Changes

← Older edit

Team Lion F2017

2,335 bytes added, 12:53, 5 January 2018

no edit summary

[[File:Summary.PNG]]

~~===Advanced hotspot analysis===~~

[[File:Function_timmings.PNG]]

the image above shows the timings for each function

matmul_0 - represents serial version

matmul_1 - represents serial version with reverse logic

matmul_2 - uses cilk_for

matmul_3 - uses cilk_for and reducer hyperboject

matmul_4 - uses cilk_for, reducer and vectorization

====matmul_0 (Serial)====

<pre>

double matmul_0(const double* a, const double* b, double* c, int n) {

for (int i = 0; i < n; i++) {

for (int j = 0; j < n; j++) {

double sum = 0.0;

for (int k = 0; k < n; k++)

sum += a[i * n + k] * b[k * n + j];

c[i * n + j] = sum;

}

double diag = 0.0;

for (int i = 0; i < n; i++)

diag += c[i * n + i];

return diag;

}

</pre>

[[File:Conc-01.png]]

====matmul_1 (Serial with j-k loops reversed)====

<pre>

double matmul_1(const double* a, const double* b, double* c, int n) {

for (int i = 0; i < n; i++) {

for (int k = 0; k < n; k++) {

double sum = 0.0;

for (int j = 0; j < n; j++)

sum += a[i * n + k] * b[k * n + j];

c[i * n + k] = sum;

}

double diag = 0.0;

for (int i = 0; i < n; i++)

diag += c[i * n + i];

return diag;

}

</pre>

[[File:Conc-11.png]]

====matmul_2 (Cilk Plus with cilk_for)====

<pre>

double matmul_2(const double* a, const double* b, double* c, int n) {

cilk_for (int i = 0; i < n; i++) {

cilk_for (int j = 0; j < n; j++) {

double sum = 0.0;

for(int k = 0; k < n; k++) {

sum += a[i * n + k] * b[k * n + j];

}

c[i * n + j] = sum;

}

double diag = 0.0;

for (int i = 0; i < n; i++)

diag += c[i * n + i];

return diag;

}

</pre>

[[File:Conc-21.png]]

====matmul_3 (+array notation, reducer)====

<pre>

double matmul_3(const double* a, const double* b, double* c, int n) {

cilk_for(int i = 0; i < n; i++) {

cilk_for(int j = 0; j < n; j++) {

double sum = 0.0;

for (int k = 0; k < n; k++) {

sum += a[i * n + k] * b[k * n + j];

}

c[i * n + j] = sum;

}

cilk::reducer_opadd <double> diag(0.0);

cilk_for(int i = 0; i < n; i++) {

diag += c[i * n + i];

}

return diag.get_value();

}

</pre>

[[File:Conc-31.png]]

====matmul_4 (+vectorization)====

<pre>

double matmul_4(const double* a, const double* b, double* c, int n) {

cilk_for(int i = 0; i < n; i++) {

cilk_for(int j = 0; j < n; j++) {

double sum = 0.0;

#pragma simd

for (int k = 0; k < n; k++) {

sum += a[i * n + k] * b[k * n + j];

}

c[i * n + j] = sum;

}

cilk::reducer_opadd <double> diag(0.0);

cilk_for(int i = 0; i < n; i++) {

diag += c[i * n + i];

}

return diag.get_value();

}

</pre>

[[File:Conc-41.png]]

===Locks & Waits===

~~===HPC Performance Characterization===~~* Best for locating causes of low concurrency, such as heavily used locks and large critical sections.* Locks are when threads are waiting too long on synchronization objects.* Uses user-mode sampling and tracing collection to identify processes. * This analysis shows time spent waiting on synchronizations.

~~==Microarchitecture==~~

~~===General Exploration===~~[[File:Lock1.png]]

[[File:Lock2.png]]

~~===Memory Access===~~[[File:Lock3.png]]

==references==

https://software.intel.com/en-us/vtune-amplifier-help-locks-and-waits-analysis

https://software.intel.com/en-us/~~vtune-amplifier-help-hpc-performance-characterization-analysis~~ ~~https://software.intel.com/en-us/vtune-amplifier-help-general-exploration-analysis~~vtuneampxe_hotspots_win_c

https://software.intel.com/en-us/~~vtune-amplifier-help-memory-access-analysis~~vtuneampxe_locks_win_c

Jsbhamber2

60

edits

CDOT Wiki β

Changes

Team Lion F2017

CDOT Wiki ^β