[svn-r5434] Purpose:

    Code improvement.

Description:
    Propagated the "fill time" property into the parallel chunk allocation
    routine, allowing it to avoid writing fill values to each new chunk
    allocated.  This improves the performance of chunked datasets in parallel
    I/O to be on par with contiguous datasets again (on modi4).

Platforms tested:
    IRIX64 6.5 (modi4)

1 files changed, 39 insertions, 26 deletions

diff --git a/src/H5Fistore.c b/src/H5Fistore.c
index 1d7feeb..a0f7497 100644
--- a/src/H5Fistore.c
+++ b/src/H5Fistore.c
@@ -2411,6 +2411,10 @@ done:
  * 		Quincey Koziol, 2002-05-16
  *		Rewrote algorithm to allocate & write blocks without using
  *              lock/unlock code.
+ *
+ * 		Quincey Koziol, 2002-05-17
+ *		Added feature to avoid writing fill-values if user has indicated
+ *              that they should never be written.
  *-------------------------------------------------------------------------
  */
 #ifdef H5_HAVE_PARALLEL
@@ -2422,6 +2426,7 @@ H5F_istore_allocate(H5F_t *f, hid_t dxpl_id, const H5O_layout_t *layout,
     hsize_t	chunk_size;     /* Size of chunk in bytes */
     H5O_pline_t pline;          /* I/O pipeline information */
     H5O_fill_t fill;            /* Fill value information */
+    H5D_fill_time_t fill_time;  /* When to write fill values */
     H5F_istore_ud1_t udata;	/* B-tree pass-through for creating chunk */
     void *chunk=NULL;           /* Chunk buffer for writing fill values */
     H5P_genplist_t *dx_plist;   /* Data xfer property list */
@@ -2429,7 +2434,7 @@ H5F_istore_allocate(H5F_t *f, hid_t dxpl_id, const H5O_layout_t *layout,
     int         mpi_rank;       /* This process's rank  */
     int         mpi_size;       /* Total # of processes */
     int         mpi_round=0;    /* Current process responsible for I/O */
-    unsigned    chunk_allocated=0; /* Flag to indicate that chunk was actually allocated */
+    unsigned    blocks_written=0; /* Flag to indicate that chunk was actually written */
     int		carry;          /* Flag to indicate that chunk increment carrys to higher dimension (sorta) */
     int		i;              /* Local index variable */
     unsigned	u;              /* Local index variable */
@@ -2452,6 +2457,8 @@ H5F_istore_allocate(H5F_t *f, hid_t dxpl_id, const H5O_layout_t *layout,
         HGOTO_ERROR(H5E_STORAGE, H5E_CANTGET, FAIL, "can't get fill value");
     if(H5P_get(dc_plist, H5D_CRT_DATA_PIPELINE_NAME, &pline) < 0)
         HGOTO_ERROR(H5E_STORAGE, H5E_CANTGET, FAIL, "can't get data pipeline");
+    if(H5P_get(dc_plist, H5D_CRT_FILL_TIME_NAME, &fill_time) < 0)
+        HGOTO_ERROR(H5E_PLIST, H5E_CANTGET, NULL, "can't retrieve fill time");
 
     /* Get necessary properties from dataset transfer property list */
     if (TRUE!=H5P_isa_class(dxpl_id,H5P_DATASET_XFER) || NULL == (dx_plist = H5I_object(dxpl_id)))
@@ -2472,23 +2479,26 @@ H5F_istore_allocate(H5F_t *f, hid_t dxpl_id, const H5O_layout_t *layout,
         chunk_size *= layout->dim[u];
     } /* end for */
 
-    /* Allocate chunk buffer for processes to use when writing fill values */
-    if (NULL==(chunk = H5F_istore_chunk_alloc(chunk_size)))
-        HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed for chunk");
+    /* Check if fill values should be written to blocks */
+    if(fill_time != H5D_FILL_TIME_NEVER) {
+        /* Allocate chunk buffer for processes to use when writing fill values */
+        if (NULL==(chunk = H5F_istore_chunk_alloc(chunk_size)))
+            HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed for chunk");
 
-    /* Fill the chunk with the proper values */
-    if(fill.buf) {
-        /*
-         * Replicate the fill value throughout the chunk.
-         */
-        assert(0==chunk_size % fill.size);
-        H5V_array_fill(chunk, fill.buf, fill.size, chunk_size/fill.size);
-    } else {
-        /*
-         * No fill value was specified, assume all zeros.
-         */
-        HDmemset (chunk, 0, chunk_size);
-    } /* end else */
+        /* Fill the chunk with the proper values */
+        if(fill.buf) {
+            /*
+             * Replicate the fill value throughout the chunk.
+             */
+            assert(0==chunk_size % fill.size);
+            H5V_array_fill(chunk, fill.buf, fill.size, chunk_size/fill.size);
+        } else {
+            /*
+             * No fill value was specified, assume all zeros.
+             */
+            HDmemset (chunk, 0, chunk_size);
+        } /* end else */
+    } /* end if */
 
     /* Retrieve up MPI parameters */
     if ((mpi_rank=H5FD_mpio_mpi_rank(f->shared->lf))<0)
@@ -2513,15 +2523,18 @@ H5F_istore_allocate(H5F_t *f, hid_t dxpl_id, const H5O_layout_t *layout,
             if (H5B_insert(f, H5B_ISTORE, layout->addr, split_ratios, &udata)<0)
                 HGOTO_ERROR(H5E_IO, H5E_WRITEERROR, FAIL, "unable to allocate chunk");
 
-            /* Round-robin write the chunks out from only one process */
-            if(mpi_round==mpi_rank) {
-                if (H5F_block_write(f, H5FD_MEM_DRAW, udata.addr, udata.key.nbytes, dxpl_id, chunk)<0)
-                    HGOTO_ERROR(H5E_IO, H5E_WRITEERROR, FAIL, "unable to write raw data to file");
+            /* Check if fill values should be written to blocks */
+            if(fill_time != H5D_FILL_TIME_NEVER) {
+                /* Round-robin write the chunks out from only one process */
+                if(mpi_round==mpi_rank) {
+                    if (H5F_block_write(f, H5FD_MEM_DRAW, udata.addr, udata.key.nbytes, dxpl_id, chunk)<0)
+                        HGOTO_ERROR(H5E_IO, H5E_WRITEERROR, FAIL, "unable to write raw data to file");
+                } /* end if */
+                mpi_round=(++mpi_round)%mpi_size;
+
+                /* Indicate that blocks are being written */
+                blocks_written=1;
             } /* end if */
-            mpi_round=(++mpi_round)%mpi_size;
-
-            /* Indicate that a chunk was allocated */
-            chunk_allocated=1;
         } /* end if */
 	
         /* Increment indices */
@@ -2536,7 +2549,7 @@ H5F_istore_allocate(H5F_t *f, hid_t dxpl_id, const H5O_layout_t *layout,
     } /* end while */
 
     /* Only need to block at the barrier if we actually allocated a chunk */
-    if(chunk_allocated) {
+    if(blocks_written) {
         /* Wait at barrier to avoid race conditions where some processes are
          * still writing out chunks and other processes race ahead to read
          * them in, getting bogus data.