Ugh, PowerPC? Whenever I hear /anything/ about that particular architecture, my mind leaps (with a profound groan) to the old PowerMacs of the late 1990s and early 2000s. IIRC the reason that line died was part popular idiocy and part architecture limitations -- PowerPC couldn't scale its clock cycles fast enough to match x86's stuff, regardless of actual performance and comparisons thereof (to the extent that they can be made!), and people wouldn't buy them as a result, because they thought that clock speed in hundreds of MHz (and, eventually, in GHz) directly correlated with performance and capabilities the way horsepower does (for the most part) with a car, even though clock speed hasn't worked like that basically since computers stopped being the size of file cabinets... as a local friend of mine pointed out, a 1MHz 6502 can run rings around a 4MHz Z80 (and, probably, a 4.77MHz 8086!) -- and that goes back to the mid- and late-/Seventies/!

To be fair, the friend I mention is a Commodore /nut/ and thus rather biased (as am I, in the same direction, albeit somewhat less so... I've become rather more fond of the RCA CDP1802, as of late... ah, but *that* architecture was weed-baked-sloth kinds of slow...).

So from my perspective, PowerPC was a flash in the pan, and now it's basically irrelevant. Sort of like Ralph Nader, you know? He had his fifteen minutes (a couple times over, really -- first with the Chevy Corvair, and then again when he ran for President in 2000) but they're over now and these days he's basically all alone in the corner by himself because he's old news. I've not heard of anything both innovative and relevant being done with PowerPC since Apple dropped it like the radioactive potato that it was... so forgive the extreme skepticism here and now as well.

Mind you, I'm not saying that that project should dry up and blow away -- I don't think that, even for a hot minute -- but I really don't think their choice of architecture is wise, nor do I think they're going to capture even a meaningful portion of the /hobbyist/ computing market, and that's a niche kind of a segment to begin with... not to mention that popular idiocy with regard to technology is at an all-time high and climbing fast (ugh ugh ugh)...

Hey, I've been wrong before. Maybe they're /totally/ the future of computing and they start this huge revolution that completely changes everything forever and kills off x86 and Windows the way x86 and DOS knocked off CP/M and 8080/Z80-based computing. Or... not. /My/ money's on "or not", and that they either fizzle out without anyone else really noticing, or spend their efforts only to become instantly mired in irrelevancy, or (most likely) some unfortunate combination of both -- but I guess we'll just have to wait and see, to know for sure...

...good lord, I'm only 31 and I'm already turning into that overly-opinionated uncle everyone has at their holiday table that just can't shut up about, well, dang near everything... I'll pipe down for a while now. (Sorry, everybody... maybe I really do need one of those On-And-On Anon twelve-step meetings for people who just can't shut up...)

...we now return to your regularly-unscheduled programming...

Forgive my own stupidity on the subject, but what's the difference between PowerPC as implemented in the PowerMacs of yore, and this "Power9" thing you mention? I assume there *are* differences? I haven't really paid that stuff attention (mostly because I didn't think it was worth it!) in well over a decade.

...although, that does sound like it's got some promise to it...

...okay, I understood probably about 3/4 of what I read, but I am rather impressed. Sounds like they've come a ways since Apple dumped them a decade or so ago. I still think that very few members of the general public will be interested, but that's mostly because most people cling hard to what they're familiar with, and (sadly) a massive portion of that is Windows...

On Tue, Sep 12, 2017 at 12:34 AM, Luke Kenneth Casson Leighton < lkcl@lkcl.net> wrote:

---

crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68

On Tue, Sep 12, 2017 at 5:32 AM, Christopher Havel laserhawk64@gmail.com wrote:

...

Forgive my own stupidity on the subject, but what's the difference

between

...

PowerPC as implemented in the PowerMacs of yore,

probably power8.

...

and this "Power9" thing you mention? I assume there *are* differences? I haven't really paid that stuff attention (mostly because I didn't think it was worth it!) in well over a decade.

google them. wikipedia. look up Talos II

---

arm-netbook mailing list arm-netbook@lists.phcomp.co.uk http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to arm-netbook@files.phcomp.co.uk

I'm not sure if you're talking to me or Luke... if it's me, the only product line of Intel's that I'm interested in is now a ghost... namely the Atom SoCs, like the Z3735F, which they stupidly killed off because I guess they don't like all those Chinese clones of the Compute Stick or something. Dunno. Given all of those clones that are on eBay (as sticks and as so-called "MiniPC" boxes), it's not like they weren't shifting enough chips or anything... and the only thing I don't like about them (other than poor Linux support, which has mostly been rectified already) is that you can't backport them to Win7 because of the low-level interface crap that W7 doesn't have drivers for because nobody took up the bother.

...for the record, I'll use Win7 for a few things... but don't ask me to use anything newer, I don't trust it and I can't stand the look...

On Tue, Sep 12, 2017 at 06:33:13AM +0100, Luke Kenneth Casson Leighton wrote:

no the reason is: they make a specific batch of the processors, making it a large batch so that they can tune one of the wafer manufacturing machines over a period of several months to get it up to as close to a 100% yield as they can, then stockpile them.

Wasn't the power processor used in Sony's playstations a while back, with something like five or six CPUs and some other thing to coordinate them? I heard they got around the yield problem by making them all with extra processors so they could make a subset of them that happened to work available in the actual delivered machine.

-- hendrik

On Tue, Sep 12, 2017 at 5:21 PM, mike.valk@gmail.com mike.valk@gmail.com wrote:

Small defects hinder speed. Thus sell them with a lower speed rating. That's why overclocking yields different results for different cpu's the CPU's are sold batched on the lowest common.

With bigger defects they can overcome by adjusting microcode to compensate.

With the rise of multi core. Defect cores are disabled and those CPU's are sold as cpu's with less cores. Sometimes you can get lucky and can re-enable core's.

So same CPU is sold in different guises. Just according to quality and sometimes demand.

In theory this is correct, in practice, at least on the x86 area intel's desktop cpus are mostly quad cores, and on laptops all the dual cores so far have been SoCs with the chipset on the die, while the mobile quad cores were not. So it's not possible for intel to ship dual cores derived from quadcores. On gpus this is common practice, to the point that I remember some amd card that some cores were disabled via firmware which people just flashed to reenable them with varying success given that in most cases those cores were disabled for a reason.