Add cycle count getter for more platforms.

LICENSE.md

@@ -36,6 +36,7 @@ Julia includes code from the following projects, which have their own licenses:
 - [MINGW](https://sourceforge.net/p/mingw/mingw-org-wsl/ci/legacy/tree/mingwrt/mingwex/dirname.c) (for dirname implementation on Windows) [MIT]
 - [NetBSD](https://www.netbsd.org/about/redistribution.html) (for setjmp, longjmp, and strptime implementations on Windows) [BSD-3]
 - [Python](https://docs.python.org/2/license.html) (for strtod and joinpath implementation on Windows) [BSD-3, effectively]
+- [Google Benchmark](https://github.com/google/benchmark) (for cyclecount implementation) [Apache 2.0]
 
 The following components included in Julia `Base` have their own separate licenses:
 

src/timing.c

@@ -55,12 +55,12 @@ jl_timing_block_t *jl_pop_timing_block(jl_timing_block_t *cur_block)
 
 void jl_timing_block_start(jl_timing_block_t *cur_block)
 {
- _jl_timing_block_start(cur_block, rdtscp());
+ _jl_timing_block_start(cur_block, cycleclock());
 }
 
 void jl_timing_block_stop(jl_timing_block_t *cur_block)
 {
- _jl_timing_block_stop(cur_block, rdtscp());
+ _jl_timing_block_stop(cur_block, cycleclock());
 }
 
 #else

src/timing.h

@@ -35,11 +35,29 @@ void jl_timing_block_stop(jl_timing_block_t *cur_block);
 #define JL_TIMING(owner)
 #else
 
-static inline uint64_t rdtscp(void)
+// number of cycles since power-on
+static inline uint64_t cycleclock(void)
 {
- uint64_t rax,rdx;
- asm volatile ( "rdtscp\n" : "=a" (rax), "=d" (rdx) : : "rcx" );
- return (rdx << 32) + rax;
+#if defined(_CPU_X86_64_)
+ uint64_t low, high;
+ __asm__ volatile("rdtsc" : "=a"(low), "=d"(high));
+ return (high << 32) | low;
+#elif defined(_CPU_X86_)
+ int64_t ret;
+ __asm__ volatile("rdtsc" : "=A"(ret));
+ return ret;
+#elif defined(_CPU_AARCH64_)
+ // System timer of ARMv8 runs at a different frequency than the CPU's.
+ // The frequency is fixed, typically in the range 1-50MHz. It can be
+ // read at CNTFRQ special register. We assume the OS has set up
+ // the virtual timer properly.
+ int64_t virtual_timer_value;
+ asm volatile("mrs %0, cntvct_el0" : "=r"(virtual_timer_value));
+ return virtual_timer_value;
+#else
+ #error No cycleclock() definition for your platform
+ // copy from https://github.com/google/benchmark/blob/v1.5.0/src/cycleclock.h
+#endif
 }
 
 #define JL_TIMING_OWNERS \
@@ -107,7 +125,7 @@ STATIC_INLINE void _jl_timing_block_start(jl_timing_block_t *block, uint64_t t)
 }
 
 STATIC_INLINE uint64_t _jl_timing_block_init(jl_timing_block_t *block, int owner) {
- uint64_t t = rdtscp();
+ uint64_t t = cycleclock();
  block->owner = owner;
  block->total = 0;
 #ifdef JL_DEBUG_BUILD
@@ -127,7 +145,7 @@ STATIC_INLINE void _jl_timing_block_ctor(jl_timing_block_t *block, int owner) {
 }
 
 STATIC_INLINE void _jl_timing_block_destroy(jl_timing_block_t *block) {
- uint64_t t = rdtscp();
+ uint64_t t = cycleclock();
  _jl_timing_block_stop(block, t);
  jl_timing_data[block->owner] += block->total;
  jl_timing_block_t **pcur = jl_current_task ? &jl_current_task->timing_stack : &jl_root_timing;