Eliminate templates from jrt_read_from_buffer_by_offset and jrt_write_to_buffer_by_offset

[rtakacs]

Refactor jrt_read_from_buffer_by_offset and jrt_write_to_buffer_by_offset to eliminate template.

JerryScript-DCO-1.0-Signed-off-by: Roland Takacs [email protected]

[ruben-ayrapetyan]

Perhaps, now it is better to move implementations of the functions from header to a module like jrt.cpp.

[rtakacs]

Done.

[ruben-ayrapetyan]

Looks good to me

[egavrin]

@rtakacs please share performance measurements and changes in binary size.

[rtakacs]

I measured some things on a rpi2 board. The master revision was 140982e

Section	Size
.text	190732 -> 190900 (-0,088)
.rodata	8168 -> 8172 (-0,04)
.bss	2363840 -> 2363904 (-0,003)

Note: the other sections are not modified.

Benchmark	RSS (+ is better)	Perf (+ is better)
3d-cube.js	132 -> 132 (0)	3.742 -> 3.756 (-0.374)
access-binary-trees.js	88 -> 88 (0)	1.909 -> 1.938 (-1.519)
access-fannkuch.js	48 -> 48 (0)	9.812 -> 9.655 (1.6001)
access-nbody.js	64 -> 64 (0)	4.539 -> 4.534 (0.1102)
bitops-3bit-bits-in-byte.js	40 -> 40 (0)	2.672 -> 2.613 (2.2081)
bitops-bits-in-byte.js	40 -> 40 (0)	3.727 -> 3.705 (0.5903)
bitops-bitwise-and.js	36 -> 36 (0)	4.296 -> 4.235 (1.4199)
controlflow-recursive.js	220 -> 220 (0)	1.727 -> 1.75 (-1.332)
crypto-aes.js	160 -> 160 (0)	6.097 -> 6.134 (-0.607)
crypto-md5.js	208 -> 208 (0)	35.795 -> 34.686 (3.0982)
crypto-sha1.js	144 -> 144 (0)	15.531 -> 14.946 (3.7667)
date-format-tofte.js	112 -> 112 (0)	4.009 -> 4.076 (-1.671)
date-format-xparb.js	96 -> 96 (0)	2.113 -> 2.14 (-1.278)
math-cordic.js	48 -> 48 (0)	4.009 -> 4.078 (-1.721)
math-partial-sums.js	44 -> 40 (9.0909)	2.34 -> 2.34 (0)
math-spectral-norm.js	48 -> 48 (0)	2.256 -> 2.289 (-1.463)
string-base64.js	204 -> 200 (1.9608)	---
string-fasta.js	56 -> 56 (0)	5.862 -> 5.775 (1.4841)
Geometric mean:	RSS reduction: 0.6375%	Speed up: 0.2681%

Note: all tests ran 10 times. Tests that failed are not shown in the table.

[tilmannOSG]

The results look good. Can we get this merged?

[egavrin]

@tilmannOSG depends on @galpeter ^_^

[ruben-ayrapetyan]

@rtakacs, could you, please, rebase the patch to the current master?

[rtakacs]

This is done by #791 so we can close this PR.