Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

iphoneclassic

macrumors 6502
Oct 12, 2011
375
7
USA
Please check what this guy posted:



The difference in performance: 8500/1700 = (around) 5 times
The difference in power consumption: around 20 times (even with low-end i5-2410M)

Result: ARM wins.

There are few fundamental omissions in the math.

1) No processor architecture is indefinitely scalable. ARM may not scale 5 times. Otherwise no chip designer ever will change the architecture.

2) Power consumption requirement growth is not linear. Requirement grows exponentially.

3) Ignored another major fact. Heat. Most chip designs are scaled back because there is no easy way to dissipate heat.
 

Renzatic

Suspended
Not a linear relationship: probably, ARM result would be better 3x-5x times, compared to Intel,
depend of which processors we compare. In this specific case, it was 4x.

What about Haswell, which can produce i5-i7 benchmark results, and power a (finally) decent embedded GPU on only 8w?

Your argument only works if, like dude said above, power vs. consumption is linear, and Intel doesn't improve their chip designs at all.
 

slughead

macrumors 68040
Apr 28, 2004
3,107
237
No, your wrist watch is not a general-purpose computer. :p

Neither is a freaking iPad. With a Geekbench score worse than a freaking PowerMac G5 (2004), it is woefully inadequate for desktop/laptop use by todays standards.

By your faulty calculations ignoring the relationship between flop/watt, my wrist watch could have the best processor OF ALL!
 

KnightWRX

macrumors Pentium
Jan 28, 2009
15,046
4
Quebec, Canada
In case of i5-2410M, it is 212.5 GeekBench Points per Watt.
In case of ARM, it is 850 GeekBench Points per Watt - four times better.

Where did you get your A6X TDP figures to compare from exactly as no such stat has been released ? :rolleyes:

(let me guess, your lower region).

Assuming a TDP of 17 Watts for the ULV Intel chips used in the MacBook Air (Intel Core i5-3427U) also whups your 212.5 figure by close to 2 times. So frankly, your numbers mean crap, you're cherry picking and have no definite proof of what you claim.
 

mixel

macrumors 68000
Jan 12, 2006
1,729
976
Leeds, UK
Maybe eventually. I'd rather see a separate additional arm CPU to run a sort of dashboard type layer, and maybe give the option to boot one without the other. Saves power, lets you use both app stores, doesn't interfere with power stuff.

I'm imagining Adobe doing a massive facepalm if the architecture changes again. I don't see it happening at all unless apple can really make arm chips cheap enough and fast enough to run th i86 software fast enough in Rosetta style.
 

KnightWRX

macrumors Pentium
Jan 28, 2009
15,046
4
Quebec, Canada
Everyone is assuming either x86 or ARM. But why cannot Apple use both? There are certainly issues in using both, but that is where innovation comes in :D

That would be a tremendous feat and would require a way to reboot to a different kernel, restart all user space applications using completely different binaries.

That, or run the entire system in emulation and just dynamically switch the emulator, which would come with its own performance hit.
 

Sincci

macrumors 6502
Aug 17, 2011
284
65
Finland
Windows 8 does not have an ARM version. People really need to learn the difference between Windows 8 and Windows RT. I've seen this come up so often it's amazing how much Microsoft has managed to confuse people about this.

People should also try to understand that Microsoft doesn't sell Windows RT anywhere. They only sell licenses to companies that are making Windows RT tablets/laptops. Due to that reason there's absolutely no way that you could ever run Windows RT on your ARM-based Mac since Apple won't license Windows RT from MS (and MS most likely wouldn't even sell it since they wants Windows RT to be the only operating system on those devices).
 

KnightWRX

macrumors Pentium
Jan 28, 2009
15,046
4
Quebec, Canada
Not a linear relationship: probably, ARM result would be better 3x-5x times, compared to Intel,
depend of which processors we compare. In this specific case, it was 4x.

If ARM could snap its fingers and produce chips that are 4x more efficient than Intel's while providing the same performance numbers, you'd have bunch more ARM laptops on the market right now.

----------

Maybe eventually. I'd rather see a separate additional arm CPU to run a sort of dashboard type layer, and maybe give the option to boot one without the other. Saves power, lets you use both app stores, doesn't interfere with power stuff.

Lenovo did that... 2 years ago. The Lenovo IdeaPad U1.
 

AppleMacFinder

macrumors 6502a
Dec 7, 2009
796
152
There are few fundamental omissions in the math.

1) No processor architecture is indefinitely scalable. ARM may not scale 5 times. Otherwise no chip designer ever will change the architecture.

2) Power consumption requirement growth is not linear. Requirement grows exponentially.

3) Ignored another major fact. Heat. Most chip designs are scaled back because there is no easy way to dissipate heat.

If ARM could snap its fingers and produce chips that are 4x more efficient than Intel's while providing the same performance numbers, you'd have bunch more ARM laptops on the market right now.

Since ARM CPUs have such a low power consumption, it is possible to put several of them to one machine. So, it's still valid.
 

KnightWRX

macrumors Pentium
Jan 28, 2009
15,046
4
Quebec, Canada
This is all part of Steve's 10 year plan.

We're late in the 16th year of that plan btw :

If I were running Apple, I would milk the Macintosh for all it's worth — and get busy on the next great thing. The PC wars are over. Done. Microsoft won a long time ago.

As quoted in Fortune (19 February 1996)

We're going to be celebrating the 17th anniversary of that plan in about 3 months. ;)

----------

Since ARM CPUs have such a low power consumption, it is possible to put several of them to one machine. So, it's still valid.

Parallele processing does not scale the same way. It's not valid at all. 32 CPU cores cannot replace 1 fast CPU core.

I haven't repeated myself as often as I have in this thread ever. You're either not very knowledgeable about the topic or you're just baiting for responses now. Either way, you're a big waste of everyone's time.
 

akbarali.ch

macrumors 6502a
May 4, 2011
801
681
Mumbai (India)
So....is there going to be a Mac OSX RT then? I see people getting mad their current programs won't run.

----------



Isn't Windows RT the ARM version of Windows?

i wasn't 100% sure of it at the time of writing, thanks for clarifying. So in that case it doesn't matter whats under the hood, till everything runs smoothly. isn't it?
 

KnightWRX

macrumors Pentium
Jan 28, 2009
15,046
4
Quebec, Canada

AppleMacFinder

macrumors 6502a
Dec 7, 2009
796
152
What about Haswell, which can produce i5-i7 benchmark results, and power a (finally) decent embedded GPU on only 8w?

What about ARM v8, which would be released the same time? Especially the newer Cortex CPUs?

Comparing future Intel architecture with existing ARM architecture is not fair. :rolleyes:
 

iRCL

macrumors 6502
Nov 2, 2011
284
0
Why not?

- Macs aren't even using the desktop line of x86 Intel chips
- ANYTHING written adhering to Apple's coding guidance, using toolkits like Accelerate, etc, should be platform independent.
- I dare say the PPC/Intel crossover disaster had a lot more to do with vendors not wanting to support Mac than anything else. @work we port VERY LOW LEVEL (hardware interfacing) code between PPC and Intel all the time and it's not a big deal. There's a lot more support now due to all the growth, I think you would most certainly see major apps ported right away (even Office, which already will be running on ARM..)
- Plenty of people are already coding for ARM platform on iOS -- browse issues of these developers, and you will never see an issue that pops up because of the architecture. Whether you're on Intel or ARM it is basically transparent
- Windows 8 supports ARM so bootcamp is still viable (you're going to see a shift toward ARM anyway in the next 3-5 years)
- Esp. as process gets better i.e. 10nm it's getting easier to get much higher performance, even at the point we're at now there's no real need for more processing power boost even from 1.7ghz MBA unless you're doing something silly like gaming. Even then, it's very capable. ARM chips will surpass that performance in the not too distant future
- So many "PROS" talk about how they won't be able to do their job with less performance -- face facts, what is so high performance in Apple's lineup right now? (Very) old XEON chips? Laptop grade chips in iMac/MacBook/Mini? If you're a "PRO" and are still on the Mac platform and haven't already woken up and smelled the coffee, you better get on it, and switch
 

patent10021

macrumors 68040
Apr 23, 2004
3,505
794
Uh Apple did once care that you could run windows on a Mac. Do you not recall that it was actually something that they marketed? It seems as though you weren't around for the PPC to Intel shift and all the hype and excitement about being able to run windows on a Mac.
I was there. Lead, follow or get out of the way.
 

slughead

macrumors 68040
Apr 28, 2004
3,107
237
Parallele processing does not scale the same way. It's not valid at all. 32 CPU cores cannot replace 1 fast CPU core.

I haven't repeated myself as often as I have in this thread ever. You're either not very knowledgeable about the topic or you're just baiting for responses now. Either way, you're a big waste of everyone's time.

Quoted for truth. Keep up the good work; don't let the uninformed fanboys win!
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.