A project I'd like to see from Jens...

[wawrzon]

*bow to you*

didnt intended to lecture anyone. just feel pity, because the project taking another course could be much further by now and sustain itself the motivation of the developers and the interest of te audience.

[OlafS3]

didnt intended to lecture anyone. just feel pity, because the project taking another course could be much further by now and sustain itself the motivation of the developers and the interest of te audience.

It had different goals and was basically done by one developer primarly because of personal interest. Too ambitious and finally too expensive (if ever completed).

[wawrzon]

of course thomas had a different goal, namely to build a complete system, but i was speaking of the "softcore" part of team.

what happens just now, could have happened in parallel to developing natami board, so to say in different thread to be eventually merged back in. im not saying the softcore team has lost any time, but there might be more buzz around it and therefore more motivation and more dynamics.

lets take majstas vampire project as example. the board doesnt seem technically superior to other similar solutions like fpgaarcade, tobiflex his own experimental de2 (?) accelerator, mist (or how the other fpga board is named?). it is middle range, someone else could have done it if attempted, but none did, and a lot of people argumented against. what is good about that (and apparently intentionally) is that the development happened in open. it met and has proven interest, and kicked others in to contribute. thats pretty an achivement in my eyes.

[biggun]

of course thomas had a different goal, namely to build a complete system, but i was speaking of the "softcore" part of team.

also thats where the fun is. testing, submitting reports, reading what commits have been submitted day by day.

Actually this is exactly how the Softcore is tested/developed.

[wawrzon]

Actually this is exactly how the Softcore is tested/developed.

but since it happens in some closed circle the general public doesnt get involved and cannot even observe the progress and thus may easily come to conclusion that there is no progress.

again as my favourite example: aros. even though not taking part in development one can read the developer ml to know where its aiming at and also you can observe what commits been done day by day. lets admit im a complete noob and am annoying people most of the time for no reason, surely creating unnecessary noise from developer perspective. now, even though i submitted a pfs3 problem report few days ago along with a log and i immediately see a fix submitted by toni another day, now i checked back today and can confirm the bug is gone, my a1200 booting fine all the way to aros-wanderer. i think it is quite rewarding to see and be able to support the development in what ever ways one is able to. sure it has handicaps too, as said noise generated by the noobs and unnecessary discussions, but lets face it, the community is at a point where noobs are probably in majority, but the game has to go on anyway, we have to find ways to take advantage of just any resource available.

[nicholas]

I've like to see a CPU slot board that has a fast x86 CPU emulating an 060 and/or a PPC that appear to the Amiga as a CyberStorm/CyberStormPPC.

Wishful thinking I know but I reckon it would sell if the performance was better than existing cards.

[TeamBlackFox]

> fast x86 cpu emulating

It'd be faster per Hz to emulate a PowerPC CPU via ARM, SPARC or MIPS CPU since with x86 you're having to cross endianness, and architecture so you're not only going to incur a performance hit, but power consumption and voltage requirements would likely be too high for an Amiga if it was x86. Heat is a consideration too. I'd personally vote MIPS since thats my favourite RISC architecture, but thats me.

Since there are some PowerPC chips being made ( Mostly SoCs ) it may be best to just get a cheap PowerPC with extended logic and libraries and plop that into an Amiga CPU slot. Using old Macs as donors, you could easily sell 68040/PowerPC cards for around the same price as the CSPPC goes, and I bet you they'd sell. I looked at 68060s, but according to Freescale's product page they ceased production and the last CPUs are EOL. Maybe someone has a Chinese supplier that can make them though. Making the boards in China may actually make it affordable.

[nicholas]

> fast x86 cpu emulating

It'd be faster per Hz to emulate a PowerPC CPU via ARM, SPARC or MIPS CPU since with x86 you're having to cross endianness, and architecture so you're not only going to incur a performance hit, but power consumption and voltage requirements would likely be too high for an Amiga if it was x86. Heat is a consideration too. I'd personally vote MIPS since thats my favourite RISC architecture, but thats me.

Since there are some PowerPC chips being made ( Mostly SoCs ) it may be best to just get a cheap PowerPC with extended logic and libraries and plop that into an Amiga CPU slot. Using old Macs as donors, you could easily sell 68040/PowerPC cards for around the same price as the CSPPC goes, and I bet you they'd sell. I looked at 68060s, but according to Freescale's product page they ceased production and the last CPUs are EOL. Maybe someone has a Chinese supplier that can make them though. Making the boards in China may actually make it affordable.

I said x86 only because it's cheap, easy to get hold of and faster than anything else available at the same price point.

Regarding the endian issues, an x86 JIT emulating a 68k is currently faster than a PPC JIT emulating a 68k so it's not really a issue.  See WinUAE vs E-UAE-JIT for OS4/MorphOS.

ARM would also be cheap and easy to get hold of but not as fast as x86 at the same price point.  The power consumption would be the major selling point here.

[OlafS3]

I said x86 only because it's cheap, easy to get hold of and faster than anything else available at the same price point.

Regarding the endian issues, an x86 JIT emulating a 68k is currently faster than a PPC JIT emulating a 68k so it's not really a issue.  See WinUAE vs E-UAE-JIT for OS4/MorphOS.

ARM would also be cheap and easy to get hold of but not as fast as x86 at the same price point.  The power consumption would be the major selling point here.

I do not think that this would be cheaper than this solution. Here you can use a off-the-shelf solution and create a connector, for your solution you would need a custom board and that makes it expensive again. Or do I understand something wrong?

[wawrzon]

@nicholas

I do not think that this would be cheaper than this solution. Here you can use a off-the-shelf solution and create a connector, for your solution you would need a custom board and that makes it expensive again. Or do I understand something wrong?

there was some person on this board thinking along and also there was discussion about such alternative. quite a lenghty one but i dont remember where.

however it appears nothing came out of that so better to support ongoing projects. as stated jens apparently is in favour of emulating cpu, but not on x86 but on mips, also it seems he is in favour of complete new system instead of an add on, which is probably partly due to easier servicing and warranty, which is likely quite important element from his perspective.

[nicholas]

@nicholas

there was some person on this board thinking along and also there was discussion about such alternative. quite a lenghty one but i dont remember where.

however it appears nothing came out of that so better to support ongoing projects. as stated jens apparently is in favour of emulating cpu, but not on x86 but on mips, also it seems he is in favour of complete new system instead of an add on, which is probably partly due to easier servicing and warranty, which is likely quite important element from his perspective.

Can anyone tell me what are the fastest MIPS CPU's that can be had in quantity for reasonable prices?

[TeamBlackFox]

Can anyone tell me what are the fastest MIPS CPU's that can be had in quantity for reasonable prices?

I have an R14k powered SGI Fuel that is fully 64-bit, 600MHz that outclasses an Intel Core Solo in general purpose computing, and thats a 10 year old CPU vs a slightly younger one. I don't know what the current MIPS specs run, but the last I used was a brand new 1GHz quad configured MIPS processor (R16k?) and that was screaming fast.


New ones I'm not quite sure on specs though.

[wawrzon]

i think its this one that is taken into consideration:
http://www.broadcom.com/products/Processors/Enterprise/BCM53003

[ElPolloDiabl]

Those specs put PowerPC to shame.

[wawrzon]

Those specs put PowerPC to shame.

do they? also remember cpu emulation comes at a cost. still..

[freqmax]

What has to be taken into account is also the amount of expensive and complex support circuitry to support various CPU:s. Doing GHz signal routing between CPU and memory ain't fun!

A single chip core with CPU+RAM is way simpler to deal with.

A few vendor caveats. Watch out for Intel EOL:ing chips and Broadcom tend to have their own brand of NDA and secrecy hell.

Majastas A600 accelerator is now being done by @biggun here? and Cyclone-5 being the descendant?

Has the source for the FPGA Arcade come out yet? I think it's a great platform. But stuff that actually gets used is even better