Power Consumption

[paddy]

Thanks WizardStan ,that helped

 

[WizardStan]

http://www.open-pandora.org/ advertises the Pandora as being 600Mhz:

ARM® Cortex™-A8 600Mhz+ CPU running Linux

That's only guaranteed at OPP5. It can run at that, but uses more power always, even when scaled back. CPU scaling is automatic, voltage scaling requires manual intervention. So best is to set at OPP3, see what kind of clocking you can get out of that. If you need 600Mhz and your Pandora doesn't overclock at all, then you'll have to switch to OPP5 to get the advertised 600Mhz, which should still be guaranteed.
This is me speaking, OPT may have something different to say on the subject. Oh look, a can of worms!

 

[Fzero]

This is quality, I don't know is 'lite setting' just mweston's term, a setting available to him as a dev, or an actual setting in the menus? And if so what can you do in this lite mode?
12 hours is plenty plenty, the 10 hours for amiga is plenty enough for me, I'm guessing I will get a fair session when running megadrive or snes too as I will be on those more than amiga no doubt.
[I could search but if you posting anyway and you know]...wonder how long it takes to charge again from flat to fully juiced, I find the iPhone very fast myself, I swear I plug it for ten mins and it's boosted considerably

 

[aq21]

http://www.open-pandora.org/ advertises the Pandora as being 600Mhz:

ARM® Cortex™-A8 600Mhz+ CPU running Linux

That's only guaranteed at OPP5. It can run at that, but uses more power always, even when scaled back. CPU scaling is automatic, voltage scaling requires manual intervention.

No. The kernel manages frequency and voltage together, changing one with the other.

 

[WizardStan]

No. The kernel manages frequency and voltage together, changing one with the other.

Yes and no. The kernel needs to set something to change the voltage and set the *maximum* frequency. It doesn't need to do this together, it can change one without touching the other, but generally it has to. Once that max frequency is set, the CPU scaling is entirely automatic within the CPU itself. If the kernel sets OPP3 with max CPU of 500Mhz, the CPU will automatically scale down or up as needed, up to 500Mhz, with no intervention from the kernel. It will never change the voltage, however.
To achieve the advertised 600Mhz in a way guaranteed by Texas Instruments, the kernel needs to set to OPP5 and max CPU to 600Mhz. The CPU will then run at OPP5 even when the CPU scales back to its lowest setting, but it will at least reach 600Mhz in a way guaranteed to be stable.
Keep in mind that that's just the statements on paper. OPT has had success exceeding 600Mhz at OPP3, so it shouldn't really a problem. *knock ontouch wood*

 

[aq21]

There is not such thing "CPU scaling is entirely automatic within the CPU itself". At most the CPU can turn the clock off when idle. But when the kernel sets the clock to 600MHz, this is the only clock the CPU runs in. Not any lower clock. I don't know where you got your misinformation, but it is still misinformation.

 

[WizardStan]

There is not such thing "CPU scaling is entirely automatic within the CPU itself". At most the CPU can turn the clock off when idle. But when the kernel sets the clock to 600MHz, this is the only clock the CPU runs in. Not any lower clock. I don't know where you got your misinformation, but it is still misinformation.

I don't know what to tell you other than yes it does. It says so in the OMAP3530 datasheet. It's quite a read, if you're interested.
here's an example of how to activate it. It's for webOS, but should be basically the same for us.
You set the max (and min), and the CPU automatically goes faster or slower depending on how CPU usage.

edit:fixed url tags

 

[MWeston]

Unless I see the tests I asked for on youtube I'd just label this as "marketing speech" as it's an unrealistic case.

I don't do marketing. I was just showing that the fuel gauge was calibrated properly when compared to a piece of reliable test equipment so the time shown can be trusted. I ran a couple of Quakes since I still have the ammeter setup. Placing in code block so my spacing is maintained:

Quake 2 - Lowest LCD brightness (a bit too low for actual gaming without some gamma adjustments) - ranged from 350 - 410mA
        - Maximum LCD brightness -  peaked at 530mA

Quake 3 - Lowest LCD brightness (Q3 is brighter so actually playable here)  - 370mA peak (lower than Q2 but probably because my build is screwy and it draws through walls in some spots)
        - Maximum LCD brightness - 500mA peak, usually around 470mA

If you are on the go and you don't have the ability to recharge you won't browse the internet unless you use an USB dongle - that would most likely draw more power than the internal chip.

Why would you switch to a USB dongle when internal wifi works? I don't understand what you mean here. Also, from looking at ED's video, you can see that internal wifi will use a lot less power under typical usage which is great.

@MWeston:
If you switch to 700MHz, nothing will immediately happen.
AFAIK, Linux and the OMAP scales. You set the MAX speed, but as long as the CPU usage isn't too high, the CPU won't clock to 700MHz. At least that's what I think - but I might be in error there .)

No, that isn't quite accurate. Yes, it is possible but code that enables DFS and DVS started in 2.6.32 (I think) and so we don't have it until it is either back ported to .27 or the entire Pandora specific driver set is moved forward to the latest kernel. Then we can use it like the N900 does which will create even more battery time.

 

[aq21]

I don't mean to offend you, but you don't seem to understand what you read. The OMAP datasheet specifically explains how the software can change the frequency (see section 4.7.6 in the file called spuf98e.pdf). It never says that the CPU changes the clock by itself. If you think it does, please give exact location.

The second link explains how to enable Linux CPU frequency scaling, in which the Linux kernel changes frequency between defined minimum and maximum according to load. When using this system, with OMAP3 cpu frequency driver, voltage and frequency are changed by the driver together.

 

[WizardStan]

I don't mean to offend you, but you don't seem to understand what you read.

None taken. It wouldn't be the first time. I could have sworn I saw something about sending "idle" nops through the CPU to trigger it to automatically scale down, but now I don't remember where. I may have confused myself with something else entirely, though I've no idea what.

 

[matzesu]

Mey Mp3 Player is playing Musik about 22 Houers, so i think 20 Houars on the Pandora is a good thing

And at the evening i dont need the highest britgness from the lcd, so about 11 Houers of gamig is fantastik.. :lol:

 

[VRAndy]

I don't mean to offend you, but you don't seem to understand what you read. The OMAP datasheet specifically explains how the software can change the frequency (see section 4.7.6 in the file called spuf98e.pdf). It never says that the CPU changes the clock by itself. If you think it does, please give exact location.

I believe that he's describing the SmartReflex stuff. I can't tell if he's describing it accurately. I can't find a TI document that describes it in non-marketing terms.

 

[Monk]

http://www.open-pandora.org/ advertises the Pandora as being 600Mhz:

ARM® Cortex™-A8 600Mhz+ CPU running Linux

That's only guaranteed at OPP5. It can run at that, but uses more power always, even when scaled back. CPU scaling is automatic, voltage scaling requires manual intervention. So best is to set at OPP3, see what kind of clocking you can get out of that. If you need 600Mhz and your Pandora doesn't overclock at all, then you'll have to switch to OPP5 to get the advertised 600Mhz, which should still be guaranteed.
This is me speaking, OPT may have something different to say on the subject. Oh look, a can of worms!

I should open a bait shop, the number of cans full of worms you find around here. I'm just interjecting that the Pandora is advertised and sold as a 600MHz device - so as far as I and the sales literature for the final product are concerned, the "default" clock rate is 600MHz. If they're going to be shipping it at 500MHz... I'll just start tying some hooks now, shall I?

If you are on the go and you don't have the ability to recharge you won't browse the internet unless you use an USB dongle - that would most likely draw more power than the internal chip.

Why would you switch to a USB dongle when internal wifi works? I don't understand what you mean here. Also, from looking at ED's video, you can see that internal wifi will use a lot less power under typical usage which is great.

It may well be that WiFi Internet Access Points are cheap or free and widespread on your side of the pond, but over here in the UK (for example) when you are "on the go" (i.e. moving from place to place) the chances are excellent that you won't have access to a Wifi network. At home yes, easily, but out and about - possibly not. Certainly not without paying more than I do, for example, for 3G access. Do some would switch to a USB dongle, reluctantly, because WiFi doesn't work for them... and many times each Wifi broadcaster needs a silly login sequence, possibly a sign-up-and-login sequence that is a PITA to go through when you're just trying to follow a Google map. So it could be far, FAR easier to stick with one simple 3G ISP. This is why I'd have preferred a built-in 3G capabilty, or at least an internal USB port. :shrug.

 

[JayFoxRox]

If you are on the go and you don't have the ability to recharge you won't browse the internet unless you use an USB dongle - that would most likely draw more power than the internal chip.

Why would you switch to a USB dongle when internal wifi works? I don't understand what you mean here. Also, from looking at ED's video, you can see that internal wifi will use a lot less power under typical usage which is great.

Well you have shown the Pandora browsing the internet using its internal wifi. How often do you actually use the internal wifi for that? I'm in germany and hardly find any open wifi spots. I usually find them at places where I spend my time anyway (school, home, ..) - but these are the places were I also charge my Pandora, so battery usage is not a problem.
I also don't spend my entire time browsing at these places.

However, when on the go its likely that you use tethering or a UMTS / 3g dongle for example to keep browsing. And this is actually when battery life matters because these are the times where you don't have the option to recharge the battery.

I hope this makes it a bit clearer


Thanks a bunch for the values with Q2 / Q3!

 

[Gusarapo]

The Pandi looks every day better.

 

[MWeston]

However, when on the go its likely that you use tethering or a UMTS / 3g dongle for example to keep browsing. And this is actually when battery life matters because these are the times where you don't have the option to recharge the battery.

Ah, sorry, I don't know why it didn't occur to me that you meant 3G.

Unfortunately I don't have one of those dongles or a plan or I'd get you that number too. The 5V supply on the host port has been tested to be 91% efficient at 650mA (charging my Archos 605) and probably more like 93% efficient at smaller loads like 300-350mA (probably the most a 3G dongle would need?), but I'm just speculating and haven't really investigated it. From those numbers, you can get an idea of how battery life will be affected. I don't know if those USB 3G adapters use smart power management which would help it work even closer to an internal solution, power wise. Without any management, you could expect the battery life to be nearly cut in half.

 

[Dead1nside]

Thanks for the additional numbers MWeston, the original video was very valuable. You've pretty much covered all my usage scenarios.

 

[juniorm33]

I should open a bait shop, the number of cans full of worms you find around here. I'm just interjecting that the Pandora is advertised and sold as a 600MHz device - so as far as I and the sales literature for the final product are concerned, the "default" clock rate is 600MHz. If they're going to be shipping it at 500MHz... I'll just start tying some hooks now, shall I?

I've actually discussed this very issue myself.
Although OPT didn't reply, the replies I did get was that it's clever marketing which I still disagree with. But I'm pretty sure OPT guarantee it too 600Mhz at OPP3, but it's still besides the point as to what it's default speed is, and why it confuses people. So I'll just agree to disagree.

 

[jottt]

I should open a bait shop, the number of cans full of worms you find around here. I'm just interjecting that the Pandora is advertised and sold as a 600MHz device - so as far as I and the sales literature for the final product are concerned, the "default" clock rate is 600MHz. If they're going to be shipping it at 500MHz... I'll just start tying some hooks now, shall I?

I've actually discussed this very issue myself.
Although OPT didn't reply, the replies I did get was that it's clever marketing which I still disagree with. But I'm pretty sure OPT guarantee it too 600Mhz at OPP3, but it's still besides the point as to what it's default speed is, and why it confuses people. So I'll just agree to disagree.

It's actually 600+ wich reminds me of good old Athlon XP 3200+ ratings By the way...who canned all the poor little worms? I say free them!

 

[kuru]

So, basically, fuck thank you very much for using (recent?) coypyrighted music in the clip, so it's not friggin watchable in Germany.