It looks like it may be done that way to minimize distance to what's in between them, which I'm guessing is shared L2 cache. The individual cores themselves look distinct and identical, much like the Cortex-A9 cores.
Next-Gen iPhone & iPhone Nano Speculation
- Thread starter Arun
- Start date
It looks like it may be done that way to minimize distance to what's in between them, which I'm guessing is shared L2 cache. The individual cores themselves look distinct and identical, much like the Cortex-A9 cores.
anexanhume
Veteran
My guess too.
It's interesting that they have the RAM directly below the CPU. Like they are trying to take advantage of good parts of PoP DRAM without the thermal disadvantage.
ltcommander.data
Regular
That makes sense.
http://www.anandtech.com/show/5686/apples-a5x-floorplan
Anand also notes the DRAM controllers have been moved from bracketing the CPUs to bracketing the GPUs possibly to give the bandwidth and latency advantage to the GPUs. There are also 4 DRAM controllers up from 2. Those early Geekbench memory tests didn't show a noticeable change from the A5 so presumably it's not 4x32-bit. I wonder what are the benefits with moving from a 2x32-bit setup to a 4x16-bit setup? Some interleaving efficiencies? Are the the 2 DRAM chips actually stacked 2x2?
I'm not seeing how moving the memory controller proxiity would change bandwidth, unless it was latency limited from performing at peak. GPUs are supposed to be much better at hiding latency than CPUs, even something like SGX543MP4. Especially against Cortex-A9, which is going to be fairly latency sensitive. Doesn't seem like the right trade-off to make.
But with mobile latencies being as high as they are despite much shorter trace lengths than with desktop memory suggests that something else is masking the travel time entirely.
But with mobile latencies being as high as they are despite much shorter trace lengths than with desktop memory suggests that something else is masking the travel time entirely.
What are the typical numbers?
I measured about 120ns on Tegra2's nVidia dev board.
Given that two areas marked as graphics are identical this means neither of them includes video encode/decode as those would not be duplicated. I was therefore assuming this was the video encode/decode
french toast
Veteran
Ive read somewhere on this forum a discussion about it, and the consensus was that Android was initially developed for a keyboard and mouse, that was before iphone became a success, once Google took over and saw the craze for multitouch...they went that route..but all the UI or something had already been settled, and that carried right over through the apps...basically multitouch input was a 'bolt on' rather than a design requirement from the start ala WP7/IOS...
So you could say its more an upgraded modern windows mobile..just with stollen code..
Here's some interesting stuff which discusses the 'fluidity' of Android vs others such as iOS and WP7:
https://plus.google.com/100838276097451809262/posts/VDkV9XaJRGS
Both the original post and the linked information provides an interesting overview of the way things work.
https://plus.google.com/100838276097451809262/posts/VDkV9XaJRGS
Both the original post and the linked information provides an interesting overview of the way things work.
So I haven't pulled the trigger, though supply at Apple Stores near me seems pretty good. Either they ramped up higher volumes or the demand isn't as strong or a little bit of both.
One thing that makes me hold back a little is that they had to pack a bigger battery, resulting in a heavier and thicker design. They get the same battery life but takes longer to charge.
Makes you think that the A6 (along with a more efficient LTE module) will result in a lighter and slimmer design in the next iteration.
One thing that makes me hold back a little is that they had to pack a bigger battery, resulting in a heavier and thicker design. They get the same battery life but takes longer to charge.
Makes you think that the A6 (along with a more efficient LTE module) will result in a lighter and slimmer design in the next iteration.
The Galaxy Nexus was originally going to have its CPU and GPU clocked 20% higher (1.5Ghz/384Mhz respectively) but either Google wanted better battery life or TI wasn't able to hit those clocks but regardless the Nexus is slower because of it. It has to use the same GPU as the Droid RAZR but deal with a 720p screen, which certainly doesn't help.
The new screen is also a pretty big contributor to the increase in power consumption.
ltcommander.data
Regular
http://forum.xbmc.org/showthread.php?tid=125840
The Apple TV 3 has also been opened up now. The single core A5 doesn't seem to be physically different supporting the theory that it's a harvested part.
The Apple TV 3 has also been opened up now. The single core A5 doesn't seem to be physically different supporting the theory that it's a harvested part.
I've been wondering myself about that. If wifi usage isn't that much longer than LTE, then it's hard to justify blaming LTE. I suppose refreshing 4x the pixels has its cost.Helmore said:
I pulled the trigger one the day of introduction (wifi only). IMO it's totally worth it. The additional weight is noticeable in side-by-side comparison. It doesn't bother me, but bothers my wife a little bit. The additional thickness is very noticeable, but only bothers because of esthetic reasons. ("Things are supposed to get thinner not thicker.")wco81 said:
I don't see a need for an A6.french toast
Veteran
Thanks for info..so i did remember that right then! it seems Android is always going to be struggling with UI responsiveness, ICS and powerfull chipsets help negate it, so does better app development. (Tegra 3 also has the 'shadow' core dedicated to UI) but Android is the only operating system that hasn't been designed from the ground up since IOS...ala windows mobile 6.5.3
They had to double the LED emitters. I would guess the high-res display is less efficient and disperses more light.
french toast
Veteran
A few more benchmarks for new ipad and a comparison against Transformer prime;
http://blog.laptopmag.com/tablet-chip-showdown-nvidia-tegra-3-vs-the-new-ipads-a5x
..Interesting that the Tegra 3 beats out the A5X in geekbench memory subtests....would have thought the A5X with dual channel controller and more 'optimised' chip would fair better in this area..
http://blog.laptopmag.com/tablet-chip-showdown-nvidia-tegra-3-vs-the-new-ipads-a5x
..Interesting that the Tegra 3 beats out the A5X in geekbench memory subtests....would have thought the A5X with dual channel controller and more 'optimised' chip would fair better in this area..