SoundBlaster Sound Card

[JoyfulNoise]

In all fairness, the transition away from PPC was due to IBM's inability to scale down the 970's (G5's) power and heat requirements to work in a notebook. Even the fastest Power Mac G5s required liquid cooling to get around this problem - which, in 2004, was pretty exotic outside of enthusiast PC builds. By the time the transition was in full effect, such as when the dual Xeon Mac Pros were released, the Intel parts were spanking anything IBM had provided to Apple up to that point. The only place a comparable Intel Mac wasn't outperforming a G4/G5 Mac was when an application was running under Rosetta translation rather than a Universal Binary.

Prior to the Intel transition, Apple had been in contract talks with PA Semi to use their more efficient POWER-derived processors, which actually at the time offered better performance-per-watt than the earliest Core Duos, but Apple ended up with Intel in the end. A side note, Apple actually later bought PA Semi and brought them in-house to develop the A* mobile processors and the Apple Silicon stuff they derived from it.

As for TSMC, they are simply the fab. Apple designs the processors in-house and TSMC builds them. Outside of that, they have no involvement in Apple Silicon and unlike the AIM alliance, they are not co-designing the processors with Apple.

Always two sides to the story . Were you on the Apple side of the desk or the IBM side ? I can assure that the foundry could have easily reached Apple’s specs … if they were willing to pay for it

 

[dfwnoob]

By the time the Intel transition rolled around, the newest Mac I owned was an old G3 iMac DV SE I rarely used, so I had no real side in the fight. I had moved back to Wintel boxes since I was doing more gaming than anything "creative"-related. The issue truly was not money; despite multiple major revisions of the 970 - including a move to 90nm lithography - IBM could not make a power-efficient processor. When they did manage to manufacture a mobile-friendly 30w TDP G5 processor, it was basically as fast as the G4 it was intended to replace. If anything, I think IBM wanted out of the "prosumer" business and wanted to focus on their enterprise stuff which actually made $$$. Making low-yield, low-power CPUs solely for Apple couldn't have been a financially prudent plan for big blue.

Apple probably saw the writing on the wall for AIM/PPC, and had been co-developing an x86-compatible version of Mac OS dating all the way back to the earliest Mac OS X Server builds (see Rhapsody), so they were ready to move if the time came. Frankly the only thing that surprised me in the x86 transition is that Apple chose Intel and not AMD, though the rumor is that AMD could not ramp production fast enough to meet Apple's demands.

Somewhat ironically, the Intel transition is what got me back into Macs since occasionally running Windows apps is a requirement in my general workflow. After a couple of dual-boot Hackintosh builds, I ended up buying a MacBook Pro years ago and most recently an M2 Air. I have a soft spot for the 68k and PPC Macs I grew up with, but being able to run full-speed virtualized Windows apps on my Mac when I needed it was magical to me and made my Intel MBP a daily driver for many years. If you had ever suffered through VirtualPC running Windows back in the PPC days, you know the pain I felt.

 

[JoyfulNoise]

As for TSMC, they are simply the fab. Apple designs the processors in-house and TSMC builds them. Outside of that, they have no involvement in Apple Silicon and unlike the AIM alliance, they are not co-designing the processors with Apple.

This was the right point ... and the death knell of IBM Manufacturing. I got out long before IBM spun off its chip business but the era of contract manufacturing is what brought it all about. IBM focussed a lot on high quality but also high-yield - every silicon wafer that came off the line had to have most of it's real estate yielding successful product. Failures couldn't be tolerated because the server/mainframe business was too important and every wafer was too expensive to throw in the trash. Changes and new designs took forever to qualify because there was no sacrificing yield. Enter the Intel's and TSMC's of the industry and their volumes were through the roof. So if they had a wafer with a 20% yield ... who cares, just crank out more. There was just no way to compete with that States-side.

I think Apple made the right call when they jettisoned Intel and started doing their own chip designs and custom manufacturing. Their current line of computers, especially the M3 I'm currently typing away on, are fabulous ... and no liquid cooling necessary (unless you want to have a glycol pump running on your work desk ...).