| /netbsd-src/external/apache2/llvm/dist/llvm/docs/ |
| H A D | Vectorizers.rst | 89 Many loops cannot be vectorized including loops with complicated control flow, 96 ``-Rpass=loop-vectorize`` identifies loops that were successfully vectorized. 123 …no_switch.cpp:4:5: remark: loop not vectorized: vectorization is explicitly enabled [-Rpass-missed… 126 switch statement cannot be vectorized. 130 …no_switch.cpp:4:5: remark: loop not vectorized: loop contains a switch statement [-Rpass-analysis=… 295 Access to global structures can also be vectorized, with alias analysis being 300 ignored (as other compilers do) are still being left un-vectorized. 338 example, the loop below will be vectorized on Intel x86 if the SSE4.1 roundps 384 trip count loops to still execute in vectorized code. The diagram below shows 385 the CFG for a typical epilogue vectorized loop with runtime checks. As [all …]
|
| H A D | TransformMetadata.rst | 75 vectorized, for instance being unrolled. 94 As an example, the following instructs a loop to be vectorized and only 149 for (; i + 3 < n; i+=4) // vectorized/interleaved loop 158 the vectorized loop. If not specified, ``llvm.loop.isvectorized`` is 160 vectorized multiple times.
|
| /netbsd-src/external/apache2/llvm/dist/llvm/docs/Proposals/ |
| H A D | VectorizationPlan.rst | 30 1. Legal Step: check if a loop can be legally vectorized; encode constraints and 45 vectorizer. The output IR contains code that has been vectorized or "widened" 103 A model of a vectorized candidate for a given input IR loop or loop nest. This
|
| H A D | VectorPredication.rst | 75 vectorized scalar instructions (reduce, shuffles, ..)
|
| /netbsd-src/crypto/external/bsd/openssl/lib/libcrypto/arch/alpha/ |
| H A D | sha1-alpha.S | 51 srl $0,24,t10 # vectorized byte swap 110 srl $2,24,t10 # vectorized byte swap 169 srl $4,24,t10 # vectorized byte swap 228 srl $6,24,t10 # vectorized byte swap 287 srl $8,24,t10 # vectorized byte swap 346 srl $10,24,t10 # vectorized byte swap 405 srl $12,24,t10 # vectorized byte swap 462 srl $14,24,t10 # vectorized byte swap
|
| /netbsd-src/crypto/external/bsd/openssl.old/lib/libcrypto/arch/alpha/ |
| H A D | sha1-alpha.S | 51 srl $0,24,t10 # vectorized byte swap 110 srl $2,24,t10 # vectorized byte swap 169 srl $4,24,t10 # vectorized byte swap 228 srl $6,24,t10 # vectorized byte swap 287 srl $8,24,t10 # vectorized byte swap 346 srl $10,24,t10 # vectorized byte swap 405 srl $12,24,t10 # vectorized byte swap 462 srl $14,24,t10 # vectorized byte swap
|
| /netbsd-src/external/gpl3/gcc/dist/gcc/config/rs6000/ |
| H A D | rs6000.opt | 635 unavailable vector resources. We collect the cost for vectorized statements 636 and non-vectorized statements separately, check the proportion of vec_cost to 658 for statements to be vectorized, check the proportion of load statements, and
|
| /netbsd-src/external/gpl3/gcc.old/dist/gcc/doc/ |
| H A D | optinfo.texi | 119 successfully vectorized.
|
| H A D | tm.texi | 5949 vectors of type @var{type} in vectorized code. This might be less than 5985 vectorized variant of the function with the @code{combined_fn} code 5987 The return type of the vectorized function shall be of vector type 5993 vectorized variant of target built-in function @code{fndecl}. The 5994 return type of the vectorized function shall be of vector type 6094 …info} is non-NULL, it identifies the loop being vectorized; otherwise a single block is being vect… 6140 in vectorized loops in current function, or non-negative number if it is
|
| /netbsd-src/external/gpl3/gcc/dist/gcc/doc/ |
| H A D | optinfo.texi | 119 successfully vectorized.
|
| H A D | invoke.texi | 12416 With the @samp{unlimited} model the vectorized code-path is assumed 12418 guards the vectorized code-path to enable it only for iteration 12425 scalar code that is being vectorized. For example, if each iteration 12426 of a vectorized loop would only be able to handle exactly four iterations 14044 The minimum number of iterations under which loops are not vectorized 15045 in an inner loop relative to the loop being vectorized. The factor applied 15247 rates into account when deciding whether a loop should be vectorized 15248 using SVE, vectorized using Advanced SIMD, or not vectorized at all. 15255 would be beneficial to unroll the main vectorized loop and by how much. This 16083 For every non-vectorized conditional branch that survives gimple [all …]
|
| /netbsd-src/external/gpl3/gcc.old/dist/gcc/ |
| H A D | tree-vect-slp.c | 3268 bool vectorized = false; in vect_slp_bb_region() local 3307 vectorized = true; in vect_slp_bb_region() 3352 if (vectorized in vect_slp_bb_region() 3358 return vectorized; in vect_slp_bb_region()
|
| H A D | target.def | 1667 in vectorized loops in current function, or non-negative number if it is\n\ 1791 /* Returns a built-in function that realizes the vectorized version of 1796 vectorized variant of the function with the @code{combined_fn} code\n\ 1798 The return type of the vectorized function shall be of vector type\n\ 1803 /* Returns a built-in function that realizes the vectorized version of 1808 vectorized variant of target built-in function @code{fndecl}. The\n\ 1809 return type of the vectorized function shall be of vector type\n\ 1828 vectors of type @var{type} in vectorized code. This might be less than\n\ 2028 "non-NULL, it identifies the loop being vectorized; otherwise a single block " 2029 "is being vectorized.", [all …]
|
| /netbsd-src/external/apache2/llvm/dist/llvm/lib/Target/X86/ |
| H A D | X86.td | 339 // The idea is that throughput bound code is likely to be vectorized, so for 340 // vectorized code we should care about the throughput of SQRT operations.
|
| /netbsd-src/external/gpl3/gcc/dist/gcc/ |
| H A D | tree-vect-slp.cc | 5998 bool vectorized = false; in vect_slp_region() local 6083 if (!vectorized && dump_enabled_p ()) in vect_slp_region() 6087 vectorized = true; in vect_slp_region() 6149 if (vectorized in vect_slp_region() 6155 return vectorized; in vect_slp_region()
|
| H A D | params.opt | 1181 …ectorizer applies to the cost of statements in an inner loop relative to the loop being vectorized.
|
| H A D | ChangeLog-2020 | 2444 we vectorized a stmt. 2446 when we vectorized the stmt. 4157 * optabs.def (OPTAB_D): Define vectorized widen add, subtracts. 7764 (vect_schedule_slp): Call it. Remove ad-hoc vectorized 10128 "basic block vectorized". 12628 pattern root, track visited vectorized stmts. 13810 (vect_bb_slp_scalar_cost): Adjust for vectorized live lanes. 13821 * tree-vect-stmts.c (vect_transform_stmt): Remove vectorized 13824 helper to set vectorized cycle PHI latch values. 13827 Call maybe_set_vectorized_backedge_value for each vectorized [all …]
|
| /netbsd-src/external/apache2/llvm/dist/llvm/docs/Frontend/ |
| H A D | PerformanceTips.rst | 164 If a hot successor block contains instructions which can be vectorized
|
| /netbsd-src/external/gpl3/gcc.old/dist/gcc/config/i386/ |
| H A D | x86-tune.def | 397 written vectorized code which use i.e. _mm_set_epi16. */
|
| /netbsd-src/external/gpl3/gcc/dist/gcc/config/i386/ |
| H A D | x86-tune.def | 426 written vectorized code which use i.e. _mm_set_epi16. */
|
| /netbsd-src/external/apache2/llvm/dist/clang/docs/ |
| H A D | LanguageExtensions.rst | 3164 A vectorized loop performs multiple iterations of the original loop 3167 widths. This restricts the types of loops that can be vectorized. The vectorizer 3226 remainder loop (the tail) to be folded into the main vectorized loop. This 3283 beneficial for example when the entire loop cannot be vectorized but some of the 3300 second loop containing S2 will be vectorized.
|
| /netbsd-src/sys/external/isc/libsodium/dist/ |
| H A D | ChangeLog | 128 vectorized implementations.
|
| /netbsd-src/external/gpl3/gcc/dist/gcc/config/epiphany/ |
| H A D | epiphany.md | 331 /* lower_subreg has a tendency to muck up vectorized code.
|
| /netbsd-src/external/gpl3/gcc.old/dist/gcc/config/epiphany/ |
| H A D | epiphany.md | 331 /* lower_subreg has a tendency to muck up vectorized code.
|
| /netbsd-src/external/gpl3/gcc/dist/gcc/config/aarch64/ |
| H A D | aarch64-simd.md | 603 ;; This can be auto-vectorized to 625 ;; (vector) Dot Product operation and the vectorized optab.
|