#include <EcalLinearCorrectionsGPU.h>

Classes
struct	Product

Public Member Functions
	EcalLinearCorrectionsGPU (EcalLinearCorrections const &)

Product const &	getProduct (cudaStream_t) const

	~EcalLinearCorrectionsGPU ()=default

Static Public Member Functions
static std::string	name ()

Private Attributes
std::vector< float, cms::cuda::HostAllocator< float > >	p1_

std::vector< float, cms::cuda::HostAllocator< float > >	p2_

std::vector< float, cms::cuda::HostAllocator< float > >	p3_

cms::cuda::ESProduct< Product >	product_

std::vector< edm::TimeValue_t, cms::cuda::HostAllocator< edm::TimeValue_t > >	t1_

std::vector< edm::TimeValue_t, cms::cuda::HostAllocator< edm::TimeValue_t > >	t2_

std::vector< edm::TimeValue_t, cms::cuda::HostAllocator< edm::TimeValue_t > >	t3_

Detailed Description

Definition at line 13 of file EcalLinearCorrectionsGPU.h.

Constructor & Destructor Documentation

◆ EcalLinearCorrectionsGPU()

EcalLinearCorrectionsGPU::EcalLinearCorrectionsGPU ( EcalLinearCorrections const & values )

Definition at line 6 of file EcalLinearCorrectionsGPU.cc.

References mps_fire::i, p1_, p2_, p3_, t1_, t2_, t3_, and contentValuesCheck::values.

     : p1_(values.getValueMap().size()),
       p2_(values.getValueMap().size()),
       p3_(values.getValueMap().size()),
       t1_(values.getTimeMap().size()),
       t2_(values.getTimeMap().size()),
       t3_(values.getTimeMap().size()) {
   // fill in eb
   for (unsigned int i = 0; i < values.getValueMap().barrelItems().size(); i++) {
     p1_[i] = values.getValueMap().barrelItems()[i].p1;
     p2_[i] = values.getValueMap().barrelItems()[i].p2;
     p3_[i] = values.getValueMap().barrelItems()[i].p3;
   }
 
   // fill in ee
   auto const offset_laser = values.getValueMap().barrelItems().size();
   for (unsigned int i = 0; i < values.getValueMap().endcapItems().size(); i++) {
     p1_[offset_laser + i] = values.getValueMap().endcapItems()[i].p1;
     p2_[offset_laser + i] = values.getValueMap().endcapItems()[i].p2;
     p3_[offset_laser + i] = values.getValueMap().endcapItems()[i].p3;
   }
 
   //   Time is a simple std::vector
   //       typedef std::vector<EcalLaserTimeStamp> EcalLaserTimeStampMap;
   for (unsigned int i = 0; i < values.getTimeMap().size(); i++) {
     t1_[i] = values.getTimeMap()[i].t1.value();
     t2_[i] = values.getTimeMap()[i].t2.value();
     t3_[i] = values.getTimeMap()[i].t3.value();
   }
 }

◆ ~EcalLinearCorrectionsGPU()

EcalLinearCorrectionsGPU::~EcalLinearCorrectionsGPU ( )

default

Member Function Documentation

◆ getProduct()

EcalLinearCorrectionsGPU::Product const & EcalLinearCorrectionsGPU::getProduct ( cudaStream_t cudaStream ) const

Definition at line 37 of file EcalLinearCorrectionsGPU.cc.

References cms::cuda::copyAsync(), EcalLinearCorrectionsGPU::Product::p1, p1_, EcalLinearCorrectionsGPU::Product::p2, p2_, EcalLinearCorrectionsGPU::Product::p3, p3_, product_, EcalLinearCorrectionsGPU::Product::t1, t1_, EcalLinearCorrectionsGPU::Product::t2, t2_, EcalLinearCorrectionsGPU::Product::t3, and t3_.

                                                                                                          {
   auto const& product = product_.dataForCurrentDeviceAsync(
       cudaStream, [this](EcalLinearCorrectionsGPU::Product& product, cudaStream_t cudaStream) {
         // allocate
         product.p1 = cms::cuda::make_device_unique<float[]>(p1_.size(), cudaStream);
         product.p2 = cms::cuda::make_device_unique<float[]>(p2_.size(), cudaStream);
         product.p3 = cms::cuda::make_device_unique<float[]>(p3_.size(), cudaStream);
         product.t1 = cms::cuda::make_device_unique<edm::TimeValue_t[]>(t1_.size(), cudaStream);
         product.t2 = cms::cuda::make_device_unique<edm::TimeValue_t[]>(t2_.size(), cudaStream);
         product.t3 = cms::cuda::make_device_unique<edm::TimeValue_t[]>(t3_.size(), cudaStream);
         // transfer
         cms::cuda::copyAsync(product.p1, p1_, cudaStream);
         cms::cuda::copyAsync(product.p2, p2_, cudaStream);
         cms::cuda::copyAsync(product.p3, p3_, cudaStream);
         cms::cuda::copyAsync(product.t1, t1_, cudaStream);
         cms::cuda::copyAsync(product.t2, t2_, cudaStream);
         cms::cuda::copyAsync(product.t3, t3_, cudaStream);
       });
 
   return product;
 }