benchmarks(3avr)

✨ NAME ✨

benchmarks — Benchmarks, bestie. That's it, that's the name. We're in the 3avr part of the AVR-LibC manual, and this page is the receipts: how many flash bytes, stack bytes and MCU clocks certain library functions are gonna cost you. 💖

okay so real talk: the results below can only give a rough estimate of the resources necessary for using certain library functions. There's a whole list of factors that can either increase or reduce the effort required, and you need to know all of them:

• expenses for preparation of operands and their stack are not considered 🙈
• in the table, the size includes all additional functions (for example, the function to multiply two integers) — but they are only linked from the library
• expenses of time of performance of some functions essentially depend on the parameters of a call. for example, qsort() is recursive, and sprintf() receives its parameters in a stack. it's giving chaos 💀
• different versions of the compiler can give a significant difference in code size and execution time. for example, the dtostre() function, compiled with avr-gcc 3.4.6, requires 930 bytes. after the transition to avr-gcc 4.2.3, the size become 1088 bytes. not me sobbing in hex 😭

---

🎀 A few of libc functions 🎀

Avr-gcc version is 4.7.1 💅

The size of a function is given in view of all picked up functions. By default AVR-LibC is compiled with the -mcall-prologues option. In brackets, the size without taking into account modules of a prologue and an epilogue is resulted. Both of the sizes can coincide, if the function does not cause a prologue/epilogue. no cap.

so the layout is: each function, then Units (Flash bytes, Stack bytes, MCU clocks), and then the number for Avr2, Avr25 and Avr4 — in that order, every time, periodt.

• atoi ("12345")
  • Flash bytes: Avr2 82 (82), Avr25 78 (78), Avr4 74 (74)
  • Stack bytes: Avr2 2, Avr25 2, Avr4 2
  • MCU clocks: Avr2 155, Avr25 149, Avr4 149
• atol ("12345")
  • Flash bytes: Avr2 122 (122), Avr25 118 (118), Avr4 118 (118)
  • Stack bytes: Avr2 2, Avr25 2, Avr4 2
  • MCU clocks: Avr2 221, Avr25 219, Avr4 219
• dtostre (1.2345, s, 6, 0)
  • Flash bytes: Avr2 1116 (1004), Avr25 1048 (938), Avr4 1048 (938)
  • Stack bytes: Avr2 17, Avr25 17, Avr4 17
  • MCU clocks: Avr2 1247, Avr25 1105, Avr4 1105
• dtostrf (1.2345, 15, 6, s)
  • Flash bytes: Avr2 1616 (1616), Avr25 1508 (1508), Avr4 1508 (1508)
  • Stack bytes: Avr2 38, Avr25 38, Avr4 38
  • MCU clocks: Avr2 1634, Avr25 1462, Avr4 1462
• itoa (12345, s, 10)
  • Flash bytes: Avr2 110 (110), Avr25 102 (102), Avr4 102 (102)
  • Stack bytes: Avr2 2, Avr25 2, Avr4 2
  • MCU clocks: Avr2 879, Avr25 875, Avr4 875
• ltoa (12345L, s, 10)
  • Flash bytes: Avr2 134 (134), Avr25 126 (126), Avr4 126 (126)
  • Stack bytes: Avr2 2, Avr25 2, Avr4 2
  • MCU clocks: Avr2 1597, Avr25 1593, Avr4 1593
• malloc (1)
  • Flash bytes: Avr2 768 (712), Avr25 714 (660), Avr4 714 (660)
  • Stack bytes: Avr2 6, Avr25 6, Avr4 6
  • MCU clocks: Avr2 2157, Avr25 2017, Avr4 2017
• realloc ((void *)0, 1)
  • Flash bytes: Avr2 1284 (1172), Avr25 1174 (1064), Avr4 1174 (1064)
  • Stack bytes: Avr2 18, Avr25 18, Avr4 18
  • MCU clocks: Avr2 3051, Avr25 2861, Avr4 2861
• qsort (s, sizeof(s), 1, cmp)
  • Flash bytes: Avr2 1252 (1140), Avr25 1022 (912), Avr4 1028 (918)
  • Stack bytes: Avr2 42, Avr25 42, Avr4 42
  • MCU clocks: Avr2 21996, Avr25 19905, Avr4 17541
• sprintf_min (s, "%d", 12345)
  • Flash bytes: Avr2 1224 (1112), Avr25 1092 (982), Avr4 1088 (978)
  • Stack bytes: Avr2 53, Avr25 53, Avr4 53
  • MCU clocks: Avr2 1841, Avr25 1694, Avr4 1689
• sprintf (s, "%d", 12345)
  • Flash bytes: Avr2 1614 (1502), Avr25 1476 (1366), Avr4 1454 (1344)
  • Stack bytes: Avr2 58, Avr25 58, Avr4 58
  • MCU clocks: Avr2 1647, Avr25 1552, Avr4 1547
• sprintf_flt (s, "%e", 1.2345)
  • Flash bytes: Avr2 3228 (3116), Avr25 2990 (2880), Avr4 2968 (2858)
  • Stack bytes: Avr2 67, Avr25 67, Avr4 67
  • MCU clocks: Avr2 2573, Avr25 2311, Avr4 2311
• sscanf_min ("12345", "%d", &i)
  • Flash bytes: Avr2 1532 (1420), Avr25 1328 (1218), Avr4 1328 (1218)
  • Stack bytes: Avr2 55, Avr25 55, Avr4 55
  • MCU clocks: Avr2 1607, Avr25 1446, Avr4 1446
• sscanf ("12345", "%d", &i)
  • Flash bytes: Avr2 2008 (1896), Avr25 1748 (1638), Avr4 1748 (1638)
  • Stack bytes: Avr2 55, Avr25 55, Avr4 55
  • MCU clocks: Avr2 1610, Avr25 1449, Avr4 1449
• sscanf ("point,color", "%[a-z]", s)
  • Flash bytes: Avr2 2008 (1896), Avr25 1748 (1638), Avr4 1748 (1638)
  • Stack bytes: Avr2 86, Avr25 86, Avr4 86
  • MCU clocks: Avr2 3067, Avr25 2806, Avr4 2806
• sscanf_flt ("1.2345", "%e", &x)
  • Flash bytes: Avr2 3464 (3352), Avr25 3086 (2976), Avr4 3070 (2960)
  • Stack bytes: Avr2 71, Avr25 71, Avr4 71
  • MCU clocks: Avr2 2497, Avr25 2281, Avr4 2078
• strtod ("1.2345", &p)
  • Flash bytes: Avr2 1632 (1520), Avr25 1536 (1426), Avr4 1480 (1480)
  • Stack bytes: Avr2 20, Avr25 20, Avr4 21
  • MCU clocks: Avr2 1235, Avr25 1177, Avr4 1124
• strtol ("12345", &p, 0)
  • Flash bytes: Avr2 918 (806), Avr25 834 (724), Avr4 792 (792)
  • Stack bytes: Avr2 22, Avr25 22, Avr4 28
  • MCU clocks: Avr2 956, Avr25 891, Avr4 794

---

🌈 Math functions 🌈

This table contains the number of MCU clocks to calculate a function with a given argument(s). The main reason of a big difference between Avr2 and Avr4 is a hardware multiplication. the Avr4 girlies are thriving fr fr 🔥

• __addsf3 (1.234, 5.678) — Avr2 113, Avr4 108
• __mulsf3 (1.234, 5.678) — Avr2 375, Avr4 138
• __divsf3 (1.234, 5.678) — Avr2 466, Avr4 465
• acos (0.54321) — Avr2 4411, Avr4 2455
• asin (0.54321) — Avr2 4517, Avr4 2556
• atan (0.54321) — Avr2 4710, Avr4 2271
• atan2 (1.234, 5.678) — Avr2 5270, Avr4 2857
• cbrt (1.2345) — Avr2 2684, Avr4 2555
• ceil (1.2345) — Avr2 177, Avr4 177
• cos (1.2345) — Avr2 3387, Avr4 1671
• cosh (1.2345) — Avr2 4922, Avr4 2979
• exp (1.2345) — Avr2 4708, Avr4 2765
• fdim (5.678, 1.234) — Avr2 111, Avr4 111
• floor (1.2345) — Avr2 180, Avr4 180
• fmax (1.234, 5.678) — Avr2 39, Avr4 37
• fmin (1.234, 5.678) — Avr2 35, Avr4 35
• fmod (5.678, 1.234) — Avr2 131, Avr4 131
• frexp (1.2345, 0) — Avr2 42, Avr4 41
• hypot (1.234, 5.678) — Avr2 1341, Avr4 866
• ldexp (1.2345, 6) — Avr2 42, Avr4 42
• log (1.2345) — Avr2 4142, Avr4 2134
• log10 (1.2345) — Avr2 4498, Avr4 2260
• modf (1.2345, 0) — Avr2 433, Avr4 429
• pow (1.234, 5.678) — Avr2 9293, Avr4 5047
• round (1.2345) — Avr2 150, Avr4 150
• sin (1.2345) — Avr2 3353, Avr4 1653
• sinh (1.2345) — Avr2 4946, Avr4 3003
• sqrt (1.2345) — Avr2 494, Avr4 492
• tan (1.2345) — Avr2 4381, Avr4 2426
• tanh (1.2345) — Avr2 5126, Avr4 3173
• trunc (1.2345) — Avr2 178, Avr4 178

---

heads up bestie: these are rough estimates only. don't come for me when your compiler version eats your flash. 💀

that's the whole page, we ate and left no crumbs 🫶 this is benchmarks(3avr) from the AVR-LibC Version 2.2.1 manual — that's the version stamped on the page, and yes, that's the exact footer vibe too. mwah, go benchmark responsibly 💋⭐