Ah, it's possible that Racket mode sets a limit on Racket's memory use, just like DrRacket does. That kind of memory limit (enforced within Racket, instead of by the OS) is based on GCable memory, so it doesn't limit 'raw allocation.
At Sun, 28 Jul 2019 11:12:04 -0700 (PDT), Peter Samarin wrote: > I just rerun the example on the commandline and it works just fine with > larger numbers of bytes. > I use emacs and racket-mode by Greg Hendershot. > I will check how that starts racket at some other time. > > Thanks Matthew! > > On Sunday, July 28, 2019 at 8:06:09 PM UTC+2, Matthew Flatt wrote: > > > > There should be no special limit on 'atomic memory, and I haven't been > > able to replicate this behavior, yet. > > > > Can you say more about how you're running (e.g., in DrRacket versus > > command-line `racket`)? Also, although it's difficult to see how it > > could matter, which Linux distribution do you use? > > > > At Sun, 28 Jul 2019 10:45:42 -0700 (PDT), Peter Samarin wrote: > > > Hi all, > > > > > > I am using Racket v7.3, Linux x86_64 version. > > > I want to allocate large (tens of gigabytes) chunks of memory with > > Racket's > > > malloc and use them for some BLAS/LAPACK routines. > > > The downloaded version of Racket lets me allocate only 2^31 "atomic" > > bytes. > > > 2^31+1 bytes produce an "out of memory" error, even though I still have > > > plenty of memory left (much more than requested 2GB) on my machine. > > > > > > In raw mode, I can go to 2^33 bytes, but not 2^34. My machine has 16GB > > RAM > > > with swap disabled. > > > Is that a bug or is the amount of allocated atomic memory limited to > > 2^31 > > > bytes? > > > > > > Here is the code to reproduce the problem: > > > > > > (require ffi/unsafe) > > > > > > (malloc (expt 2 31) _uint8 'atomic) ; => #<cpointer> > > > (malloc (add1 (expt 2 31)) _uint8 'atomic) ; => out of memory > > > > > > (define mem (malloc (expt 2 33) _uint8 'raw)) ; => #<cpointer> > > > (memset mem 0 #x00 (expt 2 33) _uint8) ; ok > > > > > > --Peter > > > > > > -- > 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/80d40912-6f55-4906-a0f9-b10cd78b > 423c%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/5d3de71c.1c69fb81.51656.1e17SMTPIN_ADDED_MISSING%40gmr-mx.google.com.
