Opened 12 years ago
Closed 12 years ago
#38792 closed defect (fixed)
atlas +mpclang33 build fails
Reported by: | ryandesign (Ryan Carsten Schmidt) | Owned by: | Veence (Vincent) |
---|---|---|---|
Priority: | Normal | Milestone: | |
Component: | ports | Version: | 2.1.3 |
Keywords: | Cc: | cooljeanius (Eric Gallager), mf2k (Frank Schima) | |
Port: | atlas |
Description
Following the fixes to the compiler checking, I tried to build atlas +mpclang33 but it failed; see log; I'm not sure what the relevant part is but I do see these errors:
cc1: error: unrecognized command line option "-fvectorize" cc1: error: unrecognized command line option "-fslp-vectorize"
Attachments (2)
Change History (20)
Changed 12 years ago by ryandesign (Ryan Carsten Schmidt)
Attachment: | main.log.bz2 added |
---|
comment:1 Changed 12 years ago by cooljeanius (Eric Gallager)
Cc: | egall@… added |
---|
comment:2 follow-up: 8 Changed 12 years ago by Veence (Vincent)
That should be fixed in r105258. Please try again, sorry for the inconvenience. Ryan, on what system did you compile? CPU? Thanks! Vincent
comment:3 Changed 12 years ago by Veence (Vincent)
r105311 should have fixed all bugs (I hope). Please try again.
comment:4 Changed 12 years ago by mf2k (Frank Schima)
It still fails for me with the same error on an iMac late 2009 with Core 2 Duo processor. Mac OS X 10.8.3. Xcode 4.6.2. Note also that I had to install gcc47 for a fortran compiler.
Changed 12 years ago by mf2k (Frank Schima)
comment:7 follow-up: 10 Changed 12 years ago by mf2k (Frank Schima)
@vince: My main.log is attached.
comment:8 follow-up: 9 Changed 12 years ago by ryandesign (Ryan Carsten Schmidt)
Replying to vince@…:
Ryan, on what system did you compile? CPU?
MacBook Pro with 15" Retina Display 4 × 2.6GHz Core i7.
comment:9 Changed 12 years ago by Veence (Vincent)
Replying to ryandesign@…:
Replying to vince@…:
Ryan, on what system did you compile? CPU?
MacBook Pro with 15" Retina Display 4 × 2.6GHz Core i7.
Thanks Ryan. I hope your LCD is fine ;)
comment:10 Changed 12 years ago by Veence (Vincent)
Replying to macsforever2000@…:
@vince: My main.log is attached.
Thanks Franck. You hit the same clang/llvm bug as #38820. I’ve committed a patch that should solve it (workaround).
comment:11 Changed 12 years ago by mf2k (Frank Schima)
Note that it's still building after a number of hours. So you must have fixed that issue. I'll let you know when it finishes. Probably tomorrow.
comment:12 Changed 12 years ago by mf2k (Frank Schima)
FYI, atlas did finally build successfully for me! This bug is fixed for me.
comment:13 Changed 12 years ago by Veence (Vincent)
Franck, you were lucky. What version of clang are you using?
comment:14 Changed 12 years ago by mf2k (Frank Schima)
$ port installed clang-3.3 The following ports are currently installed: clang-3.3 @3.3-r179575_2+analyzer+assertions+python27 (active)
comment:15 Changed 12 years ago by Veence (Vincent)
That’s the latest revision I suppose. Well, the bug some encountered despite the workaround does not seem to be triggered in your case. But you have a Core2Duo machine, haven’t you?
comment:16 Changed 12 years ago by mf2k (Frank Schima)
Yes, the machine where it built was the Core2Duo iMac I specified in comment:4. Note that I was careful to build cctools and ld64 with llvm 3.3.
$ port installed cctools ld64 The following ports are currently installed: cctools @836_0+llvm33 (active) ld64 @134.9_1+llvm33 (active)
comment:17 Changed 12 years ago by Veence (Vincent)
I don’t think it is related to cctools and ld64. The bug Ryan and other stumbled on seems to be linked to an AVX specific vectorization.
Ryan, you confirm you hit the same bug?
comment:18 Changed 12 years ago by Veence (Vincent)
Resolution: | → fixed |
---|---|
Status: | new → closed |
No more relevant since the removal of clang variants. Fixed for the time being.
Cc Me!