意外发现Cortex-M内核带的64bit时间戳，比32bit的DWT时钟周期计数器更方便，再也不用担心溢出问题了

TSG_CNTCR = 0x5C005000  --bit0 用于使能
TSG_CNTFID0 = 0x5C005020  --主频设置
TSG_CNTCVL = 0x5C005008  --低32bit
TSG_CNTCVU = 0x5C00500C  -- 高32bit
 
pg_init()
 
re = pg_write32(TSG_CNTCR, 0x00000000)  if(re == 0) then print("1") end
re = pg_write32(TSG_CNTCVL, 0x00000000)if(re == 0) then print("2") end
re = pg_write32(TSG_CNTCVU, 0x00000000)if(re == 0) then print("3") end
re = pg_write32(TSG_CNTCR, 0x00000001)if(re == 0) then print("4") end
re = pg_write32(TSG_CNTFID0, 200000000)if(re == 0) then print("5") end
 
for i=0,500,1 do
   a = pg_read32(TSG_CNTCVL)
   b = pg_read32(TSG_CNTCVU)
   print(string.format("%d, %d", a, b))
   delayms(100)
end