FYI, back to where the latest profiler work started (after nsec clock)
The goal: compare performance "map(in, 'v:val')" vs "map(in, Map1)". Map1 is
def Map1(_, val: number): number
return val
enddef
Turns out for processing a list with 3 items, 1.049 usec vs 1.590 usec;
at least that's the current results. (And not 1.166 vs 103)
Current vim:
1x33 1x65 1x99 2x33 2x65 2x99 3x33 3x65 3x99 :
nInputXnLoop (cells: usec/op)
0.608 0.625 0.641 0.999 0.957 0.954 1.237 1.179 1.166 :
map(in, 'v:val') ###-2
0.068 0.072 0.071 0.071 0.068 0.069 0.070 0.070 0.070 :
result = [] ###-4
0.144 0.151 0.146 0.171 0.166 0.164 0.194 0.190 0.189 : in =
input->copy() ###-5
2.784 6.023 10.140 27.808 36.471 45.324 77.819 92 103 :
map(in, Map1) ###-6
Upgraded function profiling patch
0.599 0.574 0.568 0.902 0.795 0.807 1.121 1.066 1.049 :
map(in, 'v:val') ###-2
0.071 0.072 0.071 0.072 0.069 0.071 0.071 0.070 0.070 :
result = [] ###-4
0.152 0.151 0.149 0.179 0.175 0.173 0.200 0.200 0.205 : in =
input->copy() ###-5
0.807 0.771 0.723 1.179 1.182 1.165 1.663 1.598 1.590 :
map(in, Map1) ###-6
-ernie
On 23/03/23 7:31 PM, Ernie Rael wrote:
Opened draft PR: https://github.com/vim/vim/pull/12192
Initial results are promising, one measure is how few loops are
required for the results to stabilize.
And the results show *faster times* when user defined functions
are involved. It's interesting how good that feels, even though
it doesn't mean anything is actually running faster in production.
On 23/03/23 8:19 AM, Bram Moolenaar wrote:
Ernie Rael wrote:
You might want to store the hash number with the function to avoid the
overhead of computing it.
This still needs to be done. To make sure I understand,
this is saving `hash_hash(name)` in ufunc_T and has_profiling has an
additional argument which is the hash. And if this argument is zero
then the hash would be computed in has_profiling.
The overhead of looking up the key in the
table can't be avoided. Also, when more function names are being cached
updating the hash table can also take a noticable amount of time.
Yes, but it also would usually indicate that the table has been very
useful.
I don't know most use cases; does the number of things being profiled
change dynamically or very often?
-ernie
--
--
You received this message from the "vim_dev" maillist.
Do not top-post! Type your reply below the text you are replying to.
For more information, visit http://www.vim.org/maillist.php
---
You received this message because you are subscribed to the Google
Groups "vim_dev" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to vim_dev+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/vim_dev/911e2e0b-ab7b-8d34-e205-6c6cbeea1cc8%40raelity.com
<https://groups.google.com/d/msgid/vim_dev/911e2e0b-ab7b-8d34-e205-6c6cbeea1cc8%40raelity.com?utm_medium=email&utm_source=footer>.
--
--
You received this message from the "vim_dev" maillist.
Do not top-post! Type your reply below the text you are replying to.
For more information, visit http://www.vim.org/maillist.php
---
You received this message because you are subscribed to the Google Groups "vim_dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to vim_dev+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/vim_dev/f61e3ca8-180e-85c7-b050-24a7f7203b6d%40raelity.com.
