Intel® Array Building Blocks (Archived)

End of Beta program of Intel(r) Array Building Blocks

Thank you for participating in the Intel Array Building Blocks (Intel ArBB) beta program. We would like to inform you that the Intel ArBB Beta program is completed as of today, November 11th, 2011.

We really appreciate your participation in the program and greatly value the feedback you provided. Based on all the input we received, we have decided to transition Intel ArBB to our research and exploration portal at whatif.intel.com to broaden exposure of the technology to a broader audience.

The Intel ArBB Virtual Machine Specification is now available!

The first draft of the IntelArray building Blocks (IntelArBB) Virtual Machine (VM) API specification is now available on our documentation page. For more information, including links to a video presentation hosted by UPCRC Illinois, check out the related knowledge base article.

Parallel Program Made in Array Building Block runs slower than my sequencial program

I have made a program of LU Decomposition using Gauss Elimination method in Intel Array Building Block but it is running quite slow.

For a 512 x 512 matrix time for my serial and parallel programs are

Serial execution : 76 ms

Parallel Execution using Arbb : 230 ms

My Arbb code is

#include<arbb.hpp>
#include<stdio.h>
using namespace arbb;

pointers inside arbb space?

Hello allI'm just figuring out how to implement a multi-column, multi-type data structure. There are about five types (known at compile-time) and each column can have a different one. The number of columns and the type for each column is available at runtime only. 2-dim dense containers or a combination of dense containers inside nested containers doesn't fit well. There will be added sophisticated operations on these data structures later. Is any "pointering" inside arbb space/custom types possible?

Dynamic Compiler Internal Error in 1.0.0.030

Dear Array Building Blocks builders,

after a year since last time I looked into ArBB I decided to give it a new try.
I still think that the ArBB is an interesting one, and that it provides expressing algorithm in a clear way while at the same time being (very) fast at run-time.

My new attempt is a fail.
I got the code to compile and run with ARBB_EMULATE=1 but when I switch to ARBB_EMULATE=0 I receive an

Internal error: CTE_COMPILER_ERROR COMP_ERROR: Dynamic Compiler Internal Error

ArBB Performance Problems

I'm running ArBB v1.0.0.030 on my 64bit ubuntu 10.04 with this simple test:

dense maa=fill(3, ARRAY_SIZE);
/*
{
const_range r=maa.read_only_range();

for(int i=0;i<10;i++)
{
printf("maa[%d]=%d\\n",i,maa[i]);
}
for(int i=ARRAY_SIZE-10;i
{
printf("maa[%d]=%d\\n",i,maa[i]);
}
}
*/
dense maa2=maa+1;

dense maa3=maa2*maa2;

for(int p=0;p
{
maa3=maa3*maa3;
}

where ARRAY_SIZE is in the order of 2^10 and PDS_ITER some 10^4.

Iscriversi a Intel® Array Building Blocks (Archived)