Ram & eMMC costs / power drain


IIRC he said it was something they'd look into, but my reading of the situation was that getting the timings right to keep everything in sync are likely to turn out to be a bit of work, so don't expect it from the first release is my advice.
[doublepost=1458682198,1458681579][/doublepost]

Technically, if you RAID 0 them together, your total size will be the size of the smallest unit (probably the eMMC, discounting the zram, cos RAID 0 certainly can't cope with drives of variable size) times the number of drives, so say the eMMC is 4GB you'll only get at most 20GB of RAIDed space, because RAID 0 means striped.

But some sort of union mount almost certainly exists to add drives as a single mount point that has all drives be r/w, and hopefully one that wear levels too.

Or you could just use the eMMC to boot off, and RAID everything else. Get matching 256GB SD cards and USB sticks, and the smallest SATA drive you can find and that's 1270GB of unified RAID0 space.

Using more than 4Gb Swapspace is pointless, zram disks have fixed size, there are MicroSD Cards with 200Gb(these are however slower then eMMC) and:

Code:
#truncate -s 14M tst1

#truncate -s 10M tst2

#losetup -f tst1

#losetup -f tst2

#mdadm --build md127 -n2 -l0 /dev/loop0 /dev/loop1

#fdisk -l /dev/md127
Disk /dev/md127: 24 MiB, 25165824 bytes, 49152 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 65536 bytes / 131072 bytes

So you can combine Disks of different Sizes. However, the Performance may be lower if you use the upper part of the resulting Array(except if the Chunks on the bigger Disk(s) are bigger as well).
 
Last edited:
So for the price of a ram upgrade, you can buy another SD card, not get to use slot2, and get what is in comparison a miniscule speed upgrade running RAID 0 for your OS instead of ramdisk. Nevermind the comparable additional power spent doing so, I guess nobody cares so long as its optional.

As for uSD and eMMC, you would, even with softwareswitching between them, be running off the same bus, so you are making no headway on anything.
 
That something is 'mobile' is just as arbitrary a limit as the 2GB barrier.
Of course its arbitrary, thats why usage scenarios are there in the first place, to seperate. And we clearly have to seperate usage scenarios here. In the same line of thinking I could argue that the Pyra absolutely needs space for a 2,5" sata hdd. It needs a lot of space and power and would severly impact mobility ? Who cares? Mobility is "arbitrary" and with such storage capacity I can finally copy my whole desktop hdd to the Pyra, because different uses cases and paradigms are for schmucks that have no real arguments. I rest my case, who needs arguments when I can throw around "arbitrary" thus needing no other "argument" then my singluar usage scenario.
...
The more people you can sell devices to, the better the fate of the whole project.
Seems a rather odd argument (and a again a thought-terminatingcliché), targeting a very blurry cloud that may (or not) have other expectations that the Pyra probably won't met either. It's a good idea to make the Pyra appealing but its always about the costs and possibilities.
Coming up with phones may not be a good idea in this case either, how many modern phones actually have 4GB of RAM ?

Edit: "Maybe a few years down the road" is the whole point. We all have increasing demands, for a device that is meant to last quite a few years. While a 2GB limit is above the pain-treshold for some now, it will certainly be for more people given enough time.
As nonsensical as your first sentence: given enough time we can skip the project as it is, and start planning with hardware that has not even been invented yet. But how about doing it the other way round? The Pandora is doing OK with its amount of RAM for quite some time yet. There is the browsing part of course, but I can't see were this can't be absorbed by the additional 1.5 GB of RAM, or is there a rapid increase of RAM usage by browsers expected in the next few years ? Again, please no silly "but I have 700+ tabs open" usecase - in the same vain, I can argue that the Pyra needs at least 8GB of RAM, because appeal, future and I want to be able to open 17 browser instances with at least 300 tabs open in each

I would rather call it a laptop and apply my appreciation of such a device in smaller form-factor than start making excuses.
What excuses ?

What I would like to see is a rational argument that explains why it is better to make the decision without also knowing the cost / battery implications.....
Is there a chance that we will be in such a situation at all ? Has ED mentioned he would invest further money in exploring this ?
 
Usage scenarios arent arbitrary. If you start fitting them to something arbitrary, like an artificial limit, then they no longer hold water. Im doing one, you are doing the other. Then you blame me for how you make the premise void.

A 2,5" is physically an impossible object to incorporate, whereas more dense ram-chips weigh exactly the same and use the same space.

If you buy a pyra because it fits your use, then there is your argument for incorporating such a use. If someone wants to do all their desktop work, which is what you allude to, after saying mobility is arbitrary, then all the power to you. Notice how bringing up use-cases is arbitrary only to the type of argument you tried to make.

Coming up with a phone is a great idea. Mp3 players become laptops, laptops become phones. Phones have 4GB ram nowadays, yes.

Galaxy S7 and S7 Edge
Galaxy Note 5 and Galaxy S6 Edge Plus
Elephone P9000 and P9000 Lite
Huawei Mate 8 4GB
OnePlus 2 4GB
Oppo R7 Plus 4GB
Letv Le 1 Pro
LG V 10
Asus ZenFone 2 4GB Model
Xiaomi Mi Note Pro
ZTE Nubia Z9 Exclusive Edition
ELEPHONE VOWNEY
Lenovo K80 4GB model

There are also +100 phones with >=3GB memory

If you look at the poll, you will see the demand for >2GB is anything but cloudy.

Sure, if selling more devices to the current marketplace is as nonsensical as selling nothing, to an imaginary marketplace.
One is a small tweak, the other one a mental leap that you liken to it.

If you are fine with openpandora performance, then buy an openpandora. There will be plenty on the market to buy. Not having a vast improvement over current products isn't a very sound business-plan.

Your mock argument was the only valid one you made.

Laptops also have at least 4GB ram, if you want to call it a laptop, it has to have that. Calling it a mini subnotebook, or netbook, doesnt have the same appeal.
 
Last edited:
(from thatgui) Is there a chance that we will be in such a situation at all ? Has ED mentioned he would invest further money in exploring this ?

It would be a great return on a fairly minimal investment to accurately examine the battery implications IMHO.

We know that Pyra has 4 slots for Ram chips, depending on their size it should be fairly easy to make a prototype or two with different configurations of Ram.

There would be no investment needed for the cost aspect to be discussed.

It would also be good to know :

The size of the Ram chips for Pyra, i.e. are they 512MB, 1GB, 2GB?
If it is possible to use a power saving state where some modules are turned off and therefore do not use battery.

To my mind the call for 2GB without understanding cost & battery implications is sounding increasingly like Dogma & I'm inherently suspicious of slavishly following Dogma....
 
To my mind the call for 2GB without understanding cost & battery implications is sounding increasingly like Dogma & I'm inherently suspicious of slavishly following Dogma....

To my mind arguments for increased specs this close to production launch is worse than pointless. It is actually, effectively, talking the project down. To make it even worse, it is an argument for additional specs based on a combination of ignorance and selfishness.

Damaging the project:
By continually harping on what you're perceiving as and framing as a negative, this line of argument leads to potential buyers distress of whether or not the device will work for them. There is no point of realistic reference being used - but plenty of fear being spread nice and thick. These threads of, "2 GB RAM isn't enough!" are starting to pile up. So far, there have been no actual realistic usage cases presented where 2GB is not adequate, but the sheer negative drumbeat is likely to or has already started to cause an impact.

Ignorance:
The tests HAVE been done on the dev board to determine how much RAM makes sense with the OMAP5 that the project centered around 2+ years ago. Please go back and check the relevant posts by TrashyMG. 2GB is not only adequate to the tasks the Pyra is designed for, it is on the upper end of practical usability with the OMAP5. Beyond that, the RAM would simply be near-line storage waiting for the processor to catch up. Additional RAM would also DOUBLE the time needed to go to and recover from standby mode (low power mode with RAM contents dumped to storage). There are costs involved, not just in the parts but in re-engineering what has already been set up and parts that have already been purchased. These calls for changes completely ignore the product development stage. ED has already done a pre-production board population run. This means that spools of chips and many Euros in set up time have already been solidified. These decisions are already done and gone for this round of this device. Could a new/bigger/better spec come along 6+ months in for a processor board with 4GB or 8GB of RAM on it? Sure, why not? That upgrade capability is part of the design spec. Having room in the product production for easily created upgrade parts as the components become more available and less expensive is a historically validated business model.

Selfishness:
To continually harp on an irrational specification chase for an outcome without any actual realistic usability gain, ignoring actual experts AND practical results, is simply a matter of feeding your own ego. On multiple occasions you've referenced the 'other side' as being for 2GB. You have that incorrect. We are for ED having the freedom to use what makes sense from his and others dev board testing, experience, design and intention for the device. Why we are fighting against your campaign is because you are picking an arbitrary number, comparing the Pyra to an illogical field of other devices, and claiming that failure is inevitable if we don't all see it your way.

2GB of RAM on the Pyra IS GREATER THAN 4GB of RAM on your laptop.
The Pyra won't run Windows. It will not have the immediate sacrifice of 2.5GB of RAM to the OS that Windows systems encounter. 2GB of RAM on the Pyra should leave ~1.8 GB free. 4GB of RAM on a Windows machine would leave about 1.5 GB free.

Android uses RAM differently.
My phone may have 3GB of RAM, but frankly beyond about 256MB of it only gets used by applications that I have left (can't close anything for real) that are simply sitting in near-line storage if/should I ever want to access them again before rebooting. The reality is that 95%+ of Android phone applications run the same on any Android phone from 256MB on up. The original 128MB RAM Android phones still work just fine. The spec chase for these are simply justification for 'new' and cleats to attach marketing to the 'new'. The additional RAM isn't 'bad', it simply does not add performance on the scale that people seem to credit it for.

More is not always better. There is a lot more to go into the RAM decision.
Lets consider the NAND for a moment. The knee jerk reaction for people is that they want it to be as large as possible. However, there are actually three factors to consider: size, speed and cost. We can assume that cost needs to be controlled to under a given level - so it can more or less be held constant. The device has the uSD slot that it can boot from that is capable of up to 200GB at 90MB/s using current uSD cards (theoretical max speed of 104MB/s). The NAND on the other hand is capable of theoretical speeds of up to 208MB/s and the largest NAND I have seen that would fit is 128GB. I have voiced that I prefer for ED & team to figure out their rough price point then select the FASTEST NAND that they can find within their budget that is large enough to run the OS & swap from. 4GB NAND might fit the OS, but doesn't leave much space. 8GB is probably fine. 16GB is plenty - capacious even. Anything larger than that starts getting expensive due to the size and would require lowering the speed to be able to afford the size. So, hopefully the project is finding the fastest 8GB or 16GB eMMC NAND that will fit the budget. Note, I'm not wanting it to be HUGE and MAX speed. I'm asking for it to be within reason - but weighted toward maximizing advantage of the NAND's interface speed. I hope it works out that way. Maybe there is more to consider in the RAM decision than just, "4GB is more than 2GB". Like the NAND, less may very well make more sense than more for reasons you and I might not even be aware of.

I know you're going to continue on your "I/we need 4GB" holy war, regardless of what any voices of reason say. I would prefer that ED & team have the flexibility to decide quantity and type of device based on facts and reason instead of hype and intimidation.

So, can we let ED & team make the choice? You have stated your point, it's clear that they heard you.

IMHO it is time to back off and support the project, in whatever ED & team decide is best, and let the egos relax. Can you?
 
On the Devboard I was running around 80-95MB idle with aTc's pyraOS (Debian Jessie) image. So plenty of free.
 
While I am generally in the "more is better" camp, it occurs to me (judging by somewhat suspect, but probably good enough in this case, Instructions Per Second (IPS) figures) that the Pyra's CPU is roughly equivalent in power to a desktop PC's in 2003, whereas the Pandora 1GHz's CPU is roughly equivalent in power to a desktop PC's in 1999. In 2003, a decent computer might have 512 MiBs of RAM, while in 1999 the figure would be 128MiBs -- 2GiBs became common around 2008 and I bought myself a £1000 laptop with 4GiBs of RAM in 2010, which had a processor much more powerful than the Pyra's (perhaps 10x more powerful). Assuming that RAM use scales somewhat with processor power, 2GiBs for a 2003-era-equivalent CPU should be fine. 4 GiB would be nice for marketing purposes/bragging rights, though.

12_15_RAM_Prices.jpg


Aside: my home desktop has 32GiBs of RAM and it is glorious -- until Chromium fills it all up.
 
Aside: my home desktop has 32GiBs of RAM and it is glorious -- until Chromium fills it all up.
I can't even all of the 4GBs of RAM in my current Linux Desktop even with Chromium...
 
I know you're going to continue on your "I/we need 4GB" holy war, regardless of what any voices of reason say.
As mentioned earlier there is no 'holy war' from me for 4GB rather than 2GB, that's entirely in your head.

Once again : All I'm saying is I'd like to make a choice about how much Ram I'd like to see in the Pyra based on my intended usage combined with information relating to cost & battery life. That desire to me seems entirely reasonable & sensible. I can understand your (and other peoples) willingness to make up your mind without cost & battery information but I'm willing to bet there are plenty who would like that info before deciding what they would like.

I imagine you're getting a little confused with your needless repetition of how Ram works in different OS's because some of my (and other peoples) intended usages for Pyra do benefit from 4GB rather than 2GB of Ram such as Image editing of large multi layer images with a long edit history. However what I'm not saying is the Pyra must have 4GB of Ram at all costs, either to suit my needs or to compete with the GPD Win as a serious computing platform, but that I'd like to know the various costs before deciding what would be my ideal spec for Pyra.

I can see many ways an open discussion about Ram & eMMC size would help the project, for instance if some of the valid and interesting points you brought up about the eMMC size vs speed vs cost were also discussed.

Always question Dogma....
 
I can't even all of the 4GBs of RAM in my current Linux Desktop even with Chromium...

All it takes is a few thousand tabs. On a Pyra, though, I doubt that would occur: the small screen would make the multitude of tabs unmanageable; the CPU may not be up to it, either.
 
In 2003, a decent computer might have 512 MiBs of RAM
And in 2003 websites were static pages with some text, images, and buttons. Modern web pages are full blown programs requiring substantial RAM... unless...
All it takes is a few thousand tabs.
is it just me then?
I fill 8GB, sometimes 12 (and on a rare occasion, all 16, 8 real and 8 swap), on just 40-50 tabs. :( That also includes about 1GB of other stuff running but that shouldn't be an issue.
Nothing particularly special, that's my morning webcomic read, load them all up, close them one by one. Is there something seriously wrong with my system to be consuming that much RAM? It was the same between Firefox and Chromium; I switched to Chromium because it would let me see and kill particularly RAM hungry tabs before my swap filled up and my system came to a grinding (literally, the hard drive goes nuts) halt.
 
And in 2003 websites were static pages with some text, images, and buttons. Modern web pages are full blown programs requiring substantial RAM... unless...

is it just me then?
I fill 8GB, sometimes 12 (and on a rare occasion, all 16, 8 real and 8 swap), on just 40-50 tabs. :( That also includes about 1GB of other stuff running but that shouldn't be an issue.
Nothing particularly special, that's my morning webcomic read, load them all up, close them one by one. Is there something seriously wrong with my system to be consuming that much RAM? It was the same between Firefox and Chromium; I switched to Chromium because it would let me see and kill particularly RAM hungry tabs before my swap filled up and my system came to a grinding (literally, the hard drive goes nuts) halt.

I find that most websites gain little from being more than text + images. I also greatly prefer sites that are well handled by Lynx.

Hmmm... currently I have a few hundred tabs open, taking up 12 GiBs. To take up all 32 GiBs would take over 1000, at least. Howeer, when I do have many thousands of tabs open they tend to come from a few sites: Wikipedia, reddit, TVTropes, forums, BBC News etc.. It may be that there is sufficient data sharing to cut down on the RAM use, or that the sites I tend to binge on are relatively light.

Why in holy hell would you guys keep so many tabs open. Seriously it's what browsing history is for...

Browsing history must be actively checked. A tab left open (for "later"... or never) may just be glanced at to activate another web-surfing session. Also, I tend to open many many tabs at once -- while I used to follow links and then return using the back button, I now read through an article and middle click on each link as it comes to open it in a new tab, to be checked after I have finished the current article. If, like with Wikipedia, all these articles have many links, then the number of open tabs can increase very quickly.
 
I have Iceweasel with NoScript and ~70 Tabs Open and it just needs 600Mb RES. + Claws-mail + HexChat + xpdf showing the 6000Pages OMAP5 technical reference + LibreOffice Impress = ~1Gb for the whole System under Debian Testing.
 
I find that most websites gain little from being more than text + images. I also greatly prefer sites that are well handled by Lynx.
Browsing history must be actively checked. A tab left open (for "later"... or never) may just be glanced at to activate another web-surfing session. Also, I tend to open many many tabs at once -- while I used to follow links and then return using the back button, I now read through an article and middle click on each link as it comes to open it in a new tab, to be checked after I have finished the current article. If, like with Wikipedia, all these articles have many links, then the number of open tabs can increase very quickly.
There is seriously not enough interesting things on the internet to keep more than 50 let alone 1000 tabs open for so long. If I need to get back to it, I'll actively search for it. More than 20-30 I find myself closing out tabs to keep the clutter down.
 
I've got two gmail tabs open, they're taking almost 400MB each, a chicken recipe tab open, that's another 300MB, a slack chat tab for 200, my calendar at 150, facebook for 130 (although I've seen this exceed 1GB before), this forum is pretty low at only 120, gog I occasionally come back to only needs 70. Then I've got actual low hitters like wikipedia and imdb taking up less than 20MB each, and a couple of phpBB tabs at a whopping 15MB total.
So I basically run the whole gamut. Question is, is any of this normal? Has my concept of what enough RAM is (and therefore how much RAM it would be nice to have in the Pyra) been skewed heavily by this?
 
I've got two gmail tabs open, they're taking almost 400MB each, a chicken recipe tab open, that's another 300MB, a slack chat tab for 200, my calendar at 150, facebook for 130 (although I've seen this exceed 1GB before), this forum is pretty low at only 120, gog I occasionally come back to only needs 70. Then I've got actual low hitters like wikipedia and imdb taking up less than 20MB each, and a couple of phpBB tabs at a whopping 15MB total.
So I basically run the whole gamut. Question is, is any of this normal? Has my concept of what enough RAM is (and therefore how much RAM it would be nice to have in the Pyra) been skewed heavily by this?
It seems pretty bizarre to me, I couldn't manage to use all of the 2GB on the devboard with a ton of tabs open on various websites including gmail, had a metric ton of large PDF files, Libreoffice, games and etc running at once.
 
Once again : All I'm saying is I'd like to make a choice about how much Ram I'd like to see in the Pyra based on my intended usage combined with information relating to cost & battery life. That desire to me seems entirely reasonable & sensible.

And that is the issue that seems to be so difficult to get through. It is not your decision to make.

Those who's decision it was to make have had access to dev boards, early OS releases and have already gotten the facts that they require through actual testing. They likely made this decision many MONTHS ago as (some) parts have been purchased and board population set-up has been tested.

The buzz is that the Pyra team's work has most likely resulted in a decision to have 2GB of RAM. I do not know 100% if this is or is not the case. Maybe I will be surprised. BUT what I do know is that the decision is being made by people directly on the project who have the skill set needed to make this decision. It is not my decision to make, and I trust them to have made the decision based on solid experience based information.

What you are now doing is damage through negative hype based on your personal opinion that somehow you deserve to make this decision. Or is it that you think that the decision that was made is, "wrong"? Either way, you apparently feel your opinion, without any basis in testing or actual facts, should override the actual development team, based on "needs" that you have so far been unable to articulate beyond spec-number chasing advertising.

TRUST ED. At this point that involves stepping back and letting them produce what is in their plan. Specs should be 'locked down' at this point. He (ED) and his team, more than any of the rest of us, wants this to be a success.
 
What you are now doing is damage through negative hype based on your personal opinion that somehow you deserve to make this decision. Or is it that you think that the decision that was made is, "wrong"?
I don't know if the 'decision' is wrong for me without knowing cost & battery impact. Please understand that is the purpose of this thread, to understand that information, not to agitate for more ram just because I would find it useful.

I strongly believe that a significant number of potential customers would rather have 4GB than 2GB Ram if cost & battery impact is minimal.

From everything I understand there would be no real production issues doing batches of Pyra with different sizes of Ram.

If that could be done and it leads to more sales or higher profits from more expensive units then that will only be good for the project IMHO.

"needs" that you have so far been unable to articulate beyond spec-number chasing advertising.

Now you're just being silly, please explain how the benefit gained from additional Ram when Image editing large multi layer 16BitsPerPixel images with a long edit history (or the other real world examples given) is 'spec-number chasing advertising'. If you carefully read through the thread you'll see that the additional marketing kudos from 4GB vs 2GB was listed as a very minor consideration.

Try looking at it this way, let's say your ideal Pyra would have 2GB of ram and it seemed likely that the default model would ship with 1GB of ram, given that some of your usage scenarios would benefit from 2GB, would you not want to have some understanding of the following :

What's the battery impact of 1GB vs 2GB?
What's the cost impact of 1GB vs 2GB?
Is it easy to make a 'premium' Pyra with 2GB?

If you thought that lots of other people might also be interested in 2GB and sales / profits could be lost would you not mention it?

Or would you follow the Dogma?....
 
Back
Top