Some benchmarks using swsusp2:
I wanted to compare compressed and uncompressed suspend/resume times. So, I made some benchmarks. Feel free to time your machine as well, and add your info!
Hardware: 1.3GHz Pentium M @ 600MHz, 512MB DDR, 4200RPM IDE hdd, 800MB swap partition; swsusp2-2.1.8.2 running on 2.6.11
|
|
No LZF compression |
LZF enabled |
RAM: 505, Swap: 150M (used) |
Suspend: |
35s |
|
|
Resume: |
36s (7 kernel + 29 swsusp) |
|
RAM: 508, Swap: 60M (used) |
Suspend: |
|
26s |
|
Resume: |
|
23s (7 kernel + 16 swsusp) |
Hardware: Athlon-XP 2600, 512MB DDR, 7200RPM IDE hdd, 1G swap partition; swsusp2-2.2.5 running on 2.6.16 (Kraymer)
|
|
No LZF compression |
LZF enabled |
RAM: 335 (used), Swap: 394M (used) |
Suspend: |
28s |
|
|
Resume: |
31s (10 kernel + 21 swsusp) |
|
RAM: 335 (used), Swap: 394M (used) |
Suspend: |
|
21s |
|
Resume: |
|
25s (10 kernel + 15 swsusp) |
How did I time this: Suspend was measured from the suspend text ui screen kicking in until the machine turned off/rebooted. Resume was measured from after the boot loader until until swsusp starts resuming, and then until the first signs of X can be seen. I ran these several times actually, but after finding out that the times didn't vary too much (less than human error caused by me and my stop watch, I think), I decided that this is exact enough. No fancy statistics.
Points Regarding Timing
Not enabling the swap partition's hard disk DMA will dramatically slow down Suspend2. Suspend/resume times will span a couple of minutes.
User interface progress reporting (FBSplash and even the text UI a little) will slow down Suspend. UI's are pretty, but maybe your time is really precious.