Ram & eMMC costs / power drain

[Mr_Loon]

Hi ED, as we're getting close to the prototyping stage it seems like a logical time to find out what your customers would like to see in terms of Ram & eMMC size.

Answers to the following would help :

1) What are the cost implications for Pyra for Ram sizes of 1GB, 2GB & 4GB?
Answer : price is about 15 EUR for the 2GB (from ED here)

2) What are the battery implications for 1GB, 2GB & 4GB of Ram?

3) What are the cost implications for Pyra for eMMC sizes of 8GB, 16GB, 32GB, 64GB?
Answer : From ED in this thread
8GB : €3-€4
16GB : €6-€7
32GB : €10
64GB : €20

4) Does the size of the eMMC have an impact on battery use? If so what impact does it have?
Answer : The size of the eMMC shouldn't have a huge impact on the battery life (from ED below)

Edit : A few of supplementary questions :

5) What is the base size for Pyra Ram? i.e. 1 x 2GB or 2 x 1GB etc.?

6) Is it possible to disable ram to save battery ?

7) What is swap performance like when system Ram use is 100%+?

8) How difficult would it be to make a 'Premium' Pyra with 4GB of Ram (& possibly a larger eMMC size?)?

A clear understanding of the above questions would really help your customers make a sensible decision re Ram / eMMC size.

 

[EvilDragon]

We don't have any prices yet, as they change constantly and will be checked when we're asking for the mass production prices.

About battery: Well, I know that with 2GB RAM we have a max of 586mW (that's when doing something like md5sum, so the real maximum case), but we'd need to check what power usage 4GB has.
We probably still stick with the 2GB RAM though - it's most probably the best solution for RAM, Battery life and costs.

The size of the eMMC shouldn't have a huge impact on the battery life - you just need to make sure to get a good one.
There are eMMCs which use more power than others.

 

[Mr_Loon]

Thanks ED, it would be great to know the actual cost implications when you get them, especially for the eMMC size.

 

 

[Mr_Loon]

We don't have any prices yet, as they change constantly and will be checked when we're asking for the mass production prices.

@ED : As we are hopefully getting towards the time when specs for Pyra should be finalised I was wondering if you are now in a position to answer the questions in the initial post?

 

[EvilDragon]

Mass ordering of the parts will happen once the prototypes have been tested and are fully working, then we'll decide on the eMMC.

 

[Mr_Loon]

Mass ordering of the parts will happen once the prototypes have been tested and are fully working, then we'll decide on the eMMC.

Thanks ED, how about the costs / power implications for the Ram, looking at the poll here over 50% of users would like more than 2GB of Ram, knowing the actual cost / battery implications would be a big help IMHO.

 

[Grench]

Thanks ED, how about the costs / power implications for the Ram, looking at the poll here over 50% of users would like more than 2GB of Ram, knowing the actual cost / battery implications would be a big help IMHO.

In the same poll 54% said 2GB is enough for them and 27% didn't know, which is a very nice and honest admission for them to make.


Frankly, the question you're quoting from should have had an alternate answer of, "I don't know - whatever ED and the project team decides as they would know best and any numeric value that the uneducated masses pick on this question would be an arbitrary result."

 

[Exophase]

About battery: Well, I know that with 2GB RAM we have a max of 586mW (that's when doing something like md5sum, so the real maximum case), but we'd need to check what power usage 4GB has.
We probably still stick with the 2GB RAM though - it's most probably the best solution for RAM, Battery life and costs.

Is this based on a performed measurement? If so, could you give more details as to how this was performed?

 

[notaz]

It's just an estimate based on data from the datasheet.

 

[Mr_Loon]

In the same poll 54% said 2GB is enough for them and 27% didn't know, which is a very nice and honest admission for them to make.


Frankly, the question you're quoting from should have had an alternate answer of, "I don't know - whatever ED and the project team decides as they would know best and any numeric value that the uneducated masses pick on this question would be an arbitrary result."

I think that the poll gave a good indication that people might be interested in more than 2GB of Ram, for customers to make a good decision about what they'd like in terms of Ram we need to know the cost & battery implications.

The poll results would probably be pretty different if 4GB compared to 2GB was an extra $10 and little impact on battery life than if it was $50 and a big impact on battery life.

From my perspective one of the strengths of the Pyra project is that ED and the team are willing to listen to and directly communicate with the users when it comes to this sort of thing.

 

[Mr_Loon]

@EvilDragon : thought this might be worth a bump

Mass ordering of the parts will happen once the prototypes have been tested and are fully working, then we'll decide on the eMMC.

I know you're not quite at the actual mass ordering of parts stage but : Are we at a stage where a reasonable estimation of costs / power implications for Ram & eMMC can be given?

I think it would be prudent to let people know the actual additional cost/battery implications of 4GB Ram rather than 2GB, as mentioned above even when this was not known over 50% of people who responded to the survey (your core customer base) were interested in more than 2GB. Without this info an informed decision cannot be made IMHO. Another good reasons being that the Pyra is attempting to compete with GPD Win as a serious computing device.

If the cost & battery implications for 4GB are not severe I think it would be a no-brainer to go for 4GB.

When was the last time anyone used a computer and thought I wish 'I had a bit less Ram'....

Why would I want 4GB of ram? Image editing, Running VM's, Future proofing, Desktop replacement use, Ram intensive gaming, Running a Ramdisk, etc.

I can certainly see why 2GB would make sense if the costs / battery drain was high but taking desktop Ram as an example 2GB is approx £10 & 4GB is approx £15. Which would you go for : a £500 system with 2GB, or a £505 system with 4GB? I've no idea if ram for a SOC is more or less expensive than desktop ram...
_______________________

As for the eMMC size could you run a benchmark along the lines of crystalmark showing benchmarks for the various storage options for Pyra (eMMC, USB2, USB3, SD Card, Micro SD Card) this would let people know how much benefit they would get from eMMC compared to other storage options, thus, along with cost info, help them decide what size of eMMC would be best.

 

[levi]

From the latest news thread:

Just a quick addition to the memory:
Memory needs power. Quite a bit of it.

Each DRAM we have on the Pyra eats 200mW, and we got 4, so that's 800mW just for the memory.
I wouldn't want to double the usage, I hope we can even disable half of the memory to save power, as for me, 1GB should be good enough in most cases.

This is a good point. Battery life is a big concern to me. I've never wished explicitly for less ram, but I have wished for more runtime frequently on devices other than my Pandora.

However, one of the best upgrades I made to this laptop was replacing the 1GB sodimm with a 2GB one. That's on 32-bit x86 though, and I do run several processes concurrently on it commonly, and a browser with dozens of tabs on (hurrah for TabGroups!). On my Pandora I don't tend to run much at the same time (except occasionally doing a PNDManager update while testing out games), and 512MB has been sufficient most of the time, and perhaps my Pyra usage will be similar (though a lot of people are talking about using their Pyra as a proper computer/communications device).

 

[Mr_Loon]

Each DRAM we have on the Pyra eats 200mW, and we got 4, so that's 800mW just for the memory.

Any idea what that means in terms of hours of 'normal' usage between charge?

To test I guess you could see how long the battery lasts whilst benchmarking/playing video with a 2GB configuration compared to a 4GB configuration or extrapolate from a 2GB compared to a 1GB with power saving enabled. I upgraded a laptop from 2GB to 4GB and saw no obvious impact on battery life, would this be somehow different for Pyra?

 

[lukey]

Laptops play in a completely different League. My Laptop battery has 56Wh @ 11.11V (~5Ah) which equals to 13000mAh @ 4.2V. Also for them the CPU is the main "consumer". For those systems only large amount of Ram (>32G) make an difference. Also interesting: http://superuser.com/questions/738848/does-more-ram-usage-cause-more-battery-usage

 

[levi]

Three big watts per RAM module? Versus 800mW in the Pandora, blimey!

 

[WizardStan]

Any idea what that means in terms of hours of 'normal' usage between charge?

6000mAh * 3.3v = about 20'000mW
20'000/10h = 2000mW/h implies that each RAM chip consumes about 10% of the total power. If 2000mW/h represents 2GB, 2400mW/h would represent 4GB
=> 20'000 / 2400 = 8.3h
So using some hand wavy math, 2GB = 10 hours of usage, 4GB = 8.3 hours of usage.

 

[TrashyMG]

them Maths

 

[rygD]

Not doubting @WizardStan since that could be accurate. What we really need is actual battery life of the prototype with "typical" use (whatever that is). I could probably tolerate 8.3 hours instead of 10 hours, however 6 hours is somewhat iffy, and I would prefer not to go much lower than that. If we can get between 8 and 10 hours (or more) with 4GB, and it isn't too expensive, I still say go for it. I won't press the issue if most others want to stick with 2, especially since that is probably the best middle ground, and what ED may prefer.

 

[WizardStan]

Not doubting @WizardStan since that could be accurate

It's not. It's a very big approximation under the assumption that you're actually using the RAM. If you're using a web browsers with a dozen tabs open, that's probably close to what it'll be. If you're playing, like, an SNES emulator or something else low RAM, the impact shouldn't be anywhere near that, half that at most.

 

[CommanderB]

But are the larger modules still being considered in the current development stage?

The actual impact on battery life could depend on much more factors then the given average power drain per module.

IMHO only a practical test might answer the battery life question