zzh
/
gatk-3.8
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
gatk-3.8/public/VectorPairHMM/src/main/c++/define-sse-float.h

174 lines
4.4 KiB
C
Raw Normal View History

Added public license text to all C++ files
2014-03-04 01:04:00 +08:00
/*Copyright (c) 2012 The Broad Institute
*Permission is hereby granted, free of charge, to any person
*obtaining a copy of this software and associated documentation
*files (the "Software"), to deal in the Software without
*restriction, including without limitation the rights to use,
*copy, modify, merge, publish, distribute, sublicense, and/or sell
*copies of the Software, and to permit persons to whom the
*Software is furnished to do so, subject to the following
*conditions:
*The above copyright notice and this permission notice shall be
*included in all copies or substantial portions of the Software.
*THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
*EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
*OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
*NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
*HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
*WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
*FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR
*THE USE OR OTHER DEALINGS IN THE SOFTWARE.
*/
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#ifdef PRECISION
Clean up
2014-01-27 03:36:06 +08:00
#undef PRECISION
#undef MAIN_TYPE
#undef MAIN_TYPE_SIZE
#undef UNION_TYPE
#undef IF_128
#undef IF_MAIN_TYPE
#undef SHIFT_CONST1
#undef SHIFT_CONST2
#undef SHIFT_CONST3
#undef _128_TYPE
#undef SIMD_TYPE
#undef AVX_LENGTH
#undef HAP_TYPE
#undef MASK_TYPE
#undef MASK_ALL_ONES
Parallel version of the JNI for the PairHMM The JNI treats shared memory as critical memory and doesn't allow any parallel reads or writes to it until the native code finishes. This is not a problem *per se* it is the right thing to do, but we need to enable **-nct** when running the haplotype caller and with it have multiple native PairHMM running for each map call. Move to a copy based memory sharing where the JNI simply copies the memory over to C++ and then has no blocked critical memory when running, allowing -nct to work. This version is slightly (almost unnoticeably) slower with -nct 1, but scales better with -nct 2-4 (we haven't tested anything beyond that because we know the GATK falls apart with higher levels of parallelism * Make VECTOR_LOGLESS_CACHING the default implementation for PairHMM. * Changed version number in pom.xml under public/VectorPairHMM * VectorPairHMM can now be compiled using gcc 4.8.x * Modified define-* to get rid of gcc warnings for extra tokens after #undefs * Added a Linux kernel version check for AVX - gcc's __builtin_cpu_supports function does not check whether the kernel supports AVX or not. * Updated PairHMM profiling code to update and print numbers only in single-thread mode * Edited README.md, pom.xml and Makefile for users to pass path to gcc 4.8.x if necessary * Moved all cpuid inline assembly to single function Changed info message to clog from cinfo * Modified version in pom.xml in VectorPairHMM from 3.1 to 3.2 * Deleted some unnecessary code * Modified C++ sandbox to print per interval timing
2014-03-18 02:42:19 +08:00
#undef VEC_EXTRACT_UNIT
#undef VEC_INSERT_UNIT
#undef SET_VEC_ZERO
#undef VEC_OR
#undef VEC_ADD
#undef VEC_SUB
#undef VEC_MUL
#undef VEC_DIV
#undef VEC_BLEND
#undef VEC_BLENDV
#undef VEC_CAST_256_128
#undef VEC_EXTRACT_128
#undef VEC_EXTRACT_UNIT
#undef VEC_SET1_VAL128
#undef VEC_MOVE
#undef VEC_CAST_128_256
#undef VEC_INSERT_VAL
#undef VEC_CVT_128_256
#undef VEC_SET1_VAL
#undef VEC_POPCVT_CHAR
#undef VEC_LDPOPCVT_CHAR
#undef VEC_CMP_EQ
#undef VEC_SET_LSE
#undef SHIFT_HAP
Clean up
2014-01-27 03:36:06 +08:00
#undef MASK_VEC
Parallel version of the JNI for the PairHMM The JNI treats shared memory as critical memory and doesn't allow any parallel reads or writes to it until the native code finishes. This is not a problem *per se* it is the right thing to do, but we need to enable **-nct** when running the haplotype caller and with it have multiple native PairHMM running for each map call. Move to a copy based memory sharing where the JNI simply copies the memory over to C++ and then has no blocked critical memory when running, allowing -nct to work. This version is slightly (almost unnoticeably) slower with -nct 1, but scales better with -nct 2-4 (we haven't tested anything beyond that because we know the GATK falls apart with higher levels of parallelism * Make VECTOR_LOGLESS_CACHING the default implementation for PairHMM. * Changed version number in pom.xml under public/VectorPairHMM * VectorPairHMM can now be compiled using gcc 4.8.x * Modified define-* to get rid of gcc warnings for extra tokens after #undefs * Added a Linux kernel version check for AVX - gcc's __builtin_cpu_supports function does not check whether the kernel supports AVX or not. * Updated PairHMM profiling code to update and print numbers only in single-thread mode * Edited README.md, pom.xml and Makefile for users to pass path to gcc 4.8.x if necessary * Moved all cpuid inline assembly to single function Changed info message to clog from cinfo * Modified version in pom.xml in VectorPairHMM from 3.1 to 3.2 * Deleted some unnecessary code * Modified C++ sandbox to print per interval timing
2014-03-18 02:42:19 +08:00
#undef VEC_SSE_TO_AVX
#undef VEC_SHIFT_LEFT_1BIT
Clean up
2014-01-27 03:36:06 +08:00
#undef MASK_ALL_ONES
Parallel version of the JNI for the PairHMM The JNI treats shared memory as critical memory and doesn't allow any parallel reads or writes to it until the native code finishes. This is not a problem *per se* it is the right thing to do, but we need to enable **-nct** when running the haplotype caller and with it have multiple native PairHMM running for each map call. Move to a copy based memory sharing where the JNI simply copies the memory over to C++ and then has no blocked critical memory when running, allowing -nct to work. This version is slightly (almost unnoticeably) slower with -nct 1, but scales better with -nct 2-4 (we haven't tested anything beyond that because we know the GATK falls apart with higher levels of parallelism * Make VECTOR_LOGLESS_CACHING the default implementation for PairHMM. * Changed version number in pom.xml under public/VectorPairHMM * VectorPairHMM can now be compiled using gcc 4.8.x * Modified define-* to get rid of gcc warnings for extra tokens after #undefs * Added a Linux kernel version check for AVX - gcc's __builtin_cpu_supports function does not check whether the kernel supports AVX or not. * Updated PairHMM profiling code to update and print numbers only in single-thread mode * Edited README.md, pom.xml and Makefile for users to pass path to gcc 4.8.x if necessary * Moved all cpuid inline assembly to single function Changed info message to clog from cinfo * Modified version in pom.xml in VectorPairHMM from 3.1 to 3.2 * Deleted some unnecessary code * Modified C++ sandbox to print per interval timing
2014-03-18 02:42:19 +08:00
#undef COMPARE_VECS
Clean up
2014-01-27 03:36:06 +08:00
#undef _256_INT_TYPE
#undef BITMASK_VEC
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#endif
#define SSE
#define PRECISION s
#define MAIN_TYPE float
#define MAIN_TYPE_SIZE 32
#define UNION_TYPE mix_F128
#define IF_128 IF_128f
#define IF_MAIN_TYPE IF_32
#define SHIFT_CONST1 3
#define SHIFT_CONST2 4
#define SHIFT_CONST3 0
#define _128_TYPE __m128
Clean up
2014-01-27 03:36:06 +08:00
#define SIMD_TYPE __m128
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define _256_INT_TYPE __m128i
#define AVX_LENGTH 4
Clean up
2014-01-27 03:36:06 +08:00
//#define MAVX_COUNT (MROWS+3)/AVX_LENGTH
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define HAP_TYPE UNION_TYPE
#define MASK_TYPE uint32_t
#define MASK_ALL_ONES 0xFFFFFFFF
Got rid of the MMX instructions in the SSE version of the code. Handling the mask operations in a class, which is defined for each version of SSE and AVX implementations separately.
2014-01-22 01:30:15 +08:00
#define MASK_VEC MaskVec_F
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_EXTRACT_UNIT(__v1, __im) \
Clean up
2014-01-27 03:36:06 +08:00
_mm_extract_epi32(__v1, __im)
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
Clean up
2014-01-27 03:36:06 +08:00
#define VEC_INSERT_UNIT(__v1,__ins,__im) \
_mm_insert_epi32(__v1,__ins,__im)
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_OR(__v1, __v2) \
Clean up
2014-01-27 03:36:06 +08:00
_mm_or_ps(__v1, __v2)
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_ADD(__v1, __v2) \
Clean up
2014-01-27 03:36:06 +08:00
_mm_add_ps(__v1, __v2)
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_SUB(__v1, __v2) \
Clean up
2014-01-27 03:36:06 +08:00
_mm_sub_ps(__v1, __v2)
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_MUL(__v1, __v2) \
Clean up
2014-01-27 03:36:06 +08:00
_mm_mul_ps(__v1, __v2)
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_DIV(__v1, __v2) \
Clean up
2014-01-27 03:36:06 +08:00
_mm_div_ps(__v1, __v2)
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_CMP_EQ(__v1, __v2) \
Clean up
2014-01-27 03:36:06 +08:00
_mm_cmpeq_ps(__v1, __v2)
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_BLEND(__v1, __v2, __mask) \
Clean up
2014-01-27 03:36:06 +08:00
_mm_blend_ps(__v1, __v2, __mask)
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_BLENDV(__v1, __v2, __maskV) \
Clean up
2014-01-27 03:36:06 +08:00
_mm_blendv_ps(__v1, __v2, __maskV)
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define SHIFT_HAP(__v1, __val) \
Clean up
2014-01-27 03:36:06 +08:00
_vector_shift_lastsses(__v1, __val.f)
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_CVT_128_256(__v1) \
Clean up
2014-01-27 03:36:06 +08:00
_mm_cvtepi32_ps(__v1.i)
#define VEC_SET1_VAL(__val) \
_mm_set1_ps(__val)
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_POPCVT_CHAR(__ch) \
Clean up
2014-01-27 03:36:06 +08:00
_mm_cvtepi32_ps(_mm_set1_epi32(__ch))
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_SET_LSE(__val) \
Clean up
2014-01-27 03:36:06 +08:00
_mm_set_ps(zero, zero, zero, __val);
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_LDPOPCVT_CHAR(__addr) \
Clean up
2014-01-27 03:36:06 +08:00
_mm_cvtepi32_ps(_mm_loadu_si128((__m128i const *)__addr))
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_SSE_TO_AVX(__vsLow, __vsHigh, __vdst) \
Clean up
2014-01-27 03:36:06 +08:00
__vdst = _mm_cvtpi32x2_ps(__vsLow, __vsHigh)
1. Integrated Mohammad's SSE4.2 code, Mustafa's bug fix and code to fix the SSE compilation warning. 2. Added code to dynamically select between AVX, SSE4.2 and normal C++ (in that order) 3. Created multiple files to compile with different compilation flags: avx_function_prototypes.cc is compiled with -xAVX while sse_function_instantiations.cc is compiled with -xSSE4.2 flag. 4. Added jniClose() and support in Java (HaplotypeCaller, PairHMMLikelihoodCalculationEngine) to call this function at the end of the program. 5. Removed debug code, kept assertions and profiling in C++ 6. Disabled OpenMP for now.
2014-01-21 00:03:42 +08:00
#define VEC_SHIFT_LEFT_1BIT(__vs) \
Clean up
2014-01-27 03:36:06 +08:00
__vs = _mm_slli_epi32(__vs, 1)
Got rid of the MMX instructions in the SSE version of the code. Handling the mask operations in a class, which is defined for each version of SSE and AVX implementations separately.
2014-01-22 01:30:15 +08:00
class BitMaskVec_sse_float {
Clean up
2014-01-27 03:36:06 +08:00
MASK_VEC combined_ ;
public:
inline MASK_TYPE& getLowEntry(int index) {
return combined_.masks[index] ;
}
inline MASK_TYPE& getHighEntry(int index) {
return combined_.masks[AVX_LENGTH/2+index] ;
}
inline const SIMD_TYPE& getCombinedMask() {
return combined_.vecf ;
}
inline void shift_left_1bit() {
VEC_SHIFT_LEFT_1BIT(combined_.vec) ;
}
Got rid of the MMX instructions in the SSE version of the code. Handling the mask operations in a class, which is defined for each version of SSE and AVX implementations separately.
2014-01-22 01:30:15 +08:00
} ;
#define BITMASK_VEC BitMaskVec_sse_float