Limited RAM in Apple's A5 chip in iPad 2, iPhone 4S motivated by battery life concerns - Page 2

Yes because you know more about this subject than the ENGINEERS at 2 of the biggest companies in the world....

I have always said this to my cousin who seems to love the phone with the best specs. I would just say "wats the point of having these 'SUPER GREAT SPECS' if your really not going to use them all that much on the phone, its more efficient to put in the specs NEEDED to perform the tasks that it can"....my point is that all these android fans love their specced out phones but is it really worth it to having a shitload of memory if your batter life is going to be shit? Apple knows how to make use of what they have and somehow still beat out the competiton with better specs, and then to put the their foot a little further up Android's ass they have way better battery life.

What do you expect. This place is crawling with Apple-fanatic-idiots. There are quite a number of good Apple folks here too.

There was a time somewhere between Jaguar and Tiger that OS updates actually felt faster on the same hardware (the RISC days with IBM.) Between Tiger and Leopard, with all of the changes it felt as though Apple forgot how to write sleek code. Then SL came, and people thought that with "no new features" it was a flop, until you realized the bloat was gone and the machines were fast again, all without a hardware upgrade.
I truly hope for all of us MS has finally gotten the hint; that good code is better than hardware advances. If so, we all win.

Hardly. The price paid for going off die, even within the same package is an access time increase of at least 30x or so. Keeping memory on die will eventually be the way of the future, and it will make computers that 30x or so faster all at once. We've got more than a few years before we get there yet.

The main benefit of packaging the way Apple does is that power requirements are significantly lower because the connections can be better engineered compared to memory off-package connections. Better engineered connections mean less noise and less noise means less signal strength needed which is why the power consumption gets lower.

So no, it's not semantic issues at all. It's hard fact and precise terminology that immediately identifies the performance regimes.

Those reasons are why Apple provides so many stock GUI elements. Sharing the graphics makes an order of magnitude or more improvement in memory efficiency. It's also why Apple suspends programs between invocations as the default way to do business. That saves both from eliminating background running code when it isn't necessary as well as further lowering overall memory requirements, which allows less RAM and again less power draw.

A slow poke of a computer by today's standards. On the other hand I just loved that little machine as it highlighted why people would want to have such a machine in the house.

It did however highlight another issue, that is Apples reluctance to offer enough RAam in their machines.

Obviously we won't get the area to implement system RAM but we should have enough die space for video RAM for the GPU. Even if it is just a frame buffer it ought to help a great deal with power savings. After all the GPU constantly accesses memory.

I suppose the modern approach would be to make the cache bigger. I'm sure that would lead to an interesting discussion amongst engineers. My thing is that moving to retina type displays in an iPad means a lot of data gets accessed for each screen update. Putting that frame buffer outside of normal memory channels just means less congestion on a regular cache.

For anybody interested there is a series on ITunes that goes into chip design. Unfortunately I don't have the exact name but it was offered through Stanford and was part of their computer science offerings.

In a nut shell you pay big time in access speed and power when going off chip. the question for Apple is the move to the 20nm nodes enough of a shrink to provide room for a reasonably sized RAM array. I suspect not, at least not for general system access. Especially after adding cores and other functionality.

While you are correct on the technology I think the big advantage for Apple is the ability to crunch a 32 bit computer down to the size of the tiny motherboard inside iPhone. In otherwords physical space is a driving force for this type of hardware. That doesn't mean the advantages you point out aren't important as the have a huge impact on the system. It is just that without the stacked dies iPhone would likely be a much fatter device.

I'm not happy to see people wallow in ignorance either. For the non technical person simply understanding that there are multiple chips in that A5 package ought to be good enough. In the future we may have 3D technologies that obscure the issues here. For now people need to realizes that A5 is a package of chips.

It took me about 15 seconds to google for that link:

http://blogs.msdn.com/b/b8/archive/2...windows-8.aspx

I wonder how you want to do "fact checking" without even being able to find that link yourself.

But then why work yourself when others can do the job for you...

The problem with a system that uses dynamic RAM is that it constantly has to be refreshed. That ends up being a significant power draw for something that might spend most of its day in your pocket. I honestly think the guy from Microsoft was addressing those loads that impact idle battery lifetimes. Let's face it when an app is being ran or used the power demand can vary widely and more importantly the user expects power to be used when the app is in use. What the user doesn't like is his cell phone going dead in his pocket.

Now here one has to expect that a device will have significant idle periods which many iOS devices do. The always on nature of most iOS devices though require that idle power be minimized as much as possible.

Thanks for helping the tall youngster, Andreas. It's sad really. In the words of Jed Clampett, "pitiful, just pitiful".

Your comment is wrong on so many levels, it's almost as if you tried.

3G has nothing to do with RAM. The only data being uses over 3G, WiFi, or any other connection is data that could otherwise be stored via Flash, but isn't because it either needs to be fresh data (what is the weather, stock price, news), or because it takes up too much space (the Netflix library, Wikipedia, Google Maps, etc...).

Using 512MB or 1GB of RAM would have no impact on connectivity required resources.

Furthermore, even 3rd party apps are allowed to cache incoming data to Flash, so even assets that are time sensitive, but haven't expired, or are too large, but only represent what was previously needed, can be written and read from Flash.

The bottom line is that this article is spot on. There are multiple reasons for the decision to use less RAM: 1) Cost 2) Power Savings 3) Size 4) Speed. The last point, speed, is a tricky one, because less RAM will actually increase the speed of the device *IF* the upper threshold of memory usage is not reached and is only of practical significance if the amount of installed RAM that you're comparing it against is overkill.

deleted

deleted

no, heavy is not better. use a laptop if you want more ram and weight.

exactly. this is a big beef with me and android hardware. i could care less if it has a 6 core cpu and 4 gig ram blah, blah, blah, blah. get the f***** battery life and weight in order first!

deleted

deleted

A Macbook Air doesn't last any longer on battery with 1GB RAM than it does with 4GB. If anything, it's the opposite because not being able to maintain data in RAM requires you to download content again and redraw the screen using wifi, CPU, GPU and Flash memory.

There is something to be said for efficient use of memory but a lack of RAM is not something we should hope for. Competing tablets/phones can have lower battery life for any number of reasons - using inefficient code, using highly clocked CPUs and GPUs, Adobe Flash support, larger displays with less efficient illumination methods.

I reckon iPad 3 will have 1GB RAM and I also reckon it won't affect battery life.

Would you be kind enough to explain what you mean by "true multitasking" and why iOS does not have it?

ty

What you say is undoubtedly true of hard drive based systems, but we are talking about the iPhone here and Dynamic RAM will most likely always consume more power than Flash RAM. Therefore accessing secondary storage will most likely have very little power hit (especially if it is just read cycles) compared to the power hit from increasing Dynamic RAM. As the article points out, Dynamic RAM consumes power whether the RAM is being used or not, whether it is actually holding any data or not. Flash RAM only consumes power when an actual read or write is in progress.

I reckon you are correct, but I also reckon that the iPad3 won't be using iPad2 power management or battery size to give the same battery life.

phone wise: nexus one, htc HD, htc sensation, motorola droid. same gripe: terrible battery life.
weight wise i bled over into android tablets....but same sentiment. screw the specs, build a better designed product all around. thats what apple does.
as far as the samsung galaxy products weight, in particular the tablet, it is acheived with plastic instead of metal. they feel a bit 'cheaper' than apple counterparts.

The ability to have your games running in the background as you have a movie running in the background as you type a document as you answer texts.

No

[QUOTE=MacRulez;1965093Do you believe that network connectivity is without energy cost?[/QUOTE]
No


Of course connectivity, and any other operation, uses RAM, that's not the point. Increasing the RAM from 512MB to 1GB isn't going to have any impact on the amount of 3G usage, that's the point.

As you yourself state, connectivity is second only to display for energy consumption, so accordingly, reading data from storage uses less energy than connectivity (which is not only true, but a whole lot faster as well).

The original point of this thread, which was highly flawed, was that with insufficient RAM, resources would be read / data swapped via 3G, which is just incredibly false.

Whether RAM energy consumption versus needing to swap resources and data to flash consumes more energy is a worthy debate for any specific platform. If you're an engineer on the project could argue that the platform either can stay under the memory threshold for energy savings or would go over.

Looking at the iPhone now as a user, it's speculative as a whole, but as an individual, I've witnessed very little swapping of data and resources do to memory limitations on iOS 4. I'm jailbroken and have monitored my memory use. If iOS 5 uses the same or less memory, then it makes sense, for me, that 512MB would use less power than 1GB.

Memristors can't get here fast enough. Ridiculously fast, single feature persistent storage doing the job of 6 transistors in a SRAM IC.

Purdue also stumbled on some neat persistent no power when not in use memory too, but it's between flash and DRAM in speed.

I don't expect memristors to be routine for around 10 years though, maybe some prototype stuff with it 5 or so. HP and IBM independently found out how to build them repeatably in the past year. Now it's onto production quality and how to incorporate a previously only theoretical circuit element into an IC.

The difference is completely obviated by the power cord. The power draws really aren't that different from 1GB to 4GB in an absolute sense. But when you put the system on a hard power budget at the milli level, not the 15 Amp 1500 W/H level those little differences become very significant.

Now spread that relatively very significant fixed milli-budget difference over a week with a source that gets less efficient as it gets depleted and we magnify power draw differences even more.

The reason a potential 1GB iPad3 won't affect apparent battery life is that other places will be engineered to offset, like add a little more battery mass and manufacture the greater amount of RAM in the next smaller stepping to reduce it's power consumption by a third or so. The Apple engineers know with with pretty good certainty what the state of the art for iPad3 production will give in power performance because Apple isn't operating on the bleeding edge. They are adapting well known tech into previously unused configurations because they have a product that supports the R&D for the relatively new production configurations.

I'm going to ignore the latent Microsoft and Windows bashing. What I find interesting is that traditionally, the screen has been the biggest power hog in most modern electronics. Clearly robust processors with numerous cycles and extended memory are also a factor (as anyone who has ever done significant overclocking will tell you) in sucking battery, but I'm sort of shocked that RAM is that big of a factor. I guess it certainly is a bigger factor than the screen in the "sitting on your coffee table" mode described in the article, but on, wouldn't the display be the issue?

The RAM is an issue because it draws power at a more or less constant rate even when the screen is off and the processor is sleeping. If you are actively using the device RAM power draw is not a significant issue, but the hours and hours you aren't using it are where it is a big deal.

Sorry but this is a pet peeve of mind. Flash is not accessed randomly and in fact is more disk like than RAM like.


I suspect it would be more correct to say it depends upon the app and user.

Even that is not entirely correct, if it is turned on flash will consume some power. How much that some is will likely depend upon its size. On the whole though flash burns a considerable amount of power during writes which can't be discounted if an app creates data.

As others have pointed out the lack of RAM can cause power usage due to reloading data from the net when it might not have too. The difference in behavior between my old 3G and iPhone 4 was dramatic and directly due to more RAM. I don't deny the validity of this story, I'm just saying that the value in having more RAM is dependent upon usage. Further more RAM can actually extend run times in some situations. Right now I just think Apple is shooting for maximum idle time and cost control more than anything.

And your point is wrong. I can see a difference in reload patterns just between my iPad and iPhone 4. More RAM means that Apple doesn't have invalidate page data to free up memory for the page you are going to. This means a greater likely hood that when you hit the back button the page you are going to gets reloaded from RAM not 3G.

So why did you say the above? You are right in the above statement so you have us confused.

I'm not sure how you even came up with that position. 3G comes into play when using net centric apps and yes the lack of RAM can result in re-reading data over 3G. I really don't understand how you can not see that point.

one thing we can agree on is that it is not as simple as some would imply. In apples case though they are very biased to minimal memory machines and have been for years. I'm not convinced Apple made the decision based on merits related to power usage. It could simply be an issue of margins.

It will vary by user but I suspect you don't know what you think you know.

QUESTION.

Apple's loaded like an Russian officer's gun during the siege of Moscow, why don't they do some research into production-quality MRAM to remove any question of power requirements, allowing capacities to skyrocket?