2I64:$src),
+    (VCREATE_v2i32 (LOADCONST_i32 0x000000FF)))
+  ) >;
+
+
+def ltos_v2i16:Pat < (v2i16 (trunc GPRV2I64:$src)),
+(IL_ASV2SHORT_v2i32
+   (BINARY_AND_v2i32
+(LLO_v2i64 GPRV2I64:$src),
+    (VCREATE_v2i32 (LOADCONST_i32 0x0000FFFF)))
+  ) >;
+
+
+def ltoi_v2i32:Pat < (v2i32 (trunc GPRV2I64:$src)),
+(IL_ASV2INT_v2i32
+   (BINARY_AND_v2i32
+(LLO_v2i64 GPRV2I64:$src),
+    (VCREATE_v2i32 (LOADCONST_i32 0xFFFFFFFF)))
+  ) >;
+
+
+
+
+def actos_v4i16:Pat < (v4i16 (anyext GPRV4I8:$src)),
+(IL_ASV4SHORT_v4i32
+ (USHRVEC_v4i32
+  (SHLVEC_v4i32
+(IL_ASV4INT_v4i8 GPRV4I8:$src),
+   (VCREATE_v4i32 (LOADCONST_i32 24))),
+  (VCREATE_v4i32 (LOADCONST_i32 24)))) >;
+
+
+def uctos_v4i16:Pat < (v4i16 (zext GPRV4I8:$src)),
+(IL_ASV4SHORT_v4i32
+ (USHRVEC_v4i32
+  (SHLVEC_v4i32
+(IL_ASV4INT_v4i8 GPRV4I8:$src),
+   (VCREATE_v4i32 (LOADCONST_i32 24))),
+  (VCREATE_v4i32 (LOADCONST_i32 24)))) >;
+
+
+def sctos_v4i16:Pat < (v4i16 (sext GPRV4I8:$src)),
+(IL_ASV4SHORT_v4i32
+ (SHRVEC_v4i32
+  (SHLVEC_v4i32
+(IL_ASV4INT_v4i8 GPRV4I8:$src),
+   (VCREATE_v4i32 (LOADCONST_i32 24))),
+  (VCREATE_v4i32 (LOADCONST_i32 24)))) >;
+
+
+def actoi_v4i32:Pat < (v4i32 (anyext GPRV4I8:$src)),
+(IL_ASV4INT_v4i32
+ (USHRVEC_v4i32
+  (SHLVEC_v4i32
+(IL_ASV4INT_v4i8 GPRV4I8:$src),
+   (VCREATE_v4i32 (LOADCONST_i32 24))),
+  (VCREATE_v4i32 (LOADCONST_i32 24)))) >;
+
+
+def uctoi_v4i32:Pat < (v4i32 (zext GPRV4I8:$src)),
+(IL_ASV4INT_v4i32
+ (USHRVEC_v4i32
+  (SHLVEC_v4i32
+(IL_ASV4INT_v4i8 GPRV4I8:$src),
+   (VCREATE_v4i32 (LOADCONST_i32 24))),
+  (VCREATE_v4i32 (LOADCONST_i32 24)))) >;
+
+
+def sctoi_v4i32:Pat < (v4i32 (sext GPRV4I8:$src)),
+(IL_ASV4INT_v4i32
+ (SHRVEC_v4i32
+  (SHLVEC_v4i32
+(IL_ASV4INT_v4i8 GPRV4I8:$src),
+   (VCREATE_v4i32 (LOADCONST_i32 24))),
+  (VCREATE_v4i32 (LOADCONST_i32 24)))) >;
+
+
+def astoi_v4i32:Pat < (v4i32 (anyext GPRV4I16:$src)),
+(IL_ASV4INT_v4i32
+ (USHRVEC_v4i32
+  (SHLVEC_v4i32
+(IL_ASV4INT_v4i16 GPRV4I16:$src),
+   (VCREATE_v4i32 (LOADCONST_i32 16))),
+  (VCREATE_v4i32 (LOADCONST_i32 16)))) >;
+
+
+def ustoi_v4i32:Pat < (v4i32 (zext GPRV4I16:$src)),
+(IL_ASV4INT_v4i32
+ (USHRVEC_v4i32
+  (SHLVEC_v4i32
+(IL_ASV4INT_v4i16 GPRV4I16:$src),
+   (VCREATE_v4i32 (LOADCONST_i32 16))),
+  (VCREATE_v4i32 (LOADCONST_i32 16)))) >;
+
+
+def sstoi_v4i32:Pat < (v4i32 (sext GPRV4I16:$src)),
+(IL_ASV4INT_v4i32
+ (SHRVEC_v4i32
+  (SHLVEC_v4i32
+(IL_ASV4INT_v4i16 GPRV4I16:$src),
+   (VCREATE_v4i32 (LOADCONST_i32 16))),
+  (VCREATE_v4i32 (LOADCONST_i32 16)))) >;
+
+
+
+def sctof_v4f32:Pat < (v4f32 (sint_to_fp GPRV4I8:$src)),
+(v4f32
+ (ITOF_v4f32
+  (SHRVEC_v4i32
+   (SHLVEC_v4i32
+(IL_ASV4INT_v4i8 GPRV4I8:$src),
+    (VCREATE_v4i32 (LOADCONST_i32 24))),
+   (VCREATE_v4i32 (LOADCONST_i32 24))))) >;
+
+
+def uctof_v4f32:Pat < (v4f32 (uint_to_fp GPRV4I8:$src)),
+(v4f32
+ (UTOF_v4f32
+  (USHRVEC_v4i32
+   (SHLVEC_v4i32
+(IL_ASV4INT_v4i8 GPRV4I8:$src),
+    (VCREATE_v4i32 (LOADCONST_i32 24))),
+   (VCREATE_v4i32 (LOADCONST_i32 24))))) >;
+
+
+def ftosc_v4i8:Pat < (v4i8 (fp_to_sint GPRV4F32:$src)),
+(v4i8
+ (IL_ASV4CHAR_v4i32
+  (BINARY_AND_v4i32
+(FTOI_v4i32 GPRV4F32:$src),
+   (VCREATE_v4i32 (LOADCONST_i32 0x000000FF))))) >;
+
+
+def ftouc_v4i8:Pat < (v4i8 (fp_to_uint GPRV4F32:$src)),
+(v4i8
+ (IL_ASV4CHAR_v4i32
+  (BINARY_AND_v4i32
+(FTOU_v4i32 GPRV4F32:$src),
+   (VCREATE_v4i32 (LOADCONST_i32 0x000000FF))))) >;
+
+
+def sstof_v4f32:Pat < (v4f32 (sint_to_fp GPRV4I16:$src)),
+(v4f32
+ (ITOF_v4f32
+  (SHRVEC_v4i32
+   (SHLVEC_v4i32
+(IL_ASV4INT_v4i16 GPRV4I16:$src),
+    (VCREATE_v4i32 (LOADCONST_i32 16))),
+   (VCREATE_v4i32 (LOADCONST_i32 16))))) >;
+
+
+def ustof_v4f32:Pat < (v4f32 (uint_to_fp GPRV4I16:$src)),
+(v4f32
+ (UTOF_v4f32
+  (USHRVEC_v4i32
+   (SHLVEC_v4i32
+(IL_ASV4INT_v4i16 GPRV4I16:$src),
+    (VCREATE_v4i32 (LOADCONST_i32 16))),
+   (VCREATE_v4i32 (LOADCONST_i32 16))))) >;
+
+
+def ftoss_v4i16:Pat < (v4i16 (fp_to_sint GPRV4F32:$src)),
+(v4i16
+ (IL_ASV4SHORT_v4i32
+  (BINARY_AND_v4i32
+(FTOI_v4i32 GPRV4F32:$src),
+   (VCREATE_v4i32 (LOADCONST_i32 0x0000FFFF))))) >;
+
+
+def ftous_v4i16:Pat < (v4i16 (fp_to_uint GPRV4F32:$src)),
+(v4i16
+ (IL_ASV4SHORT_v4i32
+  (BINARY_AND_v4i32
+(FTOU_v4i32 GPRV4F32:$src),
+   (VCREATE_v4i32 (LOADCONST_i32 0x0000FFFF))))) >;
+
+
+
+
+
+def stoc_v4i8:Pat < (v4i8 (trunc GPRV4I16:$src)),
+(IL_ASV4CHAR_v4i32
+   (IL_ASV4INT_v4i16
+(BINARY_AND_v4i16 GPRV4I16:$src,
+     (VCREATE_v4i16 (LOADCONST_i16 0x000000FF))))
+  ) >;
+
+
+def itoc_v4i8:Pat < (v4i8 (trunc GPRV4I32:$src)),
+(IL_ASV4CHAR_v4i32
+   (IL_ASV4INT_v4i32
+(BINARY_AND_v4i32 GPRV4I32:$src,
+     (VCREATE_v4i32 (LOADCONST_i32 0x000000FF))))
+  ) >;
+
+
+def itos_v4i16:Pat < (v4i16 (trunc GPRV4I32:$src)),
+(IL_ASV4SHORT_v4i32
+   (IL_ASV4INT_v4i32
+(BINARY_AND_v4i32 GPRV4I32:$src,
+     (VCREATE_v4i32 (LOADCONST_i32 0x0000FFFF))))
+  ) >;
+
+