1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
|
// =============================================================================
// === GPUQREngine/Source/GPUQREngine_Internal.cpp =============================
// =============================================================================
//
// GPUQREngine_Internal is the call-down from the dense and sparse polymorphic
// wrappers. This code is responsible for maintaining the Scheduler and
// coordinating the factorization in a main loop.
//
// =============================================================================
#include "GPUQREngine_Internal.hpp"
#include "GPUQREngine_Scheduler.hpp"
#include "GPUQREngine_Stats.hpp"
QREngineResultCode GPUQREngine_Internal
(
size_t gpuMemorySize, // The total available GPU memory size in bytes
Front *fronts, // The list of fronts to factorize
Int numFronts, // The number of fronts to factorize
Int *Parent, // The front-to-parent mapping
Int *Childp, // Front-to-child column pointers
Int *Child, // Child permutation
// (Child[Childp[f]] to Child[Childp[f+1]] are all
// the front identifiers for front "f"'s children.
QREngineStats *stats // An optional parameter. If present, statistics
// are collected and passed back to the caller
// via this struct
)
{
bool ok = true;
/* Create the scheduler. */
Scheduler *scheduler = (Scheduler*) SuiteSparse_calloc(1,sizeof(Scheduler));
if (scheduler == NULL)
{
return QRENGINE_OUTOFMEMORY;
}
new (scheduler) Scheduler(fronts, numFronts, gpuMemorySize);
ok = scheduler->memory_ok && scheduler->cuda_ok;
/* If we encountered problems initializing the scheduler: */
if(!ok)
{
bool memory_ok = scheduler->memory_ok ;
bool cuda_ok = scheduler->cuda_ok ;
if(scheduler)
{
scheduler->~Scheduler();
scheduler = (Scheduler*) SuiteSparse_free (scheduler) ;
}
if(!memory_ok) return QRENGINE_OUTOFMEMORY;
if(!cuda_ok) return QRENGINE_GPUERROR;
}
bool completed = false;
while(!completed)
{
// #ifdef GPUQRENGINE_RENDER
// scheduler->render();
// #endif
scheduler->fillWorkQueue();
scheduler->transferData();
// Launch the kernel and break out of the loop if we encountered
// a cuda error.
scheduler->launchKernel();
if(!scheduler->cuda_ok) break;
completed = scheduler->postProcess();
scheduler->toggleQueue();
}
/* Report metrics back to the caller. */
if(stats)
{
stats->kernelTime = scheduler->kernelTime;
stats->numLaunches = scheduler->numKernelLaunches;
stats->flopsActual = scheduler->gpuFlops;
}
/* Explicitly invoke the destructor */
scheduler->~Scheduler();
scheduler = (Scheduler*) SuiteSparse_free(scheduler);
return QRENGINE_SUCCESS;
}
|
