@pavlor

Quote:

pavlor wrote: @Kremlar Quote: Sounds like a single core 3.46 P4 released in 2004 would blow it out of the water, especially since AmigaOS does not support dual core processors. 3.4E GHz Pentium 4 (single core) 1484 SpecInt2000 1492 SpecFp2000 TDP: above 100W (not counting needed chipset on motherboard of course) Next try?

Current Intel Xeon E7-8870 (10 SB core) @ 2.4Ghz has 130 watts TDP i.e. ~13 watts for each SB core.

Last edited by Hammer on 21-Apr-2011 at 05:09 AM.
Last edited by Hammer on 21-Apr-2011 at 05:08 AM.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A, KS 3.2, PiStorm/RPi 3a/Emu68)

@jingof

Quote:

jingof wrote: @Forcie-NatamiTeam Quote: For one, you can claim real Amiga cred there -- doing things the Amiga way. You're not just a PC with a PowerPC CPU and some VGA chip trying to make the claim of being an Amiga. This logic is just plain wrong - and surprising coming from a Sage of Amiga lore. I've heard this many times before. That "next generation" Amiga's aren't really Amiga's simply because they are so different from the classics. Or because it shares components with 'evil' PCs. If that logic holds, then today's Macs are not TRUELY Macs because they are so different from classic Macs. Heck, if Dave Haynie and the original Amiga team were still producing new Amigas, they would likely be unrecognizably different by now. Distance from the classics cannot be the measure of "Amiganess." Otherwise, a "modern" future for this platform (or any other) is an impossibility. Evolve or Die.

At last something worth reading... exactly! Exactly!

And these very same people who go on about an Amiga not being an Amiga because it hasn't got custom chips turn around and go, who needs to use floppy disks anymore? Yes, I'm pointing at you! LOL

The Amigas custom chips were just a design you know and each chip did something cool for the system. Eventually all the chips would have been phased out just like RTG as the years went by...

I believe most people who truely believe this notion are simply finding a justification to create there own closure on the Amiga... it suits them for whatever reason not to pursue it any longer and that's fine but FFS don't run the modern movement down... It ain't your ride anymore so bow our gracefully eh?

Like I've preached for many years now, I have never left the Amiga, I have masses of old Amiga models and have 3 newer systems soon to be 4 and then 5 actually since I need to get a 460 for my collection too...

I've never stopped enjoying computing, you don't see me post day after day rising issues do you? When I get my X1000 I'll start the new road of programming and I tell you, that's very exciting Instead of just mucking around compiling I'm going to actually be able to understand what things are doing and directly change...!

Take a leaf out of my book, accept what you are and be happy and leave others alone.

@Jingof well done! I was beginning to worry I was one of the few with this kind of insight left on these boards... as you see, I don't post very often anymore...

_________________
~Yes I am a Kiwi, No, I did not appear as an extra in 'Lord of the Rings'~
1x AmigaOne X5000 2.0GHz 2gM RadeonR9280X AOS4.x
3x AmigaOne X1000 1.8GHz 2gM RadeonHD7970 AOS4.x

Deleted - duplicate post

Last edited by gregthecanuck on 21-Apr-2011 at 05:44 AM.

Quote:

@Hammer Current Intel Xeon E7-8870 (10 SB core) @ 2.4Ghz has 130 watts TDP i.e. ~13 watts for each SB core.

That E7-8870 on Intel's current price list will set you back a cool $4,616.00.

Also of note is that Intel has barely changed their pricing for ages. With AMD not really competing in the mid-to-high-end segments like they once did the Intel pricing has remained stagnant. No surprise Intel's profits are up lately, even with PC shipments down.

@gregthecanuck

Quote:

gregthecanuck wrote: Quote: @Hammer Current Intel Xeon E7-8870 (10 SB core) @ 2.4Ghz has 130 watts TDP i.e. ~13 watts for each SB core. That E7-8870 on Intel's current price list will set you back a cool $4,616.00. Also of note is that Intel has barely changed their pricing for ages. With AMD not really competing in the mid-to-high-end segments like they once did the Intel pricing has remained stagnant. No surprise Intel's profits are up lately, even with PC shipments down.

E7-8870 was launched in the month April i.e. in Q2 2011 not Q1 2011 results.

_________________
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB
Amiga 1200 (Rev 1D1, KS 3.2, PiStorm32lite/RPi 4B 4GB/Emu68)
Amiga 500 (Rev 6A, KS 3.2, PiStorm/RPi 3a/Emu68)

Quote:

More importantly, the MorphOS guys have always vehemently denied that they stole any code.

The MorphOS project came from Phase V... after Phase V ended, former employees started up BPlan (eg, "Plan B"), who eventually merged with a couple folks from Thendic France to form Genesi.

Now, I do not have direct personal knowledge of all MorphOS sources. But back when Phase 5 was working on their "C Exec" and other things, starting to re-create AmigaOS themselves in the mid-1990s, I was working with Andy Finkel at Amiga Technologies. The Phase 5 guys were really after AT to use tome of their stuff (and pay them, natch). Andy did a code review of the C Kernel, and found it was copied from AmigaOS source code. In fact, even the comments were copied, assembler to C.

Maybe MorphOS is clean, maybe not. Maybe no one actually knows. But that's such a transgression, I wouldn't trust anyone involved in Phase 5, or any code that can be traced back to Phase 5.

@itix
They still don't understand clean room development. If you have seen the Amiga source code, you cannot produce a legally separate work-alike. So any copied comments are absolute proof that the code is dirty. And they're not rejecting my claim, if you go back into those linked documents, that the comments were copied.

Point in fact -- I just don't care about MorphOS. It's not AmigaOS, it might as well be Windows for all I care. If you like it, I'm pretty certain at this point no new legal entanglements are going to happen. If the MorphOS people would like to swear in public that not a line of code or comment is copied from the AmigaOS sources or derived from the Phase 5 code (fruits of a poisonous tree, in legal terms), I will not mention MorphOS again.

Quote:

COBRA wrote: @WolfToTheMoon Quote: x86 on commodity hardware would bring in more hobbyists and more devs Would it? Why would someone even remotely consider using and developing for an OS on their x86 hardware that has no memory protection, supports only a handful of old gfx cards with limited or no 3D support with old outdated APIs and hardly any software of interest and it even costs quite a bit of money around 100 EUR, when there are plenty of alternatives with far more capability, many of which are free?

It would, simply because anyone who wanted to run AmigaOS could do so on a $200 piece of hardware. And perhaps, it AmigaOS was actually selling to anyone in non-trivial numbers, TPTB might have a little money for updating it to work on actual 1990s-vintage hardware.

In short, precisely the same reasons exist to run AmigaOS on x86 or ARM as any other platform, but the barrier to entry is significantly less.

Quote:

AROS is available for x86 and it doesn't look like it's winning over the PC developer community.

Nothing like this happens overnight. And the very fact there have been so many different Amiga and Amiga-like projects means that many efforts have been repeatedly duplicated. Sure, this happens... BSD vs. Linux, for example.

The fact is, people do run alternative OSs on x86 hardware all the time. There was a time when Linux was smaller than AmigaOS, as a community. Now it's on more machines than Windows. So it is possible for these things to grow. As it stands todays, the AmigaOS community is shrinking... people give up and quit every year. New ones rarely if ever join in and take their place.

MacOS is a religion. That's actually a pretty good thing for Apple.

There is no cheap Mac hardware, it's all overpriced. Ok, maybe if you buy something used and ancient... PPC Macs haven't been made in 6-7 years. And what you get for that expense is MacOS. If you're not interested in MacOS, you don't buy a Mac. If you're doing serious code development, you don't use an ancient computer, either.

@itix

Oh... and by-the-way, comments ARE part of the copyright of a code. If you copy the comments, you have infringed on the copyright.

@Hammer

Quote:

Current Intel Xeon E7-8870 (10 SB core) @ 2.4Ghz has 130 watts TDP i.e. ~13 watts for each SB core.

Our discussion with Kremlar was about year 2004 technology... I don´t think Xeon E7-8870 was available in 2004.

Quote:

This CPU was only used in aerospace and military applications

Many of which typically use very underpowered CPUs that have some other figure of merit (rad hardened, for example).

Quote:

we have a few benchmarks VS G5, made by an aerospace company using optimized code which resulted in the PA6T outperforming the G5 by several orders of magnitude (and by extension also x86 CPUs younger than 2004) but I don't mention it because it tells us nothing at all.

I for one would be quite interested in seeing the PA6T performance 100x or more the speed of a PPC970. THAT would be something to brag about. I'm pretty sure an i7 can't claim such performance.

@redrumloa

Welcome back!

Wow, that paper is from pre-Amino time. Not much changed since then in terms of missed deadlines.

@hazydave


Quote:

I for one would be quite interested in seeing the PA6T performance 100x or more the speed of a PPC970.

DAX only misintepreted one of the benchmarks (source code for PA6T was much more optimised). These benchmarks show that performance of PA6T is comparable to 970FX. It was nice CPU in its time (2007/2008) and one of the most powerful (maybe the best) PowerPC CPUs even today.

@KingKong

Quote:

Interesting presentations, thanks. These show that the PA6T is very usable for military, industry, office and of cource home-PC.

I'm somewhat in agreement. As I mention, even the QorIQ chips are ever-so-slightly faster per core than a modern ARM (at least until the A15 is out), and that may fast enough for some kinds of home uses. It's not a high end gamer's CPU, and it's not for people doing the modern version of the Amiga's classic strengths in multimedia work. But just look at home much computing folks are getting done on iPads and other tablets these days.

Quote:

How much faster are current CPU? AmigaOS can easily regain factor 2-3 and even more in comparison to MSwindows (just guessing, ask the AmigaOS-profis).

Not really, when you actually need the performance. But some of it's just psychological, anyway. Like your desktop performance. Windows has some amazingly bad ideas still in it, which do tend to make things clunky. I have a Tegra 2 tablet running Android, and it's plenty fast for the stuff it's doing, based on dual 1GHz ARM Cortex A9s. The average Intel Atom is about as fast as the A9. Most netbooks are running 1.6GHz or faster, and at least some are dual core. And yet, they crawl under many Windows applications. And much of that crawling is just user impatience for the OS getting around to them.

Intuition does all kinds of magic to speed up the user interface in very effective ways.

Usability, clear design, endurance (works still in 10 years), security/stability, ... such things are much more important than some percent extra performance. One single bug (in hardware or software) can null the best performance.

Quote:

BTW: be careful of too small structures (65 nm is fine, below 45 nm it may become too risky) - only guessing ... what does hazydave think?

I think the move to 32 and 28nm is going along just dandy. It's not quite to the many fabless companies yet, but Intel and AMD have been shipping in 32nm; Intel since late 2009 I think.

They've had issues with leakage current becoming dominant at some Leff nodes, but so far there have been ways around this and other limits: copper interconnects, stressed silicon, more exotic substrates. And the fact is, at each new node, you have fewer foundries. Commodore was left behind in the 80s due to the costs of new IC process development. Then it was only large companies hitting the new nodes, and now it's either giants, like Intel or Samsung (#1 and #2), or groups of companies working together.

@-pekr-
Quote:

-pekr- wrote: The problem with the long wait for the new HW is always the same - you are trumped by the next release of the "competition". X1000 has XMOS. I wonder, how will it compare to - Tabula : http://www.engadget.com/2011/04/19/tabula-scores-108-million-to-bring-cheap-programmable-chips-to/

The XMOS chip is just a network of little CPUs... it's not this wonder chip some people think it is. It may have some interesting applications in embedded control. Maybe. I did a robot architecture once that used lots of small CPUs to "cook" things like sensor data. But that was only practical because these particular CPUs (Cypress PSoC) had programmable macroblocks in both analog and digital, which, while not FPGA-like, definitely let you do things that seems to be much like custom hardware (for example, you could do SPI, UART, or I2C in digital, a DAC, an ADC, an amplifier or comparator, etc. in the same analog blocks).

The Tabula looks, at least from the EE Times article I read, to be an actual FPGA, but with some cycle-by-cycle switching of the presumably SRAM-based control array. That would, in theory, let you very quickly sub one large block for another.

For example, take a modern radio . You often have Wifi and Bluetooth in the same device, and the radio's only half duplex (new ones are MIMO, but still half duplex). So you have BT-TX, BT-RX, Wifi-TX, and Wifi-RX blocks in the design, but only one is actually doing real work at any given time. So this would let you design the radio using only 1/4 the logic blocks, normal amount of SRAM or Flash. But the memory part is cheap in an FPGA, it's the logic blocks that you're paying for.

Anyway, that's my take on it... sounds fairly cool.

Quote:

It's not AmigaOS, it might as well be Windows for all I care.

Have you ever used MorphOS ?

Legal problems apart (I'm not saying it's a good thing, or we should forget about it...), saying MorphOS is not AmigaOS is like saying Haiku is not BeOS, or like saying ReactOS is not Windows, or like saying AmigaOS 4 is not AmigaOS... All have been rewritten from scratch (yeah, that includes OS4, it certainly doesn't use much of AmigaOS'original code...) and are binary compatible with the original. So I don't get it... What does make an OS "AmigaOS" to you ?

_________________
http://www.warpdesign.fr/

@Rassilon
Quote:

Rassilon wrote: What you said about RAW image loading, h264 decoding etc is very true, however PCs circumvent some of those issues with GPU hardware assists. But as you say a G3 class CPU won't cut it.

Actually, GPU acceleration can be a very big thing. I have a dual core, Core2 laptop at 2.4GHz, which can just barely play back standard 1080/60i video... forget 1080/60p. Even on my six core AMD desktop, 1080/60p playback can go as high as 60% of all CPU, depending on which player you use.

But run a player that uses GPU acceleration (even Windows Media Player, in Windows 7), and the CPU load drops down to about 6%. That's via the DXVA 2.0 API, which isn't general purpose computing like OpenCL, but dedicated specifically to video rendering. And this is with a nVidia 8800GT GPU... hardly the latest thing.

This is also why I loved the idea of AmigaOS for PPC on the PS3. The PS3 CPU is no match for even a cheapish desktop (3.2GHz, but very simple design for a PPC), but between the GPU and the SPEs, the PS3 can decode multiple HD streams.

Quote:

Spectre660 wrote: @DAX While we are at it I wonder if SMP support for AmigaOS 4.x will allow us to use a Powerpc PCI card as well as the built in CPU on older machines.

SMP = Symmetric MultiProcessing. The scenario you describe there is inherently asymmetric.

@hazydave

Quote:

Now, I do not have direct personal knowledge of all MorphOS sources. But back when Phase 5 was working on their "C Exec" and other things, starting to re-create AmigaOS themselves in the mid-1990s, I was working with Andy Finkel at Amiga Technologies. The Phase 5 guys were really after AT to use tome of their stuff (and pay them, natch). Andy did a code review of the C Kernel, and found it was copied from AmigaOS source code. In fact, even the comments were copied, assembler to C.

Didn't Amiga Technologies (Petro Tschytschenko) send AmigaOS source code to random developers over the world? I recall Olaf Barthel received a copy (just like many others) even when he didnt ask one.

I can send you some AROS source code just for fun of it. Should keep you away from Amiga developments forever... shouldnt it?

Quote:

I wouldn't trust anyone involved in Phase 5, or any code that can be traced back to Phase 5.

I would not trust anyone involved with the official Amiga after 1994.

Quote:

They still don't understand clean room development. If you have seen the Amiga source code, you cannot produce a legally separate work-alike.

By your logic AmigaOS dirty because they have used AROS source code. AmigaOS developers have contributed to AROS and AROS developers have contributed to AmigaOS. AROS developers have contributed to MorphOS and MorphOS developers have contributed to AROS.

So by you logic AmigaOS is dirty. In fact at least one ex-MorphOS developer (who didnt commit much AFAIK) contributes to OS4.

Quote:

Maybe MorphOS is clean, maybe not. Maybe no one actually knows. But that's such a transgression, I wouldn't trust anyone involved in Phase 5, or any code that can be traced back to Phase 5.

Maybe you are on drugs, maybe not. You can not prove you are not. Oops!

Quote:

Point in fact -- I just don't care about MorphOS. It's not AmigaOS, it might as well be Windows for all I care.

That is fine. I didnt care about Met@box.

Quote:

If the MorphOS people would like to swear in public that not a line of code or comment is copied from the AmigaOS sources or derived from the Phase 5 code (fruits of a poisonous tree, in legal terms), I will not mention MorphOS again.

I can swear I have not seen any trace of AmigaOS source code or comments in MorphOS code. But then I only have an access to parts of MorphOS I'm developing. I have seen AROS source code and comments in the source (localization stuff for example).

On the other hand I developed closed source Multiview 2 for MorphOS seeing AROS implementation of Multiview first (which was part of MorphOS 1.x). Does it fall under APL now? I can assure it does not contain AROS comments or AROS code but I wont let non-MorphOS developers review it. If someone wont like it, too bad. I dont care.

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@hazydave

Quote:

But back when Phase 5 was working on their "C Exec" and other things, starting to re-create AmigaOS themselves in the mid-1990s, I was working with Andy Finkel at Amiga Technologies. The Phase 5 guys were really after AT to use tome of their stuff (and pay them, natch).

I can relate to your angryness. To quote famous Benjamin Hermans, if you paid 5 million dollars for something, would you accept other people running off with your IP without putting up a fight?

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook