Shared publicly  - 
 
A new experimental ARM Cortex A7 board named Moose has recently been added to the open-source Chromium OS repository by MediaTek.

Source: https://chromium-review.googlesource.com/203353
101
7
Caleb Boucher's profile photoMark Bridge's profile photoJoseph Hartley's profile photoNikita Golovko's profile photo
21 comments
 
Ugh. My Chromebook uses Cortex A15, which are twice as fast as the A7, and it's so slow that it's barely usable.
 
Cortex A57 should help, as well as more ram.  Kinda frustrating that all the chromebooks are so stingy on ram.
 
I don't believe any current consumer-targeted processors have address bits sufficient to go beyond 3-4GB +Bill Broadley. Breaking the 32-bit address barrier is one of the big gains from 64-bit (which doesn't imply 64-bit addressability).
 
+Morgaine Fowle Maybe you meant ARM CPUs only?  I recently bought a $170 chromebox that comes with an intel celeron (intels lower end consumer line) which is a 64 bit chip that can directly address at at least 16GB.

Granted the 64 bit arm CPUs aren't shipping yet, except in apple products.  But I expect that to change later this year, thus my mention of the cortex A57, which is a 64-bit arm chip.

BTW, I upgraded the chromebox to 8GB ram (from 2) and it's quite nice to use.
 
Knew about x86 pae, should be less surprised it's an arm feature too. Do chips actually implement it? Intel only just this year bumped atom up to 8gb max memory, for example.
 
ARM LPAE seems pretty useless at the moment.  I don't believe there's any normal arm widget you can buy that has (or can be expanded to) more than 4GB ram.

However with the coming 64 bit chips you can skip the PAE.
 
Does this mean that MT will contribute some drivers? Because that's been a weak point of theirs up to now?
 
Great! ARM a15 is faster than a12 which is faster than a9 which is faster than a7 and the fastest a15 is in the Samsung Chromebook which is pretty slow. That means this chrome device will be super slow and not usable!
 
+Jean-Baptiste Quéru I fully agree my CB11 with the exynos a15 is slow, clearly there is some optimization that needs to be done if Google wants to continue making devices with Chrome OS on arm socs.
 
+Jean-Baptiste Quéru The Cortex A15 is twice as fast as the A7, but what happens when you have 8xA7 (one of the MediaTek octa-core designs)  in a thread heavy environment like Chrome? Any thoughts?
 
+Kenton Douglas Mediatek has A15 big.LITTLE tablet chipsets (which include A7s), too, which I suspect will be what's actually used in any Chromebook produced.
 
I think you guys are off the mark. I've been using a Samsung Chromebook 2, and the GPU seems to be the true downfall of the Samsung Series 3, which is the Chromebook I'm assuming you guys used when you're calling A15s cores slow. After loading a webpage, the Samsung 2 moves to A7 cores for almost all tasks, including video.

If the A7 cores are paired with a good GPU, it should feel smooth. Page loading speeds will suffer though.
 
+Wesley Files - I've used a Nexus 10 extensively, which uses the same SoC as my Chromebook but pushes almost 4 times as many pixels, and that doesn't feel slow, so I know that that specific CPU+GPU combination. My Moto G, however, with a dual A7 paired with a different GPU, does struggle at times.
 
Not sure it matters, but the kernel specify's support for the MediaTek MT8127 tablet processor. This is quad-core, 1.5Ghz but still slow as molasses, even against the original Exynos in the Series 3 and Hp11
 
moto g is a quad a7.

regarding the comparison between a nexus 10 and the series 3/hp11, i'd say it is apt in some regards but they also use a different upper software stack (inc. wm/desktop). CROS in superficial ways is also more multitasking intensive, IMO.  Which is a way of saying more [4-8] a7 cores instead of 2 a15 cores i think would have some potential, in additional to a lower market price. (how are those chromebook 2's working out WRT cores? +Wesley Files seems to think it is fine. Same idea, except crbk2 is higher end)

finally, isn't an a7 ~2 dmips/mhz? a15 is ~3, i wouldn't say that's twice as fast.

side note, i scooped this a couple days ahead of francois (hurray for me :-) )
https://plus.google.com/106810288520604103752/posts/3PbFv5vbax9 and crossposted to the chromebooks community
 
+Ted Phillips It's 1.9 vs approx. 3.5-4.0 in Dmips. But, I'd suggest the multiple cores (4-8) would offer advantages in a thread-heavy/multi-talking environment like Chrome.  I suspect that we would end up with something between the Samsung Chromebook 1 & 2 (tending towards the newer device) in terms of performance - depending on the GPU. It could also mean a $150 dollar Chromebook.
 
Chrome is multithreaded, but isn't mostly one thread per tab?  If so, considering that you can only see a single tab (usually) not sure a bunch of slow CPUs is going to make for a particularly nice web experience.
 
+Bill Broadley It seems to work that way, yes. I mentioned it earlier, but I think page loading speeds would be among the slowest we've seen on Chromebooks because of the A7 cores. With a good GPU, it would smooth out pages and overall I don't think it would make a bad impression.

Chrome apps may be able to take advantage of the multicore design, as well as SIMD in JavaScript.
Add a comment...