glibc/sysdeps/aarch64/multiarch/Makefile

ifeq ($(subdir),string)
sysdep_routines += memcpy_generic memcpy_thunderx memcpy_falkor \
		   memmove_falkor memset_generic memset_falkor
endif
aarch64: Thunderx specific memcpy and memmove * sysdeps/aarch64/memcpy.S (MEMMOVE, MEMCPY): New macros. (memmove): Use MEMMOVE for name. (memcpy): Use MEMCPY for name. Change internal labels to external labels. * sysdeps/aarch64/multiarch/Makefile: New file. * sysdeps/aarch64/multiarch/ifunc-impl-list.c: Likewise. * sysdeps/aarch64/multiarch/init-arch.h: Likewise. * sysdeps/aarch64/multiarch/memcpy.c: Likewise. * sysdeps/aarch64/multiarch/memcpy_generic.S: Likewise. * sysdeps/aarch64/multiarch/memcpy_thunderx.S: Likewise. * sysdeps/aarch64/multiarch/memmove.c: Likewise. 2017-05-24 23:46:48 +00:00			`ifeq ($(subdir),string)`
aarch64: Optimized implementation of memmove for Qualcomm Falkor This is an optimized memmove implementation for the Qualcomm Falkor processor core. Due to the way the falkor memcpy needs to be written, code cannot be easily shared between memmove and memcpy like in case of other aarch64 memcpy implementations due to which this routine is separate. The underlying principle is the same as that of memcpy where it tries to use registers with the same lower 4 bits for fetching the same stream, thus optimizing hardware prefetcher performance. The memcpy copy loop copies 64 bytes at a time using the same register pair since that's the way to train the hardware prefetcher on the falkor core. memmove cannot quite do that since it needs to avoid overlaps, so it does the next best thing, i.e. has a 32 byte loop with a 32 byte end (prefetch a loop ahead to account for overlapping locations) with register pairs that alias so that they hit the same prefetcher. Due to this difference in loop size, they have to currently be separate implementations but efforts are on to try and get memmove to fall back into memcpy whenever it can without simply duplicating all of the code. Performance: The routine fares around 20-25% better than the generic memmove for most medium to large sizes (i.e. > 128 bytes) for the new walking memmove benchmark (memmove-walk) with an unexplained regression between 1K and 2K. The minor regression is something worth looking into for us, but the remaining gains are significant enough that we would like this included upstream as we looking into the cause for the regression. Here is a snippet of the numbers as generated from the microbenchmark by the compare_strings script. Comparisons are against __memmove_generic: Function: memmove Variant: walk __memmove_thunderx __memmove_falkor __memmove_generic ======================================================================================================================== <snip> length=16384: 12508800.00 ( 6.09%) 11486800.00 ( 13.76%) 13319600.00 length=16400: 13614200.00 ( -0.67%) 11585000.00 ( 14.33%) 13523600.00 length=16385: 13448400.00 ( 0.10%) 11732700.00 ( 12.84%) 13461200.00 length=16399: 13594100.00 ( -0.22%) 11859600.00 ( 12.57%) 13564400.00 length=16386: 13211600.00 ( 1.13%) 11503800.00 ( 13.91%) 13362400.00 length=16398: 13218600.00 ( 2.12%) 11573200.00 ( 14.30%) 13504700.00 length=16387: 13510900.00 ( -0.37%) 11744200.00 ( 12.76%) 13461300.00 length=16397: 13603700.00 ( -0.15%) 11878200.00 ( 12.55%) 13583200.00 length=16388: 13461700.00 ( -0.13%) 11558000.00 ( 14.03%) 13444100.00 length=16396: 13517500.00 ( -0.03%) 11561300.00 ( 14.45%) 13513900.00 length=16389: 13534100.00 ( 0.17%) 11756800.00 ( 13.28%) 13556900.00 length=16395: 13585600.00 ( 0.11%) 11791800.00 ( 13.30%) 13601200.00 length=16390: 13480100.00 ( -0.13%) 11685500.00 ( 13.20%) 13462100.00 length=16394: 13529900.00 ( -0.23%) 11549800.00 ( 14.43%) 13498200.00 length=16391: 13595400.00 ( -0.26%) 11768200.00 ( 13.22%) 13560600.00 length=16393: 13567000.00 ( 0.20%) 11779700.00 ( 13.35%) 13594700.00 length=32768: 71308800.00 ( -6.53%) 50220800.00 ( 24.98%) 66939200.00 length=32784: 72100800.00 (-11.55%) 50114100.00 ( 22.47%) 64636300.00 length=32769: 71767000.00 ( -7.10%) 51238400.00 ( 23.54%) 67010000.00 length=32783: 70113700.00 (-40.95%) 51129000.00 ( -2.78%) 49744400.00 length=32770: 71367600.00 ( -6.52%) 50244700.00 ( 25.01%) 67000900.00 length=32782: 64366700.00 ( 4.71%) 50101400.00 ( 25.83%) 67545600.00 length=32771: 71440100.00 ( -6.51%) 51263900.00 ( 23.57%) 67074900.00 length=32781: 66993000.00 ( 0.34%) 51108300.00 ( 23.97%) 67220300.00 length=32772: 71443900.00 (-60.50%) 50062100.00 (-12.47%) 44512600.00 length=32780: 71759100.00 ( -6.58%) 50263200.00 ( 25.35%) 67328600.00 length=32773: 71714900.00 (-33.21%) 51076600.00 ( 5.12%) 53835400.00 length=32779: 71756900.00 ( -6.56%) 51290800.00 ( 23.83%) 67337800.00 length=32774: 59689300.00 (-34.55%) 50068400.00 (-12.86%) 44363300.00 length=32778: 71847500.00 (-18.20%) 50084100.00 ( 17.61%) 60786500.00 length=32775: 71599300.00 ( -6.54%) 51278200.00 ( 23.70%) 67204800.00 length=32777: 71862900.00 (-60.85%) 51094000.00 (-14.36%) 44677900.00 length=65536: 282848000.00 ( -6.60%) 199187000.00 ( 24.93%) 265325000.00 length=65552: 243285000.00 (-41.61%) 198512000.00 (-15.54%) 171805000.00 length=65537: 255415000.00 (-23.47%) 202499000.00 ( 2.11%) 206858000.00 length=65551: 280122000.00 (-62.95%) 203349000.00 (-18.29%) 171911000.00 length=65538: 283676000.00 (-14.46%) 198368000.00 ( 19.96%) 247848000.00 length=65550: 275566000.00 (-51.76%) 198494000.00 ( -9.31%) 181581000.00 length=65539: 283699000.00 ( -6.58%) 203453000.00 ( 23.57%) 266195000.00 length=65549: 286572000.00 ( -6.65%) 202607000.00 ( 24.60%) 268712000.00 length=65540: 283710000.00 ( -6.59%) 199161000.00 ( 25.17%) 266160000.00 length=65548: 237573000.00 ( 11.48%) 198462000.00 ( 26.06%) 268395000.00 length=65541: 284150000.00 ( -6.58%) 203273000.00 ( 23.75%) 266600000.00 length=65547: 286250000.00 ( -6.70%) 202594000.00 ( 24.48%) 268263000.00 length=65542: 284167000.00 ( -6.60%) 199122000.00 ( 25.31%) 266584000.00 length=65546: 285656000.00 ( -6.59%) 198443000.00 ( 25.95%) 268002000.00 length=65543: 284600000.00 ( -6.58%) 203247000.00 ( 23.89%) 267030000.00 length=65545: 285665000.00 ( -6.40%) 202575000.00 ( 24.55%) 268472000.00 <snip> * sysdeps/aarch64/multiarch/Makefile (sysdep_routines): Add memmove_falkor. * sysdeps/aarch64/multiarch/ifunc-impl-list.c (__libc_ifunc_impl_list): Likewise. * sysdeps/aarch64/multiarch/memmove.c: Likewise. * sysdeps/aarch64/multiarch/memmove_falkor.S: New file. 2017-10-05 16:25:31 +00:00			`sysdep_routines += memcpy_generic memcpy_thunderx memcpy_falkor \`
aarch64: Optimized memset for falkor The generic memset reads dczid_el0 on every memset. This has a significant impact on falkor for a range of sizes because reading dczid_el0 is slow. The DZP bit in the dczid_el0 register does not change dynamically, so it is safe to read once during program startup. With this patch dczid_el0 is read once during startup and zva_size is cached. This is used to invoke the falkor-specific memset; the generic memset routine remains unchanged. The gains due to this are significant for falkor, with run time reductions as high as 48%. Here's a sample from the falkor tests: Function: memset Variant: walk simple_memset __memset_falkor __memset_generic ===================================================================== length=256, char=0: 139.96 (-698.28%) 9.07 ( 48.26%) 17.53 length=257, char=0: 140.50 (-699.03%) 9.53 ( 45.80%) 17.58 length=258, char=0: 140.96 (-703.95%) 9.58 ( 45.36%) 17.53 length=259, char=0: 141.56 (-705.16%) 9.53 ( 45.79%) 17.58 length=260, char=0: 142.15 (-710.76%) 9.57 ( 45.39%) 17.53 length=261, char=0: 142.50 (-710.39%) 9.53 ( 45.78%) 17.58 length=262, char=0: 142.97 (-715.09%) 9.57 ( 45.42%) 17.54 length=263, char=0: 143.51 (-716.18%) 9.53 ( 45.80%) 17.58 length=264, char=0: 143.93 (-720.55%) 9.58 ( 45.39%) 17.54 length=265, char=0: 144.56 (-722.07%) 9.53 ( 45.80%) 17.59 length=266, char=0: 144.98 (-726.42%) 9.58 ( 45.42%) 17.54 length=267, char=0: 145.53 (-727.53%) 9.53 ( 45.80%) 17.59 length=268, char=0: 146.25 (-731.81%) 9.53 ( 45.79%) 17.58 length=269, char=0: 146.52 (-735.39%) 9.53 ( 45.66%) 17.54 length=270, char=0: 146.97 (-735.81%) 9.53 ( 45.80%) 17.58 length=271, char=0: 147.54 (-741.08%) 9.58 ( 45.38%) 17.54 length=512, char=0: 268.26 (-1307.85%) 12.06 ( 36.71%) 19.05 length=513, char=0: 268.73 (-1273.89%) 13.56 ( 30.68%) 19.56 length=514, char=0: 269.31 (-1276.89%) 13.56 ( 30.68%) 19.56 length=515, char=0: 269.73 (-1279.05%) 13.56 ( 30.68%) 19.56 length=516, char=0: 270.34 (-1282.24%) 13.56 ( 30.67%) 19.56 length=517, char=0: 270.83 (-1284.71%) 13.56 ( 30.66%) 19.56 length=518, char=0: 271.20 (-1286.54%) 13.56 ( 30.67%) 19.56 length=519, char=0: 271.67 (-1288.67%) 13.65 ( 30.24%) 19.56 length=520, char=0: 272.14 (-1291.04%) 13.65 ( 30.22%) 19.56 length=521, char=0: 272.66 (-1293.69%) 13.65 ( 30.23%) 19.56 length=522, char=0: 273.14 (-1296.13%) 13.65 ( 30.20%) 19.56 length=523, char=0: 273.64 (-1298.75%) 13.65 ( 30.23%) 19.56 length=524, char=0: 274.34 (-1302.16%) 13.66 ( 30.20%) 19.57 length=525, char=0: 274.64 (-1297.78%) 13.56 ( 30.99%) 19.65 length=526, char=0: 275.20 (-1300.04%) 13.56 ( 31.01%) 19.66 length=527, char=0: 275.66 (-1302.86%) 13.56 ( 30.99%) 19.65 length=1024, char=0: 524.46 (-2169.75%) 20.12 ( 12.92%) 23.11 length=1025, char=0: 525.14 (-2124.63%) 21.62 ( 8.40%) 23.61 length=1026, char=0: 525.59 (-2125.36%) 21.88 ( 7.37%) 23.62 length=1027, char=0: 525.98 (-2127.14%) 21.62 ( 8.46%) 23.62 length=1028, char=0: 526.68 (-2131.10%) 21.62 ( 8.42%) 23.61 length=1029, char=0: 527.10 (-2131.70%) 21.79 ( 7.73%) 23.62 length=1030, char=0: 527.54 (-2118.51%) 21.62 ( 9.10%) 23.78 length=1031, char=0: 527.98 (-2136.37%) 21.62 ( 8.43%) 23.61 length=1032, char=0: 528.70 (-2139.38%) 21.62 ( 8.43%) 23.61 length=1033, char=0: 529.25 (-2124.37%) 21.62 ( 9.11%) 23.79 length=1034, char=0: 529.48 (-2142.95%) 21.62 ( 8.43%) 23.61 length=1035, char=0: 530.11 (-2145.13%) 21.62 ( 8.44%) 23.61 length=1036, char=0: 530.76 (-2147.10%) 21.79 ( 7.73%) 23.62 length=1037, char=0: 531.03 (-2149.45%) 21.62 ( 8.42%) 23.61 length=1038, char=0: 531.64 (-2151.87%) 21.62 ( 8.42%) 23.61 length=1039, char=0: 531.99 (-2151.63%) 21.80 ( 7.75%) 23.63 * sysdeps/aarch64/memset-reg.h: New file. * sysdeps/aarch64/memset.S: Use it. (__memset): Rename to MEMSET macro. [ZVA_MACRO]: Use zva_macro. * sysdeps/aarch64/multiarch/Makefile (sysdep_routines): Add memset_generic and memset_falkor. * sysdeps/aarch64/multiarch/ifunc-impl-list.c (__libc_ifunc_impl_list): Add memset ifuncs. * sysdeps/aarch64/multiarch/init-arch.h (INIT_ARCH): New local variable zva_size. * sysdeps/aarch64/multiarch/memset.c: New file. * sysdeps/aarch64/multiarch/memset_generic.S: New file. * sysdeps/aarch64/multiarch/memset_falkor.S: New file. * sysdeps/aarch64/multiarch/rtld-memset.S: New file. * sysdeps/unix/sysv/linux/aarch64/cpu-features.c (DCZID_DZP_MASK): New macro. (DCZID_BS_MASK): Likewise. (init_cpu_features): Read and set zva_size. * sysdeps/unix/sysv/linux/aarch64/cpu-features.h (struct cpu_features): New member zva_size. 2017-11-20 12:55:04 +00:00			`memmove_falkor memset_generic memset_falkor`
aarch64: Thunderx specific memcpy and memmove * sysdeps/aarch64/memcpy.S (MEMMOVE, MEMCPY): New macros. (memmove): Use MEMMOVE for name. (memcpy): Use MEMCPY for name. Change internal labels to external labels. * sysdeps/aarch64/multiarch/Makefile: New file. * sysdeps/aarch64/multiarch/ifunc-impl-list.c: Likewise. * sysdeps/aarch64/multiarch/init-arch.h: Likewise. * sysdeps/aarch64/multiarch/memcpy.c: Likewise. * sysdeps/aarch64/multiarch/memcpy_generic.S: Likewise. * sysdeps/aarch64/multiarch/memcpy_thunderx.S: Likewise. * sysdeps/aarch64/multiarch/memmove.c: Likewise. 2017-05-24 23:46:48 +00:00			`endif`