🔒 Locked

Hybrid sleep in win7 - a farce?

Started by speciesUnknown Sep 29, 2010 at 7:45 PM 7 replies 2.4k views

Original Post

Sep 29, 2010 at 7:45 PM

It takes my laptop, when it was booted in win7, (a core 2 duo macbook pro, with 2gb ram, but with a really slow 4200rpm hdd) 5 minutes to fully recover from a hybrid sleep. This is totally unacceptable. It takes the same laptop 2 minutes to boot up afresh. So on balance, I'm better off booting afresh and having 3 minutes to open everything again.

It takes my desktop 2 minutes to wake from hybrid sleep, and around 1 minute to freshly boot up. As a form of comparison, OS X on the same laptop can use hybrid sleep and it is indistinguishable from normal sleep; I can only be certain its using hybrid sleep if I take the battery out while it is aleep so that I can guarantee that it has to use hybrid sleep. (Please, no stupid fanboy arguments, I'm looking for a practical solution to this issue.)

What seems to happen is that, even though it has one slow hdd, it tries to resume normal operation instantly, even though every single page has gone onto the HDD, and pages being paged out are clashing with pages being paged back in, and the whole thing is just a massive traffic jam.

So, is hybrid sleep a total farce? Have I exposed an edge case here, or is hybrid sleep so badly flawed that I may as well turn it off? Or is it an issue with the windows drivers for the "emulated" old fashioned bios? I thought 64bit windows 6.x kernels could use EFI bios. Perhaps this is as a result of improper bios emulation? But that doesn't make sense, since hybrid sleep is entirely a software process.

Is there some way I can make hybrid sleep actually work properly? This laptop goes into hybrid sleep even if the battery didn't die, and I can wake from normal sleep (battery powered sleep) in 10 seconds. I have disabled hybrid sleep, so if my laptop runs out of batteries while windows is booted, it just dies instead of being unusable for 5 minutes.

Warning: If any body tries to turn this into a fanboy argument I'll steal their wheels and prop their car up on bricks. Yes, I did this to somebody before (with help from 3 friends).

benryves

Sep 29, 2010 at 7:52 PM

I thought hybrid sleep was only offered on desktops? It's intended to be a combined hibernate plus standby, so it stores RAM to disk then enters standby mode. In theory this allows for a faster wake than a full hibernate, yet can still recover when power is lost. In practice I found that it would enter standby fine, but then take an incredibly long time to come back on (five minutes or so) or not come back on at all.

A simple fix for me was to open a command prompt with administrator rights and type powercfg /hibernate on and to go back to using old-fashioned hibernation instead.

speciesUnknown

Sep 29, 2010 at 8:04 PM

Quote:
Original post by benryves
I thought hybrid sleep was only offered on desktops? It's intended to be a combined hibernate plus standby, so it stores RAM to disk then enters standby mode. In theory this allows for a faster wake than a full hibernate, yet can still recover when power is lost. In practice I found that it would enter standby fine, but then take an incredibly long time to come back on (five minutes or so) or not come back on at all.

A simple fix for me was to open a command prompt with administrator rights and type powercfg /hibernate on and to go back to using old-fashioned hibernation instead.

Yeah, that's basically what I'm experiencing here, the 5 minute mark sounds very familiar. I know that windows knows its on a laptop and not a desktop because it has installed all the advanced mobility tools. I actually found the setting to turn off hybrid sleep buried in the advanced power settings.

frob

Sep 29, 2010 at 9:03 PM

Quote:
Original post by speciesUnknown
Have I exposed an edge case here, or is hybrid sleep so badly flawed that I may as well turn it off? Or is it an issue with the windows drivers for the "emulated" old fashioned bios? I thought 64bit windows 6.x kernels could use EFI bios. Perhaps this is as a result of improper bios emulation? But that doesn't make sense, since hybrid sleep is entirely a software process.

It is a hardware compatibility issue.

I've seen similar results on different computers around the office. We identified a few cards that caused problems with it. Our IT people hate it.

One dual-Ethernet machine had the Ethernet card responsible: The extra Ethernet card in place caused a hybrid sleep computer to wake up slowly or crash; pull the card from the machine and hybrid sleep works perfectly.

Another hybrid sleep machine had an older graphics card. Resume from hybrid sleep would take approximately forever. We moved it to a newer video card and it started working correctly.

Another had an USB card that gave an extra 8 slots on the back. With the card in place it would never fully sleep, just crash while storing the hibernate data. Pull the card and hybrid sleep works great.

Laptops are hit-and-miss. Either it works flawlessly or it must be disabled.

davepermen

Sep 30, 2010 at 2:25 AM

while, yeah, it's a matter of hardware, too.. your harddisk sure doesn't help.

most likely, the hw setup you have forces windows in loading back all the ram while doing random accesses across the disk. and both the actual load back + the random accesses are very, very slow on your hdd. i had such a hdd on a laptop of mine. i replaced it with an ssd, and i'm glad i did.

but it's a hit'n'miss thing, other than that, yes.

a macbook with a 4200rpm drive? wtf is that? if it's a 2.5" drive, replace it asap. it will gain you much joy having a much faster drive in. i'd get an ssd, but a nice 7200rpm hdd would be great in your case, too.

speciesUnknown

Sep 30, 2010 at 12:31 PM

Quote:
Original post by davepermen
while, yeah, it's a matter of hardware, too.. your harddisk sure doesn't help.

most likely, the hw setup you have forces windows in loading back all the ram while doing random accesses across the disk. and both the actual load back + the random accesses are very, very slow on your hdd. i had such a hdd on a laptop of mine. i replaced it with an ssd, and i'm glad i did.

but it's a hit'n'miss thing, other than that, yes.

a macbook with a 4200rpm drive? wtf is that? if it's a 2.5" drive, replace it asap. it will gain you much joy having a much faster drive in. i'd get an ssd, but a nice 7200rpm hdd would be great in your case, too.

Actually I was initially incorrect - the hdd is a 5400rpm hard disk. It's a 2008 model which I actually bought refurbished in 2009, because I wanted a battery I could take out (I prefer 2x6 hour batteries to 1x8 hour battery)

Upgrading the hard disk would be easy as its clearly accessible, but I don't think the hard disk is totally to blame for the 5 minute wakeup time. Due to seek times, an SSD would totally fix the problem, but I'm thinking there must be a windows configuration option which can stop it from using this; it seems to be flawed in principle to attempt to resume normal operation while a wakeup procedure is still running.

davepermen

Sep 30, 2010 at 12:39 PM

one thing you could try is to disable superfetch. no clue if that reloads all sorts of things into ram on wakeup. but as it's easy, costs nothing, it's sure worth a try.

Sneftel

Sep 30, 2010 at 12:52 PM

Works great for me. Hybrid sleep should come back within a couple of seconds if the RAM's still okay, and (like hibernation) shouldn't take too long even if it isn't. I kind of doubt that you're seeing page-file thrashing, at least of the sort you described: Hibernation doesn't use the page file. After restoring, all the same virtual memory is paged in.

One caveat: Firefox sometimes gets really, really moody upon reawakening on Windows, the sudden passage of time presumably getting it quite confused. I've had it freeze up and tie up my CPU for several minutes doing... things. That's occasionally made it faster to kill then restart Firefox immediately on unsuspend, if it's obviously doing that sort of thing.

Quote:
I can only be certain its using hybrid sleep if I take the battery out while it is aleep so that I can guarantee that it has to use hybrid sleep.

You'd be surprised, actually. If you only took the battery out for a couple of minutes, the RAM was probably still being refreshed from the capacitor banks.

Quote:
So, is hybrid sleep a total farce? Have I exposed an edge case here, or is hybrid sleep so badly flawed that I may as well turn it off? Or is it an issue with the windows drivers for the "emulated" old fashioned bios? I thought 64bit windows 6.x kernels could use EFI bios. Perhaps this is as a result of improper bios emulation? But that doesn't make sense, since hybrid sleep is entirely a software process.

I wouldn't be surprised at all if it was a BIOS issue. Hibernation still works through ACPI on EFI systems, and ACPI incompatibilities are still a problem.

davepermen

Sep 30, 2010 at 1:00 PM

Quote:
Original post by Sneftel
Works great for me. Hybrid sleep should come back within a couple of seconds if the RAM's still okay, and (like hibernation) shouldn't take too long even if it isn't. I kind of doubt that you're seeing page-file thrashing, at least of the sort you described: Hibernation doesn't use the page file. After restoring, all the same virtual memory is paged in.

while this should be true, i've noticed different behaviour in some setups. very massive harddrive trashing going on, which should not happen (it should be one sequencial read of insert-your-amount-of-ram gigabytes.

on a typical harddrive, this would mean around 10 secs per gigabyte. in his case, it could be slower. still, around 30 seconds for the restore, after that, it should be 100% available again.

BUT i've seen systems which never do that.

besides, to speed up, win7 only writes those pages that actually contain data.

and yeah, some apps have massive hickups afterwards. but on a system with issues, ALL the apps have hickups and are non-responsive.

Topic Locked

This topic has been locked by a moderator. New replies are not allowed.