And suddenly it works.

[TrashyMG]

First off as of now there is no heating issue. We know the SoC gets hot, but until a fully working prototype is put together we don't know how effective the current solution for heat management is.

 

[CrazyCapra]

It depends on the size of the reactor.

 

[God Ginrai]

First off as of now there is no heating issue. We know the SoC gets hot, but until a fully working prototype is put together we don't know how effective the current solution for heat management is.

It would be more accurate to claim it is a potential heating issue rather than denying that there is one.

-God Ginrai

 

[Klumpen]

It looks like underclocking the Pyra will be as common as overclocking the Pandora, giving people a bit more power than an overclocked Pandora with even longer battery life, doesn't seem too bad in combination with the keyboard backlight, more RAM, better case, bigger screen, better nubs, more compatible USB, HDMI out, etc...
Maybe the planned dock should have some cooling system integrated to run it full speed in stationary mode if there's no room for an efficient mobile cooling system.

From what I gathered, heat and automatic underclocking is the reason why most smartphones get noticably slower after a while of use.

Just don't grill our SD cards, I've no idea how heat resistent those are.

 

[EvilDragon]

Yes, I've just read up a few articles on modern SoCs (A15 and later) about thermal management.

Basically, thermal management is being done in three different ways:

1. Try to not produce much heat in the beginning
2. Try to move the heat out of the way as fast as possible
3. Throttle CPU to prevent overheating

Yes, CPU throttling is part of the normal heat management these days.
No smartphone will run at full power for a longer period of time. The maximum CPU power is only planned to be used for peaks.

About 1.:
According to an article I found, the OMAP does a very good job to prevent to produce heat, using TIs SmartReflex technology. If you enable it, the OMAP needs less power (and produces less heat) for the same processing power (due to some smart management).
Of course, that won't work if you run a stress test using both cores at full power, but it will have an effect in normal usage.

About 2.:
We're also doing this, though we need to find out if what we're doing is enough or not

About 3.:
We need to do the other tests first, but I guess some limit will need to be implemented. We cannot do some magical stuff that other manufacturers also can't do.
However, in normal usage, throttling should never be necessary, only when stressing it too much.

Additionally, I opened up a GPD XD to how they are doing it.
The SoC in the GPD can also heat up to 90C (so about the same reach).
I'll make some pictures, but basically, they're doing the same: Using copper layers to spread the heat. There are no heatsinks, no holes in the case, nothing.

 

[Askarus]

I'll make some pictures, but basically, they're doing the same: Using copper layers to spread the heat. There are no heatsinks, no holes in the case, nothing.

Don't forget Chinese magic. You won't see it...

 

[notaz]

According to an article I found, the OMAP does a very good job to prevent to produce heat, using TIs SmartReflex technology. If you enable it, the OMAP needs less power (and produces less heat) for the same processing power (due to some smart management).

But it needs a driver and there is no OMAP5 driver for it, unfortunately. Maybe it can be salvaged from some TI's vendor kernel, but that's work.

We need to do the other tests first, but I guess some limit will need to be implemented.

Yeah, it turns out even thermal shutdown patches have been rejected for mainline, so there is no protection of any kind right now...

 

[EvilDragon]

You mean a driver for SmartReflex or the thermal limit?

 

[Neelix]

However, in normal usage, throttling should never be necessary, only when stressing it too much.

The trouble with that is that what is normal usage for a smartphone won't necessarily be normal usage for a UMPC or gaming device, and throttling effectively robs you of power when you most likely need or want it the most. (or you wouldn't be stressing the CPU in the first place)

Here's hoping that the existing plans for heat management are adequate.

-Neelix

 

[oskda]

There's no magic SoC that doesn't produce heat when they're fully used

... Well, there is a company with SOC (form mobile handset) that doesn't produce heat: NINTENDO Its SOC have so low power/speed that no heat is produced (nothing you can feel even under continuous "stress"). Obviously the reason is that maximum frequency (MHz) is low and optimal for those SOC, so they don't waste energy in heat form. For example Nintendo 3DS has dual-core ARM11 at 266MHz And original Nintendo DS had ARM 9 at 66 MHz :O
[doublepost=1457444352,1457443699][/doublepost]

This Charging whitout booting feature sounds pretty cool..

I think that is the NORMAL and CORRECT WAY; the other way -when you charge it switches on automatically- is the new and wrong/incorrect way

I had a PDA a lot of years ago, and it didn't start-up automatically when charging, nor my Zaurus SL-C1000 and SL-C3000 did it, nor even my Android smartphones.... but iPhone and stupid WindowsPhone do it. Even with WindowsPhone, when you connect it to charge it start-up system, and if you turn it off, it start-up other time and son on Totally crazy and no-sense.
[doublepost=1457444714][/doublepost]

I realize it would be pretty unnecessary to put a microfan in a Pyra, but is there potentially enough space to make it fit in the case near the CPU? I'm more so interested in doing it as a just because "5% less heat at max load is something" project.

I think it isn't a good idea: fan noise and with time problems with dust/dirt.

I don't like notebooks with fan (even worse tablets with fan like Surface Pro) because they make noise, and with time they will have problems (and some machines are no designed to be serviceable). On a desktop PC there is little problem: you can open de door, clean fans or change them as they are big and easily reachable.

On other side, you can do a mod and include fixed fan or attachable fan, so you can cool your Pyra while docked, if you like/need so.
[doublepost=1457445523][/doublepost]

They would not build liquid cooling solutions into smartphones if there is no heat issue.
Pyra will be no exception there.

Really it isn't liquid cooling but heatpipes Heatpipes are better than liquid cooling, but not the same.
[doublepost=1457445837][/doublepost]

A lot of the 'old' games on Steam like the original X-COM actually ship with DOSBOX to emulate a DOS machine on the Windows environment.

You should take a look at Open XCom: http://openxcom.org/
They have made an exact XCom clone but without bugs, and running ok in most modern SO ( http://openxcom.org/downloads-milestones/ ), including Linux, of course (you need original game data, as it is copyrighted, but easily available on the net or from friends).

It seems it is ported to Pandora: http://repo.openpandora.org/?page=detail&app=openxcom

 

[EvilDragon]

I had a PDA a lot of years ago, and it didn't start-up automatically when charging, nor my Zaurus SL-C1000 and SL-C3000 did it, nor even my Android smartphones.... but iPhone and stupid WindowsPhone do it. Even with WindowsPhone, when you connect it to charge it start-up system, and if you turn it off, it start-up other time and son on Totally crazy and no-sense.

Android does that as well, however, they have a minimal kernel that doesn't start the full OS.
Usually, you can see a battery charging logo on screen. Once that one appears, you can be sure your smartphone has booted that minimal kernel and the CPU is running.

The Pyra doesn't even need the CPU to charge.

 

[benoitb]

... Well, there is a company with SOC (form mobile handset) that doesn't produce heat: NINTENDO Its SOC have so low power/speed that no heat is produced (nothing you can feel even under continuous "stress"). Obviously the reason is that maximum frequency (MHz) is low and optimal for those SOC, so they don't waste energy in heat form. For example Nintendo 3DS has dual-core ARM11 at 266MHz And original Nintendo DS had ARM 9 at 66 MHz :O

And then Nintendo couldn't get the original 3DS to emulate the SNES properly (their virtual console SNES games are New3DS only)

 

[oskda]

Samsung is using some special kind of liquid cooling mechanism not sure if that could be done on the Pyra.

No liquid cooling, only heatpipe (heatpipe is much better than liquid cooling).

Lot of places talk about liquid cooling in phones... incorrectly.
[doublepost=1457447819,1457446991][/doublepost]

Android does that as well, however, they have a minimal kernel that doesn't start the full OS.
Usually, you can see a battery charging logo on screen. Once that one appears, you can be sure your smartphone has booted that minimal kernel and the CPU is running.

The Pyra doesn't even need the CPU to charge.

You are right

Please, has Pyra the capability to start-up on timer/clock like some computers and some phones can do? (In past all Nokia phones I remenber could star-up on clok-time, but today WindowsPhone can't do it; and in Android only some phones like some ones with Mediatek and Hisilicon SOC can do it).
[doublepost=1457448501][/doublepost]One interesting experiment, and information source, would be to run Pyra/prototype in normal computer stress -for example compiling large source code or with a demanding emulator- continuously at room temperature (for example something like 20°C degrees), and then determine the speed/frequency it can keep continuously on time after Throttling-down.

In that way we can know the speed we can maintain under those situations (obviously, those change if you use one or two cores, or GPU/etc).

 

[fusion_power]

For example Nintendo 3DS has dual-core ARM11 at 266MHz And original Nintendo DS had ARM 9 at 66 MHz :O

Yupp, always impressive what can be pulled out of limited hardware if done right. For me, it seems as more powerful an Chip is, as more lazy are the coders to optimize their stuff. Which leads into more demanding apps and more heat producing high performance processors. I would not be surprised if there are Tetris, Pac Man or Pong versions out there that need a modern Multicore CPU to run at decent speed.

 

[levi]

And then Nintendo couldn't get the original 3DS to emulate the SNES properly (their virtual console SNES games are New3DS only)

Don't mix up what's technically possible with marketing. I know of people using flashcarts to run SNES games on their original (fat) DSes. Granted, you had to disable a layer or two in certain scenes on Super Metroid to maintain framerate, but the 2011 3DS is a generation faster than that hardware.

 

[Grench]

There is currently no known thermal issue. All of this is based on conjecture. Options presented are for consideration IFF such considerations are warranted.

Heat pipes are a phase change cooling solution - liquid to gas. Though not technically liquid cooling, they do involve liquid (and gas) in their systems. They're very nice - but very expensive to design and difficult to implement.

Stating that the project should use a heat pipe would imply that you have someplace in mind for the condenser/radiator to reside. When dreaming up a location for this ~80*C+ hot spot there are a few considerations to keep in mind. It should not be anywhere that the hands will touch during normal use. The SoC board buttons up to the bottom of the keyboard - there is no direct path to the top or the back hinge of the unit thick enough to allow space for the >= 1mm thick pipes.

It sounds like the space, though it may be 0.2mm or so, between the SoC card and the primary motherboard is going to get filled with some kind of electrically non conductive thermal transmission pad or compound - though that may still be up in the air.

If the system needs more heat removal, there are lots of mid-range solutions between nothing and heat pipes.

Normal polycarbonate is a good thermal insulator. There are polycarbonate blends/formulas that increase it's thermal transfer rate. It may be that a simple shift in case materials could suffice.

Passive transfer via thermally conductive metal. A thin layer of copper or silver, with an electrically non-conductive coating where appropriate, could be sandwiched between the SoC and keyboard then out the front just under the SD slots to 'fill in' the front face of the unit with a textured passive heat exchanger (radiator).

Build a bigger sink. The lower case plastics piece that the battery door clips into and forms the lower portion of the SD slots has a lot of material. If the material of that one piece where changed to something more thermally conductive (High TC plastic/PC, aluminum) it might significantly change the picture - that would be a BIG heat sink and a broad surface for dissipation. A hybrid case with one piece of Aluminum and the rest plastic wouldn't be 'as pretty', but could be 'cooler' .

Cooler for the SoC and other internals is directly related to hotter for the outside case materials. Either as a spot (radiator in thin layer conductor example) or broad areas (where the case or major component of it is the radiator). There is no magic 'heat eliminator'. Heat cannot be destroyed. Heat can be moved either by conduction or convection.

There is currently no known thermal issue. All of this is based on conjecture. Options presented are for consideration IFF such considerations are warranted.

 

[moxie]

"There is currently no known thermal issue. All of this is based on conjecture. Options presented are for consideration IFF such considerations are warranted."

Quoted FOR GREAT TRUTH.

Let's focus on problems that exists, eh?

 

[FBnil]

Please, has Pyra the capability to start-up on timer/clock like some computers and some phones can do?

Not that I am aware of, however, on the Pandora, you can put it in low power mode, while a script wakes it up into high powermode, and additionally runs something, like playing a wav file... I ran that many image-flashes ago... it was a decent alarm clock.

http://repo.openpandora.org/?page=detail&app=wakeupalarm.dimag0g

The script really does a sleep, which is very efficient, so the commandline implementation can be done each morning, and provided you do not reboot, which, well, happens too often, it remembers the wake up.

I do not see why these features would be missing on the Pyra... being Linux and all..

There is a Wake-on-Lan too (but horribly inefficient, as you need to keep power to the USB, so it is just slowing down the CPU to 300Mhzm and disable screen... I do not remember if I tested if could do it with less Mhz). Note: I used a USB to RJ45 dongle
[doublepost=1457471302,1457470573][/doublepost]

Let's focus on problems that exists, eh?

Yes: Where is the Yen, Euro, Pound key? (let's start a keyboard discussion....). Why isn't there a ñ ?

 

[erico]

Don't forget Chinese magic. You won't see it...

Oh, you mean Chinese black magic? Absolutely! And sorcerers and ghosts as well....

 

[rSl]

some time ago i saw a video with john carmack talking about 3d-engine development on a samsung smartphone for the occulus rift.
he said, that the phone would overheat in about 20 seconds when running his engine without throtteling. he stated that he
had to completely rethink his coding for mobile devices, so the engine keeps going inside the thermal budget without triggering
the soc to throttle/slow down too much.

for me personally the ability to run software at full speed for a long time on a machine without overheating
marks the line between a 'device' (smartphone, tablet) and a 'real computer' (desktop/handheld pc).
so here is hope we can get as much 'smartreflex' and hw-cooling as possible on the pyra to run software at full speed without
throtteling or superminimal emergency throtteling only.