Quote:

I rather like the sentiment in the last paragraph. Sort of a "but this one goes to eleven" situation eh?

"Any objections? "

None what so ever. That sequence seems to be exactly what has occurred. It reminds me of why I stopped over clocking multiplier locked processors. Too much experimentation needed to get a minor boost.

> Sort of a "but this one goes to eleven" situation eh?

Admittedly, I just had to read up on the meaning of this phrase (as we don't have an equivalent in German) and made a find on the all-knowing Wikipedia:

http://en.wikipedia.org/wiki/Up_to_eleven

And yes, this seems what the situation is like :-)

Neat link. Yes, its a 'Spinal Tap' reference.

"Well it's one louder, isn't it?"

I always wondered why Hyperion ported AOS4 to the Pegasos and now it seems apparent. At least with that port they can claim hardware parity with MorphOS hardware. Or they could until the Mac ports were introduced.

Just yesterday I saw a post on Amiga.org from someone over clocking a 2.0 Ghz 7448 third party processor board. I guess that gives us about twice the raw power (more or less) tghan the best AOS system (unless someone out there has one of those rare 1.4 Ghz Pegasos processor boards).

> I saw a post on Amiga.org from someone over clocking
> a 2.0 Ghz 7448 third party processor board.

I bet you saw it here as well ;-)

http://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=7707&forum=11&start=11

> I guess that gives us about twice the raw power (more or less)
> tghan the best AOS system (unless someone out there has
> one of those rare 1.4 Ghz Pegasos processor boards).

User-sided overclocking aside (there're AmigaOnes with 1.25 GHz 7455A/B overclocked to 1.4 GHz and Pegasos II G4 with 1.0 GHz 7447 overclocked to 1.267 GHz), some AmigaOne owners have machines with genuine 1.267 GHz 7457:

http://www.amiga-news.de/en/news/AN-2007-02-00084-EN.html
http://web.archive.org/web/20080302035620/http://www.acube-systems.biz/eng/hardware.php
http://www.acube-systems.biz/index.php?page=news&id=10

And there are even AmigaOne owners with 1.267 GHz 7457 overclocked to 1.533 GHz:
http://www.a1k.org/forum/showpost.php?p=1013528
https://www.hdrlab.org.nz/benchmark/gfxbench2d/OS/AmigaOS/Result/1767
https://www.hdrlab.org.nz/benchmark/gfxbench2d/OS/AmigaOS/Result/1772
https://www.hdrlab.org.nz/benchmark/gfxbench2d/OS/AmigaOS/Result/1773


Edit: added links to 1.533 GHz AmigaOne XE G4

[ Edited by Andreas_Wolf 19.02.2017 - 12:19 ]

Those are pretty good figures. I've always felt successful if I've managed to get at least an additional 200 Mhz. Although this was much more significant when I was over clocking an AMD Duron from 600 to 800 Mhz.

Is the 7447 processor used on the Pegasos processor cards an A or B?

Quote:

Neither, it is just 7447. The 7447A is the production successor of the 7447.

> Is the 7447 processor used on the Pegasos processor cards an A or B?

1.0 GHz card uses 7447, rare 1.4 GHz card uses 7447A. 7447B (or even C) is not used on the Pegasos at all.

I've found a company that offers BGA removal, re-balling, and re-instillation.
What do you think about the potential of upgrading processor cards via such a approach?

> I've found a company that offers BGA removal, re-balling, and re-instillation. What
> do you think about the potential of upgrading processor cards via such a approach?

Usually, the speedup isn't worth the high costs. ACube's 1.267 GHz upgrade offer for the AmigaOne was 407.50 EUR including the new CPU and rework. If your CPU is broken then it might be worth the hassle to not only have it replaced but also upgraded while you're at it. Luckily, in MorphOS land we don't have to resort to such approaches. You can cheaply buy a faster Mac or a 3rd party CPU upgrade :-)

Quote:
While teh 7447A cards were überrare already it should be mentioned that there should be (have been) also a few 7448 eval cpu cards. But they have never seen the wild....

https://morph.zone/modules/newbb_plus/viewtopic.php?forum=11&topic_id=4498&sortname=&sortorder=&sortdays=&viewmode=flat&order=0&start=35

Quote:

Yes, I totally agree that there is space for an ultra-low price MorphOS2.x capable system to fit into our tiny niche market. That will only work if the port of MorphOS2.x to such a system were easy enough to be done quickly by the existing MorphOS Dev. Team so it would be worth their time and effort, or if it were subsidised by the board manufacturer, or another group, such as the work to port to the Efika was paid for by Genesi, but that did not work out too well for Genesi, so it would be hard to make it happen again.

Such a low price board would not need to be too much more powerful than the Efika is/was, but the issues of low RAM amount and poor USB performance could be fixed, as well as the IDE interface, or hopefully it would have SATA instead.

I don't doubt that some new design and CPU will be found in the not too distant future. The challenge will be for the MorphOS Dev. Team to decide to support one of the new boards and make sure that it has a chance of being produced and supported by the manufacturer for at least a few years, so it is worthwhile to expend the effort to port MorphOS2.x to it.

With the prospect of running MorphOS2.x on a few G5 Mac models and the G4 PowerBook in the future, our hardware needs should be satisfied for as long as it takes the Dev. Team to find other alternative hardware to port to, even if that means a long wait while they eventually move MorphOS2.x to x86 hardware.

Edit: I am not advocating a move to x86, but if nothing good comes along that is PPC (and therefore easy to port to), then a move to x86 will probably be unavoidable eventually.

[ Edited by amigadave on 2011/2/2 15:41 ]

> it should be mentioned that there should be (have been) also a few
> 7448 eval cpu cards. But they have never seen the wild....

Yeah, that's why I omitted those ;-)

> as well as the IDE interface, or hopefully it would have SATA instead.

The MPC5125 which Zylesea endorses provides neither PATA nor SATA. You can check the SoC's specs out right here on MorphZone, by courtesy of Zylesea:

https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=7014&forum=11&post_id=72216#72216

Quote:

Yes, when I left I thought about redrumloa's recent purchase and came to the same conclusion myself. Plus, does anyone know how the Peg's CPU card sets its multiplier and what about potential differences in cpu voltage?

Quote:
The 5125 has usb 2.0 high speed but no sata. Instead I'd go for a SDHC card (SDHC controller inbuild, 200 Mbps) for the system and some stuff. Mass storage would be accessed by usb or ethernet. It is a real complete SoC. And it costs only about US$ 11 (bulk volume, 10k). There is really not much needed to add to the 5125. With *real* bulk volume I guess a 50 US$ board including 256-512 MB RAM would be possible. In lower volume <80 US $ sould be possible though.
At 1 W it offers an e300/400 core just like the Efika 5200B. Not a rocket, but pretty useable. The inbuild gfx of the 5121 is rather weak though but goes up to WGXA reslution. For the tinkerer it has CAN, I2C and quite some other stuff. At that low cost, low wattage and simplicity I think it would gain some interest in several markets.
I think that chip literally cries for an ultra low cost board - and while I cannot make that happen myself now, I would seriously consider to join such an attempt if others participate.

> does anyone know how the Peg's CPU card sets its multiplier

https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=6970&forum=11&post_id=71675#71675

Not sure if that goes deep enough to satisfy you though ;-)

> There is really not much needed to add to the 5125.

I think audio should be there. How to add that?

> it offers an e300/400 core just like the Efika 5200B.

Should even be slightly faster.

https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=7001&forum=3&post_id=74602#74602

> The inbuild gfx of the 5121 is rather weak

Compared to the 5125's gfx it's a powerhouse ;-)

Quote:

Ooops, meant 5125 of course here. And the 8377 is generally spoken nicer, but it costs more and the design would become probably more complex.
Audio with the 5125 could either be left out on the board (a usb soundcard could fill the gap. okay requires an isochronous driver) or an audio chip may be attached to the LPC or to whatever is best to connect. The Wolfson WM9715 that is used with the TWR-5125 offers a touchsceen controller, too.

> the 8377 is generally spoken nicer

I wasn't referring to the MPC8377E per se but to the part where the posting says:

"The e300c4 that is built in the MPC8377E (and MPC5121e) has twice the cache size of the 603e and is 2-way superscalar which the 603e (as well as e300 versions before e300c4) is not."

These two improvements are also true for the MPC5125 as it has the same core version as the MPC5121e.

> an audio chip may be attached to the LPC or to whatever is best to connect.
> The Wolfson WM9715 [...] is used with the TWR-5125

Thanks for the pointer. So far I overlooked the presence of audio (even AC'97 in this case) on the TWR-MPC5125. I just had a closer look at the schematics* and found that the WM9715 is attached to the MPC5125's I²S interface**.

* http://www.freescale.com/files/microcontrollers/hardware_tools/schematics/TWRMPC5125SCH.pdf (page 2)
** http://en.wikipedia.org/wiki/I%C2%B2S

|||

[ Edited by Velcro_SP 19.04.2011 - 08:03 ]

Andreas that reference is fantastic! I have never been able to find anything on the Peg G4 processor card. I wish the options went a little higher, but that is more than enough information.

Zylesea, I always appreciate your enthusiasm, but I really don't think it possible to get the price of even a really low performing PPC board down to the price range you're discussing.
Really, a slightly redesigned efika with a memory slot instead of on board memory would be about as useful, but would still cost about twice what your asking.

The Efika has a much higher BOM. For the 5125 there is really not much needed to be added. No pci on the board reduces complexity of the board drastically.
The TWR 5125 board goes for 119 US$ already - and includes some extras (for example usb to ser and some flash).
The 50 US$ would require real bulk volume though, but I strongly guess below 80 US$ is a realistic estimate for medium production runs (that means not Acube's homeopathic numbers, but a few thousand boards each run).
Just the bare board with RAM, no extras includes. Ultra low cost for your tinkering hobby.Why not?

Quote:

But it would mean it should have a BIOS able to cope with every possible memory module the user might plug in. Something PCs do very well... at a cost.

Soldered memory solves the problem. But it's true, there's always the cost/amount compromise.

You could set up the BIOS to initially boot the system with the slowest, highest latency memory settings and the user could enter the BIOS setup and change the memory settings.
That would lower thew complexity of the memory part of the BIOS.

You're giving for granted that such a clever BIOS already exists...