IBM didn't want to design a power-efficient PowerPC chip. And then they switched from Intel once Intel stopped being able to design new power-efficient chips.
Well, that's not what the story was at the time. "You don't order enough" was the reasoning.
Further, nearly contemporaneously after apple M chips debuted intel released chips with P and E cores. Furthermore, Intel made lots of chips that have TDP under 10W, even 5W - multicore, even. I still use them to run HA VMs for emergency communications internet gateways.
"power-efficient" is a weird thing to claim, anyhow. What does it mean? PPC were much faster per socket than maybe even server class chips by intel, if you wanted power efficiency you could run them slower and get whatever FLOPS/J intel could give.
I am very sure i remember Apple not having a choice.
> "power-efficient" is a weird thing to claim, anyhow. What does it mean?
Cycles per Joule, or cycles per second per watt. While staying at the laptop-level available power envelope.
> PPC were much faster per socket than maybe even server class chips by intel
The absolutely top POWER chips were at the time _somewhat_ superior to Intel (and AMD), but not by much. And that superiority was achieved by raw strength, the POWER chips had a very large die with tons of additional cache, and ran at higher frequencies (i.e. more power dissipation).
However, laptop-class chips were absolutely underpowered. Intel ran circles around them. The same was true for consoles, PS3 had a puny underpowered CPU (with multiple co-processors that were supposed to make parallel tasks easier), and XBox360 was barely better than then current top desktop CPUs.
> The relationship between Apple and IBM has been rocky at times. Apple openly criticized IBM for chip delivery problems, though Big Blue said it fixed the issue. More recent concerns, which helped spur the Intel deal, included tension between Apple's desire for a wide variety of PowerPC processors and IBM's concerns about the profitability of a low-volume business, according to one source familiar with the partnership.
>IBM loses cachet with the end of the Apple partnership, but it can take consolation in that it's designing and manufacturing the Power family processors for future gaming consoles from Microsoft, Sony and Ninendo, said Clay Ryder, a Sageza Group analyst. "I would think in the sheer volume, all the stuff they're doing with the game consoles would be bigger. But anytime you lose a high-profile customer, that hurts in ways that are not quantifiable but that still hurt," Ryder said.
furthermore, if you look past the "Intel is just too slow for the power envelope" that you stated:
>The "bad quality assurance of Skylake" was responsible for Apple finally making the decision to ditch Intel and focus on its own ARM-based processors for high-performance machines. That's the claim made by outspoken former Intel principal engineer, François Piednoël. "For me this is the inflection point," says Piednoël. "This is where the Apple guys who were always contemplating to switch, they went and looked at it and said: 'Well, we've probably got to do it.' Basically the bad quality assurance of Skylake is responsible for them to actually go away from the platform."
So both of the claims about cycles/J being the reason are just sound bite ahistoria.
>The PowerXCell 8i powered super computers also dominated all of the top 6 "greenest" systems in the Green500 list, with highest MFLOPS/Watt ratio supercomputers in the world.
that's 2008 - when apple originally announced they'd complete the move to intel.
PS3 underpowered?
>According to Folding@Home in a statement on its website; “Using the Cell processor of the PS3, we should be able to do more folding than what one could do on a PC. Also, since the PS3 has a powerful GPU, the PS3 client will offer real time visualization for the first time.
Also if you compare like for like, the CPUs IBM put out in 2005 with 3.2ghz had a TDP of 75W. Intel CPUs in 2005 at 3.2GHZ used 130W TDP. both used 90nm processes.
You're fractally wrong. I could actually keep going with every claim you made and put paragraphs in. I'm sorry.
> furthermore, if you look past the "Intel is just too slow for the power envelope" that you stated:
Well, yes. Intel was not able to deliver fast mobile CPUs on time. So Apple decided to give up and do it on their own.
> PS3 underpowered?
Yes. It was terribly underpowered. I worked with it back then, and it was slower than a 1GHz Pentium for practical tasks. The CPU in PS3 was clocked at around 3GHz, but it had full in-order execution. So any code with branches just died.
Sony's way around it was SPUs - special coprocessors, that were even more underpowered with little local memory. But there were 8 of them, and the common data bus was pretty fast.
It worked great for graphics and for something computation-heavy like Folding@Home. Kinda like modern GPUs. But it sucked for general-purpose code.
> Also if you compare like for like, the CPUs IBM put out in 2005 with 3.2ghz had a TDP of 75W. Intel CPUs in 2005 at 3.2GHZ used 130W TDP. both used 90nm processes.
Now compare their performance for actual tasks. You'd be surprised.
Cell architecture bombed. As a result, Sony switched to a regular architecture for PS3.
