Date: Thu, 7 Aug 1997 23:00:22 -0600 (MDT) From: Atipa <freebsd@atipa.com> To: "John S. Dyson" <toor@dyson.iquest.net> Cc: dyson@FreeBSD.ORG, ggm@connect.com.au, current@FreeBSD.ORG Subject: Re: IDE vs SCSI was: flags 80ff works (like anybody doubted it) Message-ID: <Pine.BSF.3.91.970807225059.4717A-100000@dot.ishiboo.com> In-Reply-To: <199708080427.XAA02012@dyson.iquest.net>
next in thread | previous in thread | raw e-mail | index | archive | help
My thoughts exactly. Depending on the card and driver level, SCSI is
actually slower in many cases. Similar IDE drives outperformed SCSIs
unsing a 2940UW w/ 2.1.5-RELEASE. 2.2 kernels and above have noticably
faster Adaptec drivers. Haven't played much with the NCR, but I
think I heard the ncr.c does not support ultra or ultra-wide. True?
I definitely recommend IDE for workstations and medium demand servers.
However, SCSI will still have its market. When expandability and performance
are key, SCSI is the only solution (for now!). Maybe fibre-channel will
start to get more popular...
Kevin
On Thu, 7 Aug 1997, John S. Dyson wrote:
> >
> > The main advantage with SCSI is the speed of the drive; not the
> > interface. Typically SCSI drives are 5400RPM at worst, and IDEs are
> > 5400RPM at best. I do not see IDE drives getting those 7-8ms seek times
> > either. Good seeks for IDE are 10ms, roughly 30% slower than SCSI.
> >
> > Doing a very crude benchmark:
> > $ time dd if=/dev/zero of=/tmp/file1 bs=1024 count=20000
> > $ time dd if=/tmp/file1 of=/dev/null bs=1024 count=20000
> > $ time dd if=/tmp/file1 of=/tmp/file2 bs=1024 count=20000
> >
> I certainly don't disagree that SCSI drives are generally better. However,
> IDE's are very very inexpensive. Here is an interesting benchmark:
>
> This is the result of my "slow" recent, but not the latest, greatest
> IDE drive (WD 4GB drive.):
>
> Command overhead is 88 usec (time_4096 = 348, time_8192 = 607)
> transfer speed is 1.57828e+07 bytes/sec
>
> dd if=/dev/rwd1 of=/dev/null count=1600 bs=64k
> 1600+0 records in
> 1600+0 records out
> 104857600 bytes transferred in 10.881267 secs (9636525 bytes/sec)
>
> This is the result of my Hawk, SCSI drive, with an NCR 815 interface:
>
> Command overhead is 845 usec (time_4096 = 2071, time_8192 = 3297)
> transfer speed is 3.34201e+06 bytes/sec
>
> dd if=/dev/rsd0 of=/dev/null count=1600 bs=64k
> 1600+0 records in
> 1600+0 records out
> 104857600 bytes transferred in 27.336979 secs (3835742 bytes/sec)
>
> Looks like the command overhead of the IDE drive is very low. In fact,
> I doubt that an Atlas-II or a WD Enterprise with an AHA2940 will do much
> better than 200usec. Sure, tagged command queuing, etc will make SCSI under load
> out-perform an IDE system -- however, 4GB for about $300 is very tempting. A good
> 4GB SCSI drive that has the 30% higher performance that you suggest would cost
> at least $600, right? A WD Enterprise, a Seagate Barracuda, or Quantum Atlas-II
> costs at least that...
>
>
> /*
> * I think that BDE wrote this:
> */
> #include <sys/types.h>
> #include <sys/time.h>
> #include <fcntl.h>
> #include <stdio.h>
> #include <unistd.h>
>
> #define ITERATIONS 1000
>
> static int syserror(const char *where);
> static long timeit(int fd, char *buf, unsigned blocksize);
>
> int main(int argc, char **argv)
> {
> char buf[2 * 4096];
> int fd;
> long time_4096;
> long time_8192;
>
> if (argc != 2)
> {
> fprintf(stderr, "usage: %s device\n", argv[0]);
> exit(1);
> }
> fd = open(argv[1], O_RDONLY);
> if (fd == -1)
> syserror("open");
> time_4096 = timeit(fd, buf, 4096);
> time_8192 = timeit(fd, buf, 8192);
> printf("Command overhead is %ld usec (time_4096 = %ld, time_8192 = %ld)\n",
> (time_4096 - (time_8192 - time_4096)) / ITERATIONS,
> time_4096 / ITERATIONS, time_8192 / ITERATIONS);
> printf("transfer speed is %g bytes/sec\n",
> 4096 * ITERATIONS * 1000000.0 / (time_8192 - time_4096));
> exit(0);
> }
>
> static int syserror(const char *where)
> {
> perror(where);
> exit(1);
> }
>
> static long timeit(int fd, char *buf, unsigned blocksize)
> {
> struct timeval finish;
> int i;
> struct timeval start;
>
> if (read(fd, buf, blocksize) != blocksize)
> syserror("read");
> if (gettimeofday(&start, (struct timezone *)NULL) != 0)
> syserror("gettimeofday(start)");
> for (i = 0; i < ITERATIONS; ++i)
> {
> if (lseek(fd, (off_t)0, SEEK_SET) == -1)
> syserror("lseek");
> if (read(fd, buf, blocksize) != blocksize)
> syserror("read");
> }
> if (gettimeofday(&finish, (struct timezone *)NULL) != 0)
> syserror("gettimeofday(finish)");
> return (finish.tv_sec - start.tv_sec) * 1000000
> + finish.tv_usec - start.tv_usec;
> }
>
>
>
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.91.970807225059.4717A-100000>