Choose timezone
Your profile timezone:
Help us make Indico better by taking this survey! Aidez-nous à améliorer Indico en répondant à ce sondage !
Color code: (critical, news during the meeting: green, news from this week: blue, news from last week: purple, no news: black)
High priority Framework issues:
Global calibration topics:
Sync reconstruction
Async reconstruction
EPN major topics:
Other EPN topics:
Full system test issues:
Topology generation:
AliECS related topics:
GPU ROCm / compiler topics:
TPC GPU Processing
TPC processing performance regression:
General GPU Processing
Updates on ONNX runtime implementation
Updates on NN side
Updates on simulation
processBulk
version (see below).processBulk()
CPU handler to perform simultaneous fitting (1 Fit/thread) of a vector of TrackParCov
s: PR.
#include <array>
#include <iostream>
#include <experimental/meta>
template <typename T, std::size_t N>
struct struct_of_arrays {
// omitted
// ...
// ...
};
struct point {
float x;
float y;
float z;
};
int main() {
struct_of_arrays<point, 2> p;
p.x = {1.1, 2.2};
p.y = {3.3, 4.4};
p.z = {5.5, 6.6};
std::cout << p.x[0] << ", " << p.x[1] << std::endl; // output: 1.1, 2.2
return 0;
}
Remarks:
For some computations, especially the ones suited for GPU, the SoA memory layout might yield better performance.