|
From: Norwid B. <nb...@ya...> - 2022-06-23 12:33:44
|
Emanuel Berg <in...@da...> wrote:
>
> Brilliant,
>
> set format y "%.0f"
> set ytics add (0 0, \
> "1M" 1000000, \
> "2M" 2000000, \
> "4M" 4000000, \
> "6M" 6000000, \
> "8M" 8000000, \
> "10M" 10000000, \
> "12M" 12000000, \
> "14M" 14000000)
>
> works!
>
With the data available (and because the question is interesting), two
additional comments:
+ you may ease managing user-defined tics with exponential notation
like e.g., in Fortran:
```
set format y "%.0f"
set ytics font "verdana,10" add \
( 0 0, \
"1M" 1e06, \
"2M" 2e06, \
"4M" 4e06, \
"6M" 6e06, \
"8M" 8e06, \
"10M" 1e07, \
"12M" 1.2e7, \
"14M" 1.4e7)
```
+ gnuplot actually may add SI prefixes (k, M, etc.) on the scales for
you and show user-defined tics. For this to happen, drop "set
format y "%.0f" to try instead
```
set format y '%.0s %c'
set ytics font "verdana,10" add \
("25 M" 2.5e7) # only shown for the ordinate scaling by lg
set logscale y # i.e., the implicit form of "set logscale y 10"
```
The plot then retains an obvious hint something happened to
"katzeilla" and "rjc" during the first interval of recording while
the relative variation of the other traces was smaller. Credit to
stackoverflow user Christoph for the showcase[1] about the
alternative format.
Perhaps altogether with an instruction like "set yrange[1000:1e9]"
this may justify to move/reorganize the keys a little bit.
[1]
https://stackoverflow.com/questions/25123624/gnuplot-y-axis-format-convert-bytes-to-megabytes/25125788#25125788
|