Параллельные вычисления

MPI_Allgatherv with large message sizes


I'm trying to collect data with MPI_Allgatherv with a large receive buffer for which the total size is larger than 2GB. As I could understand here (http://software.intel.com/en-us/forums/topic/361060) this is not supported. Unfortunately when I try to use the -ilp64 option with mpiifort I run into several problems:

1) when using include 'mpif.h' to  include mpi, then after the following commands:

mpiifort -warn -O1 -g -traceback -check bounds -i8 -c gather.f

FAX CED (ANS, ANSam) tone detection


I am working on a project that requires me to detect FAX answer tone (CED) (2100Hz duration 2.6 to 4 seconds).

I created a tone using an online tone generator and fed raw PCM data to 'DetectTone' function multiple times, but it did not detect a tone. I tried initing USC_TD_Fxns with 'USC_ANSs_Fxns' and 'USC_ANSam_Fxns'.

Since this did not work, I did a loopback of the data generated by IPP 'GenerateTone' function the same way as above. I inited the tone generator with (USC_ANSam_Fxns) and generated USC_ANSam tone, with duration 10 and volume 1.

internal_error: 20000_27078 since updating to 2013 sp1

I get this very cryptic and unhelpful error message since updating to 2013 sp1 Update 2 (also happened with SP1 and SP1 Update 1).  In the past I have just reverted back to the 2013 suite to solve this problem, but am now getting concerned that this hasn't been fixed yet!

If I could at least get a line number out of the damned thing then I could experiment with the source file and produce a simple test case.  Any suggestions would be welcome!



The full error message is as follows (where <filename> is the C file where the problem occurs):

Documentation for kernel development?

Are there any resources available for those who might wish to recompile the Linux kernel used on the Phi? Are there any tools in MPSS to help us recompile k1om Linux and/or boot it? I'm using MPSS 3.1.2.

I could hopefully hack my way through it, but I would hate to do so having missed some existing instructions or tools which could have made things easier.

Offload error

Dear all, I'm a novice for MIC programming and I encountered a offload error when I want to run an example code on the host and offload part of the code to MIC cores for multithread-computing based on OpenMP. The offload error is: cannot offload to MIC - device is not available Out cluster has 8 mic nodes with the naming convention of nodeXX-mic0 (XX = 09~16). I think if I want to run the code, I have to specify which node to use, right? How can I do it? I searched for the internet and I didn't find an applicable answer. It would be great if I could get your help.

Why has CPU frequency ceased to grow?

All of you probably recall the rapid rate of CPU frequency advancement at the end of the last century and beginning of this one.  Tens of megahertz rapidly transformed into hundreds, and then hundreds of megahertz quickly became a full gigahertz, then a gigahertz and a bit, finally two gigs and a bit.

Xeon Phi 3120A Fan Speed

Is there a way to set the fan at a higher speed?

My fan doesn't kick in until the temperature is around 81 or 82 degrees, which concerns me because it pretty much stays at 81/82 degrees and I would like it to kick in much earlier.


(Windows 7 Ultimate 64 bit)

Resolving kernel symbols with Xeon Phi and MPSS 3.1.2

I'm trying to understand why an application is spending lots of time in [vmlinux], but I can't get VTune to resolve the MIC's kernel. I have found some suggestions (http://software.intel.com/en-us/forums/topic/387212) to tell amplxe-cl to search /lib/firmware/mic to resolve kernel symbols, but this directory doesn't exist anymore. (I imagine it did with earlier versions of the MIC tools.)

I'm currently running amplxe-cl like so:

“FUNCTION WAS VECTORIZED” but it doesn't vectorize on the place of the function call

The present question relates to an already existing question on Stackoverflow with the difference that in this case AVX is the target ISA and that the function to be vectorized is more complex. When I use the __attribute__((vector(...))) declaration in the function definition:

Подписаться на Параллельные вычисления