| Title: | Windows NT |
| Notice: | See note 15.0 for HCL location |
| Moderator: | TARKIN::LIN .com::FOLEY |
| Created: | Thu Oct 31 1991 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 6086 |
| Total number of notes: | 31449 |
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 5590.1 | MPOS01::naiad.mpo.dec.com::mpos01::cerling | I'[email protected] | Thu Jan 23 1997 07:25 | 13 | |
| 5590.2 | METALX::SWANSON | Thu Jan 23 1997 09:28 | 12 | ||
| 5590.3 | Less data to transfer from disk ???? | BBPBV1::WALLACE | john wallace @ bbp. +44 860 675093 | Fri Jan 24 1997 02:10 | 7 |
Well, I've often wondered if there might sometimes be a small benefit
due to decreased transfer time (less actual I/O to do). But it seems
obvious that the price you pay is increased CPU usage for the same
work, because the decompression isn't free.
regards
john
| |||||
| 5590.4 | Esp. if compressed fits in disk cache & uncompressed doesn't | SMURF::PBECK | Paul Beck | Fri Jan 24 1997 11:02 | 7 |
On fast machines, I believe this is a real factor; under some
circumstances (large discontiguous transfers, especially on a
fragmented disk), compressed data can be processed faster than
uncompressed data, because the extra disk head travel takes much
more time than the extra CPU time to decompress. (e.g. if the
uncompressed data occupies more file fragments than the compressed
data would)
| |||||