Weird behaviors of pmap across version of julia #10640
Closed
rszoh opened this Issue on Mar 26, 2015 · 4 comments
Projects
None yet
Labels
parallel
Milestone
No milestone
Assignees
No one assigned
4 participants
@rszoh
@kmsquire
@ihnorton
@malmaud
Notifications

You’re not receiving notifications from this thread.
@rszoh
rszoh commented on Mar 26, 2015

When I run a piece of code in Julia 0.3.5 it runs perfectly but in other version like0 .3.6 and 0.3.7 it does not. I get this error:
convert has no method matching convert(::Type{Int64...}, ::Int64), which makes not sense to me, since convert has a method like that!
@kmsquire
The Julia Language member
kmsquire commented on Mar 26, 2015

@rszoh, thanks for the issue. However, without some code demonstrating your problem, it's impossible to debug. Can you post a code snippet and the exact output that it produces? A small example would be appreciated. Thanks!
@rszoh
rszoh commented on Mar 26, 2015

I made a post before reporting the issue and here is the link https://groups.google.com/forum/#!topic/julia-users/bWvsJqCCEEk

when it works I get:

Warning: could not import Base.add! into NumericExtensions
Done!
Start here ... 
Warning: could not import Base.add! into NumericExtensions
Warning: could not import Base.add! into NumericExtensions
    From worker 2:  Done!
    From worker 3:  Done!
Done!!
2x12 DataFrame
| Row | n1   | n2   | p     | k   | m    | kap0 | pct | mu  | Log_bay  | pos | stron | ver_strg |
|-----|------|------|-------|-----|------|------|-----|-----|----------|-----|-------|----------|
| 1   | 50.0 | 50.0 | 200.0 | 0.0 | 24.0 | 0.1  | 1.0 | 1.0 | -58.6029 | 0.0 | 0.0   | 0.0      |
| 2   | 50.0 | 50.0 | 200.0 | 1.0 | 24.0 | 0.1  | 1.0 | 1.0 | -58.4965 | 0.0 | 0.0   | 0.0      |

julia> versioninfo()
Julia Version 0.3.5
Platform Info:
  System: Linux (x86_64-redhat-linux)
  CPU: Intel(R) Xeon(R) CPU           E5520  @ 2.27GHz
  WORD_SIZE: 64
  BLAS: libopenblas (DYNAMIC_ARCH NO_AFFINITY Nehalem)
  LAPACK: libopenblasp.so.0
  LIBM: libopenlibm
  LLVM: libLLVM-3.4.2

However, when it fails on Julia 0.3.6 and 0.3.7 I get the following error:

fatal error on fatal error on 2: 3: ERROR: `convert` has no method matching conv
ert(::Type{Int64...}, ::Int64)
 in convert at base.jl:13
 in convert at base.jl:21
 in deserialize at serialize.jl:447
 in handle_deserialize at serialize.jl:351
 in deserialize at serialize.jl:334
 in anonymous at serialize.jl:354
 in ntuple at tuple.jl:30
 in deserialize_tuple at serialize.jl:354
 in handle_deserialize at serialize.jl:346
 in anonymous at task.jl:853

I hope that's not too much,

Many thanks!
@ihnorton ihnorton added the parallel label on Mar 28, 2015
@ihnorton
The Julia Language member
ihnorton commented on May 19, 2015

Is this still an issue on 0.3.8 with the latest updates to Devectorize and NumericExtensions?

Reduced case (from that email thread, thanks to @binarybana)

test.jl:

using Devectorize
function myfun(pam) 
  println("Entering myfun")
  foo = ones(5)
    @devec x = sum(foo)
  println("exiting myfun")
  return x
end

test with:

parm = rand(10,10)
M = { parm[i,:] for i in 1:2} # size(parm,1)}
require("test.jl")
pmap(myfun,M)

@malmaud
malmaud commented on Oct 14, 2015

I can't reproduce this problem on the current Julia master (b6c0d95) and Devectorize (754cc06), so I'll close for now. @rszoh, please reopen if this is still an issue for you.
@malmaud malmaud closed this on Oct 14, 2015