You do realize there is a HUGE engineering penalty paid for x86 instruction backwards compatibility that ARM doesn't have, right off the bat? Frankly it's amazing what performance Intel has gotten out of their internal house of cards. What I get wistful over is what they could do if they didn't have to worry about all that literal crap in there gumming stuff up.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Macs Could Jump to ARM in 2020
- Thread starter AlphaAtlas
- Start date
You do realize there is a HUGE engineering penalty paid for x86 instruction backwards compatibility that ARM doesn't have, right off the bat? Frankly it's amazing what performance Intel has gotten out of their internal house of cards. What I get wistful over is what they could do if they didn't have to worry about all that literal crap in there gumming stuff up.
IdiotInCharge
NVIDIA SHILL
- Joined
- Jun 13, 2003
- Messages
- 14,675
Not really?
Um, Apple's been doing it for years - see bitcode: https://seekingalpha.com/article/32...ependence-and-future-proofing-the-apple-watch
Apple's already transitioned architectures completely for their desktop OS - Motorola 68K to Power PC then PowerPC to Intel x86. Internally ever since the iPhone they have supported Intel x86 and ARM concurrently. There is no one on the planet that has more real world experience at running multiple hardware architectures concurrently.
I can guarantee you, as a Mac user as long as an app is abandonware an ARM based Mac laptop wouldn't impact me in the slights. Well, other than being faster, using less power and not generating as much heat. Nothing special
Some people are so trapped and caught up in the here and now. They can't see the forest through the trees.It's called speculation...
How dare we use critical thinking with thoughts and ideas based on history and modern technological capabilities to speculate on potential future technologies.
I claimed it. So what? I linked to benchmarks also. The A11 (nevermind the A12) is within 10% of a quad core Haswell based i5. That is a fact. And in fact if I wanted to cherry pick I'm sure I could find a Haswell i5 that is in fact slower than an A11.
You are in denial. And you have nothing to back up your statements aside from conjecture and bravado. I see right through it.
I had a better article in mind but this is one of the points - backwards compatibility requires them to maintain more transistors - increases size and power use: https://www.quora.com/How-does-Instruction-Set-Architecture-affect-CPU-power
There's a few other things going on, not just size/power. If I come across the article I'm thinking of I'll post it later.
There isn't a huge penalty.
We're still waiting for proof of this bold claim. Synthetic tests on a castrated I5 fall short of it.
D
Deleted member 93354
Guest
Um, Apple's been doing it for years - see bitcode: https://seekingalpha.com/article/32...ependence-and-future-proofing-the-apple-watch
Apple's already transitioned architectures completely for their desktop OS - Motorola 68K to Power PC then PowerPC to Intel x86. Internally ever since the iPhone they have supported Intel x86 and ARM concurrently. There is no one on the planet that has more real world experience at running multiple hardware architectures concurrently.
I can guarantee you, as a Mac user as long as an app is abandonware an ARM based Mac laptop wouldn't impact me in the slights. Well, other than being faster, using less power and not generating as much heat. Nothing special
I'm not going to get into it.
IdiotInCharge
NVIDIA SHILL
- Joined
- Jun 13, 2003
- Messages
- 14,675
I'd agree that the increase in transistors/power/heat exists, but generally disagree on the scope. Pretty much every architecture decodes instructions, so the instruction set is less relevant to code running than it is to it running efficiently on a particular arch.
Feel free to come up with a better cross platform benchmark. AFAIK one does not exist.
Geekbench seems like a pretty good test to me. I'm not too sure why you get to throw direct evidence out the window and then we are all supposed to accept your conjecture as fact. That doesn't work with me. It screams of bias and subjectivity quite frankly.
Red Falcon
[H]ard DCOTM December 2023
- Joined
- May 7, 2007
- Messages
- 12,445
I thought you said you were an expert.
Really, though, I would like to know your thoughts on DocNo's post as well.
rudy
[H]F Junkie
- Joined
- Apr 4, 2004
- Messages
- 8,704
There is a massive difference between MS and apple. If apple were trying to port all phones to x86 lets say atom, that would be a big problem. A total reset of their application base. But that's not what is going on here apple is talking about converting their entire laptop / desktop line to ARM. Since apple currently holds less than a 10% market share and since apple has a notoriously accepting customer base there is much less of an issue. Combine that with the reality that there are tons of applications that aren't on macs, and a ton of applications that are on ios, you have a pretty good incentive to make it happen. You don't even have to hit top performance it just has to be passable. Afterall apple managed to not update a model for more than 5 years and their customers didn't even notice.
Windows conversion to ARM is painful because their entire strength in terms of applications is developed for win32. Second MS doesn't have a customer base that just rolls over and accepts whatever they do, look at windows 10 and that backlash, and windows 8, look at windows store. MS just isn't willing to do what it takes to force its customers into the future apple is. That is why apple can succeed where MS failed. Its not a pipe dream its pretty doable as was already pointed out they have done it a couple times before apple has the best customer base of any large tech company to take the pain.
Last edited: