| 1 |
On 10 Sep 2009, at 09:30, Alan McKinnon wrote: |
| 2 |
> ... |
| 3 |
> But I doubt the wisdom of updating an SSD netbook on the machine |
| 4 |
> itself: |
| 5 |
> |
| 6 |
> 1. Wear on the SSD itself with all those compiles |
| 7 |
> ... |
| 8 |
|
| 9 |
No harm in compiling on a hard-drive, via NFs or otherwise. I believe |
| 10 |
read speed of SSDs is fast, writes are slow. |
| 11 |
|
| 12 |
However, I am sceptical of wear claims, at least of you're using ½- |
| 13 |
decent flash memory (and SanDisk & Kingston are cheap these days, at |
| 14 |
least in "modest" but usable sizes like 4gig). |
| 15 |
|
| 16 |
I have read many people talk about wear of flash memory to be a |
| 17 |
problem, but I don't think from anyone who's actually HAD a problem |
| 18 |
with it. I have read of many people using it happily for root |
| 19 |
filesystems over periods of years. |
| 20 |
|
| 21 |
I concede that syncing the portage tree & the compilation of emerging |
| 22 |
packages results in an above-average number of writes, but I have this |
| 23 |
notion that the wear / limited writes problems have been largely |
| 24 |
overcome with modern flash memory (c.f. "write levelling"). |
| 25 |
Furthermore I have heard figures bandied about in the order of |
| 26 |
100,000s per block and such as "you'd need to write to the flash card |
| 27 |
constantly for years" in order to kill it. |
| 28 |
|
| 29 |
I would really love to hear empirical evidence either way on this |
| 30 |
matter, but I don't think the OP needs to be too cautious of wear. (Of |
| 31 |
course this advice is worth what he paid for it, and warrantied to |
| 32 |
that value). |
| 33 |
|
| 34 |
Some previous comments: |
| 35 |
http://archives.gentoo.org/gentoo-user/msg_d6e65b4d64a51c97f7c43c723e525e06.xml |
| 36 |
|
| 37 |
Stroller. |
Archives