Overview of SoC options

[thatgui]

I don't think finding space for one part is "working wonders". If I were wanting him to add a laundry list of things, then I'd totally agree with you.

Well time to fight over discuss the definition of "laundry list" then :

- two additonal shoulder buttons, they may not be big, but require a big "keep out" area to get them mechanically stable

- 3G Chip + connectors for Antennas

- RAM

- MicroHDMI out

- a big bunch of additional wires to connect that all

For me that is already a rather big list in terms of board design, your milage may vary of course,

I was under the impression that there was already stuff under the battery on the Pandora PCB. (Just not exposed) Am I wrong? If I'm not, then how would something being under the battery on the Pyra PCB increase its thickness beyond what the parts under the Pandora battery did?

Yes there are parts under there, but with the "stacked" solution you will have something that is a lot larger (height wise) than any of the other parts: thickness of the SD-Card, thickness of the Sim-Card, thickness of the mechanical parts to keep everything together.


As the battery compartment of the Pandora is a little bigger than the battery, maybe the height of the compartment could be made smaller and the battery contacts could be "raised" a little from the pcb surface to keep good contact in order to get more room between pcb surface and battery compartment - so Grenchs solution may fit after all > dunno.


Edit: Sorry, hit the wrong button, reply to the first "paragraph" expanded

 

[Neelix]

No, I don't doubt the new guy in general. I doubt that he will do a much(!) better job than MWeston did. I think it is quite possible that he will create a more space efficient layout. He has already experience in designing at least two phone pcbs, which gives him the credit of bringing up workable designs within tight spaces, and also the "judgement" from the outside, that he is doing a fine job (otherwise he may not have had a chance to design the neo900 pcb or Ed choosing him for the Pyra). I just doubt that he can work wonders.

For some reason  when reading that I can't help thinking of this:

Scotty: Do you mind a little advice? Starfleet captains are like children. They want everything right now and they want it their way. But the secret is to give them only what they need, not what they want.


Lt. Commander Geordi La Forge: Yeah, well, I told the Captain I'd have this analysis done in an hour.


Scotty: How long will it really take?


Lt. Commander Geordi La Forge: An hour!


Scotty: Oh, you didn't tell him how long it would *really* take, did ya?


Lt. Commander Geordi La Forge: Well, of course I did.


Scotty: Oh, laddie. You've got a lot to learn if you want people to think of you as a miracle worker.

(Scotty always multiplied his time estimates by 4)

- Neelix

 

[God Ginrai]

- two additonal shoulder buttons, they may not be big, but require a big "keep out" area to get them mechanically stable

These would not take any additional room on the PCB. They are being stacked above the current shoulders.

- 3G Chip + connectors for Antennas

- RAM

Viable.

- MicroHDMI out

This was already accounted for in our initial discussions about EvilDragon saying that the back of the device was already full as it is.

- a big bunch of additional wires to connect that all

 You can normally fit traces in the gaps where you can't fit other things.

At this point, it seems like the 3G Chip and RAM are the only things that had not been accounted for. 2 things barely even qualifies as a list.

Scotty: Do you mind a little advice? Starfleet captains are like children. They want everything right now and they want it their way. But the secret is to give them only what they need, not what they want.

Lt. Commander Geordi La Forge: Yeah, well, I told the Captain I'd have this analysis done in an hour.

Scotty: How long will it really take?

Lt. Commander Geordi La Forge: An hour!

Scotty: Oh, you didn't tell him how long it would *really* take, did ya?

Lt. Commander Geordi La Forge: Well, of course I did.

Scotty: Oh, laddie. You've got a lot to learn if you want people to think of you as a miracle worker.

I don't really like Star Trek, but that's a funny quote.

-God Ginrai

 

[thatgui]

These would not take any additional room on the PCB. They are being stacked above the current shoulders.

Oh, did not know that the final decision has already been made about that, if so the additional needed space would be minimal indeed.

This was already accounted for in our initial discussions about EvilDragon saying that the back of the device was already full as it is.

I know,I put on the list to emphasise the last point (traces) => 19 traces (if the micro hdmi port is fully connected) vs. 2 traces (barrel jack) + 2 (or 3?) traces for video out on the ext port

You can normally fit traces in the gaps where you can't fit other things.

Thats why I don't like the tetris analogy, traces can't be just put anywhere, they need two endpoints and often have other limiting factors regarding placement and length.As far as I know the Pandoras pcb is already a multi layer board, so there are not necessarily enough gaps to fit everything in (I know this does not automatically strengthen my point regarding traces).

Scotty: ...

This already made it in a shortened and overexaggerated form into popular culture around here - even people that are not into SciFi cite it now and then (coincidentally a colleague of mine just a few hours before your post):

Scotty: Captain, I will need at least four weeks to do that

Kirk: Scotty, you have four hours

Scotty: Ok, for you captain, I'll make it in two

 

[Klumpen]

Scotty: Captain, I will need at least four weeks to do that


Kirk: Scotty, you have four hours


Scotty: Ok, for you captain, I'll make it in two

I think that the first number was already larger than the second one.

 

[God Ginrai]

You can normally fit traces in the gaps where you can't fit other things.

Thats why I don't like the tetris analogy, traces can't be just put anywhere, they need two endpoints and often have other limiting factors regarding placement and length.As far as I know the Pandoras pcb is already a multi layer board, so there are not necessarily enough gaps to fit everything in (I know this does not automatically strengthen my point regarding traces).

This is my fault, I guess I was not clear enough. I wasn't suggesting that we just drop traces wherever there was a hole.

To my understanding, you would not place parts right up against each other so that heat can properly dissipate. Then, the gaps you left between the parts would be ideal for traces, wouldn't they? And where multiple different traces needed to cross in a similar area, you could use the gaps created from parts not fitting together exactly. This is what I was talking about.

EDIT: Just noticed the formatting blunder I made and have fixed it.

-God Ginrai

 

[thatgui]

This is my fault, I guess I was not clear enough. I wasn't suggesting that we just drop traces wherever there was a hole.

To my understanding, you would not place parts right up against each other so that heat can properly dissipate. Then, the gaps you left between the parts would be ideal for traces, wouldn't they? And where multiple different traces needed to cross in a similar area, you could use the gaps created from parts not fitting together exactly. This is what I was talking about.

I thought that I did understand you clearly there, but now I'm not so sure anymore: Board size and part footprint are fixed and the problematic parts are probably already seperated due to lenght and signal runtime optimizations. So where will the "win" in space come from ?

 

[God Ginrai]

This is my fault, I guess I was not clear enough. I wasn't suggesting that we just drop traces wherever there was a hole.

To my understanding, you would not place parts right up against each other so that heat can properly dissipate. Then, the gaps you left between the parts would be ideal for traces, wouldn't they? And where multiple different traces needed to cross in a similar area, you could use the gaps created from parts not fitting together exactly. This is what I was talking about.

I thought that I did understand you clearly there, but now I'm not so sure anymore: Board size and part footprint are fixed and the problematic parts are probably already seperated due to lenght and signal runtime optimizations. So where will the "win" in space come from ?

Basically, arrange the parts in such a way that the traces can all travel through the same gaps. Like at a rail yard, tons of different tracks all go to the same area.

-God Ginrai

 

[thatgui]

Basically, arrange the parts in such a way that the traces can all travel through the same gaps. Like at a rail yard, tons of different tracks all go to the same area.

Yeah, but whats the difference on how is it done anyway ? If there are gaps between components they are used regardless of the reasons why these gaps exist (heat dissipation, dictated by design choices, dictated my tracing optimizations). I'm sorry it looks like I'm currently hiting a wall trying to follow your thoughts, I'll guess I have to sleep over it.

 

[Linux-SWAT]

_wb_, seems that OMAP5432 uses DDR3, not DDR2 like 5430:

http://www.ti.com/lsds/ti/omap-applications-processors/omap-5-processors-products.page

I'll also add that x86 is a pretty bad architecture, that's why when you know something else than biased benchmarks on "reputed testing websites" you don't even consider x86 as an option.

Something ironic about that, the execrable M$ was unable to put together a windows ARM (as i predicted), so will stick with an execrable architecture.

Fortunately for both, trolls exist to remind us they exist.

 

[God Ginrai]

_wb_, seems that OMAP5432 uses DDR3, not DDR2 like 5430:

http://www.ti.com/lsds/ti/omap-applications-processors/omap-5-processors-products.page

I'll also add that x86 is a pretty bad architecture, that's why when you know something else than biased benchmarks on "reputed testing websites" you don't even consider x86 as an option.

Something ironic about that, the execrable M$ was unable to put together a windows ARM (as i predicted), so will stick with an execrable architecture.

Fortunately for both, trolls exist to remind us they exist.

Actually, from what I've heard, the only reason Windows RT is doing bad is due to lack of apps. (Hey, sounds like the Wii U!)

-God Ginrai

 

[Linux-SWAT]

There's plenty of reasons. To the incompetency of m$ guys, you can add the no drivers problem, the no interest from pple, etc...

 

[Exophase]

_wb_, seems that OMAP5432 uses DDR3, not DDR2 like 5430:

http://www.ti.com/lsds/ti/omap-applications-processors/omap-5-processors-products.page

No, LPDDR2 is not DDR2. It'd be better if Pyra had LPDDR2 than DDR3.

I'll also add that x86 is a pretty bad architecture, that's why when you know something else than biased benchmarks on "reputed testing websites" you don't even consider x86 as an option.

Are we really doing this again? The people who know the most in this field (including CPU architects who have designed many different kinds of CPUs) don't agree with you.

 

[Linux-SWAT]

Well, my architecture teacher was a ~60 years-old standalone museum and said it sucks.

So far he proven right on other topics too.

 

[slaeshjag]

As transistor count has exploded, the part of the x86 CPU that translates x86 instructions into more RISC-like ones is a rather small part of the CPU.

That x86 is a utterly horrible arch to work with if you have to do any kind of assembler, is mostly irrelevant today, as only the select few "sacrifices" have to spend their mental health on it.

 

[Exophase]

Well, my architecture teacher was a ~60 years-old standalone museum and said it sucks.

So far he proven right on other topics too.

And maybe what he said was right 30 years ago, but today it's wrong. I don't know when you got your education so I don't know if it's you or him that are repeating obsolete information.

 

[Linux-SWAT]

Or you.

 

[Exophase]

Linux-SWAT, until you make any kind of technical argument on this matter, like ones I've given several times now, don't bother. Deferring to the authority of some unknown professor and responding with "or YOU could be wrong" doesn't cut it.

Believe it or not, professors don't always keep up with the industry. I remember my scientific computing professor in grad school, about 7-8 years ago. Very skilled guy who was doing a lot of real work for national labs and part of other research programs. Once in class he asked how many clock cycles transcendental floating point instructions like fsin take on modern processors. I answered, dozens of cycles, he said no, actually just 2-3 cycles, make me look silly. But this wouldn't be true on anything resembling a modern CPU at that time. He then went on to say that all high-performance processors that work on vectors actually just pipeline them, they don't execute the operations in parallel. It was apparent that he was still thinking that HPC was done on old vector computers like Crays, which was way out of date.

 

[Linux-SWAT]

Having worked in the probably worst french university (Paris 8 Vincennes St Denis), i agree that some teachers are totally disconnected with the industry reality.

Fortunately, i never took courses there, and even if i can't give technical details because i'm spectrum-wide, not specialized, i'm convinced about what i'm saying.

I collect informations here and there and make my own opinion.

Years ago, when pdf was supposed to be malware-proof, i was skeptical. Same for mobile phone spying, i was told i'm paranoid. I also had a bad feeling about flash.

When i said windows arm will not happen (well, it was just a spark), i even know someone who said i was wrong .

 

[Exophase]

Having worked in the probably worst french university (Paris 8 Vincennes St Denis), i agree that some teachers are totally disconnected with the industry reality.

Fortunately, i never took courses there, and even if i can't give technical details because i'm spectrum-wide, not specialized, i'm convinced about what i'm saying.

I collect informations here and there and make my own opinion.

And when people come along who can give details that challenge your preconceived notions you're not even open to them, you just deny it. Even worse is that you probably formed your staunch opinions on this a very long time ago, so you're not open to things changing. The semiconductor industry is nothing if not in a constant state of change.

Years ago, when pdf was supposed to be malware-proof, i was skeptical. Same for mobile phone spying, i was told i'm paranoid. I also had a bad feeling about flash.

So what do we take from this? You're an oracle whose intuition should be valued over a well reasoned argument? We should trust you know because you've been right about something or other in the past?

When i said windows arm will not happen (well, it was just a spark), i even know someone who said i was wrong .

But you were wrong. Unless "will not happen" was supposed to be subject to something other than the obvious literal interpretation. In that case, yes, it's easy to always be right when you can redefine terms as you please after the fact.