regblock.h File Reference

Detailed Description

Implementation of the basic register blocking heuristic for sparse matrix-vector multiply (as described in the Sparsity SC'02 and IJHPCA'04 papers), based on block compressed sparse row format.

#include <oski/common.h>
#include <oski/modcommon.h>
#include <oski/heurexport.h>

Go to the source code of this file.

Data Structures

struct oski_regblock_t

Stores the results of evaluating the register blocking heuristic. More...

struct oski_profile_BCSRMatMult_t

Stores a register blocking profile. More...

Name mangling.

#define oski_regblock_t   MANGLE_(oski_regblock_t)

#define oski_LoadProfileBCSRMatMult   MANGLE_(oski_LoadProfileBCSRMatMult)

#define oski_UnloadProfileBCSRMatMult   MANGLE_(oski_UnloadProfileBCSRMatMult)

Defines

#define INC_OSKI_HEUR_REGBLOCK_H

oski/heur/regblock.h included.

#define PROF_BCSR_GET(P, r, c)   ((P)->perf[((r)-1)*(P)->max_c + (c)-1])

Returns the performance (Mflop/s) at a block size $r\times c$ , where $r$ and $c$ are 1-based.

#define PROF_BCSR_SET(P, r, c, v)   ((P)->perf[((r)-1)*(P)->max_c + (c)-1] = (v))

Set the performance (Mflop/s) at a block size $r\times c$ , where $r$ and $c$ are 1-based.

Functions

const oski_profile_BCSRMatMult_t * oski_LoadProfileBCSRMatMult (const char *filename)

Load the uniformly-aligned register blocking profile for sparse matrix-vector multiply.

void oski_UnloadProfileBCSRMatMult (void)

Unload the uniformly-aligned register blocking profile for sparse matrix-vector multiply.

Generated on Wed Sep 19 16:41:23 2007 for BeBOP Optimized Sparse Kernel Interface Library by

1.4.6


Data Structures
struct	oski_regblock_t
	Stores the results of evaluating the register blocking heuristic. More...
struct	oski_profile_BCSRMatMult_t
	Stores a register blocking profile. More...
Name mangling.
#define	oski_regblock_t MANGLE_(oski_regblock_t)
#define	oski_LoadProfileBCSRMatMult MANGLE_(oski_LoadProfileBCSRMatMult)
#define	oski_UnloadProfileBCSRMatMult MANGLE_(oski_UnloadProfileBCSRMatMult)
Defines
#define	INC_OSKI_HEUR_REGBLOCK_H
	oski/heur/regblock.h included.
#define	PROF_BCSR_GET(P, r, c) ((P)->perf[((r)-1)*(P)->max_c + (c)-1])
	Returns the performance (Mflop/s) at a block size $r\times c$ , where and are 1-based.
#define	PROF_BCSR_SET(P, r, c, v) ((P)->perf[((r)-1)*(P)->max_c + (c)-1] = (v))
	Set the performance (Mflop/s) at a block size $r\times c$ , where and are 1-based.
Functions
const oski_profile_BCSRMatMult_t *	oski_LoadProfileBCSRMatMult (const char *filename)
	Load the uniformly-aligned register blocking profile for sparse matrix-vector multiply.
void	oski_UnloadProfileBCSRMatMult (void)
	Unload the uniformly-aligned register blocking profile for sparse matrix-vector multiply.