From: Martin Mares <mj@ucw.cz>
Date: Fri, 24 Nov 2006 18:07:03 +0000 (+0100)
Subject: Much better parallel copy benchmark.
X-Git-Tag: holmes-import~506^2~13^2~230
X-Git-Url: http://mj.ucw.cz/gitweb/?a=commitdiff_plain;h=0fb2881e28c685ac25d7c4de00e506c9fd3a4b9d;p=libucw.git

Much better parallel copy benchmark.
---

diff --git a/debug/sorter/file-test.c b/debug/sorter/file-test.c
index c1672d29..f861bbb5 100644
--- a/debug/sorter/file-test.c
+++ b/debug/sorter/file-test.c
@@ -10,74 +10,111 @@
 #include <fcntl.h>
 #include <unistd.h>
 
+#define COPY
+#define DIRECT O_DIRECT
+
+#define P_INIT do { cnt = 0; cnt_rep = 0; cnt_ms = 1; } while(0)
+#define P_UPDATE(cc) do { \
+  cnt += cc; \
+  if (cnt >= cnt_rep) { cnt_ms += get_timer(); \
+    printf("%d of %d MB (%.2f MB/sec)\r", (int)(cnt >> 20), (int)(total_size >> 20), (double)cnt / 1048576 * 1000 / cnt_ms); \
+    fflush(stdout); cnt_rep += 1<<26; } } while(0)
+#define P_FINAL do { \
+  cnt_ms += get_timer(); \
+  log(L_INFO, "Spent %.3f sec (%.2f MB/sec)", (double)cnt_ms/1000, (double)cnt / 1048576 * 1000 / cnt_ms); \
+} while(0)
+
 int main(int argc, char **argv)
 {
   ASSERT(argc == 4);
   uns files = atol(argv[1]);
   uns bufsize = atol(argv[2]) * 1024;				// Kbytes
-  uns rounds = (u64)atol(argv[3]) * 1024*1024 / bufsize;	// Mbytes
+  u64 total_size = (u64)atol(argv[3]) * 1024*1024*1024;		// Gbytes
+  u64 cnt, cnt_rep;
+  uns cnt_ms;
   int fd[files];
-  byte *buf[files];
+  byte *buf[files], name[files][16];
+  uns xbufsize = bufsize;					// Used for single-file I/O
+  byte *xbuf = big_alloc(xbufsize);
+
+  init_timer();
 
-  log(L_INFO, "Initializing");
+#ifdef COPY
+  log(L_INFO, "Creating input file");
+  int in_fd = sh_open("tmp/ft-in", O_RDWR | O_CREAT | O_TRUNC | DIRECT, 0666);
+  ASSERT(in_fd >= 0);
+  ASSERT(!(total_size % xbufsize));
+  P_INIT;
+  for (uns i=0; i<total_size/xbufsize; i++)
+    {
+      for (uns j=0; j<xbufsize; j++)
+	xbuf[j] = i+j;
+      uns c = write(in_fd, xbuf, xbufsize);
+      ASSERT(c == xbufsize);
+      P_UPDATE(c);
+    }
+  lseek(in_fd, 0, SEEK_SET);
+  sync();
+  P_FINAL;
+#endif
+
+  log(L_INFO, "Initializing output files");
   for (uns i=0; i<files; i++)
     {
-      byte name[16];
-      sprintf(name, "tmp/ft-%d", i);
-      fd[i] = sh_open(name, O_RDWR | O_CREAT | O_TRUNC, 0666);
+      sprintf(name[i], "tmp/ft-%d", i);
+      fd[i] = sh_open(name[i], O_RDWR | O_CREAT | O_TRUNC | DIRECT, 0666);
       if (fd[i] < 0)
-	die("Cannot create %s: %m", name);
+	die("Cannot create %s: %m", name[i]);
       buf[i] = big_alloc(bufsize);
     }
   sync();
+  get_timer();
 
-  log(L_INFO, "Writing %d files in parallel with %d byte buffers", files, bufsize);
-  init_timer();
-  u64 total = 0, total_rep = 0;
-  for (uns r=0; r<rounds; r++)
+  log(L_INFO, "Writing %d MB to %d files in parallel with %d byte buffers", (int)(total_size >> 20), files, bufsize);
+  P_INIT;
+  for (uns r=0; r<total_size/bufsize/files; r++)
     {
       for (uns i=0; i<files; i++)
 	{
+#ifdef COPY
+	  uns ci = read(in_fd, buf[i], bufsize);
+	  ASSERT(ci == bufsize);
+#else
 	  for (uns j=0; j<bufsize; j++)
 	    buf[i][j] = r+i+j;
+#endif
 	  uns c = write(fd[i], buf[i], bufsize);
 	  ASSERT(c == bufsize);
-	  total += c;
-          if (total >= total_rep + 1024*1024*1024)
-	    {
-	      printf("Wrote %d MB (round %d of %d)\r", (int)(total >> 20), r, rounds);
-	      fflush(stdout);
-	      total_rep = total;
-	    }
+	  P_UPDATE(c);
 	}
     }
+#ifdef COPY
+  close(in_fd);
+#endif
   log(L_INFO, "Syncing");
   sync();
-  uns ms = get_timer();
-  log(L_INFO, "Spent %dms (%d MB/sec)", ms, (uns)(total/ms*1000/1048576));
+  P_FINAL;
 
   log(L_INFO, "Reading the files sequentially");
-  total = total_rep = 0;
+  P_INIT;
   for (uns i=0; i<files; i++)
     {
       lseek(fd[i], 0, SEEK_SET);
-      for (uns r=0; r<rounds; r++)
+      for (uns r=0; r<total_size/xbufsize/files; r++)
 	{
-	  uns c = read(fd[i], buf[i], bufsize);
-	  ASSERT(c == bufsize);
-	  total += c;
-          if (total >= total_rep + 1024*1024*1024)
-	    {
-	      printf("Read %d MB (file %d)\r", (int)(total >> 20), i);
-	      fflush(stdout);
-	      total_rep = total;
-	    }
+	  uns c = read(fd[i], xbuf, xbufsize);
+	  ASSERT(c == xbufsize);
+	  P_UPDATE(c);
 	}
       close(fd[i]);
     }
-  ms = get_timer();
-  log(L_INFO, "Spent %dms (%d MB/sec)", ms, (uns)(total/ms*1000/1048576));
+  P_FINAL;
 
+  for (uns i=0; i<files; i++)
+    unlink(name[i]);
+#ifdef COPY
+  unlink("tmp/ft-in");
+#endif
   log(L_INFO, "Done");
   return 0;
 }