iRex Iliad was not designed for Suspend/Hibernate

[TadW]

Originally Posted by Henry Loenwind

If I had to guess why they cannot do S3/"suspend to RAM" or even "suspend to flash", I'd say the hardware buttons have no way to wake up the CPU. So the CPU has to run to check if a button is pressed.

Sounds like a hardware design mistake.

[Mike Kostousov]

I think that problem with suspend could be unsolvable just if iRex made special hardware which wakes up cpu each 1milisecond. But then iRex needed special conclusion and effort for that.
I also read some papers about PXA255. It is cleary said that processor can have idle state and idle 33mhz state just by it self, without spesial hardware. As I understood, cpu can also have sleep mode, but there must be some hardware which awakes cpu. And, as I saw in /proc, there is no clock scale code for PXA255 in kernel. The good using of power scaling can consume at least half energy.
The question for scotty1024. Can you create some code in ipdf and idjvu which will just measure how much time it uses to render a page, and all cached pages? How much time isn't it doing anything?

[NatCh]

Originally Posted by jæd

Well... Then you're back to what "standby", "suspend" and "hibernate" actually mean...

I don't know ... if they put it out there with no qualification, then I think they intended us to assume that it meant what we basically assumed that it did mean.

Maybe it comes down to their B2B focus -- as others have commented, 10 hours is fine for most business applications -- especially compared to laptop/tablet battery life.

In any case, it's another indication that the iLiad and the Sony Reader aren't really directly competing products, they're simply aimed at different targets.


On another note, CommanderROR, I think you have generally conducted yourself in an extremely civil manner, even when you've been well and truly torqued. Sure you've expressed ... discontent on occasion, but you haven't actually gotten nasty about it that I've seen. For that I commend you.

[CommanderROR]

Thanks NatCh...:-)

I really was(am) pretty angry...staying civil isn't always easy then...

[segatang]

Hi,
As my knowledge, the Windows XP supports

* Conventional turn-off *
"The OS should close all data/processes" then the PC turns off all power of its peripherals.
When you press the power button of the PC again, it means you "restart(reboot)" the data/processes of OS from HDD to RAM from "zero condition".

It is like that a man lost his memory.
He should get his blank memory back from his dairy, etc (HDD) to his memory part of brain (RAM).

So, turn-off/restart(reboot) always cost a lot time about mins, you known.

* Hibernate mode *
"All recent data/processes in the RAM is written into HDD" for a while; then the power is entirely shut down like conventionally turning-off mode (note: no power consumed anymore after hibernate.)

It is like that a man sleeps all night. [/i]
His brain and his body full into deep sleep all night (no power consumed), but he can remember everything soon (about 30 seconds), when he wakes up (resume).

It just needs the time for all recent (actually, it became "previous", now) data/processes is written back from HDD to RAM. The OS is rebuild by the resent data/processes, not from the zero condition.

* Standby(suspend) mode:
Very low power is consumed to retain "all recent data in the RAM", but power of other peripherals such as monitor, HDD, fan, etc is stopped.

It is like that a man is just dozing/taking a nap.
Therefore, you can resume PC (wake up him) to do the previous work very quickly in about several seconds after other peripherals powered up.

Because the RAM continuously keeps the recent data/processes, theoretically the stanby/suspend mode will beresumed very very very fast, but time delay still exists due to delay of the peripherals.
*************************************

I believe the PPC or Palm's standby/suspend mode is as the same as above because PPC or Palm stiil consumes power after into the mode.
No reason why the same X-scale CPU based iLiad cannot make it.

[NatCh]

Originally Posted by CommanderROR

I really was(am) pretty angry...staying civil isn't always easy then...

Which is exactly why I consider it worth commenting on.

[Henry Loenwind]

Originally Posted by segatang

No reason why the same X-scale CPU based iLiad cannot make it.

You (and many others, I think) gut the wrong impression here. The iLiad CAN do all the modes you described. "Turn-off" is implemented. "Hibernate" is not, as the build in flash is to small, but nothing stops iRex or any third-party hacker from implementing it with a memory card. "Standby" is what the iLiad does between key presses, although it is not yet complete.

What the iLiad can not do is "Suspend to RAM", that is Standby + disabling the CPU, so the only components that would still consume power would be the RAM, the realtime clock, the memory controller and the keyboard controller. There simply are no memory or keyboard controllers (as I understand), so the CPU must do that work.

However, I'd guess the difference in power consumption between dedicated memory plus keyboard controllers and the CPU in 33Mhz mode is not very big.

[Malder1]

I'm sorry that I revealed such problem. I still have a hope that iRex will find some solution maybe pseudo-suspend mode. Or they at least will make automaticall loading of last opened PDF file.

[NatCh]

Originally Posted by Malder1

I'm sorry that I revealed such problem.

Hey, if they don't get revealed then there's less motivation for iRext to fix them!

Besides, it's better for folks to buy the device knowing its limitations than to be surprised by them after the fact, which is actually what folks are finding so annoying.

[kusmi]

What would be interesting to know, what the difference in power consumption between full suspend (to RAM, to FLASH) and just maximum power cycling (33MHz and slow RAM refresh). Because at the end, it might not be a big difference?

I'm asking, because after all, even if you would turn of your iLiad completely, the battery usually discharge itself over time, so perhaps there is not so much difference between this normal battery-discharge or the very minimal power-consumption during power-cycling, or what do the experts think :-) ?

[VillageReader]

Originally Posted by Henry Loenwind

You're quite right, just one little detail: A PC has multiple standby modes, S3 is (I think) "suspend to RAM", while S1 leaves the CPU running and only switches off some other hardware.

So what iRex is aiming for is "Standby S1", with the CPU still running but with a reduced frequency.

If I had to guess why they cannot do S3/"suspend to RAM" or even "suspend to flash", I'd say the hardware buttons have no way to wake up the CPU. So the CPU has to run to check if a button is pressed.

Well, I'm not a designer or hardware expert, but since they have the 'lock' button now, could that be used? Locking keys would go into some low power state (suspend, hibernate, whatever) and unlocking would wake the unit up.

[k2r]

I don't see a plausible reason yet why the iLiad shouldn't be capable of suspend-to-flash _hardwarewise_ .
If iLiad would provide us with complete access to the devices we own I was quite confident that there will be someone who figures it out.

[CommanderROR]

I think it must be something to do with the wakeup key...

And about the Power-Consumption comparison...I guess we really would have to get an anwer to that, but I'm afraid it won't be the same...
I can't imagine any CPU running at even 33Mhz that consumes so little power that the battery lasts for at least a few weeks (which is bout the times it takes for my PDA to run out of juice if I just leave it lying around without useage).

[k2r]

Originally Posted by CommanderROR

I think it must be something to do with the wakeup key...

Suspend to flash in my definition would mean that the state of the system (RAM and devices) is saved to flash and that the system is restored the next time it is started.
Of course this would mean using the power-switch to start up from suspend to flash and it would just mean faster "boot-up"-times.

[CommanderROR]

Suspend to flash should work, and it would already be very helpful...Suspend to RAM would of course be a lot better!