@WolfpackN64

$59, huh?
Good price, pity it isn't clocked higher.

_________________
Amiga! "Our appeal has become more selective"

@Beans

Well, it's a microcontroller. They also have a core IP for heavier embedded and consumer use, but they haven't produced a chip of that IP yet.

This is comparable to the ARM Cortex-M MCU's.

@WolfpackN64

Good comparison, and fast for a microcontroller.

_________________
Amiga! "Our appeal has become more selective"

@Beans

Certainly considering it's still fabricated on 180nm.
If AOS4 EVER is going to do an ISA shift, it'd better be to RISC-V. Because an open platform just gives you a lot more flexibility and security.

@WolfpackN64

Well, if OS4 were to transition to another ISA, it would make sense not to jump on the X64 bandwagon.
AROS is already there, and the MorphOS development team has announced it as a target.
An open platform like Power8/9 or RISC V makes sense.

But PPC still has some time left before its eol.
That would give some time for the 64bit variant of RISC V to mature.

_________________
Amiga! "Our appeal has become more selective"

@Beans

In that time anything can happen, and no one is in a hurry in a Amigaland.

@Beans

Quote:

Well, if OS4 were to transition to another ISA, it would make sense not to jump on the X64 bandwagon. AROS is already there, and the MorphOS development team has announced it as a target. An open platform like Power8/9 or RISC V makes sense.

So the presence of two inconsequential offerings would dictate processor support?

How about learning from this whole PPC debacle and going platform agnostic instead of fantasizing about niche market chips?

@WolfpackN64

Quote:

WolfpackN64 wrote: @cdimauro Oh Intel is twisting and turning in all kind of directions with Ryzen now. it's pathetic.

"Business is war" (cit.)

@Beans

Quote:

Beans wrote: @BigD Intel? Who care? Never liked them, really hated the company when the P4s were being made, enjoying AMD's push to compete. Threadripper cpus will be available months before Intel introduces a product in response to them.

Intel introduced Skylake-X before AMD's Threadripper. Only for high-end models, they will introduced one month later.

Anyway, I don't see the problem: competition is good.

@TRIPOS

Quote:

TRIPOS wrote: @cdimauro They bought PA Semi for its human resources, that later brought them their first in-house custom ARM core (of desktop class design), and which has improved a lot over the years. I'm confident that Apple could (should they want to) build CPU's for desktops and laptops based on their current ARM core design. Package a few more cores in the chip, some more controllers like PCIe etc, soup up the clock to some 3+ GHz with active cooling, and they would suddenly be controlling their entire eco system in-house! Microsoft is going ARM. And not some limited "creators edition" version for Raspberry Pi or some "Windows CE", no, the full consumer desktop version! And it will come with a x86->ARM JIT for existing x86 apps: https://youtu.be/A_GlGglbu1U Please note that this is a standard "old" processor, and not one of the new more souped-up ones, like for example (one of many): Broadcom BCM58800 8 x A72 cores at 3GHz! 2MB L2 per core 8MB L3 shared 3 x DDR4-2400 controllers PCIe 3 x16 USB 3.0 8 x SATA 3 100Gb Ethernet Crypto and RAID support.

Let's see to concrete benchmarks when such ARM-based device arrives.

Anyway, Microsoft is not planning to abandon Intel.

Last edited by cdimauro on 07-Aug-2017 at 05:47 AM.
Last edited by cdimauro on 07-Aug-2017 at 05:46 AM.
Last edited by cdimauro on 07-Aug-2017 at 05:46 AM.

@WolfpackN64

Quote:

WolfpackN64 wrote: @BigD Moral of the story, we should all just transition to RISC-V.

@WolfpackN64

Quote:

WolfpackN64 wrote: @Beans Certainly considering it's still fabricated on 180nm. If AOS4 EVER is going to do an ISA shift, it'd better be to RISC-V. Because an open platform just gives you a lot more flexibility and security.

@Beans

Quote:

Beans wrote: @WolfpackN64 Well, if OS4 were to transition to another ISA, it would make sense not to jump on the X64 bandwagon. AROS is already there, and the MorphOS development team has announced it as a target. An open platform like Power8/9 or RISC V makes sense. But PPC still has some time left before its eol. That would give some time for the 64bit variant of RISC V to mature.

Or switch to my ISA

Which is a mixture of x86/x64 and our beloved 68Ks: "only" with 32GP registers, 64/128 SIMD registers, many addressing modes, and state-of-the-art SIMD unit. 32 & 64-bit of course, and... no prefixes or double memory indirect modes.

And it's a CISC, of course, which can give points to the most compact ISAs in terms of code density.

Last edited by cdimauro on 07-Aug-2017 at 05:49 AM.
Last edited by cdimauro on 07-Aug-2017 at 05:48 AM.

@ne_one

Platform agnostic just means supporting as many ISA's as possible. Seeing how even AROS is struggeling to keep ARM support, that seems highly unlikely for the Amiga(like) systems.

@WolfpackN64

Quote:

Seeing how even AROS is struggeling to keep ARM support,

Struggling only because no-one is working on ARM at the moment, not because of the architecture itself.

@michalsc

That's my point. Being platform agnostic is easy to say, but maintaining several ISA's at once is a lot of work and even some larger Linux distro's don't maintain ALL ISA's. It's not a knock against ARM or AROS.

@cdimauro

Quote:

Intel introduced Skylake-X before AMD's Threadripper. Only for high-end models, they will introduced one month later.

Actually, neither company has 'introduced' anything, but they do appear to be willing to offer Amiga-like pre-orders.
The multi-die solutions AMD is offering are a bit cruder, but there's no doubt that their Ryzen cpu set this current development war off, and the competition IS a good thing.

@ne_one

Quote:

How about learning from this whole PPC debacle...

And what 'debacle' would that be?
Look, I've always favored alternatives.

When Amiga initially shifted to PPC, it was the apparent direction that other 68K users like Apple and the 68K's manufacturer Motorola were headed.
And the processors held their own for quite some time.
If IBM had gotten the 970 to scale higher, Apple might have continued development longer.

When it was available, the G5 was entirely competitive with X86, in fact with a better fpu and early adoption of 64 bit, it was actually ahead of the competition.

So, 'debacle'? Hardly. We still rarely see ARM cpus clocking higher, and the massive development dumped into X64 is a bit ridiculous. Even Intel didn't originally think this ISA was the way forward.
Its a bit like 'jacking up the car and driving a new car under it'.

And the argument about code density? A bit disingenuous, considering how massive code has become. Its not like size matter anymore when I load video card driver packages in the 50MB range (that, btw, wa for an older package I downloaded from AMD yesterday).

If YOU want mainstream, goes use Windows or OSX, they certainly have distinct advantages over Amigoid OS'.

In the meanwhile, I'm enjoying the distinction of having a different platform. When we were first discussing an ISA shift, I favored ARM, as at least it was licensable.
Again, I've never liked Intel, and never thought it should have become the default standard.
In that regard, the 68K was significantly better when the IBM PC was introduced.
And I hope we hold out as long as possible before becoming 'assimilated'.

Last edited by Beans on 07-Aug-2017 at 10:48 AM.
Last edited by Beans on 07-Aug-2017 at 10:48 AM.

_________________
Amiga! "Our appeal has become more selective"

@Beans

Quote:

And I hope we hold out as long as possible before becoming 'assimilated'.

Use Intel, we'll make it worth your while



They've hardly set the mobile world alight!

_________________
"Art challenges technology. Technology inspires the art."
John Lasseter, Co-Founder of Pixar Animation Studios

@Beans

Quote:

@ne_oneQuote: How about learning from this whole PPC debacle... And what 'debacle' would that be? Look, I've always favored alternatives. ...When it was available, the G5 was entirely competitive with X86, in fact with a better fpu and early adoption of 64 bit, it was actually ahead of the competition.

I'm curious if you agree with my impression of how things played out. Would you mind evaluating this explanation?

PowerPC failed to succeed because of the following factors:

1) Apple was badly mismanaged during the 90s, i.e. when PowerPC was launched.

2) Mass adoption of Windows, which is tied to the x86 architecture, and especially of Microsoft Office, whose Windows version was incompatible with the Mac version (even until recently, if not still).

3) Software lock-in leads to hardware lock-in. Everyone needs a computer for business, and Apple wasn't seen as a serious computer for business. By the end of the century, Apple looked dead, and the newly-returned Jobs' candy-colored iBooks and iMacs looked ridiculous.

4) Neither Freescale nor IBM cared that much for the desktop market. Freescale was hugely successful in the embedded market, and IBM was hugely successful in the server market (hence the strange abandonment of the Cell processor).

5) The Megahertz myth.

6) Universities increasingly taught assembly programming & CPU engineering using the cheap, ubiquitous, and readily-available x86 architecture, rather than other, superior architectures. (A CS colleague at my university says this is why he teaches assembly w/x86 rather than 68K, which he considers much more elegant.)

7) Combine (1)-(6) and you end up with Intel so flush with profits from their immense economies of scale that they can invest hugely into R&D. The rise of AMD and Cyrix gives them motivation to do so. That makes it impossible for PowerPC chips to maintain their initial superiority, or even to maintain parity. That R&D enabled Intel to diversify x86 to the point where they could undermine IBM's dominance of the server market & Freescale's dominance of the embedded market, and Intel now really is everywhere.

...except mobile. The reason ARM owns that is that no one -- not Intel, not Freescale, and certainly not IBM! -- offered a chip so well-suited for the mobile device market, and ARM didn't mind acting as a "design house" that licensed their designs to, uh, Intel, Freescale, and Apple, who now offer their own versions of ARM.

So this is my general impression, and I'm curious how accurate it is.

_________________
I've decided to follow an awful lot of people I respect and leave AmigaWorld. If for some reason you want to talk to me, it shouldn't take much effort to find me.

@simplex

Mostly, but Intel completely blundered into the embedded market, even when paying companies to adopt their chips. In the general server market, Intel still reigns supreme, yet they're becoming more squeezed by ARM on the low end (even though their Atom low powered server chips were good, they had a massive bug that basically destroyed the servers recently which hurt Intel) and POWER on the high end. IBM holds on to Big Iron for dear life and especially in financial calculations and exascale computing, the POWER architecture absolutely demolishes Intel's architecture.

Intel still spends a massive amount on R&D, but it seems they've gotten stuck in a bureaucratic mess. They simply fail to decently scale the core architecture up or down enough to break into markets other then desktops and normal servers. While AMD with only a fraction of R&D spending has a more solid footing in the embedded market and is putting pressure in the server and desktop space.

On the embedded end, ARM reigns supreme, but for networking and especially for military use, PowerPC is still firmly entrenched despite a seeming lack of progress. MIPS is dying and RISC-V is still a wildcard at this point, but could threaten ARM in the long run.

The semiconductor market is in a constant state of flux so it's hard to predict. Add Intel's AME and AMD's PSP and you have a good argument as a consumer for just ditching x86 whenever possible.

And LibreOffice is better then MS Office anyway

@cdimauro

Quote:

But PCWorld reported tests and concrete data that even my kid can look and judge how was the situation.

Maximum PC did multiple benchmarks too, with similar results as PCWorld... only their interpretation was different.

Quote:

and an historian should know it.

Historian should read and judge sources, not censure them, if he doesn´t like, what they say.

Maximum PC is really good source for PC point of view. PC Magazine (28 October 2003, First Looks, p. 38) shares its conclusion, InfoWorld too (5 January 2004, p. 24).

Quote:

In fact, I prefer facts.

Fact is, even PC magazines admited it was - after long time - the first CPU in Apple computers competitive to x86.


For an historian, important is not only what really happened (what CPU was faster in this regard), but also how, how this "reality" was perceived by contemporaries and posterity (what reported magazines of the day about G5).

@simplex

Sounds pretty spot on.
If IBM had known Intel was going to eat this far into their server market, they might have put up a better fight.

And curiously, having competition from AMD and Cyrix actual strengthened Intel as it encouraged them to invest in R&D, and the additional X86 cpus on the market helped overwhelm the competition on pure numbers.
Of course those number encouraged software developers to focus on the X86 market, to the exclusions of all others, reinforcing the market for X86 cpus, because if you wanted a particular software package you were usually limited to X86.

Frankly, considering the rather poor start, (in the 8088/8086 family), X86 has advanced farther than I ever would have thought possible.

Then again, its RISC under the hood these days anyway.

_________________
Amiga! "Our appeal has become more selective"

@simplex

It failed because IBM was not able to deliver a power efficient G5 for laptop so Apple ditch them.
The rest is history.
Kamelito