Intel® Many Integrated Core Architecture

Customize the uOS


for my masters thesis I am supposed to make the MIC runable with a Kernel Version of 3.13 or above. This also includes to use the Kernel mic_card Module on the MIC so that the Linux Kernel mic_host and mic_card cann communicate.

I already looked into porting the Intel Kmod Module, but this is not the goal.

Core pinning with pthread

Hi all, I did some follow-up on my last topic 

Long story short, I'm trying to emulate SIMD behavior in MIC environment,

and I actually decided to implement the offloaded segment in native code without OpenMP pragmas,

which means I'm implementing thread pinning to individual cores with pthread_attr_setaffinity_np.

The logic is all there from start to finish. Here's what I did:

MPI with hpcg application killed signal 9

I bought Intel Parallel Studio XE 2015 Cluster version for running MPI framework.

After make for co-processor only mode using -mmic option, run hpcg applicaiton.

Command that I used : mpiexec.hydra -iface mic0 -host mic0 -n 200 ./xhpcg

mic0 uses network file system sharing the same directory which I run.

But just I returned APPLICATION TERMINATED WITH THE EXIT STRING: Killed (signal 9)

So, I used -genv I_MPI_HYDRA_DEBUG option.

Why Xeon Phi always got bad efficacy?

I tried to run a for loop 1,000,000,000 times on Xeon E5 and Xeon Phi, and measurement time to compare their efficacy, I'm so surprise I got the following result:

On E5 (1 Thread): 41.563 Sec
On E5 (24 Threads): 22.788 Sec
Offload on Xeon Phi (240 Threads): 45.649 Sec

Why I got the bad efficacy on Xeon Phi? I do nothing on the for loop. If my Xeon Phi coprocessor didn't had any problem, what work for Xeon Phi can get good efficacy? Must be vectorization? if not vectorization, can I do any thing on Xeon Phi use its threads to help me something?

Fail to flash SMC (MPSS3.4.1)


I could not update my flash, time out on Centos 7.0 (64bit).

sudo micflash -update -device all
mic0: Flash image: /usr/share/mpss/flash/EXT_HP2_C0_0390-02.rom.smc
mic0: Flash update started
mic0: Flash update done
mic0: SMC update started
then time out.

Here is micinfo:

MicInfo Utility Log
Created Tue Nov 18 17:25:01 2014

Intel® Xeon Phi™ Coprocessor enabled applications on Stampede

This is intended to be an informational post -- I received the following announcement as I am on a TACC mailing list, wanted to pass the word around about the fact that several applications are readily available from their Stampede cluster -- more below (what follows is their announcement in entirety)


New Xeon Phi Applications on Stampede

Intel® MPSS - changes in release cadence and support

Up until now, Intel has been releasing its Manycore Platform Software Stack (Intel® MPSS) on a quarterly cadence, with each release being supported for 1 year from the date it was issued.

Beginning October 2014, the release timing and the support lifetime of Intel® MPSS is changing, namely to support divergent community needs:

Mounting over NFS with MPSS 3.4.1

Dear all,

I must apologise for this seemingly simple question.

I am having a problem mounting anything over nfs since I installed a new MIC card and installed MPSS 3.4.1.  I have a working NFS setup on another machine with MPSS 3.2.1 but the same script is not working for me with my new MIC.

I am using micctrl like this:

sudo micctrl --addnfs= --dir=/micnfs

I then ran:

ofed-mic cannot load ipb_server kernel module on SL6.5


Does MPSS 3.4.1 support kernel 2.6.32-431.29.2? From the documentation I see it should. We have installed OFED-3.5-2-MIC as recommended on Scientific Linux 6.5 and ofed-mic driver is built successfully (MPSS 3.4.1), but it cannot be started because ibp_server cannot be loaded. With other kernels it works fine although (but in that particular case we need to use that kernel for now).


running offload code on node without a mic

Hi again, i'd like to run my #pragma offload code on a node that does not have a mic present. When I try I get the following error:

"offload error: cannot offload to MIC - device is not available"

Is there a flag to tell the software to run the CPU only based version? I thought the binary had both code paths, so it should just choose a sensible one for what is available?

Thanks for your help.




Subscribe to Intel® Many Integrated Core Architecture