Intel® Concurrent Collections for C++

Prescribing work based on global state

I have been reading through the documentation of Intel CnC and am trying to map Intel CnC into my sequential image stitching algorithm.

The algorithm consists of three steps and two item collections.

Steps: (1) Read image, (2) Compute the image's FFT, and (3) Compute a phase correlation image alignment method between two image's FFTs.

Item collections: (1) Image, (2) Pair of Images

One thing to note, for the item collections I plan to use C++ classes. In these classes they have pointers to the image data as well as the FFT data.

 

Question:

Call for Abstracts: The Seventh Annual Concurrent Collections Workshop (CnC 2015)

Call for Abstracts:

CnC 2015: The Seventh Annual Concurrent Collections Workshop

https://engineering.purdue.edu/plcl/cnc2015

September 7-8, 2015 at Raleigh, NC (co-located with LCPC'15)

______________________________________________________

Important Dates

Abstracts Due: August 10, 2015

Acceptance Notification: August 13, 2015

Call for Participation for CnC-2014: The Sixth Annual Concurrent Collections Workshop

Call for Participation

CnC-2014: The Sixth Annual Concurrent Collections Workshop, http://cass-mt.pnnl.gov/cnc2014

September 18-19, 2014 at Intel Corporation, Hillsboro, OR (co-located with LCPC'14)

____________________________________________________________________________________

Address: 2111 NE 25th Ave, Hillsboro, OR 97124

(Jones Farm Campus)

Workshop Agenda:

Thursday, September 18th, 2014: (JFCC 119)

Name of the C++ compiler help needed

I have Parallel Studio 2013 SP rev 3 recently installed in my Windows 64 bit computer with a Intel CPU.  I am trying Wolfram Mathematica use the Intell C++ compiler.

I need the directory name which if correct is at

"C:\\Program Files (x86)\\Intel\\Composer XE 2013 SP1"

 

But now, I am suppose to enter the compiler name with extention. I am not sure whi one numerous files is the actual compiler and the correct extension. 

Segmentation fault with prime example

I downloaded both Intel Concurrent Collections for C++ binary and source and then installed on 4 VMware images. Each of them has Fedora 17 installed. I attempted to run prime/prime example. It always ended with a segmentation fault. The segmentation occurred in CnC::Internal::chronometer::step_timer::start_put() method which is around the call to cnc_rdtsc method. See attached file for details.

I compiled the libraries myself and still observed the same issue. Any suggestion?

Minh-Nhut

Call for Abstract for CnC-2014: The Sixth Annual Concurrent Collections Workshop

____________________________________________________________________________________

Call for Abstracts:

 

 CnC-2014: The Sixth Annual Concurrent Collections Workshop

http://cass-mt.pnnl.gov/cnc2014

 

September 18-19, 2014 at Intel Corporation, Hillsboro, OR (co-located with LCPC'14)

____________________________________________________________________________________

 

Important Dates

Abstracts Due: August 8, 2014

Notification: August 22, 2014

Intel(R) CnC (Concurrent Collections) 1.0 available in open source

The new version 1.0 of Intel(R) Concurrent Collections for C++ is available at

https://software.intel.com/en-us/articles/intel-concurrent-collections-f...

For the first time it has also been released as open source. Check out it's new home-page on git-hub: https://icnc.github.io

Other news of this release are

  • Support for re-use

Limit the Memory Consumption

I'm running my program in a HPC environment (no hard drives, no swap). The system kills any program that is using too much memory. I used the garbage tuners on my collection, but I'm afraid that at some point where I reach my maximum memory capacity, your scheduler launches a step that produces more data thus increasing the memory footprint instead of launching a step that consumes data thus decreasing the memory footprint.

Is there anyway to limit the memory consumption?

Subscribe to Intel® Concurrent Collections for C++