From a0a718f9fb721be2b3677fc136ea4434b4b806d0 Mon Sep 17 00:00:00 2001
From: DingoBits <107956274+DingoBits@users.noreply.github.com>
Date: Thu, 3 Aug 2023 13:22:25 -0700
Subject: [PATCH 1/7] refactor: add leading underscore in asm for mach-o
 linking

---
 CoreClockChecker/BoostClockChecker_arm.s |   3 +
 MemoryLatency/MemoryLatency_arm.s        |  15 +++
 instructionrate/arm_instructionrate.s    | 113 ++++++++++++++++++++++-
 3 files changed, 128 insertions(+), 3 deletions(-)

diff --git a/CoreClockChecker/BoostClockChecker_arm.s b/CoreClockChecker/BoostClockChecker_arm.s
index 1fdd912..8746b6d 100644
--- a/CoreClockChecker/BoostClockChecker_arm.s
+++ b/CoreClockChecker/BoostClockChecker_arm.s
@@ -1,7 +1,10 @@
 .text
 .global clktsctest
 
+.global _clktsctest
+
 /* x0 = iterations, return elapsed TSC in x0 */
+_clktsctest:
 clktsctest:
   sub sp, sp, #0x40
   stp x10, x11, [sp, #0x10]
diff --git a/MemoryLatency/MemoryLatency_arm.s b/MemoryLatency/MemoryLatency_arm.s
index d0e99df..8c449f9 100644
--- a/MemoryLatency/MemoryLatency_arm.s
+++ b/MemoryLatency/MemoryLatency_arm.s
@@ -8,9 +8,18 @@
 .global stlftest128
 .global matchedstlftest
 
+.global _latencytest
+.global _longpatternlatencytest
+.global _preplatencyarr
+.global _stlftest
+.global _stlftest32
+.global _stlftest128
+.global _matchedstlftest
+
 /* x0 = ptr to arr
    x1 = arr len
    convert values in array from array indexes to pointers */
+_preplatencyarr:
 preplatencyarr:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -30,6 +39,7 @@ preplatencyarr_loop:
 /* x0 = iteration count
    x1 = ptr to arr
    do pointer chasing for specified iteration count */
+_latencytest:
 latencytest:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -49,6 +59,7 @@ latencytest_loop:
    x1 = ptr to arr
    do pointer chasing with longer pattern, given different patterns
    within each cacheline */
+_longpatternlatencytest:
 longpatternlatencytest:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -86,6 +97,7 @@ longpatternlatencytest_loop_inc:
 
 /* x0 = iteration count
    x1 = ptr to arr. first 32-bit int = store offset, second = load offset */
+_stlftest:
 stlftest:
   sub sp, sp, #0x40
   stp x14, x15, [sp, #0x10]
@@ -114,6 +126,7 @@ stlftest_loop:
   add sp, sp, #0x40
   ret
 
+_stlftest32:
 stlftest32:
   sub sp, sp, #0x40
   stp x14, x15, [sp, #0x10]
@@ -144,6 +157,7 @@ stlftest32_loop:
 
 /* x0 = iteration count
    x1 = ptr to arr. first 32-bit int = store offset, second = load offset */
+_stlftest128:
 stlftest128:
   sub sp, sp, #0x40
   stp x14, x15, [sp, #0x10]
@@ -172,6 +186,7 @@ stlftest128_loop:
   add sp, sp, #0x40
   ret
 
+_matchedstlftest:
 matchedstlftest:
   sub sp, sp, #0x40
   stp x14, x15, [sp, #0x10]
diff --git a/instructionrate/arm_instructionrate.s b/instructionrate/arm_instructionrate.s
index 00a2e95..70e9e41 100644
--- a/instructionrate/arm_instructionrate.s
+++ b/instructionrate/arm_instructionrate.s
@@ -17,6 +17,23 @@
 .global rortest
 .global mixmulrortest
 
+.global _clktest
+.global _addtest
+.global _addmultest
+.global _addmul21test
+.global _mixaddjmp21test
+.global _mul32test
+.global _mul64test
+.global _latmul64test
+.global _noptest
+.global _fusejmptest
+.global _jmptest
+.global _mixmuljmptest
+.global _mixmuljmptest21
+.global _mixaddjmptest
+.global _rortest
+.global _mixmulrortest
+
 .global vecadd128test
 .global latvecadd128test
 .global vecmul128test
@@ -43,6 +60,32 @@
 .global scalarfmatest
 .global latscalarfmatest
 
+.global _vecadd128test
+.global _latvecadd128test
+.global _vecmul128test
+.global _latvecmul128test
+.global _mixvecaddmul128test
+.global _faddtest
+.global _latfaddtest
+.global _latfmultest
+.global _latvecfadd128test
+.global _latvecfmul128test
+.global _vecfadd128test
+.global _vecfmul128test
+.global _mixvecfaddfmul128test
+.global _mixaddvecadd128test
+.global _mix3to1addvecadd128test
+.global _mix1to1addvecadd128test
+.global _mixmulvecmultest
+.global _mixvecmulfmultest
+.global _mixvecaddfaddtest
+.global _mixjmpvecaddtest
+.global _mixjmpvecmultest
+.global _vecfma128test
+.global _latvecfma128test
+.global _scalarfmatest
+.global _latscalarfmatest
+
 .global mixvecfaddfma128test
 .global mixvecfmulfma128test
 .global loadtest
@@ -51,6 +94,14 @@
 .global vecloadtest
 .global vecstoretest
 
+.global _mixvecfaddfma128test
+.global _mixvecfmulfma128test
+.global _loadtest
+.global _mixloadstoretest
+.global _mix21loadstoretest
+.global _vecloadtest
+.global _vecstoretest
+
 //renamer tests
 .global indepmovtest
 .global depmovtest
@@ -58,7 +109,14 @@
 .global movzerotest
 .global subzerotest
 
+.global _indepmovtest
+.global _depmovtest
+.global _xorzerotest
+.global _movzerotest
+.global _subzerotest
+
 /* x0 = arg = iteration count. all iteration counts must be divisible by 10 */
+_clktest:
 clktest:
   sub sp, sp, #0x30
   stp x14, x15, [sp, #0x10]
@@ -94,6 +152,7 @@ clktest_loop:
   add sp, sp, #0x30
   ret
 
+_noptest:
 noptest:
   sub sp, sp, #0x30
   stp x14, x15, [sp, #0x10]
@@ -141,6 +200,7 @@ noptest_loop:
   add sp, sp, #0x30
   ret
 
+_addtest:
 addtest:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -194,6 +254,7 @@ addtest_loop:
   add sp, sp, #0x50
   ret
 
+_addmultest:
 addmultest:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -238,6 +299,7 @@ addmultest_loop:
   add sp, sp, #0x50
   ret
 
+_addmul21test:
 addmul21test:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -287,7 +349,7 @@ addmul21test_loop:
   add sp, sp, #0x50
   ret 
 
-
+_mul32test:
 mul32test:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -332,6 +394,7 @@ mul32test_loop:
   add sp, sp, #0x50
   ret
 
+_mul64test:
 mul64test:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -376,6 +439,7 @@ mul64test_loop:
   add sp, sp, #0x50
   ret
 
+_latmul64test:
 latmul64test:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -416,6 +480,7 @@ latmul64test_loop:
   ret
 
 /* needs an additional parameter passed in x1 - ptr to array of 4 floats */
+_vecadd128test:
 vecadd128test:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -453,6 +518,7 @@ vecadd128test_loop:
   add sp, sp, #0x20
   ret
 
+_latvecadd128test:
 latvecadd128test:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -485,6 +551,7 @@ latvecadd128test_loop:
   add sp, sp, #0x20
   ret
 
+_faddtest:
 faddtest:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -522,6 +589,7 @@ faddtest_loop:
   add sp, sp, #0x20
   ret
 
+_latfaddtest:
 latfaddtest:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -554,6 +622,7 @@ latfaddtest_loop:
   add sp, sp, #0x20
   ret
 
+_latfmultest:
 latfmultest:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -586,6 +655,7 @@ latfmultest_loop:
   add sp, sp, #0x20
   ret
 
+_latvecmul128test:
 latvecmul128test:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -618,6 +688,7 @@ latvecmul128test_loop:
   add sp, sp, #0x20
   ret
 
+_vecmul128test:
 vecmul128test:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -655,6 +726,7 @@ vecmul128test_loop:
   add sp, sp, #0x20
   ret
 
+_scalarfmatest:
 scalarfmatest:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -696,6 +768,7 @@ scalarfmatest_loop:
   add sp, sp, #0x20
   ret
 
+_latscalarfmatest:
 latscalarfmatest:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -728,6 +801,7 @@ latscalarfmatest_loop:
   add sp, sp, #0x20
   ret
 
+_vecfma128test:
 vecfma128test:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -769,6 +843,7 @@ vecfma128test_loop:
   add sp, sp, #0x20
   ret
 
+_mixvecfmulfma128test:
 mixvecfmulfma128test:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -810,6 +885,7 @@ mixvecfmulfma128test_loop:
   add sp, sp, #0x20
   ret
 
+_mixvecfaddfma128test:
 mixvecfaddfma128test:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -851,6 +927,7 @@ mixvecfaddfma128test_loop:
   add sp, sp, #0x20
   ret
 
+_latvecfma128test:
 latvecfma128test:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -883,6 +960,7 @@ latvecfma128test_loop:
   add sp, sp, #0x20
   ret
 
+_vecfadd128test:
 vecfadd128test:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -920,6 +998,7 @@ vecfadd128test_loop:
   add sp, sp, #0x20
   ret
 
+_latvecfadd128test:
 latvecfadd128test:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -952,6 +1031,7 @@ latvecfadd128test_loop:
   add sp, sp, #0x20
   ret
 
+_vecfmul128test:
 vecfmul128test:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -989,6 +1069,7 @@ vecfmul128test_loop:
   add sp, sp, #0x20
   ret
 
+_latvecfmul128test:
 latvecfmul128test:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -1021,6 +1102,7 @@ latvecfmul128test_loop:
   add sp, sp, #0x20
   ret
 
+_mixvecfaddfmul128test:
 mixvecfaddfmul128test:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -1060,6 +1142,7 @@ mixvecfaddfmul128test_loop:
   add sp, sp, #0x20
   ret
 
+_mixvecaddmul128test:
 mixvecaddmul128test:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -1099,6 +1182,7 @@ mixvecaddmul128test_loop:
   add sp, sp, #0x20
   ret
 
+_mixaddvecadd128test:
 mixaddvecadd128test:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -1161,6 +1245,7 @@ mixaddvecadd128test_loop:
   add sp, sp, #0x50
   ret
 
+_mix3to1addvecadd128test:
 mix3to1addvecadd128test:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -1233,6 +1318,7 @@ mix3to1addvecadd128test_loop:
   add sp, sp, #0x50
   ret
 
+_mix1to1addvecadd128test:
 mix1to1addvecadd128test:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -1310,6 +1396,7 @@ mix1to1addvecadd128test_loop:
   add sp, sp, #0x50
   ret
 
+_mixmulvecmultest:
 mixmulvecmultest:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -1364,6 +1451,7 @@ mixmulvecmultest_loop:
   add sp, sp, #0x50
   ret
 
+_mixvecmulfmultest:
 mixvecmulfmultest:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -1401,6 +1489,7 @@ mixvecmulfmultest_loop:
   add sp, sp, #0x20
   ret
 
+_mixvecaddfaddtest:
 mixvecaddfaddtest:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -1438,6 +1527,7 @@ mixvecaddfaddtest_loop:
   add sp, sp, #0x20
   ret
 
+_mixjmpvecaddtest:
 mixjmpvecaddtest:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -1485,6 +1575,7 @@ mixjmpvecaddtest_jellydonut:
   add sp, sp, #0x20
   ret
 
+_mixjmpvecmultest:
 mixjmpvecmultest:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -1522,6 +1613,7 @@ mixjmpvecmultest_jellydonut:
   add sp, sp, #0x20
   ret
 
+_vecloadtest:
 vecloadtest:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -1553,6 +1645,7 @@ vecloadtest_loop:
   add sp, sp, #0x20
   ret
 
+_vecstoretest:
 vecstoretest:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -1589,6 +1682,7 @@ vecstoretest_loop:
   add sp, sp, #0x20
   ret
 
+_loadtest:
 loadtest:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -1626,6 +1720,7 @@ loadtest_loop:
   add sp, sp, #0x50
   ret
 
+_mixloadstoretest:
 mixloadstoretest:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -1663,6 +1758,7 @@ mixloadstoretest_loop:
   add sp, sp, #0x50
   ret
 
+_mix21loadstoretest:
 mix21loadstoretest:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -1710,6 +1806,7 @@ mix21loadstoretest_loop:
   add sp, sp, #0x50
   ret
 
+_jmptest:
 jmptest:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -1742,6 +1839,7 @@ jmptest_jellydonut:
   add sp, sp, #0x20
   ret
 
+_fusejmptest:
 fusejmptest:
   sub sp, sp, #0x20
   stp x14, x15, [sp, #0x10]
@@ -1772,6 +1870,7 @@ fusejmptest_jellydonut:
   add sp, sp, #0x20
   ret
 
+_mixmuljmptest:
 mixmuljmptest:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -1817,6 +1916,7 @@ mixmuljmptest_jellydonut:
   add sp, sp, #0x50
   ret
 
+_mixmuljmptest21:
 mixmuljmptest21:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -1872,6 +1972,7 @@ mixmuljmptest21_jellydonut:
   add sp, sp, #0x50
   ret
 
+_mixaddjmptest:
 mixaddjmptest:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -1917,6 +2018,7 @@ mixaddjmptest_jellydonut:
   add sp, sp, #0x50
   ret
 
+_mixaddjmp21test:
 mixaddjmp21test:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -1963,7 +2065,7 @@ mixaddjmp21test_jellydonut:
   add sp, sp, #0x50 
   ret
 
-
+_mixmulrortest:
 mixmulrortest:
   sub sp, sp, #0x80
   stp x14, x15, [sp, #0x10]
@@ -2020,7 +2122,7 @@ mixmulrortest_loop:
   add sp, sp, #0x80
   ret
 
-
+_rortest:
 rortest:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -2066,6 +2168,7 @@ rortest_jellydonut:
   add sp, sp, #0x50
   ret
 
+_depmovtest:
 depmovtest:
   sub sp, sp, #0x40
   stp x14, x15, [sp, #0x10]
@@ -2103,6 +2206,7 @@ depmovtest_loop:
   add sp, sp, #0x40
   ret
 
+_indepmovtest:
 indepmovtest:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -2142,6 +2246,7 @@ indepmovtest_loop:
   add sp, sp, #0x50
   ret
 
+_xorzerotest:
 xorzerotest:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -2180,6 +2285,7 @@ xorzerotest_loop:
   add sp, sp, #0x50
   ret
 
+_movzerotest:
 movzerotest:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]
@@ -2218,6 +2324,7 @@ movzerotest_loop:
   add sp, sp, #0x50
   ret
 
+_subzerotest:
 subzerotest:
   sub sp, sp, #0x50
   stp x14, x15, [sp, #0x10]

From 476e55c4a297ac335c15621b0b1ff8921433e472 Mon Sep 17 00:00:00 2001
From: DingoBits <107956274+DingoBits@users.noreply.github.com>
Date: Thu, 3 Aug 2023 13:25:29 -0700
Subject: [PATCH 2/7] refactor: 4-byte alignment for aarch64 functions

---
 CoreClockChecker/BoostClockChecker_arm.s | 2 ++
 MemoryBandwidth/MemoryBandwidth_arm.s    | 2 ++
 MemoryLatency/MemoryLatency_arm.s        | 2 ++
 instructionrate/arm_instructionrate.s    | 2 ++
 4 files changed, 8 insertions(+)

diff --git a/CoreClockChecker/BoostClockChecker_arm.s b/CoreClockChecker/BoostClockChecker_arm.s
index 8746b6d..ba6dac5 100644
--- a/CoreClockChecker/BoostClockChecker_arm.s
+++ b/CoreClockChecker/BoostClockChecker_arm.s
@@ -3,6 +3,8 @@
 
 .global _clktsctest
 
+.balign 4
+
 /* x0 = iterations, return elapsed TSC in x0 */
 _clktsctest:
 clktsctest:
diff --git a/MemoryBandwidth/MemoryBandwidth_arm.s b/MemoryBandwidth/MemoryBandwidth_arm.s
index fad94bc..dcdc65c 100644
--- a/MemoryBandwidth/MemoryBandwidth_arm.s
+++ b/MemoryBandwidth/MemoryBandwidth_arm.s
@@ -18,6 +18,8 @@
 .global _flush_icache
 .global _readbankconflict
 
+.balign 4
+
 /* x0 = ptr to array (was rcx)
  * x1 = arr length (was rdx)
  * x2 = iterations (was r8)
diff --git a/MemoryLatency/MemoryLatency_arm.s b/MemoryLatency/MemoryLatency_arm.s
index 8c449f9..028b5b6 100644
--- a/MemoryLatency/MemoryLatency_arm.s
+++ b/MemoryLatency/MemoryLatency_arm.s
@@ -16,6 +16,8 @@
 .global _stlftest128
 .global _matchedstlftest
 
+.balign 4
+
 /* x0 = ptr to arr
    x1 = arr len
    convert values in array from array indexes to pointers */
diff --git a/instructionrate/arm_instructionrate.s b/instructionrate/arm_instructionrate.s
index 70e9e41..2750ba0 100644
--- a/instructionrate/arm_instructionrate.s
+++ b/instructionrate/arm_instructionrate.s
@@ -115,6 +115,8 @@
 .global _movzerotest
 .global _subzerotest
 
+.balign 4
+
 /* x0 = arg = iteration count. all iteration counts must be divisible by 10 */
 _clktest:
 clktest:

From 6a86a91c01f5d816232930a8f521b27dc14657fa Mon Sep 17 00:00:00 2001
From: DingoBits <107956274+DingoBits@users.noreply.github.com>
Date: Thu, 3 Aug 2023 13:31:08 -0700
Subject: [PATCH 3/7] refactor: add conditional for numa-related code

---
 MemoryBandwidth/MemoryBandwidth.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/MemoryBandwidth/MemoryBandwidth.c b/MemoryBandwidth/MemoryBandwidth.c
index ee883a4..5a0b744 100644
--- a/MemoryBandwidth/MemoryBandwidth.c
+++ b/MemoryBandwidth/MemoryBandwidth.c
@@ -18,9 +18,10 @@
 #include <sched.h>
 #include <math.h>
 #include <sys/mman.h>
-#include <sys/sysinfo.h>
 #include <errno.h>
+
 #ifdef NUMA
+#include <sys/sysinfo.h>
 #include <numa.h>
 #endif
 
@@ -39,7 +40,9 @@ typedef struct BandwidthTestThreadData {
     uint64_t start;
     float* arr;
     float bw; // written to by the thread
+    #ifdef NUMA
     cpu_set_t cpuset; // if numa set, will set affinity
+    #endif
 } BandwidthTestThreadData;
 
 float MeasureBw(uint64_t sizeKb, uint64_t iterations, uint64_t threads, int shared, int nopBytes, int coreNode, int memNode);

From 7fef74265e0c27e85eea108054e944e06cc018d1 Mon Sep 17 00:00:00 2001
From: DingoBits <107956274+DingoBits@users.noreply.github.com>
Date: Thu, 3 Aug 2023 13:33:43 -0700
Subject: [PATCH 4/7] fix: compile madvise-related code for linux only

---
 MemoryLatency/MemoryLatency.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/MemoryLatency/MemoryLatency.c b/MemoryLatency/MemoryLatency.c
index cc5fa8f..5dd2957 100644
--- a/MemoryLatency/MemoryLatency.c
+++ b/MemoryLatency/MemoryLatency.c
@@ -204,7 +204,7 @@ int main(int argc, char* argv[]) {
         fprintf(stderr, "Usage: [-test <c/asm/tlb/mlp>] [-maxsizemb <max test size in MB>] [-iter <base iterations, default 100000000]\n");
     }
 
-#ifndef __MINGW32__
+#ifdef __linux__
     if (hugePages) {
        size_t hugePageSize = 1 << 21;
        size_t testSizeKb = singleSize ? singleSize : default_test_sizes[testSizeCount - 1];

From c45aca47055fe7aad4496682d32a294225706731 Mon Sep 17 00:00:00 2001
From: DingoBits <107956274+DingoBits@users.noreply.github.com>
Date: Thu, 3 Aug 2023 13:34:10 -0700
Subject: [PATCH 5/7] fix: typo in latencytest

---
 MemoryLatency/MemoryLatency.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/MemoryLatency/MemoryLatency.c b/MemoryLatency/MemoryLatency.c
index 5dd2957..7acb38a 100644
--- a/MemoryLatency/MemoryLatency.c
+++ b/MemoryLatency/MemoryLatency.c
@@ -17,6 +17,7 @@
 #include <numaif.h>
 #include <sys/sysinfo.h>
 #endif
+
 #include <errno.h>
 #include <sched.h>
 
@@ -595,10 +596,11 @@ float RunAsmTest(uint32_t size_kb, uint32_t iterations, uint32_t *preallocatedAr
     // Run test
     gettimeofday(&startTv, &startTz);
     #ifdef LONGPATTERN
-    if (longpattern) sum = longpatternlatencytest(scaled_iterations, A);
-    else 
-    #endif
+    if (longpattern)
+        sum = longpatternlatencytest(scaled_iterations, A);
+    else
         sum = latencytest(scaled_iterations, A);
+    #endif
     gettimeofday(&endTv, &endTz);
     uint64_t time_diff_ms = 1000 * (endTv.tv_sec - startTv.tv_sec) + ((endTv.tv_usec - startTv.tv_usec) / 1000);
     float latency = 1e6 * (float)time_diff_ms / (float)scaled_iterations;

From 147c41ab332bff2fc5dc75790ed1995705d279fb Mon Sep 17 00:00:00 2001
From: DingoBits <107956274+DingoBits@users.noreply.github.com>
Date: Thu, 3 Aug 2023 13:39:45 -0700
Subject: [PATCH 6/7] fix: explicitly declare checkExtensionSupport

---
 GpuMemLatency/instruction_rate.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/GpuMemLatency/instruction_rate.c b/GpuMemLatency/instruction_rate.c
index 2c3c264..6aa6b0d 100644
--- a/GpuMemLatency/instruction_rate.c
+++ b/GpuMemLatency/instruction_rate.c
@@ -214,6 +214,8 @@ float instruction_rate_test(cl_context context,
         float4_element_count, a_mem_obj, result_obj, A, result, opsPerIteration);
     fprintf(stderr, "INT8 G Multiplies/sec: %f\n", int8_mul_rate);
 
+    short checkExtensionSupport(const char *extension_name);
+    
     if (checkExtensionSupport("cl_khr_fp64")) {
         fp64_instruction_rate_test(context, command_queue, thread_count, local_size, chase_iterations, float4_element_count,
             a_mem_obj, result_obj, A, result);
@@ -462,4 +464,4 @@ float fp16_instruction_rate_test(cl_context context,
     fprintf(stderr, "FP16 G FMAs/sec: %f : %f FP16 GFLOPs\n", gOpsPerSec, gOpsPerSec * 2);
 
     return gOpsPerSec;
-}
+}

From 7d5de37838b13c85d8c3a979e152e9e80826f88c Mon Sep 17 00:00:00 2001
From: DingoBits <107956274+DingoBits@users.noreply.github.com>
Date: Thu, 3 Aug 2023 13:39:59 -0700
Subject: [PATCH 7/7] chore: remove old binary

---
 instructionrate/arm_instructionrate | Bin 31536 -> 0 bytes
 1 file changed, 0 insertions(+), 0 deletions(-)
 delete mode 100755 instructionrate/arm_instructionrate

diff --git a/instructionrate/arm_instructionrate b/instructionrate/arm_instructionrate
deleted file mode 100755
index 452fdef64e18695e2c07f21bd2823555f8c6533e..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 31536
zcmeHwe|%NdmG8c}0i*m16#<n=PCzsugd~Jt)*m+^5O5$OpwG6>+}tD=h$J_8$&CfI
z?e!OBMyBN!Udv~&yl9<ygsC%h#7gJs^V&KcMv&$yO_}!?8q`i1rEL@xQo+o3oxRt+
z=d62f2uSC>KRo&Dth2xSyVqWOoxS(j`~13JTd-tdmd~en^Q#w>9UG&OfJrEhGDE3=
znyW6y{|V|^H570*Lck<TdB&(_nPYf<NXj=iO9N6~Yj`Z#LXtyFnwPK^Xq2VWwXy$`
ztzAoWde;)Ik)>@6l5)r>d(-J;9_8yX`Fcz~mde=AlIurmD6BN;QdT03HkQPtZP~A)
z^|Ze+K(TarvvCAV&UYvBk)4Cxat(30DKC>>is?XMV`++4Ga4IeXEfGNZ)}LPubUnU
zwbrefT~^fATvVd#P0fe0R#o4nw*EAG+{an>-v9XjY53b!n=WsE&ia?%lig$k=_3wp
z#J|km(>SGZc?KSo@T*~JBOeGy4!;@;2fH=?5z;^B;PCEt>{uLeNXk9Y5$9d1i<h2R
z8Ssh>_zS=-1i7sMFj|6$&p_}h_ou)ugtRRftXmffu4;&c8XF!6t6;FDwILE+6|7sc
z77-lPx?o#0)EW&og&HC%RNLGdRjb3%XhT!Dc~yOAJxEBeTT2;&t3nNpE?{+gsI^{^
z7HSK#m3IrNTC%ueez3HtWRBw9nPn=tc;(VyeYiEex}hx^Ze6)_eq(bayfReVNM^2X
zYK|Dof|}|{)b-6myYZVpANrvmalQGl*Nj~{UKYZ4;@?KM$ySeJpEe&lau|~Phfep~
z>4fdE`B`c+k+Jdk;Wr&I_bQDaHsb{22{WF~y+d2Vcv-RL2L!&mRO8D8{$iQN?-BTu
zM$a07FEayZi@<AUYdz}(zU>-~#{_=D@HYzlRce6Pwh4UN)f(R^@bSj}-2x9_+@Ni*
zz_(4%ctYTT%Qe1F;D@NgVe1ljrNIvi{IT(x-y`th#{b6!UO7|q&j|dKsc()MU-<ZP
ze39Y%&`x;FCQUUC+<9Kx?7&S`?SeZT_!Zg|`+u(kAMU`<RO|MZg+k3xu2T)4aA&zW
z4%~S@$#vj7Mw6oP4%|89S`OT4f1v{(=CCv0^8ftYgD-cU^Q-vrEVcFAgRksN_=DZO
zz;-LOdGrHSeYrpWeB8>)?<Ot>JMF7TdrPO;@kbEPy4mL1@ePPibj1H0@d`)$8;I9B
z;@?925l8&nh(E344wLQOa}WNg^J&;ec2{&&^_?81`i|d_)pU7oQQmwN=zDvP>hr7E
z8^_*O@t#S(_;cX?vcLa3s^{@H)bzjk;3>b_dd%;O_tg2~|3diYt~Y+wx8se(<l`SG
zwR%U-*DE{2_r3VRJKtQ;`B%|5KX`onJ0H|3-~Pvx+Ft-&mWnNZ<oJVch96q6y+UR0
zF9-zA;fVJ>l9TmBK}}x#ug+F&`QF(__ty^3dSa29v?J%0iFrAf*W~3~9m>lougxpG
zx-d@x`yrKGZsk1%ovP-$Z>VE0d{A(ps?So14?YhYRbpZu>80`tRDOJy7ypFj&yf6@
z{P+$p{&CHpB>9Q___w_HJ2iit<Oi(y1~2|Qnm<hP6IT2IFaBQ5???M{)z>PBN4)q6
z%|9vmv4VK57ynhw|Fz`z6vXfL;y<bR|5x&BCdO~~;(uTBe<t~fiSb*#_=h$BCz2nV
z6fgDS_iFwNk{`G(e!UmJL-YSe^3~+{HD3I)n*ZM=KQTEz(u@D1=5LYwn(O0PUi?Fv
z|8>b%H^e{s#H0P6()<S{zvhPcyN2KP!>T^+FUR~>?xOzEEcvQ1e$?=ZOFZgd#I2Xy
zSYiB42N!*Ax7`;LSH+2ckL0T<@qhB-$Hv(FC6XVT691oG{DkHQB;P8H|D6{<rtK+9
z<4=u$&x@ZJYui&O`Rc~_cf9y&oXwvg`5iaLpYY-bG(T7JYo^7&=EaX`ezxQXrpMd7
z_zBI&I@Ucl1d8ITz4&UpZU2ch{*3sSy!Zjl?@8kq$M5js$29+t<i|?l^S$^9&EF^a
zJ*DwEZvJMwy{W5h`(I4)yY$#!7QfMpAJF{Wl5fq5=X>#En*WUC2WH2w^x`Kpf1Bj@
z%#L5?#aGwZ_HUB>*qr#s=X77EzKH&%>l@JgnB=Ru@!xszW11h8{KVY&+lK!z&X;_i
zFE~=w*8r|{wMFMq?aSD_IGtyH!TK>Lu<gOTpg>&}{}ai#ZnF9Nz4(Ev;&)1ZkLJJ3
zd}n=??vIr+Psiu&GX9?PIBg1yjn9?*z|A)Quf6!OvGHk=uWqsVf608OO$qJOeB?3b
zU+t&FE%9$K-)T?m+V~i$zvfn3|0B$I>Q~p=ZQ?Si-?}Y+zZXBC`5!BB+&XTv%V_Y*
zqbAs8{1$m!bBKV7|A`krq04wn@-1KdPVi&YhA^*uYfhlghxufM^1bK7eDW1sT2h#c
zog5axe6ynI+=KsQ&zF}V56_JUEB^@k@(_OISG$h+1AT8zQt^MoeAX_zr^=qcW+Qzv
zbl=Q2?W^hwU^!fNu74cO)48k;>Q4cG{3*x+$lH{9F9BO0Mm-9isp<<MPp{v%OJN>Q
z`6Ryp^V*4LzxHPN@Q0r$I9tBu74RQH-ik9-eYJMptYls)^W1~~%U(0kyqo6MH5K#c
z1$=5&jQqB{st<CHT}F1GukURCcMfA-K;u|Vg?}FT?%abvw*AKTAL;KO$NMCoqIxJ_
z9G3;gIOBUn{2CGerikB~?aog+r-H-!DZWU=f8ckgpAhk{i};U49BY!K{t8k4dJ+Ez
z5&zi4pGN)B=HlfDQ--K9&Uh8#Z2v0kJN@$)BECz+C#tOYD~|lX^hiJJNGJX^LsR9E
zzxE1${Z|pkyfA4W@$(%1p!h+N-itV`=cv8$dhYm@>Re91$~(Ej*Le!<mez46(N6;C
zi#=G|p%1=6>$>~Uwt6tHi)G!`&Fi@^uy>xQ?0g&RxnnZ}eRWvJeGh9mTH`&6bsVkn
z3RKp9Kh}9~%?R|>WKG&J>7n`Cv#@rgwH(%Xy#cK4>ad1W;DydUyr0IDK#qFiQH%-J
zaP@=_Yr10?8){(>biJ1adkVlQ$f?UK7+;fD5U9;7A77AXjW49N-O7_I$~%u=892w|
zMF2MT|3v-o-`?iyJTWJ$a|ZH%BP;gCv%V3pJ?k6w#H9>Mjoe>zuK)QHb1FK2ds}%Y
z%8k<)Aj?1K*X29=;eX~n*=e2iuyejI>8E#R_&bl^kZt>m?cL-X{@PmKu-9()UGdr!
z-{rRdHo=E$5pGAAV*L4s#@PPjH&k}MJ7YeMtMT`IpZ2#$J?=2aeJLc><BuV!9*ZFt
zAw2t)QP<jbL)vvQ@f;Js&~$DK7i!~3w2kxGNcOEY_Bm~(eX^J8LHlg;VEJTozEbZW
zINQJSLhYtL^&9`FT_+IULwFbA-w}=@*ml4!viWv|DF|fe#o`z1+w<7CebZM)ZHJB9
zVdHk#xE(fbhm9A7e_9)#{mKZGed)q^?Riv>k>sOdmAIt@p%h^zLK(s=gxn3gt4lWg
zpnCa+zprlF@LcuQ4STAe-|&3(#o}y}Jyt$y#!6ua>AsZV{AC7SD#;GKRGJfbY3A_2
zOJyqX(k#StrxxdxOfAV<KD9KjZR*Utty9bLo}W4^?_%-6*g?84Wf)YMSD?&G7|vfN
z+SLvm<4YI@Q>JJ8h7Bb<(7tz|eeXc~-huYL1MT}F@r$)%Hu~Begt-XwsNYrO=9erS
zoxl9{&*isOUzxvk`I!9Y@4hO(`-@}qPygw-{PA_;^XIL;I)AXd!PrH*E@c=@nO?^v
z%@+rYt4j`+R4+eRTHSVVX7$#CW!29goK@X@aCY_SgLA6K|7>pcyr0dh&K)H0lG=5&
zBx2ijbY{f1>*%bAZP(E`5!<d`&WqS~4T5(`?dmrE>NfuBHvZ~1{`!US*C2R<v5V?>
zDZ^mO^lCe~-DoRRr*5<rs(Uxu3hI8b_+acH-Ip>9rcBSeA1#iQP&+$X8fl~Uc(g3?
zJlbRKMdE|8gLGfYFqksE>ds?^T~|8(<O4eX<O6E^M`uS)qwQa4`lq&s&LcE`qVo&Q
zpXmI8`BV3fA6B2<@Pa<iTtK>~JsxWxTJvUF!{zvLUduG!V?WXw*}10U`JTjiUMO*%
zCk__x-k|+K^`mv_g{(!Wt|N!&^|f5DUT9sZo=%(T*g4}=UuT?-sWVRXcE&G-KJrP=
z@Bj0CLbg*IARo|rpXyHQO0M(&%k;tegs%bdwGF-|kZIleYph+**EeV#OY=io)6(3L
z*0<EZX{}3hNUSNZrG7qG9{1^tZC<R8?!^3%)@dscXq|Q!f<3O$JeI~in#a<(NAp-3
z`!51#n+95^dD=neUmDNo{7d5<oquV)c#-%(cF_3db*!jek_`#J+QZ{h4fd&TQW_o_
z+GE>@_S})B_SkkJ4fl!na9-TAOy<S?vt(Wi`<yp{>+qbn2K!_yja>uzOrE=_9gz)m
zjA|gM9a&@(jv?7V?WhBicr{cucw__ddZ>Kxa4*2-u?=)R@<MFz^5I~yfsRk6>k7+M
z?3OzbmLse{Sc&jyFi#si&WU8tR9sWO5n&p_bc7iQp9bTyGxps9?5I}CwuL0wLy~MF
zNp_JW+enhlBs1|Y#t!N~TzA&Z<^Q2^vUkAqykpEupg&L_i6PV=1Q09)g+P5IfzW{v
zL#ROrAXo?rf%-@Sf%-@cp$36{m8p*;ugl}QXwN78E1mHg5$`!C(oLMopzEqnA?_;2
zqdgqQ{N<k!rXtWa>PCdWK=?ZZYO{20lCDwGwMx1+$=4|*9z%U=5CRAmf<ov)eG>>B
z2r+~j1g`&uwAq1dq;@22$z)>!$E*V(hERhLK(G)LLJ#asAao!MR^NHs(Z%>1a~zuQ
z(A<aSa&(U)2f@Ck4#{{7cGe&S5G(|R&;$P_5UdaT_b7y(Q;;-&d;h}bZv*-Jw?4X$
zp=~6;u{|9qpKT?Zm_s%(CuWX6wapV~n;F}HKP&CN(8(1;IzRl<(9Y9!mvs(g-${SY
zu2TpfB7B5!8sQ8AwdL`b9o}+!=r6W(5AELa{Lr6o**f&l7O!LN#iMKUlFM!RbX^r}
zTiAJW#RA;J68DJwSsCvU4O9=i?+tlv%i1CPw`>}6V9T>ZUf(ih$Ukq%8M1H7?E;ta
zOgy$X%|^bD<#*qYqqPm$LH91{KFhoQ;k$n0zXJDLM(p~Xf0%vm<zg|m#dSZ;@5|^b
zr!91^=RN;Olt04U^BHO1|G8L<ZOP~tkABSd{LVM(wM!X3jV+^aye?(<G`7Ge`!8iU
zpDlEr<mcV_S$Bnv-anyjK^xw<2}Q&8R$XIr-CC=yB^<85$(meLy2`3r^g!~_|ML3!
zHmfDv3ffwotdpvnTRic_k@|2;eK-=enwszTU{r=%!k}bK4>z}3({HtkWtq_CiMO@a
zdg9g1Q7al+8;)4_HF;TjdwWxhwW__Xp*dnj!);NnTggo&R#U?|_$A`jFvDtUZ%nUH
z$xWpf$j_)h<mKuC^k?9+ek9L;$<mSb(O0T8S<iuatvDVYhhL@qgfeDiVktGVbb4(A
z8bc;SW|s||-c}cC47FOTT4t77ur$q<lG3@hg8RdD7Tz<<z)~umr;joWi#D`0uD3#w
zdanYU`7#;-pQiDWioLw&#kU(>GFJw39LLRg#%=|~c3P0x6El;?{o?Ycj{7CbpK2hf
z<B(pJq+g^rQ}#GB*+(jw$J;x}WU12bAp_{8X6UWbqk($%UyMC%gI6K!ag0j?=t1q$
z`U)v>da1EF<UP0xUR$)eHJm>5*>;)!n^9HDk51-{j4D&((%|^1@k2MCMx4_kb?eQU
z%s#iZ(HYBG?^N%|#P;HoM>F$R=y7Ucd1a-wB!dm)FozxRs%7&tYfa}by=%z;R(SBs
zJ)1$A9ZSmxRI9jt;Cc2JNBTh3ahx}#Czh7au)X1_{yh1}E8h9w^2D$&c9;9E^?#n;
zX}=!8-hKW3^b-8KH~afn0^0|<2J#H#Ly&V1_4jXv+z7b`avLPxYfw8Ok3jB)?1kJ1
ziT69yVaV~2$04Udo`DQN=KLJ`A;&|mfh>f42y!muX2?a5&p_S-iDymK8ps2X>mYj|
zH$t9*+y<G0mn(NdPJrACN$<<=gRF!+40#XaamYr<GmtUJobLYqt&rm(cS06Iz6d!N
zvI}w%<U5e}K%RkI138@h0%<{Rge-%kEmsTWdtkZBUYF;)c6h#x(d(@f-T;r%%P-3#
z-ZX@JfE5#ljj#Yh6xhv>v{?u%5jub^f>h-rt?y><@K<cfsy_O!Z;m90@~uNS0FE)k
z#x?=l2aMlCC0+-@s{+GfUnK;#57>(WI|6L4z<Po00Y<uMBbkk3zFT19fqh?Kcqc*a
z6d0}nsAmL5ZI9~jEQ8wNlLDhQxeb`JzSKUq3JmYMsLcY~3v83X4glK-%vo*^u*U>;
z3Rq0!%|Uy7NMLw}O05%EF)%vb&N3>2wFn;GH&cy**9dHlz+%Aa1-2DfjlgySyH{W@
z0;Bfmw51E!N`butY?%Y2c6<id5@4sO5tNT?`R>RaBPxe1x}svs@U6q5SD>jL8oFqt
zDyMgiiC>6*wB%LvJ38qY{(>#Tkz>f>%l+TYq8vjP4P9)qfb`k+nFl`ez-J!#%mbf!
z;4=^W|Ly_){tka{2U8?_qi_1q#&P-v5N$kX<L^a{(6CxJNAH-=x+K5zV?L*&_**CD
zb2*B?_hUX+lT82clYYCXZJ4Lf#@|}$#XN*I{>BQeFKFWnyFGSH;Y(I}GoPtZ=JP@Z
zO?}g)hrjb;Qg#|S#z_8V%QzG7F#zKMV;I+~!`Q>$ZsGFtOn&Bl+^_2gzuM3zCf;in
z1pOv%)Tv$*?=tCqCeC&l^?z{DVJdK;lE3@L-)-XWG_k?_eWoXi8PSrz!^GcR;_oc+
zca``%N}P+opTu$gZW4bdiNA})-$CN<9#xv}9PxLJ_&Y|GGg*_C{JkPZ`8!1%AFQPQ
zM8}qzGp_*$o0oX6A6RlLR)M#@eqdVQCm${fTu5HxI?JXeTs43GO;+Juwe68;yH!$B
zR9aL#y<}dy9WR+;kn<EmUkjVd4c5W+qcpl+AUFEb*g8)<TXDa4=J8_`;QsE$hbkTy
z-1udR`->aTQQXhmIKIxv1C1lWkI{wuy#t2znZB1dTyg(!^G7IdcW!)Sa@=&|qts?I
zp1See<aZ9;_-M7S+==LYKfZLw-&=F0`qh=m@z;%y!FL;Z#80L8F^ckdnL?=zr+*)i
z(*0_j>RRTdheySf7Gh<q_xn?b-qYh~dOM->4}GhS+mB^3!V+va8}R^caII&wIwJg|
z@pM1a`OC{5)?X&{Ti|D*T*f(B^V8dj#?z1c^a0Bqrnd131U%YU&tsyUlp$>(`z>)o
zWIt>Zdbqx_-7-HGAS+Uv#>c9FIPMxBt(H}!xU_dW!a#N|V@H~7cR`Y$CH`Lw9us`F
zM~>6?89vW<jCvKrj?d9<x%Fy2>HTGs%}>Xl15S4Gyb4i!(|XeT<I70(dGu5ETd!xk
zP48bvp=Tg}W{a-(@3vJV)C7TlWbh{?ZjaX^)V(~xvw7(2wxdR<LY~+e{7Qo#5cp(+
zKgJV7!=J-?ctU9KGK23G__vKcOX!27j%}ggCj|aW!(YS`GLz?C!+%lWO$IN<!%WWY
zs|J5Y;13&on&o8Udzl7*Rp46<KDWTlKW6YQf$uW-nu%`ys|Ht--1t$0M+JV`;AaI+
z*G6glY`M;@XQc7l8G#Qoc;jR@f1<(93j7v>Z=xF(j_q!PSKi<Nv5RklVH=@}3*Go8
z=1+0sod&O&>c(F-_$h&R8~mXg-TdDe+`jnfNb=K!4fe$7qNrobH~9AjKFi?QG_X6i
zI}JWT;I#(t5co>w7dez*H)`-_1pem+pEkqIpJq=sN2o4=-)8VaG5>zr=m`k?IfHK&
z_-h9568OIwys*S=&qoH23VfKc=aj&&H+Z1bt!JLWw+Vc?!QT=1YJ*P`^ZQ2({-nS+
z8hi<zSRLDcGx#$C-(~Q_0^e!ydb8eORrrn!wh^jV;QI~!&}?REdBos3;(VRye9GsM
z2@BkMegi!|d?8ul#PT|i;QKb<zuzzChuK3YNBX>U9B|?no9~wri#9#4NS`N92R|2n
z;B}&AD|HGL_#E(rnYZMcgn5Q9*OQA4-Xr?;7lD6{l{(~?Z5;kL`e}Yi+qZyQ_MZFS
z-x@xq3iihK%*BNP%VR$LRoGAU;&m=-`ziF~I_4XUAF}ZgN}jKe7+kKC-!=9Gv}2Rk
zXFdTw8uld2e2ne9%-C~EoPUSn!pcN8!XYi-Wd8w?xW>lEsl#Hv_XTsjb{l)R+@qMk
zlAi0td3KhqXKdPf;bv1VUk6~5CL8-VTa0MA5PWJUHNp>f5uTMk-~N)(&({}N+ecSu
zI}649VYSUquWy^t!`CNRPt4%=PG&^QZyB7gV{qyY;ADTX@Y{G?&?3Jbp2SQo-!=Zs
zvE2A`Mo*XMFE1Fs$?GNuY`$9u#TJ<3m=N~7lYxKMl$&e%#VMn5vB6{FxC}>sq53Wp
z{mwFYhbVV8@LUB@w72;-j`OlOZ~hNcZj124GFyLoeK(o<?iTz;V^6QJ=P^@m@AYn7
z+kjKQi-~^sw9!)|>iZ*uQ`KnW@#he5%SPP)mYZ=vUhjI_F4xVqV`wL%v&Jg94jKWR
z{GThzEwJ%)J*A{arPj;#X$Z7(;M6|9FZ%UTgLjGcUuW=aF(3O!(?0pS9a|8!`RV;`
zz0qTddHBuj&oNp9pN1bd{N18of7{?=#Js4};C%g%?fH?xYs7JX-Nw`H{JG)ZBhE*^
zHuzRy{~4o4UO&k;{^9GGY|m)ZZh4(<-=|07Si|S*oXnqS@WKK{w47=5#6<hAFnE_}
z|D%kbi$#0B+wgk?|2|W0mpJYZ8Jw@fvQ>|w|5Cd;bR8pF{+ZFgS=hP5=;7<Qoa?WF
zQ-9>^5_~@Tf#LJ@Ugp1SaC!arb%P(dftgzViu9v@$>a4taOz(j!p@;Mzu4oR=y%r`
zJRtm81l)2LX~z};C;!xo^THPlUM9}h5u@J{$K_GO&lTh6Hw}NY;D68HJ;KfnMo+d_
zkNktd<@NJ_H25aL?<W0O>DSGU5RUq`2zyQ#J-vl)CuN)S<|eT&`-y3{mazXS!><wT
z?T?NACq;k31B9*7lA>mHT|ItmBUIZEjD}XL>((j24S@0ZWMlBl;ShfFBSKgNSg5rr
z*br%pwzk(r@tl4u9;+;>qr|po`>Ivgtxx`#M=;tH#KWGEa2vGOHwRZYHrIw4gY|gE
zvMm^DM={M!EsbHk?@(0i&7hwwX$Xc|TSM!Ecs-$Yy;{{8Y6=JI+nbu`M@bwp7}XGU
zQS0hfv^L(|9HNReuBG=0f{o41Eh^I7l8WHzV0z{_nW*EXB`E~SGfPuZJV*&-c9{#o
zgYL+PhbLX=DqwEFRTMNKA0U}eKg7L?-UMNr@mxH(t3U?q;+qr{B~!Hi45){>rhtsv
zg`v7&=#e!*0#FZ^AYo>-xg=v^$v}xHMSDv+#rBEw5GD<C`yVjgI?-mibhy#`!i|mV
z>zgC(QI*VrB0O?1>c|1@^w4#R48VS>I)y-~{aiLZzU@LwOFVGA3=@WZDWVf_6b!8}
z$HjC*nYg7Tx>LB=0Ns2XGO61Ed2wAm$Zil&U_dt?cH>QzR1G=eU^USDOL*BRMOxR~
z`arm~IRzlq4d5LlHxLXiTwcC(L2yBJWe`6mRQZMK@}-OCV-UeFQ3V$)GO3Fym#bja
zk~=EOmjv%vxNyaSmBE$e6-yQbmBaJYGIafQjZ!TIDA@$9;zyMNj^8{&ZrKmM+!|`Z
zv(f|p91^mjIZ>n5eL9&ZSliZ?{FNj#)R9T47U43G^R+cZ)|R)TOxKS?B@M;zG|3Wm
z)x!0mXh=c19u}2p^sLq%vz6m`pa3!uZ<xXFupT*$+jxXZjx5S?)J*AAT?E$dG|A4!
z3C`cNa@N85GgeN#DI98RZw)VOkJ#ri<s6M!Ckk`)N=9)mg9Eg~sq)DVj*7Z}CCgQw
z7{OT^%0Ts?2LH{_t5&tFL?fbOqpOP~I1Q<eOmeW~B>AIR4wLNP>2lhHnzS_s*WkSw
z{Mr_ZLLo2&|5=&+OJJ@Xi|>HK`i4jl0~3DpD-`8H*40>22h^$6G@pphhhl6%JKj-y
zWN~=i=4dTu2r4a+ZqbTO3j<esGijOg&>Cv8$50hui9Vkz`{&ObV+*Ca&c<A0c91jO
zfx!TGAJs;oAUcKHHaN#L(n?vh6IhR#rzpG`voh5jz%hO>N6d_u6qBQhsVsg}&5U}c
z9N6G)Owi~)@S5nh$Ns4{&);8TwLCN7wqVy+fW`-0NfgUxOv;r!1#&ttSOJ#X7R1<*
z&Rn{}G!smAOot-bWo$OGCu?s5u47>%s;F&!Q#4cy8Ew^a4e#OLgj-uw5nccb7p;!8
z7uB{m;P>wu>J3z0v3NR7s_nEjp|&-usD6C}`m~I;>XiG#t@PHbD-uLnYq&8)8O&Zw
zV^kH{)hmjI*I}RL$4G0ohvcI08Z#qaQ;)nH(>e6iSm)q<C=N9>)S=X7=+ZjSGgJ{K
z=1rJyyH_I@AEV2(J_P%=6y#j9!%|$vwYkpHXnBYwPclD0kHb<dcW~|Bz0DT+`MC?0
zmYbfIzY^&m;h8nIf1BB1$+vZJS<BvV`NOqUBZj3}(*C_BKTG_+hP~l(0iFsm57j~P
z?d!V@&gJv-LM(fb?%a&JB=o1)qi6CsKR;K*l5cOb{mf&z7WwHJKaTVBM=Y0_lE{AI
z$ofA3jE)KC=jW7Ic9;Y%pYwD5A2azYAZg?0nON@BFrKhdik8GO|2Ke<Psl#n__<G(
zT#ioFw(>s7KSOZl=jWqXZg!Hf&(~)d-YW9*buN~3#r1;#XEgFJMSgxxm8H2%n9L>H
ze}~D>{^#eiSkm=6=O&_x{7d!c{2K$B&GKf^;AHtf6#4l%FP8M4oO6@qCy+Q2foc3a
z82^5)<a3<m0puO!$j{I19$o?v8}oE56=(b~61Z;?hwmTq?*p@AxP5UxZU@JZhg`(=
z^K)oDG%#SZg#S1{%YOsMnLl~|z;wJWTTAL+Ahm1noxp$6#pUyLzAlrW%i(Oo9Q8gD
z2rFiSk(~c}fpdPR<hJ9ao-yU~{T3RBsIRj<+%It%GHJUrf6f{myMYE*$HwI{G#Wcx
Scg9H0Xw$LZJ3;Kr{Qm(iDu53F