How much of an affect will rpm speed have on a system?

[chipbgt]

My Hdd is a 7200 rpm, does this affect speed much over 5400? Is it worth the extra cash?

[Mntsnow]

Truthfully it isnt quite what it's all cracked up to be...Just like the ATA66 versus ATA33. Yes you will see "some" improvement but not much! If you really want to see improvement you need to get drives that have a faster ACCESS time. I would rather have a ATA33 or ATA66 drives that had faster access times than a drive that had lower access times but ran at 7200 or 10,000.... Also remember that the faster the speed the greater the heat!

Mntsnow

[chipbgt]

Hmm...see I thought RPM would affect the access time....so its could be possible to have a 5400 drive with a lower access time than a 7200?

and yeah Scsi would be nice but I am not ready to step into that yet..maybe in my next system http://sysopt.earthweb.com/forum/smile.gif Thanks for the response.

[Underclocked]

Sorry guys, trying to link to some reviews at StorageReview showing some pretty significant differences (goofed up the links). The new IBM 75GXP sold me (had to order one http://sysopt.earthweb.com/forum/wink.gif )

[This message has been edited by Underclocked (edited 06-05-2000).]

[DEMENTEDEVIL]

Man its a bunch of bull, not that much of a performance gain between 7200 and 5400, like doing 55mph and 75mph...both ain't gonna give you no rush. unless your system is hooked up with the latest driver and all....save your money and get some ice-cream! :-)

[BBA]

I beg to differ there is a difference.

The performance of an older 7200RPM drive was around 14M/sec 9-10msec access time. The newer ones are 7-9msec and 19-25M.sec.

The performance of 5400RPM drives is around 2-5M/sec with 10-15msec access time.

If you ever loaded Q2 or Q3 in a lan environment, you know what that means between switching maps...major advantage in faster loading times.

Personally, I will never buy or recommend any 5400RPM drive again.

[ctaylor]

I would give you a typical response....DEPENDS UPON WHAT YOU WANT TO DO.

If you are using this machine for file storage, office, or home use I would think that 5400 RPM would be fine.

If you want to run multimedia presentations, super graphics applications, or anything else where big amounts of data should be transferred fast for a smooth visual display, then a faster speed would be fine.

For my purposes I am willing to wait an extra 1/2 second for the drive to get up to speed and send me my data.

SIDE COMMENT OR BEFUDDLED RAMBLING: Also, you can improve your read time by choosing a faster file system. FAT16 and FAT32 writes data in clusters to a hard drive. At the end of each cluster there is a marker indicating which cluster is used to continue the file. The FAT system cannot allow relative spacing (I.E. moving 3 sectors forward, or 12 sectors backwards) to read the file. FAT16 and FAT 32 file systems require that the hard drive start at zero and manually count out each sector until it finds the one it wants to continue reading the file - which may be several a large number of sectors into the drive, but only a few sectors away relative to the previous sector. You lose performance each time the hard drive controller has to "count."

[BBA]

So what file system do you recommend?

NTFS is way slower than FAT is...I did performance testing for my company comparing fat, fat32 and NTFS. Fat 16 gave the overall fastest performance, but it's limits suck. FAT32 is the next best, and NTFS was the slowest.

Maybe a NON-MS file system is better?

[ctaylor]

I don't have much direct experience with file systems as Ted Ts'o (Kerberos ecncription software/file systems developer for VA Linux) taught me most of what I know last week.

Ted likes the ext2 and ext3 (*nix) file systems (but he also wrote huge chunks of the O/S loaders for each of these file systems so he is most likely biased). Both of these use relational indexing to find clusters which should reduce the seek time.

Ted's opinion of M$ file systems was basically "they work." NTFS is really a reinvention of the Unix based "inode" method of indexing clusters on a drive.

Wish I had something else I could comment on the subject. *SHOULDER SHRUG*



[This message has been edited by ctaylor (edited 06-06-2000).]