[dejanh]

They develop their own chip. Its made for the OS and only for that OS and not an off the shelf offering from TI or Qualcomm.

I'm always so fascinated how people always bash on Apple hardware, yet they don't understand that Apple produces the most optimized devices around. Everything works in unison.

Posted via CB10

[Daveensannie]

Here is the latest, when will they do a native BlackBerry App to kick a$$ of those haters ?

http://browser.primatelabs.com/geekbench3/810568

Posted from my flawless BlackBerry Z30

It's behind the iphone 6 with his dualcore clocked at 1.4............. OMG :P Apple did it again it is even faster then the Galaxy S5....and the Note 4 is only slightly better with multicore, singlecore Apple is way better

Posted via CB10

[dvarnai]

I'm not quite sure if it's really faster or they just cheat. Samsung cheated before, they overclocked the cpu for the time the app was running. Also iphone hasn't got true multi tasking and they prioritize the Ui to run as smooth as it can, therefore if the geekbench app runs the benchmarks in the main thread, it may simply get more cpu power prioritized for it, while others phone would have apps running in the background as well. To properly test performance you should open 4 apps and test how well they perform in a true multi tasking environment. Oh wait... you can't do that on an iphone.

BlackBerry Q10 SQN100-3

[Daveensannie]

I'm not quite sure if it's really faster or they just cheat. Samsung cheated before, they overclocked the cpu for the time the app was running. Also iphone hasn't got true multi tasking and they prioritize the Ui to run as smooth as it can, therefore if the geekbench app runs the benchmarks in the main thread, it may simply get more cpu power prioritized for it, while others phone would have apps running in the background as well. To properly test performance you should open 4 apps and test how well they perform in a true multi tasking environment. Oh wait... you can't do that on an iphone.

BlackBerry Q10 SQN100-3

Dont try to talk things good that is not good lol admit it and accept it hahaha

Posted via CB10

[dvarnai]

Dont try to talk things good that is not good lol admit it and accept it hahaha
Posted via CB10

ok so your iphone6's cpu is faster in a synthetic test. but it can run less stuff at the same time because the OS lacks multi tasking. maybe you should also run the benchmark standalone without an OS, im sure it will be faster as well. also, im still not sure if the iphone is running any 'adjustements' while these benchmarks are running. it wouldnt be the first time a company cheats on these tests:

Note 3?s benchmarking ?adjustments? inflate scores by up to 20% | Ars Technica

edit: btw if you care about synthetic tests, how about that you have 2 gigs less memory than the passport? let me guess, youll say ios doesnt need more lmao

[Daveensannie]

Can't stop laughing lol, the so called active frames is that what you mean by multitasking? What is there so active on the frames?, when I have facebook on active frame and I re-open it, it has te reload again to see new posts, nothing active about it, same thing for IOS it just leaves it "open" but only refreshes it when you actually open it soooooo we don't need the 3Gig of ram. I just look at numbers and the numbers are saying and you can not deny that that the Iphone 6 does a fantastic job with the dualcore clocked at 1.4

Posted via CB10

[dvarnai]

Run a youtube video in the browser, it will keep playing. Pretty easy example of true multi tasking. Also you are clearly an isheep, bb10 is super smooth even without quad core etc and it doesn't require more than 1gb of ram to just run a single app either. But unlike with your iphone6 I can run more because theres free ram. So if your iphone has something super useless it's epic and awesome and revolutionary but if it lacks something every other phone has its not needed? Isheep right there. I won't be checking this thread anymore cause it's pointless

BlackBerry Q10 SQN100-3

[vader42]

Remember, the apple chip is 64 bit. A 32 bit chip needs 4 instructions to perform a 64 bit instructions. So a dual core 64 bit behaves like a quad core with twice the clockspeed (for 64 bit operations).

This would mean the performance of the a8 is similar to a 2.8ghz quad core 32 bit cpu.

For 64 bit optimised apps (ie. Benchmarks) it would be fast.

Posted via CB10

[vader42]

Forgot to mention, for 32 bit apps (if compatible) there would be no advantage unless the 64 bit chip could emulate 4 cores. You would only double the performance - so you get a quad core 1.4ghz. Not that fast.....

Posted via CB10

[Daveensannie]

Forgot to mention, for 32 bit apps (if compatible) there would be no advantage unless the 64 bit chip could emulate 4 cores. You would only double the performance - so you get a quad core 1.4ghz. Not that fast.....

Posted via CB10

Thank you for this explanation

Posted via CB10

[anon(3993749)]

Remember, the apple chip is 64 bit. A 32 bit chip needs 4 instructions to perform a 64 bit instructions. So a dual core 64 bit behaves like a quad core with twice the clockspeed (for 64 bit operations).

This would mean the performance of the a8 is similar to a 2.8ghz quad core 32 bit cpu.

For 64 bit optimised apps (ie. Benchmarks) it would be fast.

Posted via CB10

Hahaha. No. A 64 bit CPU is not twice as fast as a 32 bit CPU.

Here are the 2 main benefits of having a 64-bit CPU:

1. More bits means the CPU can do things more accurately: more decimals, larger numbers etc. (has no impact on mobile phones)
2. The CPU can address a larger number of memory locations: more than 3.1GB which is the limit for 32-bit CPUs (again no impact on mobile phones)

The A8 is fast, but not because it's 64-bit. The old 32-bit A6 was also pretty fast for a dual core SoC.


Posted via CB10

[dvarnai]

Not sure where you got 3.1gb from, the limit is 4gb but using pae you can address 64gb of memory on a 32bit cpu

BlackBerry Q10 SQN100-3

[anon(3993749)]

Pretty sure the limit is around 3GB (it varies depending on device): http://en.m.wikipedia.org/wiki/3_GB_barrier

Posted via CB10

[dvarnai]

Technically it is capable, operating systems like windows tend to reserve 512mb for the system that's why you would see 3.5gb on windows. But using pae as mentioned by your link as well, you can reach 64gb on a 32 bit cpu

BlackBerry Q10 SQN100-3

[anon(3993749)]

Technically it is capable, operating systems like windows tend to reserve 512mb for the system that's why you would see 3.5gb on windows. But using pae as mentioned by your link as well, you can reach 64gb on a 32 bit cpu

BlackBerry Q10 SQN100-3

Yes, I know it's possible using PAE, but we don't know if mobile devices support PAE. I would be willing to bet they don't yet.

Posted via CB10

[Zedi Master]

Can't stop laughing lol, the so called active frames is that what you mean by multitasking? What is there so active on the frames?, when I have facebook on active frame and I re-open it, it has te reload again to see new posts, nothing active about it, same thing for IOS it just leaves it "open" but only refreshes it when you actually open it soooooo we don't need the 3Gig of ram. I just look at numbers and the numbers are saying and you can not deny that that the Iphone 6 does a fantastic job with the dualcore clocked at 1.4

Posted via CB10

That's strange. I have 3 applications running on my Zed at the moment. One if them is Facebook. When I switch to it, it is exactly where it was. Part way through my news feed. And not reloading anything. Seems active to me. If I wait long enough, I'm sure it will show new posts when I scroll up. Multi-tasking.

Posted via CB10

[vader42]

Hahaha. No. A 64 bit CPU is not twice as fast as a 32 bit CPU.

Actually, as a lot of modern apps use 64bit arithmatic, you need at least 4 instructions to perform it on a 32 bit. For example, a 64 bit integer multiplication is (result is 128 bits):

on a 64 bit machine
----------------------------
result = a x b (each 64 bit) - depending on architecture, optimised to 1 pipeline clock (you can have an answer every clock cycle after the pipeline fills)

on a 32 bit machine
---------------------------
(<< means shifted left, or in other words the larger part of the 64 bit number)
a = a1<<32 + a2
b = b1<<32 + b2

so axb = a2*b2 + (a1*b2)<<32 + (a2*b1)<<32 + (a2*b2)<<64

Now with modern processors, you can shift and add a result, so you really only have the 4 multiplication:

clock 1: a2*b2 => register
clock 2: (a1*b2)<<32 => added to register
clock 3: (a2*b1)<<32 => added to register
clock 4: (a1*b1)<<64 => added to register

Hence, a 64 bit CPU does the multiply in one pipeline clock, the 32 bit does it in 4. If you optimise your program for 64 bit calculations, you will gain up to 4 times the performance. That is why 32 bit code has no benefit. Unless your 64 bit CPU can emulate 2 32 bit cores (like many 64 bit desktop CPUs), then it will not benefit from any speed up. So no 32 bit emulation, it is a clock for clock.

No emulation (ie. only one 32 bit instruction per 64 bit core): A8 equivalent dual core 1.4G
Emulation of 2 32 bit cores on a 64 bit core: A8 equivalent quad core 1.4G
Optimisation to 64 bit coding: 4 x speedup which is either dual core 5.6G (4 times the clock), or quad core 2.8G

There are lots of other factors, like cache, pipeline structure, memory bandwidth etc. The bottom line is that doubling the number of bits squares (4x) the performance if the app can make use of the full bit space.

Benchmarks compiled for the apple will be 64 bit optimised - hence great results. Real world 32 bit apps - not so fast

[Turnda643]

Personally, I could care less about these test. For me, when the device is in my hands and it performs as I want it to, then that's the device for me.

Plain and simple!

Subscribe to Tri County Catering and Party Idea Channel. Pin #C00123065

[Dave Bourque]

Samsung is known to cheat these tests. Why some of you think Apple wouldn't either is beyond me.

Z10STL100-3/10.2.1.3247

[Dave Bourque]

You're right. I didn't look at that one. Still, it's a good result for that SoC version.

Posted via CB10

That's because Nexus uses the bare bones Android which makes sense.

Z10STL100-3/10.2.1.3247

[app_Developer]

Pretty sure the limit is around 3GB (it varies depending on device): http://en.m.wikipedia.org/wiki/3_GB_barrier

Posted via CB10

ARM is different from x86.

ARMv8 processors (such as Apple's A7 and A8) have a sizable speed advantage from the considerable extra register space, if you have code compiled to take advantage of this. Most of Apple's frameworks and shared libraries are compiled to take advantage of this. So are most iOS apps submitted over the past year.


Sent from my iPhone using Tapatalk

[jfpoole]

Samsung is known to cheat these tests. Why some of you think Apple wouldn't either is beyond me.

Z10STL100-3/10.2.1.3247

One of the Geekbench developers here. We've been unable to find any indication that Apple manipulates benchmark scores.

[offyoutoddle]

i've always thought benchmarks are about willy waving and nothing more. To be honest if your phone does what you want it to do, doesn't appear to slow down appreciably, who cares what tricks of multitasking or suspending are going on? I'm not even sure how reliably a high number on one phone measures this against another phone.

If it can meet your needs surely that's what matters? There are times when ios suspending apps is a good thing I reckon. Certainly battery optimisation is pretty good. Some apps don't fully suspend also, certain threads are left able to run I thought also, such as music , video? Multitasking or not, hardly matters really in real terms.