SoC: Back and forth!

[SNESFAN]

I meant WizardStan was against change

Lies and slander. I love change, I'm against bad arguments.

Pretty much the same point, just greatly expanded

Not the same point. Your entire argument, the one and only thing you said, was that the OMAP5 was weaker. After I pointed out the fault in just using that as an argument you effectively re-iterated the same flawed argument. SNESFAN saw the flaw in the argument and made additional points that make the A80 specifically worth waiting for, not just because it is better but because of features it and it alone has.

Yes, I wouldn't even go as far as to say the omap is weaker, it might be, but that's not really the point I'm making as Stan is pointing out. My argument is at lower processing power levels the a80 is going to literally crush the omap in energy consumption and heat production.

I strongly feel its going to be night and day difference. Like the difference between a normal use runtime of double on the same battery. Or setting up governors that will allow for epicly long use cases. A quad a7 android tab with 8 inch screen and 6000mah battery getting 16 hours of heavy use. Epically good, vs trying to squeek out 7 hours by running low intensive tasks. I have a feeling that the numbers that ED finds with the omap are going to be more towards the latter than the former.

One of the first things that the Pandora was known for is its best in class battery life. Almost above all else, that's its strongest feature to me.In a mobile PC there are long spans of time doing insignificant amounts of processing, this will be highly beneficial. To me, our use case and the a80 hardware are made for each other. Software wise might be up for debate.

 

[slaeshjag]

The A80 is probably a lot better from a performance and battery perspective. However, with the Pandora being EOL now, and going for A80 would delay the pyra by at least 6 months and cause an unideal driver situation, it would be suicide to wait for the A80.

 

[Natsu]

Also, I want to point out, if the a80 is release soon after then this all kind of moot. Would bother early adopters and devs,tho.

I love change

Not the impression I got when you were shooting down GG for even discussing another button layout, but I digressed.

Not the same point. Your entire argument, the one and only thing you said, was that the OMAP5 was weaker. After I pointed out the fault in just using that as an argument you effectively re-iterated the same flawed argument. SNESFAN saw the flaw in the argument and made additional points that make the A80 specifically worth waiting for, not just because it is better but because of features it and it alone has.

In my mind, I was including power and features in one package, especially when we already compare the A80 to OMAP5 from past discussions. Stronger tech tend to mean better features as well.
Admittedly, I scanned (I will go back), but it's because being weaker is part of reason that we may see less devs support and less customers trying make that decision between this and other devices that the OMAP5 is nowhere near.

At least we can used everything of the A80 and are closer to recent tech. This isn't about getting the best tech around,

it's about not being too far behind.

We may not have the exactly the same reason, but we are coming to the same conclusion that the OMAP5 is so far, may even too far, behind rather that be in power or features.

800 bucks is my top bid, anything more than that, it would be a tough purchase.

Then wouldn't a stronger SOC (at the same price) be ideal to make the purchase easier?

 

[WizardStan]

Not the impression I got when you were shooting down GG for even discussing another button layout, but I digressed

Encouraging and making suggestions is shooting down? I agreed that six buttons was good, I even agreed that one of his layouts would work well if it could be made to work: I had my needs, he had his, and he found a compromise. When askarus tried it and found that it couldn't be made to work comfortably in the space provided that's when I stopped encouraging his button layout. What should I have done to not "hate change" then?edit: also to remind you, it's not like it's my decision anyway. Any shooting down I did was because the layout wasn't my personal preference, and I always tried to give reasons why something wasn't the way I liked it.

In my mind,

Which no one can read. If you want to say something you need to use words.

 

[EvilDragon]

Powersaving depends A LOT on software as well. The Pandora received a HUGE battery boost after notaz implemented the power saving features. If those are not yet existant for Linux for the A80, it could even be that it uses MORE battery than the OMAP5.

 

[fusion_power]

It looks like there actualy nothing "real" exists at all for the A80 software-side, not much support, data, no community... I'm sure the OMAP5 will be the much better SoC for the Pyra, at least for the start. Thanks to the good daughter-board design idea there is always a door open for other Chipsets.

 

[Binky]

it's about not being too far behind.

There's got to be a proverb for this, but you really shouldn't wait for next year's technology.

If we keep on waiting, there will always be something new, but there will never be a Pyra.

The great thing here is that the processor is on a separate board. You can have your Pyra and wait for next year's SOC. Of course there will be an even better one on the horizon by next year. but that's not ED's  problem.

 

[SNESFAN]

Powersaving depends A LOT on software as well. The Pandora received a HUGE battery boost after notaz implemented the power saving features. If those are not yet existant for Linux for the A80, it could even be that it uses MORE battery than the OMAP5.

Absolutely, no contest there, proper standby in drivers is key to maximizing battery life. Android drivers have also excellent standby features though.


I question that proper standby would be more impactful to active battery use as having a set of a7 cores to throw low intensive processes onto. But without any measurable data from either its complete speculation on both our parts.


But the key to the argument is that one might be correctable in time/effort if it in fact a flaw, the lack of particular hardware won't change though. One is an unknown possible correctable negative where the other is an absolute negative with no possible correction without designing and replacement of hardware.


Its not a game ender, but if finalized hardware decisions aren't made by the time the a80 comes out, it would in my humble opinion, be benifical to relook at the options again at that time. While if I was in your position I would maybe not wait out this unknown and go with the known with caution, I would also maybe be willing to cut losses and go after the big fish if it's within reach at the right time.

 

[WizardStan]

I would also maybe be willing to cut losses and go after the big fish if it's within reach at the right time

Also remember the SoC is on a daughterboard specifically designed to be upgradable. He can release an OMAP5 version soon, and then 6 months later, once the A80 is out and all the unknowns are known, start selling alternate versions.

 

[SNESFAN]

Based off things I've read, he wants to sit on a hardware platforms for years rather than months and have upgrade paths that way. It would make more financial sense to do it that way to burn off on hand parts, that's also why I asked him how many omap chips he planned on getting. The a80 is targeting a cheaper market than the omap I imagine. Selling the stronger performer at a higher premium even if less was invested might also help him out. But again if I was a customer and saw that my early adopter investment was improved greatly less than a year later I would be slightly irritated. Speaking as a consumer of course.

 

[Askarus]

But again if I was a customer and saw that my early adopter investment was improved greatly less than a year later I would be slightly irritated

I would not as long as the to processors can run nearly the same.

I would be happy if I could pay some money to upgrade to a more recent processor.

Tons of cheap Android devices will come up with the faster processor I would wish to have as well.

I wouldn't mind an upgrade every year.

x86 and ARM split does worry me a lot more.

But the same architecture with the same Cortex isn't that frightening to me.

When programs can run on thousands of Android devices we can make them run on 2 Linux computers.

One processor is slower but rather have some more frequent upgrades than have an old processor (2 Years is pretty old for me).

 

[SNESFAN]

In the same product line it is frustrating, and happens everywhere. Even if you are fine with it, you are not most people. Even with mainstream apple products, brand loyalty will only take you so far, and they expect the annual upgrades. 18 months as a minimum would be my suggestion on any major revision outside of storage. Especially for a small scale manufacturer. The firmware won't even be finished before its "abandoned" its a recipe for ridicule. The only redeeming quality of the omap is current availability in comparison. This will be irrelevant a year from now and will be abandoned if something that is across the board better and possibly slightly cheaper. The timing and delay of the a80 could be rather unfortunate, but if he follows a 18 month upgrade cycle it could be an appealing option.

 

[WizardStan]

I see what you're saying but I'm not sure I agree. Assuming ED were to drop the OMAP5 entirely in favour of the A80 I agree that it is annoying, but I'm imagining selling both, the same way he's been selling the Rebirth and 1Ghz Pandoras side-by-side. If the OMAP5 version continues to sell for about the same price and the A80 at a premium, even less than a year later, there's less to complain about. This is especially true if there's a certified upgrade path. And even if he does stop production on the OMAP5 in favour of the A80, it's not like it's been kept a secret that this may happen: early adopters are paying their price to get something cool before anyone else, and sometimes that price is the risk that something cooler comes along a little while later.

Point is that it's not really so clear cut and if you have issues ED may be able to work out a good compromise that doesn't leave you feeling slighted. Of course, the A80 may turn out to not be viable at all due to software restrictions, in which case the entire discussion is moot.

 

[SNESFAN]

Yea, that was something I mentioned, but you definitely have it at a smarter angle for sure. The premium upgrade to generate profit. And it would be a significant and appealing upgrade path, even with the availably relatively close to each other, but it could be worked on when available and released as an option say a year or so later. Let's hope the pyra gets released soon so it would work out nice. Hopefully by Christmas this year or something.

That would be a pretty smart move financially too. And the extra time would allow Linux support to mature to the level of that of the omap, less risky as well. I like it. But the longer the pyra takes to market the less appealing that cycle will look.

 

[notaz]

Personally I hate this upgradeable daughterboard idea, it will be developer's nightmare. 3 OMAP revisions were bad enough, GPUs are behaving different on each of them (just look all at the pain ptitSeb has to go trying support all 3 variants) and I had sunk a lot of time for ironing kernel issues that only show up on DM3730 and not OMAP3 and vice versa. Trying to keep compatibility between different vendor SoCs will be a huge strain, I can already feel the pain by just thinking about it.

 

[EvilDragon]

Well, the daughterboard is mainly to safe costs, reduce the failure rate and offer upgrades in the far future. It was not meant to upgrade each year.

 

[SNESFAN]

Then if possible hope for the best and plan for the worst. I think that omap is going to disappoint people in the long run compared to soon the be available options if you're in it for the long haul.

 

[FaeMinx]

Alternatively, we could just make the best of the omap for the next 3 - 4 years, finally shaking off the financial issues this project has had so far, focus on the software, and then upgrade to something far superior to the A80 with less financial risk.

=~. O =

 

[fusion_power]

Personally I hate this upgradeable daughterboard idea, it will be developer's nightmare. 3 OMAP revisions were bad enough, GPUs are behaving different on each of them (just look all at the pain ptitSeb has to go trying support all 3 variants) and I had sunk a lot of time for ironing kernel issues that only show up on DM3730 and not OMAP3 and vice versa. Trying to keep compatibility between different vendor SoCs will be a huge strain, I can already feel the pain by just thinking about it.

Is it really that bad onto the ARM SoC front? Or is this also a general problem? Because onto the PC side, it seems everything is more or less compatible since Ages "IBM x86 compatible". Not sure how difficult it is to optimize for different AMD/Intel processors but most dev talk is usualy about graphic cards those days.  :unsure:   CPUs also have alot of special stuff for multimedia and even gaming enhancement, so is the situation there better or is this just because of much more devs and/or better documentation?  Or is it even worse to keep stuff "compatible" onto x86 PC's?

 

[Klumpen]

Page 49

People are still debating about which SoC to use?