Re: [racket-users] How can I increase expansion+compilation speed?

[Ryan Kramer]

Thanks Ryan, the macro profiler is great! It quickly identified a macro 
that I was able to convert to a procedure. (I'm not sure why I made it a 
macro in the first place.) And the .zo file decreased from about 910KB to 
130KB, and the compilation speed went from 40 seconds to about 8 seconds 
(with debugging). I can probably do even better, but fixing this first big 
problem was quite a relief!

On Monday, July 20, 2020 at 8:42:57 AM UTC-5, Ryan Culpepper wrote:
>
> One thing to check is the size of the resulting bytecode file. When I 
> compiled it, I got a 911 KB .zo file. So the most likely reason is that 
> your macros are just producing a lot of code.
>
> You can run the macro profiler (eg, `raco macro-profiler aw-schema.rkt`) 
> to get a summary of what macros contribute most to the compiled code size. 
> See 
> https://docs.racket-lang.org/macro-debugger/index.html#(part._.Macro_.Profiler)
>  
> for more details. I also gave a talk about the macro profiler called "The 
> Cost of Sugar" a few years ago at RacketCon. The slides are at 
> http://www.ccs.neu.edu/home/ryanc/talks/racket18-sugar.pdf, and there 
> should be a recording online somewhere.
>
> The usual fix is to use helper functions. There are some examples in the 
> slides.
>
> Ryan
>

-- 
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 racket-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/racket-users/1b403370-3382-42a4-9ce1-8a168e3121cao%40googlegroups.com.

Re: [racket-users] How can I increase expansion+compilation speed?

[Ryan Culpepper]

One thing to check is the size of the resulting bytecode file. When I
compiled it, I got a 911 KB .zo file. So the most likely reason is that
your macros are just producing a lot of code.

You can run the macro profiler (eg, `raco macro-profiler aw-schema.rkt`) to
get a summary of what macros contribute most to the compiled code size. See
https://docs.racket-lang.org/macro-debugger/index.html#(part._.Macro_.Profiler)
for more details. I also gave a talk about the macro profiler called "The
Cost of Sugar" a few years ago at RacketCon. The slides are at
http://www.ccs.neu.edu/home/ryanc/talks/racket18-sugar.pdf, and there
should be a recording online somewhere.

The usual fix is to use helper functions. There are some examples in the
slides.

Ryan


On Mon, Jul 20, 2020 at 2:08 AM Ryan Kramer 
wrote:

> Using DrRacket, the following file takes 18 seconds to compile without
> debugging, or about 40 seconds to compile with debugging:
> https://raw.githubusercontent.com/default-kramer/plisqin-tutorials/e844825b48137553246c64e73516d880b9068825/define-schema-answer-key/aw-schema.rkt
>
> When I say "compile", I mean that I click "Run" in DrRacket and wait for
> it to give me REPL access. But that file just defines and provides a slew
> of functions, it doesn't really do anything at runtime.
>
> (Using the command-line `racket that-file.rkt` takes only 7 seconds, which
> is OK. But many people, myself included, primarily use DrRacket.)
>
> Admittedly, this is 612 dense lines of code. So it might just be a case of
> "yeah, that's about how long it takes." But it's also possible (or
> probable?) that I have some inefficient macros somewhere. Are there any
> tools/techniques I can use to help me reduce the compilation time?
>
> --
> 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 racket-users+unsubscr...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/racket-users/40e40eff-a0be-4850-9360-c9648cb5b8d9o%40googlegroups.com
> 
> .
>

-- 
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 racket-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/racket-users/CANy33q%3D1d1ojXVrcsM1mYCa2%2BcLkN2-x7x0WM197BukAA%3DRuzA%40mail.gmail.com.

Re: [racket-users] How can I increase expansion+compilation speed?

[Linus Björnstam]

40s to compile seems rather excessive, considering macros like match that also 
do a lot of work are instant, almost regardless of their size (I once managed 
to generate a 2000 line (match ...) and hardly noticed). Is there anything in 
define-schema that makes it non-linear? Do you process tables or columns over 
and over? How much code does it expand to?

I might have some computer time tonight, and will be able to test it out and 
explore the library you are using.

-- 
  Linus Björnstam

On Mon, 20 Jul 2020, at 02:08, Ryan Kramer wrote:
> Using DrRacket, the following file takes 18 seconds to compile without 
> debugging, or about 40 seconds to compile with debugging: 
> https://raw.githubusercontent.com/default-kramer/plisqin-tutorials/e844825b48137553246c64e73516d880b9068825/define-schema-answer-key/aw-schema.rkt
> 
> When I say "compile", I mean that I click "Run" in DrRacket and wait 
> for it to give me REPL access. But that file just defines and provides 
> a slew of functions, it doesn't really do anything at runtime.
> 
> (Using the command-line `racket that-file.rkt` takes only 7 seconds, 
> which is OK. But many people, myself included, primarily use DrRacket.)
> 
> Admittedly, this is 612 dense lines of code. So it might just be a case 
> of "yeah, that's about how long it takes." But it's also possible (or 
> probable?) that I have some inefficient macros somewhere. Are there any 
> tools/techniques I can use to help me reduce the compilation time?
> 
>  -- 
>  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 racket-users+unsubscr...@googlegroups.com.
>  To view this discussion on the web visit 
> https://groups.google.com/d/msgid/racket-users/40e40eff-a0be-4850-9360-c9648cb5b8d9o%40googlegroups.com
>  
> .

-- 
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 racket-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/racket-users/12a6b283-5e70-4bbf-b377-24905c38ee29%40www.fastmail.com.

[racket-users] How can I increase expansion+compilation speed?

[Ryan Kramer]

Using DrRacket, the following file takes 18 seconds to compile without 
debugging, or about 40 seconds to compile with 
debugging: 
https://raw.githubusercontent.com/default-kramer/plisqin-tutorials/e844825b48137553246c64e73516d880b9068825/define-schema-answer-key/aw-schema.rkt

When I say "compile", I mean that I click "Run" in DrRacket and wait for it 
to give me REPL access. But that file just defines and provides a slew of 
functions, it doesn't really do anything at runtime.

(Using the command-line `racket that-file.rkt` takes only 7 seconds, which 
is OK. But many people, myself included, primarily use DrRacket.)

Admittedly, this is 612 dense lines of code. So it might just be a case of 
"yeah, that's about how long it takes." But it's also possible (or 
probable?) that I have some inefficient macros somewhere. Are there any 
tools/techniques I can use to help me reduce the compilation time?

-- 
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 racket-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/racket-users/40e40eff-a0be-4850-9360-c9648cb5b8d9o%40googlegroups.com.