doc clang-format

Author: ahbarnett

docs/devnotes.rst

@@ -17,6 +17,9 @@ Developer notes
 
 * There are some sphinx tags in the source code, indicated by @ in comments. Please leave these alone since they are needed by the doc generation.
 
+* Source code is now in clang format: devs should run ``clang-format --files=<editedfile> -i --style=.clang-format`` before pushing, or set up their editor to do this
+  automatically.
+
 * If you add a new option field (recall it must be plain C style only, no special types) to ``include/finufft_opts.h``, don't forget to add it to ``include/finufft.fh``, ``include/finufft_mod.f90``, ``matlab/finufft.mw``, ``python/finufft/_finufft.py``, and the Julia interface, as well a paragraph describing its use in the docs. Also to set its default value in ``src/finufft.cpp``. You will then need to regenerate the docs as in ``docs/README``.
 
 * For testing and performance measuring routines see ``test/README`` and ``perftest/README``. We need more of the latter, eg, something making performance graphs that enable rapid eyeball comparison of various settings/machines. Marco is working on that.

perftest/compare_spreads.jl

@@ -4,7 +4,7 @@ using CairoMakie
 using JLD2        # for load/save arrays to file
 using UnPack
 
-fnam = "results/master-vs-svec2l_gcc114_5700U_nthr8"   # outfile head
+fnam = "results/master-vs-svec2_gcc114_5700U_nthr1"   # outfile head
 # locations of pair of FINUFFT repos to compare...
 repo1 = "/home/alex/numerics/finufft"
 repo2 = "/home/alex/numerics/nufft/finufft-svec2"
@@ -62,7 +62,7 @@ function plot_all(fnam,ts,wstr,dims,M,N,nthr)
 end
 
 # main script...........................................................................
-nthr = 8; # 1: leave cpu freq at max (4.3GHz); for 8, lower to 2.7GHz since drops to this.
+nthr = 1; # 1: leave cpu freq at max (4.3GHz); for 8, lower to 2.7GHz since drops to this.
 # set freq lim with cpupower-gui
 # check with:   watch -n 1 sort -nr /sys/devices/system/cpu/cpu*/cpufreq/scaling_cur_freq
 dims = 1:3