If you run "raco make -v cmdline2.rkt" you will see output if .zo
compilation is happening.

Robby


On Mon, Sep 14, 2015 at 8:16 PM, John Carmack <[email protected]> wrote:
> No strace on Android, unfortunately.
>
>
>
> From: Marc Burns [mailto:[email protected]]
> Sent: Monday, September 14, 2015 8:10 PM
> To: John Carmack
> Cc: Racket Users
> Subject: Re: [racket-users] Startup times
>
>
>
> Here’s the result of `strace -c -f -- racket -l racket/base` for Racket
> 6.1.1.8 on my Linux workstation:
>
>
>
> % time     seconds  usecs/call     calls    errors syscall
>
> ------ ----------- ----------- --------- --------- ----------------
>
>  99.08    0.004000         571         7           nanosleep
>
>   0.92    0.000037           0       236           read
>
>   0.00    0.000000           0       103        23 open
>
>   0.00    0.000000           0        79           close
>
> …
>
>
>
> Does it look similar on the Note 4?
>
>
>
> On Sep 14, 2015, at 9:00 PM, Marc Burns <[email protected]> wrote:
>
>
>
> Set the environment variable PLTSTDERR=debug to get more verbose output.
>
>
>
> Startup involves traversing all the bytecode files that comprise the base
> environment. How fast is filesystem access on the Note 4 compared to PC? You
> could use strace to find the latency on different system calls made during
> startup.
>
>
>
> On Sep 14, 2015, at 8:55 PM, John Carmack <[email protected]> wrote:
>
>
>
> I am experimenting with running racket natively on Android to compare with
> my current embedded Chibi scheme implementation.  It would be convenient to
> just leave racket as a separate process and communicate over sockets/pipes
> so it exactly mimics my remote development case.
>
>
>
> The startup time to run a trivial console program is very long.  A one line
> program with #lang racket/base takes over seven seconds:
>
>
>
> root@trlte:/mnt/shell/emulated/0/Oculus/racket/bin # time ./racket
> cmdline2.rkt
>
> cmdline2.rkt
> <
>
> line 1
>
> line 2
>
> line 3
>
> line 4
>
> line 5
>
> line 6
>
> line 7
>
> line 8
>
> line 9
>
>     0m7.96s real     0m7.04s user     0m0.65s system
>
>
>
> My first test, which still had the default #lang racket, took almost a
> minute to start:
>
>
>
> root@trlte:/mnt/shell/emulated/0/Oculus/racket/bin # time ./racket
> cmdline.rkt
>
> cmdline.rkt
> <
>
> line 1
>
> line 2
>
> line 3
>
> line 4
>
> line 5
>
> line 6
>
> line 7
>
> line 8
>
> line 9
>
>     0m54.16s real     0m48.68s user     0m4.83s system
>
>
>
> On a PC, it only takes a fraction of a second.  This was on a Note 4, which
> should not be 100x slower than a PC.  Could it not be using the compiled
> library bytecode somehow?  I didn’t see any command line options for verbose
> output on startup, is there any way to force some extra information?
>
>
>
>
>
> --
> You received this message because you are subscribed to the Google Groups
> "Racket Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> For more options, visit https://groups.google.com/d/optout.
>
>
>
>
>
> --
> You received this message because you are subscribed to the Google Groups
> "Racket Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> For more options, visit https://groups.google.com/d/optout.
>
>
>
> --
> You received this message because you are subscribed to the Google Groups
> "Racket Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> For more options, visit https://groups.google.com/d/optout.

-- 
You received this message because you are subscribed to the Google Groups 
"Racket Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to