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

sam_dean

Suspended
Original poster
Sep 9, 2022
1,262
1,091
I first heard of the metric of performance per watt from Steve Jobs during 2005 WWDC.


This was used as a means to explain why they were moving from PowerPC 90nm to Intel 65nm processors. Also used as the reason to transition from Intel 14nm to Apple 5nm.


The timeline below was created so anyone wondering why performance per watt, process nodes like Intel's 14nm to Apple's 5nm and other business decisions Apple made in the past 2 decades allowed us Macs with Apple Silicon chips that are insanely great and the envy of x86 PC laptops.
On a side note... I regret buying any laptop in the last decade. I wish I did this instead

- 2011 MBP 13" 32nm > 2021 MBP 16" 5nm

This is ideal if your use case has not changed and you are only upgrading for preventive maintenance and after final Security Update.

I forgot the importance of Intel being stuck on 14nm from 2014-2020 to how performance per watt would impact laptops.

OLCP patcher does not impact preventative maintenance. 2022 macOS Ventura would perform like a dog on any Intel laptop that is more than a dozen years.
 
Last edited:
  • Like
Reactions: h.gilbert

ahurst

macrumors 6502
Oct 12, 2021
410
815
It sucked for a lot of other reasons, but one of the *benefits* of Intel being stuck at 14nm for so long is that it kept older hardware viable for a really long time (since the latest and greatest was never that much faster in comparison, at least in terms of single-core CPU). For example, my 2013 27" iMac is way more comfortably usable today than a 2001 desktop would have been in 2011, or a 2007 desktop would have been in 2017.

Now that the 14nm era is over, I'm curious whether ~10-year-old computers will continue to be relatively usable (since they're fast enough for most modern tasks), or whether the sudden jump in average performance will mean devs optimizing less and thus hardware showing its age faster... in other words, will a 2018 computer still handle basic workloads easily in 2028, or has that era of longevity now passed us by?
 

sam_dean

Suspended
Original poster
Sep 9, 2022
1,262
1,091
It sucked for a lot of other reasons, but one of the *benefits* of Intel being stuck at 14nm for so long is that it kept older hardware viable for a really long time (since the latest and greatest was never that much faster in comparison, at least in terms of single-core CPU). For example, my 2013 27" iMac is way more comfortably usable today than a 2001 desktop would have been in 2011, or a 2007 desktop would have been in 2017.

My first Mac was a 2000 iMac G3 DV SE 200nm. It's maximum macOS version it can use without OLCP patcher is 2005 macOS Tiger that received its last release in 2005. Useful life of 5 years.

I think I traded it in for a 2003 iBook 12" 130nm as we were coming out of the 90's that standardized 3 year replacement cycle established by GAAP. Like the iMac it's maximum macOS version it can use without OLCP patcher is 2005 macOS Tiger that received its last release in 2005. Wow! Only 2 years useful life! What a crime!

I still have the first Intel Mac, 2006 MBP 15" 65nm, that I bought the month the Apple Store at Fifth Avenue opened. I knew sales tax would be imposed but what a buying experience!

It's maximum macOS version it can use without OLCP patcher is 2009 macOS Snow Leopard that received its last release in 2011.

My mother used it until 2014 as it was used for Mail, Safari & VLC. After which it has stayed in cold storage.

Using my method of useful life it should be replaced after July 2011. So that Mac had 5.5 years of secure service. If I was aware of how awesome SSDs were and if they were cheap-ish I'd have upgraded its HDD to SSD by late 2008.

The length of macOS Security Update support is an indicator to me when its actual useful life has ended.

I also have the first redesigned Intel iMac, 2007 iMac 20" 65nm, that I also bought the month of release. It's maximum macOS version it can use without OLCP patcher is 2015 macOS El Capitan that received its last release in 2018. If I was aware of how awesome SSDs were and if they were cheap-ish I'd have upgraded its HDD to SSD by 2012.

Using my method of useful life it should be replaced after July 2018. So that Mac had 11 years of secure service.

Now that the 14nm era is over, I'm curious whether ~10-year-old computers will continue to be relatively usable (since they're fast enough for most modern tasks), or whether the sudden jump in average performance will mean devs optimizing less and thus hardware showing its age faster...
I measure usefulness of a device by these measures

- use case
- final Security Update

I use my 2012 iMac 27" 22nm to edit photos from Canon EOS EF bodies from 2008-2015. As my use case has not changed since 2015 and an upgrade from Fusion Drive to SSD was made in 2017 made the useful life of my iMac for my use case good until now.

But over half a year ago it received its final Security Update. That and it being more than 10 years starting yesterday makes me intent to replace it with hopefully a 2023 iMac 27" replacement.

Typical consumer does not look at process node at all. It isn't emphasized as a useful metric unless clock speeds, benchmarks then core counts.

I personally use the process node as a somewhat neutral and easy to understand metric of how advance a chip is. Some may argue that its nonsense but if you look at the time scale from a 1992 Intel i486DX2-66 800nm that was the CPU of a 1994 Compaq Presario CDS920 vs a 2023 A17 Bionic 3nm that will come in a future iPhone 15 Pro Max makes you appreciate how far we've come.

in other words, will a 2018 computer still handle basic workloads easily in 2028, or has that era of longevity now passed us by?
I have a 2019 MBP 16" 14nm, the 1st 16" MBP. I bought this the week of release and "luckily" bought it $400-off. I expect its maximum macOS version it can use without OLCP patcher would be 2026 macOS. It may receive its final Security Update as late as 2028.

With Apple Silicon I think the useful life may stretch out longer hence Apple not doubling RAM & SSD at current SKU's MSRP. If they did then the four year or longer typical Mac replacement cycle may double in length to 8 years or more.

That would spell bad news for Apple that sells 1 out of 2 Macs to new Mac buyers. Ideally 2 of 3 Mac buyers should be new customers while retaining all Mac users until they pass. Reason for this would to grow the Mac annual worldwide shipment by nearly 2x.
 
Last edited:

Tankmaze

macrumors 68000
Mar 7, 2012
1,707
351
This is a good recap for the timeline, appreciate the effort 👍


On a side note... I regret buying any laptop in the last decade. I wish I did this instead

- 2011 MBP 13" 32nm > 2021 MBP 16" 5nm

This is ideal if your use case has not changed and you are only upgrading for preventive maintenance and after final Security Update.

You forgot to add the mid 2015 15" MBP with the AMD R9, that is a solid MBP. The last MBP before the touchbar and butterfly keyboard fiasco.
 
  • Like
Reactions: sam_dean

sam_dean

Suspended
Original poster
Sep 9, 2022
1,262
1,091
This is a good recap for the timeline, appreciate the effort 👍
Welcome!
You forgot to add the mid 2015 15" MBP with the AMD R9, that is a solid MBP. The last MBP before the touchbar and butterfly keyboard fiasco.
That's my ideal personal ownership timeline if I could redo it. ;)

I was looking over the 2015 Mac desktops and laptops and the only SKUs that were using Intel 22nm chips were

- Mid-2015 iMac 27"
- Mid-2015 MacBook Pro 15"

All others were 1st year 14nm chips. :-(
 
  • Like
Reactions: Tankmaze

MrGunny94

macrumors 65816
Dec 3, 2016
1,148
675
Malaga, Spain
This is a great recap thanks for sharing! I gotta say the M1 Pro and M1 Max are outstanding chips, the only thing I need them in it's more 2-3h of battery life and I'm golden.

Hopefully with the M3 Pro we can see even more PPW and better battery life even further at least on the 14". I think the 16" already has a very good battery life tbh.
 
  • Like
Reactions: sam_dean

Gudi

Suspended
May 3, 2013
4,590
3,267
Berlin, Berlin
Typical consumer does not look at process node at all. It isn't emphasized as a useful metric unless clock speeds, benchmarks then core counts.
826B3FA3-B050-48D2-8899-3E88E13BDBAD.jpeg

top-left corner
 

Bodhitree

macrumors 68020
Apr 5, 2021
2,085
2,216
Netherlands
The advance of the last 20 years is not only process nodes though. Each process node frees up more space on-die for various circuits through shrinking the previous architecture, forcing an architectural reworking which takes on board various lessons that have been learnt about the way the system should work. Intel may have been stuck on 14 nm for a decade but they have made progress in performance in small 10% a year increments.

So in a way even Intel has been making progress in performance-per-watt. However the big leaps with Apple Silicon have been the big changes in the process nodes.
 

ian87w

macrumors G3
Feb 22, 2020
8,704
12,638
Indonesia
Nice recap. The only thing missing from the intel to AS announcement is Jobs' presentation style.

Also, on Windows side, there's AMD. While intel is stuck in 14nm for the longest time, AMD is gaining by number of cores and efficiency, to the point that AMD chips now are actually better in terms of efficiency. With intel still carrying premium price tag without anything to deliver, on the PC side, my choice is AMD.

Consumers will start noticing the battery life and heat (lack of), especially those that managed to experience Apple Silicon.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.