Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
iPad 3 to Have 1GB of RAM?
- Thread starter GoGa
- Start date
- Sort by reaction score
Ditto - my Gen 1 iPad keeps crashing out of apps lately. Even two tabs in Safari can cause issues. At least that's the excuse I'm sticking to when justifying the purchase to the missus.
Last edited:
Good point, but if we change my example to be less graphic intense, say 10mb graphical ram (40mb retina) and 100mb processing task ram (same for retina), the tables are turned.
I guess that this really points out that whether retina@1024 or non-retina@512 is better completely depends on the apps being run.
having no issues with my dual core computers since 2007.
don't see the point of quad core in a tablet. what apps would use those cores anyway?
----------
same issue with my 3GS. Takes ages to load an app, sometimes won't load an app or crash safari. But transitions, 3D animation, everything is still smooth. It is just blatantly lacking RAM. Apart from that is seems to have enough CPU/GPU power even in 2012.
How come your post is(Number 1) 9 Minutes before MacRumors Post, 08.12 v 08.21
That linked post is still wrong as snowmoon pointed out, they forgot to add the 8396 wired pages.
Does the iPad 2 have problems with memory, if not and the ipad 3 has double ram I don't think it would have problems either.
To prove him wrong, you've used an example which actually fills less than 1% of the ipad's total ram.
Also, not many graphics are full screen. In an actual application, there will be various graphical elements loaded into memory. Most of them will be smaller images (ie: sprites or text) that are re-used many times.
There is of course many different examples depending on use of vectors, 3D models, textures or videos which all scale up in memory differently. You could talk about this forever.
like you said, one could talk about this forever, but I think there are a legitimate concern, since apps can't take advantage of that whole 1GB all by themselves. for example, an iPad1 app could take up around 20mb total before it would be force-quit by the OS, whereas an ipad2 app could get away with around 35-40Mmb...
One of the recent apps I worked (still working) on is a catalog app for a fashion client, and all their artwork is pixel- based, not vector, and full screen with multiple layers of transparent images (for fade effects etc), and right now, on the iPad 2, we can buffer in around 8 full-screen images plus some simple movie on top of that, with the others being pulled into memory and released as the user swipes up and down in the catalog... Right now, hitting the flash memory is the slowest operation, so the more ram the better, but again, quadrupaling the pixels, means either buffering less and having a slower response sometimes with 1gb, or being able to buffer the same amount and have a relatively fluid interface with 2gb.... Also reading almost 4 time the memory from slow-@ss flash storage will just make things worse... So ya, in our usage case, 1gb may not be "enough"...
Slow? You mean things would be better if you read from a spinning hard disk or from the network?
When at my parents house i use there iPad 2 al the time and it never crashed(safari) on me. Well my iPad is a hole different story, it just misses his ram juices.
Thanks for your real world example. That's really insightful.Fukui said:
That's very true that while the average memory usage might remain relatively low, a raster heavy app like yours will struggle.
I expect that the main benefit on a "retina" iPad will merely be a very sharp home screen, UI and text for quite a while, and while some apps will transition easily with a relatively small hit on ram, others will have to make a difficult choice.
However, I'm keen to see the screen with my own eyes. As the pixel matrix becomes smaller, scaling is less noticeable. It may be that a middle ground 200dpi image would still look very crisp on the screen while saving you a lot of memory. 200dpi in print is often fine (I used to design book covers that would go to print at 200 and they looked excellent). Of course print is completely different as a lower dpi image scaled up to a fixed size pixel screen usually looks naff. But then I've never seen how that would look when the pixel grid is so small that it is practically invisible. ...Just thinking aloud here.
Last edited:
When at my parents house i use there iPad 2 al the time and it never crashed(safari) on me. Well my iPad is a hole different story, it just misses his ram juices.
With a bit of luck we will find out in 6 hours, but I will still probably have to wait a few months before they arrive here, but I think I will buy a MacBookPro instead.
Things would be better if they sped up the flash drive speed is what I was getting at... Especially if the ram is going to be limited.... It's not like on the Mac where ssds are super fast, iPad and other smartphones are by comparisson, a bit slower, unfortunately. RAM though, is still quite fast, but then again, power consumption makes it still slower than the Mac by a huge margin... Though its slowly getting there...
Last edited:
Technically the equation is:
(244,276 + 8396) x 4,096 = 1,034,944,512
The brackets are important.
And I do agree the article here and the original poster should have known a page is 4096 bytes. This is because 1k = 1024 bytes for memory. As we all know. A simple mistake no one till the user snowmoon noticed. Well done to him for picking up this error.
Arnold Kim the article author should have picked up this error and mentioned the correct figures in the article he wrote. Maybe I ask too much of him. I don't know. I simply don't know if he (Arnold Kim) is capable of these sort of things.
Wirelessly posted (Mozilla/5.0 (iPhone; CPU iPhone OS 5_0_1 like Mac OS X) AppleWebKit/534.46 (KHTML, like Gecko) Version/5.1 Mobile/9A405 Safari/7534.48.3)
I think the iPad 3/HD will force developers to depend less on full screen pixel art, but if they can use lossy compression (e.g. JPEG), you can increase the resolution without the 4x multiplier, and only need to store the uncompressed version of the artwork that is currently on screen, since the CPU has enough power to decompress quickly. Also, even with more RAM, if the flash memory isn't 4X faster (which it won't be), naively scaling up all your artwork 4X will make the app huge and slow to load. Exporting lots of full screen bitmaps with transitions is very easy for designers using more traditional authoring tools, but I think this is a dead end. With the powerful GPU, there are much, much more efficient ways to render high resolution content.
I think the iPad 3/HD will force developers to depend less on full screen pixel art, but if they can use lossy compression (e.g. JPEG), you can increase the resolution without the 4x multiplier, and only need to store the uncompressed version of the artwork that is currently on screen, since the CPU has enough power to decompress quickly. Also, even with more RAM, if the flash memory isn't 4X faster (which it won't be), naively scaling up all your artwork 4X will make the app huge and slow to load. Exporting lots of full screen bitmaps with transitions is very easy for designers using more traditional authoring tools, but I think this is a dead end. With the powerful GPU, there are much, much more efficient ways to render high resolution content.
You're completely misinterpreting what the parent said.
Think of it this way, you have X RAM available to the entire system. Of that, some is used by the OS, some is used by background processes, some is used by your application.
Of the RAM used by your application, (usually) only a small fraction of that is graphics-related in any way shape or form. (An image viewer *might* be the exception to that rule.)
As you showed in your post, the size of the image is quadrupled, but how many of those full-screen resolution images would have to be held in memory at once to fill up the 512MB available to the entire system? (Roughly 789.6.)
Quadrupling the pixel-count of image resources only quadruples a certain *portion* of total memory demands. In all but the most extreme edge cases, doubling the memory available to the system will more than compensate for that increased demand. (After all, you don't need to store every image in memory all the time, just the ones you're currently using.)
We’ve done compression benchmarking, and with the images we are using, jpg at near lossless quality doesn’t get much better than png... that and the cocoa touch frameworks will expand it out in memory and realign the pixels from RGBA to BGR+A(premultilplied) anyways so in terms of memory usage its a little more complicated than jpg=less RAM usage, and while grabbing a well compressed jpg from flash storage may be slightly faster depending on the image, realigning and premultiplying in real time can be slower in many cases (in our usage case)... we’ll have to see how the iPad 3 handles things... but my guess is that optimized pngs are still the best bet... that and jpegs don’t support alpha... but i think as a lot of posters are saying, in the average case 1gb will probably be just fine....
Last edited:
1GB RAM very likely
AnandTech has just posted an analysis of the new iPad, speculating that the device has 1GB of RAM - Analysis of new Apple iPad - AnandTech
AnandTech has just posted an analysis of the new iPad, speculating that the device has 1GB of RAM - Analysis of new Apple iPad - AnandTech
The Gigabyte is actually 10^9 = 1 billion bytes.
If you ask marketing people, "everything around" 1 billion bytes is a Gigabyte anyways. What you're talking about is a Gibibyte (GiB, 2^30 bytes) though.
Last edited:
Which is fine and dandy except computers dont use the IEC system (yet). The main reason people get confused is because they directly compare the two systems to each other. While IEC deemed a gigabyte as 1 billion bytes, a computer will still see otherwise. This is mainly why the box you posted claims that while "1GB = 1bill bytes; the formatted capacity will be less". Because a computer still sees 1 billion bytes as less than 1073741824 bytes; because computers still count in base-2 (whereas humans count in base 10)