Active Topics

 


Reply
Thread Tools
Posts: 735 | Thanked: 1,054 times | Joined on Jun 2010
#1
hmmmm, what do we have here, a qualcomm based cheap midrange chip sporting ARM v8 cores:

http://www.anandtech.com/show/7573/q...adreno-306-gpu

looks like a worthy successor to the qualcomm based cheap midrange chip currently in use...

what do you think; how interested is jolla in emulating the 64bit push of Tizen?

Last edited by Jedibeeftrix; 2013-12-10 at 08:47.
 
Posts: 728 | Thanked: 1,217 times | Joined on Oct 2011
#2
I always find it hard to understand what's the hype with 64bit on mobile. What does it bring to the table? Longer words? Sure... but what for? It's not like it'll make it easier to port from x86_64 or anything. More memory? Seriously? You'll run into IO issues before you hit performance issues because of lack of RAM. "Faster"? Same as before.

Somebody enlighten me please, otherwise I'd like to see efforts spent in energy efficiency, better batteries. Samsung would be doing well spending more effort in serving their customers right as well. Lots of other things to do before going 64bit in mobile...
 

The Following 8 Users Say Thank You to ggabriel For This Useful Post:
ste-phan's Avatar
Posts: 1,196 | Thanked: 2,708 times | Joined on Jan 2010 @ Hanoi
#3
Where is Tizen pushing exactly? All ten million Note 3 sold during the past 2 months were limited to Android.
Jolla should push their 32bit software platform forward where no Tizen will go.
 

The Following 2 Users Say Thank You to ste-phan For This Useful Post:
Posts: 735 | Thanked: 1,054 times | Joined on Jun 2010
#4
@ ggabriel -

http://www.anandtech.com/show/7335/t...ne-5s-review/4

"Similar to the x86-64 transition, the move to A64 comes with an increase in the number of general purpose registers. ARMv7 had 15 general purpose registers (and 1 register for the program counter), while ARMv8/A64 now has 31 that are each 64-bits wide. All 31 registers are accessible at all times. Increasing the number of architectural registers decreases register pressure and can directly impact performance. The doubling of the register space with x86-64 was responsible for up to a 10% increase in performance.

The original ARM architecture made all instructions conditional, which had a huge impact on the instruction space. The number of conditional instructions is far more limited in ARMv8/A64.

The move to ARMv8 also doubles the number of FP/NEON registers (from 16 to 32) as well as widens all of them registers to 128-bits (up from 64-bits). Support for 128-bit registers can go a long way in improving SIMD performance. Whereas simply doubling register count can provide moderate increases in performance, doubling the size of each register can be far more significant given the right workload. There are also new advanced SIMD instructions that are a part of ARMv8. Double precision SIMD FP math is now supported among other things.

ARMv8 also adds some new cryptographic instructions for hardware acceleration of AES and SHA1/SHA256 algorithms. These hardware AES/SHA instructions have the potential for huge increases in performance, just like we saw with the introduction of AES-NI on Intel CPUs a few years back. Both the new advanced SIMD instructions and AES/SHA instructions are really designed to enable a new wave of iOS apps."

@ ste-phan:

http://liliputing.com/2013/11/tizen-...hitecture.html

"In terms of 64-bit support, Tizen 3.0 will support both 64-bit ARM and Intel-based chips, which makes sense since Intel is a major backer of the project, but most phones and tablets still ship with ARM-based processors."
 

The Following 4 Users Say Thank You to Jedibeeftrix For This Useful Post:
Posts: 728 | Thanked: 1,217 times | Joined on Oct 2011
#5
Jedibeeftrix: thanks, that's a nice article. However, I still don't think that we actually need over 4GB of RAM in the mobile space any time soon. I don't even think we need more than one core, but that's another topic.

I've got an x86 laptop with 16GB of RAM. The main use of that RAM (more than 70% of it) is either disk cache or virtual machines - several virtual machines ;-) For my browser, chat, terminals, OS, widgets and the lot I definitely need <4GB, perhaps a tiny bit more. I am happy with 64bit on my laptop, don't get me wrong, and I can report that it only got stable a couple of years ago (I have everything compiled with 64bit symbols, I don't run 32bit things, except on certain VM's).

Now, I think about my N9 - 1Gb of RAM is plenty, except in rare occasions. My next phone will have 1GB of RAM as well, which will be enough for another couple of years. Double that, then double it again, it will be plenty - no - it will be more than I need unless they do something with the display+device that enables me to use my mobile phone as an actual laptop. This isn't the case today. It may be the case for some tablets, but it's pretty rare - and if you are going to have a tablet with a keyboard and everything, you may as well get a proper laptop and not be restricted to the OS.

Don't get me wrong - I like innovation, I jus think that efforts sholud be spent elsewhere. Do we want to work on 64bit to avoid the painful transition that we experienced with x86? YES!!! Do we want it RIGHT NOW? I don't think so. Would you rather have an ARM CPU that is 40% more power efficient or one that is the same as today but runs 64bit? Give me the better power management please!

As for the original topic of this thread, I appreciate that eventually we'll need more memory, I don't see the use case today, whilst in the 90s we already knew about virtual machines, we already knew about memory heavy applications and yet it took us ages to migrate to a technology that could map more RAM. Hell, we even hacked it! :-)

One last point: ARM definitely wants to get into the 64bit arena _because_ of the server business, not sure that the mobile business definitely needs it, unless there is a huge common ground where migrating to 64bit all at once makes sense (think toolchains, OS kernels and so on).
 

The Following 3 Users Say Thank You to ggabriel For This Useful Post:
Posts: 1,298 | Thanked: 2,277 times | Joined on May 2011
#6
I really hope for Intel's Silvermont (Merrifield) to emerge as a strong contender. Not only it's 64 bit, unlike sickening ARM manufacturers Intel actually makes open GPU drivers which work with Wayland. So there is hope that libhybris will be unnecessary. This ARM idiocy needs to be broken.

64 bit issues aren't limited to RAM. It allows native calculations for 64 bit integers, which are slower when they are emulated in 32 bit architectures.

Last edited by shmerl; 2013-12-10 at 17:34.
 

The Following 2 Users Say Thank You to shmerl For This Useful Post:
Posts: 728 | Thanked: 1,217 times | Joined on Oct 2011
#7
Originally Posted by shmerl View Post
64 bit issues aren't limited to RAM issues. They allow native calculations for 64 bit integers, which are slower when they are emulated in 32 bit architectures.
Yeah, good point, and some other things too... I'm still not convinced we need it right now. I think the slowness of RAM/flash outweights the speed of the CPU anyway. As an anecdote, I was playing/learning with the Raspberry Pi's assembly - and somewhere it said "oh, man, don't make a division, it's soooo expensive". Honestly, maybe it took a couple more microseconds when I tried it. Not sure we'll bump into these issues any time soon in the mobile space.

Re Intel - yeah, baby, I'd like to see some competition :-) Bring on AMD as well, although they've never been too good at low power CPU's. The best thing of this is that prices will probably drop. Intel/AMD are already 64bit capable, and I wouldn't go back for the same reason: it takes less effort to stick to what works :-)
 
Reply


 
Forum Jump


All times are GMT. The time now is 23:12.