Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 1 Jul 2012 20:07:48 +0400
From: "Elijah [W&P]" <>
Subject: Re: nvcc -o for win-builds

No I didn't tried to make a makefile target for win32 I was just lurking
through sources and tried to compile some "modules" to see if it's actually
possible. But I'm hoping for win32-cuda version to become a reality :)
>$ nvcc -c -Xptxas -v -arch sm_10 -o cuda_common.o
>ptxas info    : Compiling entry function '__cuda_dummy_entry__' for 'sm_10'
>ptxas info    : Used 0 registers
Success :)

But when it comes to solving those "include" mysteries I'm afraid I'm not
the right person to help correctly
>$ nvcc -c -Xptxas -v -arch sm_10 -o cuda_cryptmd5.o
>c:\cygwin\home\elijah\john\_cuda\src\cuda\../cuda_cryptmd5.h(10) : fatal
error C1083: Cannot open include file: 'stdbool.h': No such file or

As to the patch - here it is with explicitly set -o for all cuda-related

On 1 July 2012 18:13, Solar Designer <> wrote:

> On Sun, Jul 01, 2012 at 12:26:45PM +0400, Elijah [W&P] wrote:
> > As I can see (at least for -cuda targets) not every nvcc operation has an
> > "-o" option
> > But if you are planning to make versions for windows "-o" is needed
> because
> > "nvcc $(NVCC_FLAGS)" will give you cuda_common.obj
> instead
> > of cuda_common.o and "mv cuda/cuda_common.o cuda_common.o" will fail.
> Can you make and post a patch for that, please?
> Are you actually building our GPU stuff on Windows now?  Any success
> with that (after the fix above)?
> Thanks,
> Alexander

Content of type "text/html" skipped

Download attachment "nvcc_cc_o.diff" of type "application/octet-stream" (4938 bytes)

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.