• Home
  • History
  • Annotate
  • only in /frameworks/rs/driver/runtime/
History log of /frameworks/rs/driver/runtime/
Revision Date Author Comments (<<< Hide modified files) (Show modified files >>>)
b9675775b030b187b8528cba2d8e0e5c0a7bf8f7 14-Nov-2014 Stephen Hines <srhines@google.com> Fix implementation of vload with [u]long3/4 and double3/4.

Bug: 18380209

These implementations were accidentally attempting to return <3 x i64> or
<4 x i64>, or the double equivalents. The ABI requires that this be converted
into a stack return instead, so we transform our hand-written bitcode to do
exactly that.

Change-Id: I2be489b23bf639b16d8762a11a8430f40ea5b16c
l32/allocation.ll
0dbb70da81c442b85bfb4adf64793941afb93a23 21-Oct-2014 Tim Murray <timmurray@google.com> Fix AArch64 calling convention bug that prevents inlining.

bug 18071147

Change-Id: I062c919ba75c5aeb28e0358a671b81aa64a284c2
rch/asimd.ll
a1795f6ba2ced93b6811fc9c3e2d0e579af196d6 15-Oct-2014 Tim Murray <timmurray@google.com> Fix vload/vstore.

Functions were marked with readonly metadata, which breaks when
returning values on the stack.

bug 17615222

Change-Id: Ia600d0ded8e13d5a86a1bb0501b833846312e2e4
l64/allocation.ll
cbdb6480528fc374144df1ae9c604c9100d33268 09-Oct-2014 Tim Murray <timmurray@google.com> Fix missing relocation entries and libclcore entry points.

This fixes support for rsSetElementAt_ulong from pre-21 SDKs.

bug 17934731

Change-Id: I1598f612257b8faeec871df6b2f126395b0e7d82
s_allocation.c
483ef72ca767c24d17e0ffd11155b0f1ee6d167b 01-Oct-2014 Tim Murray <timmurray@google.com> Fix building librsrt_<arch>.bc.

By building for host, the LOCAL_SRC_FILES_<bitness> was no longer
reliable.

Change-Id: Ic03dcc4fa70d730f87c8783edfdd5233051b9db7
ndroid.mk
01ca8a4f99583e0fcaa980ca70415f21e0621b9b 24-Sep-2014 Tim Murray <timmurray@google.com> Fix vstore/vload/setelementat.

This contains two fixes. The first removes the readonly attribute
from vstore functions.

The second fixes 64-bit operations with vectors greater than 128
bits.

ll64/allocation.ll was based on ll32/allocation.ll, but the
AArch64 calling convention passes vectors greater than 128 bits
on the stack rather by value.

This fixes vload/vstore/SetElementAt.

bug 17615222

Change-Id: Ib8e2633f9b1236c8b360929b9ca5fe53ebb24703
l32/allocation.ll
l64/allocation.ll
s_allocation.c
1bb2eed69caa28cf8198d58db7d9134cc2f563f5 06-Sep-2014 Jean-Luc Brouillet <jeanluc@google.com> Improve rsMatrix* documentation, fix bugs

Improves the user-facing documentation. Fix the incorrect row & column
naming on the Get/Set API. Fix a bug where rsMatrixLoadMultiply could
not have the destination be one of the source,
e.g. rsMatrixLoadMultiply(&l, &l, &r)

Change-Id: I42207aacf4ebe815d4a79db2aaa9c44f85864696
s_matrix.c
33164686a7ac88d4eda38201be4127937e9c12b0 01-Sep-2014 Yong Chen <yong.a.chen@intel.com> Fix bugs for x86 platform

- Fix incorrect input/output pointer for blur intrinsic
- Avoid the negative value for length function in bc lib

Signed-off-by: Yong Chen <yong.a.chen@intel.com>

(cherry picked from commit a0cdfe05905a5a4d2e494665809c8af9d040c116)

Change-Id: I3c6e082dffdfef122999c93d1c235e99af8c62ba
rch/x86_sse3.ll
c8c7befb7d1be0c84ab3e26eab9ed0839de45dfc 29-Aug-2014 Stephen Hines <srhines@google.com> Build RS runtime libraries using our arm and aarch64 triples only.

Bug: 17333374

Without this change, we might end up using different (broken) calling
conventions on non-ARM architectures. This ensures that all C source
files get built the same. Any .ll files can still use the appropriate
target arch-specific features.

Change-Id: I2658c49620d297cb1ef8b96b5b5d1597f5fd7361
ndroid.mk
uild_bc_lib_internal.mk
bc1136f44f7816be807cfaf75a9e1e7983eace38 27-Aug-2014 Tim Murray <timmurray@google.com> Fix 32-bit clcore on ARM64.

32-bit clcore wasn't getting generic.c on ARM64, resulting in
missing symbols.

bug 17300428

Change-Id: I21969e06b620ecc667d507c5ee17cee739427178
ndroid.mk
uild_bc_lib_internal.mk
974467797f45a7e4ac191eb158edb78492a48e56 26-Aug-2014 Tim Murray <timmurray@google.com> Enable ASIMD math operations.

bug 16846318

Change-Id: Ia19cca77ef7ab6e14da9fefad522f79e25c810de
ndroid.mk
rch/asimd.ll
rch/clamp.c
d8b8f8a16415496acc9844a89599ce7f377bd04d 20-Aug-2014 Jason Sams <jsams@google.com> Fix ulong

The new 64 bit functions were mismatches between long and ulong
Make all functions consistent.

bug 16846318

Change-Id: I806f392c12af881b54aa287153bd66ad7aa1b81e
rch/clamp.c
s_cl.c
bdceed58826b6b4191a7f483764347895d0172d7 18-Aug-2014 Tim Murray <timmurray@google.com> Fix rsSetElementAtImpl and triple.

bug 16846318

Change-Id: Iac75d261aaf5c38dfd414adf73c8311005a189f0
ndroid.mk
l64/allocation.ll
l64/math.ll
1aa9dfc002f6b763d34d75d9f47abb4aa70584a2 06-Aug-2014 Tim Murray <timmurray@google.com> Bugfixes for AArch64.

- Disable use of runtime pointer until we can set it correctly in invokes.
- Disable YUV and ColorMatrix ASIMD intrinsics
- Modify runtime stubs to use large objects correctly
- Fix calling convention in allocation.ll
- Add appropriate rsr functions for compat lib and large objects

bug 16846318

Change-Id: I7f6a4ff66fd1b3ad40b08cfc9ecdda7d53e95c7d
s_allocation.c
d2d7c18c6ec3752ea19ca44157198f4a85d72464 18-Jul-2014 Jason Sams <jsams@google.com> Add native variants of more math lib functions.

Thunk though to normal impl for CPU path. This is intended
to support GPU hw with dedicated fast ops.

Change-Id: Ide52c2c5cbabc7128c84dbbdc60a240236616476
s_cl.c
a36c50a6ab87f4c9049318d4c6c8ec7b0a1e6e12 17-Jun-2014 Jason Sams <jsams@google.com> Add 64bit large object support

Add HAL hooks for filling in LO fields.

Change-Id: Ib9bdee714a062de519d2a9708c1016404a1b11a4

Conflicts:
driver/rsdBcc.cpp
driver/rsdCore.cpp
rsAllocation.h
rs_hal.h
s_allocation.c
461178e21e0141f56e63f345de3f833d0edf72a3 20-Jun-2014 Jason Sams <jsams@google.com> Fix rootn(-5, 0) to return inf

Function is equal to pow(-5, 1/0)

Change-Id: Ic17abcf76b9a36979accd84ee78ff3362ed91203
s_cl.c
fa6f90e9b5413fea556b1f4b92458f6cb9355635 12-Jun-2014 Tim Murray <timmurray@google.com> Switch to large objects for 64-bit.

Change-Id: I2dce35720b61c28cebac5bbb6791ea747c6c8a1d
l64/allocation.ll
16cfba29a921308efad37855c487d4278de5f571 12-Jun-2014 Tim Murray <timmurray@google.com> Update how the debug runtime is built for 64-bit.

Change-Id: I3b17836c789c5db3dab54737b790969c675b94c3
ndroid.mk
uild_bc_lib_internal.mk
be46aa62ed595c517dbf236dfbb3d22e1c194907 10-Jun-2014 Tim Murray <timmurray@google.com> Correctly build 64-bit runtime library by splitting into 32 and 64 components.

Change-Id: Ia520418c42e5a50840b72610c3dc71190eaaf1d7
ndroid.mk
llocation.ll
l32/allocation.ll
l32/math.ll
l64/allocation.ll
l64/math.ll
ath.ll
e3af53b643677c40d228ffd3624cf259f4dc68ed 10-Jun-2014 Tim Murray <timmurray@google.com> Update structs and defines for 64-bit.

Also adds RS_FIND_OFFSETS, which prints address info for various types that allow us to easily update __pad.

Change-Id: I4d57248e155d113c83ff69f4f8f64160aa24fb9b
s_structs.h
60fe47db884673ace2b41c6a037a376bbd0fd670 27-May-2014 Tim Murray <timmurray@google.com> Update runtime stubs.

Adds rsDebug for doubles. Also adds additional powf lookup.

Change-Id: I0f0a815a2b143102bcbf6d274d1ca4a0d897b651
ath.ll
1efb81921fc26d70255ab61da987a232b47ad236 20-May-2014 Tim Murray <timmurray@google.com> Add support for LOCAL_SRC_FILES_<extra> to build_bc_lib.mk.

Change-Id: I7f857b86bf0bfd165de461fe2036f48ae0ac9128
uild_bc_lib_internal.mk
3ff0fe77fdba8ad4a920dc27157d8c1786bb3661 20-May-2014 Jason Sams <jsams@google.com> Cleanup 64bit and remove deprecated functions

Change-Id: I8f4570af8ab7da05831258dd264118b28ce5d352
ndroid.mk
atrix.ll
sClamp.ll
s_allocation.c
s_core.c
s_element.c
s_matrix.c
s_mesh.c
s_program.c
s_sample.c
s_sampler.c
s_structs.h
c17fa065b6fee1d2176bd057de98b93d723b59e5 14-May-2014 Stephen Hines <srhines@google.com> Switch to use slang/rs_version.mk for consistency.

Change-Id: Ibfda53448570be58941133cb53c1bbb90391aa30
ndroid.mk
uild_bc_lib_internal.mk
c944fc899e868612d25d5e70e3f038bbdb4a73b2 13-May-2014 Jason Sams <jsams@google.com> Fix errors with native_powr and small numbers.

There are still small errors around zero, but the error is greatly reduced.

Change-Id: I4571130f758225b17281040abddf8cd825fad447
s_cl.c
53826db2ea7f26a241be881c2b454ab3e1e5dd50 08-May-2014 Jason Sams <jsams@google.com> Implement long/double min/max

Change-Id: I87a8fced5a9f33d40aa15f0c247c804fad09adb8
rch/generic.c
rch/neon.ll
s_cl.c
70404fa9781f13a8ca07d621f74f3810daab9704 05-May-2014 Jason Sams <jsams@google.com> Switch to using LLVM intrinsics for conversions

Replaces the hand written .ll files for conversion.

Change-Id: Ie8c6c8354847ddc84217f163060a9ee909496327
ndroid.mk
onvert.ll
s_convert.c
0f558365e0427d6e11396cf41384c45ca94012ca 03-May-2014 Jean-Luc Brouillet <jeanluc@google.com> Fix float to long convert bug.

(Extensive generated tests win!)

Change-Id: I411a77a054b8d2ed2ac9796eaedcdf3911683cc7
onvert.ll
69cf14d340132b99784a0d1497a3e9cbb23ca54b 29-Apr-2014 Jason Sams <jsams@google.com> Merge "Fix precision issues with half_recip, half_sqrt, and half_rsqrt"
ba92a7085bbb8916334a6571ff33355873883173 29-Apr-2014 Jason Sams <jsams@google.com> Fix precision issues with half_recip,
half_sqrt, and half_rsqrt

Change-Id: I37affd7e817dd66784a541beec811bb1bca701c2
rch/generic.c
rch/neon.ll
s_cl.c
3a830c08a468ffba0caa22da00001230117d7a84 26-Apr-2014 Stephen Hines <srhines@google.com> Fix x86_64 multiply defined symbol errors.

Change-Id: Ib4c2ffc8b9f9dfd57ec4e0ca76d61459dfbbd775
rch/generic.c
s_cl.c
8c24cd62187fb963cb1b76d2e91ae8c247c7f6fb 11-Apr-2014 Tim Murray <timmurray@google.com> Update fw/rs for Clang 3.5.

Change-Id: I0a9a1b1e61cb435f0102936ad9ffd51599c49e88
ndroid.mk
087fa6e19968b55e5fae7aa38971c8ec63b616c6 23-Apr-2014 Tim Murray <timmurray@google.com> Merge "Remove VP9-related runtime functions."
e195a3f57ace3b66d313a6ee88c6e93d5c9d87f4 13-Mar-2014 Tim Murray <timmurray@google.com> Make RS compile on ARM64.

bug 13280327

Change-Id: I79080c448ca17d0cba8df075737c22b24ce9a085
ndroid.mk
bcd5b9af756d10317faf54fa3742f89dfacef152 08-Mar-2014 Jean-Luc Brouillet <jeanluc@google.com> Enable more advanced CTS tests.

Add the possibility to make the generated random values for one argument
be dependent on another one. Add the possibility to add precision limits.
Change the way precision of results is verified.

This is a redo of the mess that's CL 84333 and 84332

Change-Id: I6c59ee4588627540b5bf1952adaf70b061ccc025
s_cl.c
8e1f8dce2a9c36a0b9bb0dca278beabc3716e088 04-Mar-2014 Jason Sams <jsams@google.com> Add unsigned atomics

bug 11523997

Change-Id: If0527c3c6aa193f40b5f9c94c517781c56c66252
s_core.c
039aa29085562aa0cf967e8584fd40ae818a20c6 01-Mar-2014 Jason Sams <jsams@google.com> Merge "Implement vector load/store."
41660c4c73fc425a2e3511e2070b2748cdd1107c 01-Mar-2014 Jason Sams <jsams@google.com> Implement vector load/store.

Support loading vec(2,3,4) vectors from scaler
buffers of the same component type.

Change-Id: Ice9f96d595c62ffe5e58e3d28b278417cea08fee
llocation.ll
s_allocation.c
d8c4983b1a6c975ef3a691bde0a62c34dda1197e 25-Feb-2014 Jason Sams <jsams@google.com> Add long & double conversions.

Change-Id: I236fb3d8cfed73fe8437c39a27e10f0d27a48ba1
onvert.ll
0b0bcff691d047da1d658889866c6a0347850f1c 26-Feb-2014 Jean-Luc Brouillet <jeanluc@google.com> Fix the sign issue on pown().

Change-Id: I9ee826c81c14f7c3ed137608643e5cfc0f078bce
s_cl.c
531ee4ef10f54dab0ba1d5f7248073da06907425 25-Feb-2014 Tim Murray <timmurray@google.com> Remove VP9-related runtime functions.

We've decided that it makes more sense to support this as an intrinsic rather than a set of runtime functions for kernels.

Change-Id: I9b99e4d9edc3ca185437a88c837405195eab18a2
ndroid.mk
s_dct.c
s_dct.h
s_fadst.c
s_fadst.h
s_iadst.c
s_iadst.h
s_idct.c
s_idct.h
s_walsh.c
3e0d1e79789df55021b459ae13590844b67aebd2 25-Feb-2014 Jean-Luc Brouillet <jeanluc@google.com> Fix normalize and fast_normalize.

Did not handle 0 length vectors correctly. Also fast_normalize did
not handle negative scalars correctly.

modified: rs_cl.c

Change-Id: Ide13b0bcb004009200173bc1f6f59a0303ef4f87
s_cl.c
b1657a5989c1d52a5c107bd787a2ea8616cf69ee 12-Feb-2014 Colin Cross <ccross@android.com> rs: build bc files for multiple architectures

Build bc libs for first and second architectures on multilib builds.

Change-Id: I0732e070e2b3d906132cad0b62a699bf65d34ca6
ndroid.mk
uild_bc_lib.mk
uild_bc_lib_internal.mk
1b0b0baa2e094de111f05e959f23cc9f589a99b6 19-Feb-2014 Colin Cross <ccross@android.com> Revert "rs: build bc files for multiple architectures"

This reverts commit 448c4db4867adeedf5860de7f89a401233393539.

Change-Id: I31e83868c09834228d54b8de9895a68220b70807
ndroid.mk
uild_bc_lib.mk
uild_bc_lib_internal.mk
448c4db4867adeedf5860de7f89a401233393539 12-Feb-2014 Colin Cross <ccross@android.com> rs: build bc files for multiple architectures

Build bc libs for first and second architectures on multilib builds.

Change-Id: I5eddb2e73088ae9aac4e8816ba9aefa39abcc97a
ndroid.mk
uild_bc_lib.mk
uild_bc_lib_internal.mk
d941ee179d5003617d1ab308d522b5951908a34f 13-Feb-2014 Jason Sams <jsams@google.com> Merge "Fix normalize(float)"
c3f2748adb72722206aa515381b7ee8d5ba99cc5 13-Feb-2014 Jason Sams <jsams@google.com> Fix mac build.

Change-Id: Id1c43976b8bd463ac59c793328b8544a3aeb0e89
s_idct.c
d22e2e22523f46ae4869982fdc8f3396cc3d0888 12-Feb-2014 Jason Sams <jsams@google.com> Clean up naming of VP9 support functions.

Change-Id: Ibfa931f3fb3a2ed2649733b7d7e679c0091dda90
s_dct.c
s_fadst.c
s_iadst.c
s_idct.c
s_walsh.c
afdb47f35f2c2df1cab652e7899943ddfe00692d 12-Feb-2014 Jason Sams <jsams@google.com> Merge "Add image computing oriented builtin functions"
197fc09254eace8d2625c23c192487e376b39d35 28-Jan-2014 Matthieu Delahaye <matthieu@multicorewareinc.com> Add image computing oriented builtin functions

The new built-in functions covers inverse DCT, ADST and walsh transforms.

Change-Id: Ie7044da4a5becf2cc80d066a258211721a5939d4
ndroid.mk
s_dct.c
s_dct.h
s_fadst.c
s_fadst.h
s_iadst.c
s_iadst.h
s_idct.c
s_idct.h
s_walsh.c
5eaa97c8fb83fecd4ed8eaa22a736dcb7d720c62 08-Feb-2014 Jean-Luc Brouillet <jeanluc@google.com> Fix half* failing to load

Change-Id: I9800e0a3dffe2d0ded6047b5d35af91e8996c3e6
rch/neon.ll
fce65a4abe55263645d468e36bcd42264ef5dfa3 08-Feb-2014 Jason Sams <jsams@google.com> Fix normalize(float)

Change-Id: I3770812f3ff64e9a1ab44ad0e3746541252d2116
s_cl.c
0dff3fc9ecb16468f335cb18fd19ae68a8f2231c 07-Feb-2014 Stephen Hines <srhines@google.com> Merge "Fix clz implementation for types with size < 4 bytes."
1788153fc5a35756d403a3a7e9c099b72c0e1d22 07-Feb-2014 Jason Sams <jsams@google.com> Merge "Fix half_recip decls."
beac9ca07271706b8429a676d071539b7e841f10 07-Feb-2014 Jason Sams <jsams@google.com> Fix half_recip decls.

Change-Id: I19c049adaa18d052a65923ad0935382334561ece
rch/neon.ll
c117d8dad895ab9bae4ba6077365f0dfd33ece47 07-Feb-2014 Stephen Hines <srhines@google.com> Fix clz implementation for types with size < 4 bytes.

Bug: 12911231

Change-Id: I2b612b34ae7d69c2793ba44d2b43ab96432e3828
s_cl.c
0ec1635641a2075c9d2349219632650401f88881 23-Jan-2014 Jean-Luc Brouillet <jeanluc@google.com> Implement version of step() that takes a scalar edge and a vector for the other argument.

Tests to be done in another CL.

Bug: 12112379

Change-Id: I28a865a9a8f272e9ac4808dacd1ad35a16731a0b
s_cl.c
c3cfa12028dc35494a03b0ffd802b800ed9981de 21-Dec-2013 Stephen Hines <srhines@google.com> Produce librsrt_<ARCH>.bc files as host shared libraries.

BUG: 12247226

These will be used with bcc_compat and the RenderScript support library to
ensure that we can generate optimized shared object files.

Change-Id: I67306867f9f1b70f0986eb4ff484a26bcecba373
ndroid.mk
uild_bc_lib.mk
e30a05aefa5ac6db3fac00cd67304126c7e0cd5e 04-Oct-2013 Stephen Hines <srhines@google.com> Fix target names for x86.

Change-Id: I8b890cc26c8cf7b5d8479f03395f082862378a5d
rch/x86_sse2.ll
rch/x86_sse3.ll
4da42506a08ed7fdb61615b3524f111df939fc6e 03-Oct-2013 Stephen Hines <srhines@google.com> Remove references/use of ARCH_X86_HAVE_* in frameworks/rs.

Bug: 11048298

These SSE-related defines are unnecessary because all x86 Android devices
have these features. We switch all the checks to __i386__ instead.

This also fixes an issue where non-ARM bitcode files were being created with
the ARM defines present. This is problematic when we then try to do things
that would be arch-specific (i.e. the presence of SSE-enabled functions for
clamp(), ...).

This also cleans up the different target architecture specified in the x86
bitcode files so that they match the platform target (i686-unknown-linux).

Change-Id: I4776bbdce360de26e8a00e05d2cb19341d94a173
ndroid.mk
rch/generic.c
uild_bc_lib.mk
s_cl.c
0052f8dcb5ebb5c9205a1d25445629fb5a772380 20-Sep-2013 Jason Sams <jsams@google.com> Fix for YUV bugs

Two issues, YUV intrinsic would repeat u&v for the edge padding
During the support for flexible YUV, u&v got swapped in the accessor.

bug 10826418

Change-Id: I1cf6e27c2114807d6aece4f9dd44257d4f4aa477
s_allocation.c
61656a7c6fc13421679d0a1cdf8b5b861e286892 04-Sep-2013 Jason Sams <jsams@google.com> Cleanup type offsets which cannot be calculated for flexible YUV.

Support flexible YUV

bug 10567550

Change-Id: I4f6e5a8d86eeee635605460f1751208f3320969b
(cherry picked from commit a75372759e288be3fb8835735a830b1f7d1a4c42)
s_allocation.c
s_structs.h
146e138f5c6eb4980ee6d85d33b951b87b6e8efe 20-Aug-2013 Stephen Hines <srhines@google.com> Refactor the libbcc runtime for x86 platform

This patch is merged/rebased from AOSP, where it was initially submitted to
frameworks/compile/libbcc by Jun Tian <jun.j.tian@intel.com>. All conflicts
have been resolved.

This patch refactors the libbcc runtime code to support x86 platform.
It removed the redundant x86 code and added the missing functions in
libclcore_x86.bc.
It resolved the RenderScript failures on the x86 platform.

Bug: 9961583
Change-Id: I2c8be0f710960ee5e0614721f5edfbaf028c67e1
ndroid.mk
rch/dot_length.c
rch/generic.c
rch/sqrt.c
rch/x86_clamp.ll
rch/x86_dot_length.ll
rch/x86_generic.c
rch/x86_math.ll
rch/x86_sse2.ll
rch/x86_sse3.ll
uild_bc_lib.mk
s_cl.c
9e9e6335fd281caa48378aee969a561eba964d88 07-Aug-2013 Stephen Hines <srhines@google.com> Use BUILD instead of HOST for bcc_strip_attr.

This is actually a merge of a separate patch that went into jb-mr2-dev libbcc:
https://googleplex-android-review.googlesource.com/#/c/337382/1

Change-Id: Ieb8acb6d7a4806cb7910e6eaf65e75d398b9345b
uild_bc_lib.mk
ddceab9a001f07a3395226c5e06e3b420720af0f 07-Aug-2013 Jason Sams <jsams@google.com> Refactor hal to remove cpuConsumer from drivers

This CL should minimize build breaks due to
BufferQueue changes in the future.

Change-Id: I565a6eae5cc25603741fef32f2cfcb31a32eb757
s_structs.h
1ed5ef9b253850a199eecff6b7941e57c408e509 29-Jul-2013 Tobias Grosser <grosser@google.com> Provide Renderscript TBAA metadata for allocations

Annote all common runtime functions that get or set data in an allocation
with Renderscript TBAA metadata.

Change-Id: Ie56117f5d094b7e9579e9ff926990fcacfe38019
ndroid.mk
llocation.ll
s_allocation.c
861cab4b1e81501a770509b19be4c53c006316ba 30-Jul-2013 Stephen Hines <srhines@google.com> Merge "Extract trivial rs(Get/Set)ElementAtImpl() method"
b18ab3a86d70dbdc492ab1ba925af1d2dc1c423d 22-Jul-2013 Tobias Grosser <grosser@google.com> Remove unused and incorrect code

This code misses the clamp after the multiply-add. To not confuse we remove
this code.

Change-Id: Ibf2cdc05065d710c4000165fba30887206b43106
s_core.c
98f1f05105146e1caeb124cae6e286c151f63151 10-Jul-2013 Tobias Grosser <grosser@google.com> Extract trivial rs(Get/Set)ElementAtImpl() method

Reduce redundant code and create the rs(Get/Set)ElementAtImpl() methods which
only load or store data, but which do nothing else. Keeping those functions
simple is important, as they may later be replaced by LLVM-IR implementations
annotated with RenderScript specific alias information.

The immediate benefit of this change is the removal of code duplication.
Instead we rely on LLVM to perform inlining and constant propagation. We
verified that LLVM performs these transformations successfully by both
examining the generated LLVM-IR as well as by running the image processing
benchmark. We did not measure any performance impact.

We also discussed the possibility, that other compilers may not be able to
do this (rather basic) transformation. However, this code will generally be
compiled by LLVM. Vendors that might use a different compiler normally have
also their own implementation of the run-time library, such that this change
does not apply for them.

Change-Id: I3d079785b2fd5ce8373cc6e5fa118611ec1f487e
s_allocation.c
582b3646d6634f74a13828cceb1414823c18e66f 01-Jun-2013 Stephen Hines <srhines@google.com> Take absolute value of value for fast_length().

Bug: 9018865
Change-Id: I3f6539a4f97d0b1276341adc0752ce510f13cda5
s_cl.c
080e2ef14e9090b38a37197d1bf5c5eba69cc77f 30-May-2013 Tobias Grosser <grosser@google.com> Build NEON run-time when targetting cortex-a15

The build of the NEON run-time library has been disabled for cortex-a15, as
performance with NEON enabled was lower than performance without NEON. We now
have ideas why this was the case. To enable testing of NEON performance and
later to switch on NEON by default, we now enabled the build of the library.

Change-Id: Iaf44e221a124d7829425c47a7e36aa43dc2fd02d
ndroid.mk
5a47020542c52af3e879c1cd67674ca979ff0a18 30-May-2013 Stephen Hines <srhines@google.com> Move RS runtime to frameworks/rs.

Bug: 7342767
Change-Id: Ia45064a5257b8ce460918f327670e3be550d4b56
ndroid.mk
rch/clamp.c
rch/dot_length.c
rch/generic.c
rch/neon.ll
rch/sqrt.c
rch/x86_clamp.ll
rch/x86_dot_length.ll
rch/x86_generic.c
rch/x86_math.ll
uild_bc_lib.mk
onvert.ll
ath.ll
atrix.ll
sClamp.ll
s_allocation.c
s_cl.c
s_core.c
s_element.c
s_matrix.c
s_mesh.c
s_program.c
s_sample.c
s_sampler.c
s_structs.h