@KimmoK

"What do Amiga users want for a CPU?"

Normally users do not buy a system because of the processor but what they can do with the system. Watching Video, hear music, do some office, play games or being creative. I do not think you can "sell" a system because of the processor.

@NutsAboutAmiga

I was actually NOT thinking about you

@KimmoK

The framebuffers themselves would reside in external memory, and the bandwidth needs to be high enough for the task (which is something I haven't calculated), but I would think a Kintex7 70T/Artix7 100T could do it with a limited amount of shaders, with a 7K160T, 7A200T or 7K325T needed for more oomph. The 70T costs 127USD in quantities of 1, 325T comes in at around 400. Noone ever pays the single piece cost for those, but Xilinx don't actually tell you what it's gonna cost without a quote. A good rule of thumb is around half of list price if you have some sort of volume.

A 7K70T and a7A100T both have 240 25x18 bit "DSP slices", aka Multiply-And-Accumulate blocks and over 4MB of block RAM. PCIe 2.0, x4 on Artix and x8 on Kintex. Would bring the total T10x2 board hardware cost up to around 150EUR.

I feel I'm getting a bit excited about this. The AC701 would be perfect as a starting point, as it has PCIe, HDMI and memory (in addition to a million things we wouldn't need, but still).

http://www.xilinx.com/products/boards-and-kits/EK-A7-AC701-G.htm

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@olegil

I read all infos about 2 FPGA projects and both want to create a core that is independent and can easily adapted to different FPGAs. So you can buy a cheap off-the-shelf FPGA, put the core on it and you have a "Amiga" (without needing custom hardware). I really hope that at least one of these projects becomes real in a not too distant future because it would change the situation. Up to this point I concentrate myself on my distribution .

@OlafS25

I know about those, it's just that being a hardware developer I'm never really satisfied with other peoples implementations

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@olegil

I know

That is a "illness" all engineers seem to have

Last edited by OlafS25 on 14-Feb-2014 at 10:28 AM.

@olegil

" Would bring the total T10x2 board hardware cost up to around 150EUR."

Interesting indeed.

"PowerPi" kind of system where that "olegilAGA" chip is on PCIex4 bus?

Would it be sane if system memory is used for frame buffer?
+ Then the resulting "screen" could be further processed as texture of ATI 3D GPU?
or... too complex...?

UPDATE:
After quick check....
PCIe card with XILINX ARTIX-7 & memory & DVI/hdmi out would be interesting geeky board to play with...
http://fpgablog.com/primages/2012/Kintex-7-FPGA-video.gif

Last edited by KimmoK on 14-Feb-2014 at 12:26 PM.
Last edited by KimmoK on 14-Feb-2014 at 12:22 PM.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@tlosm

Quote:

tlosm wrote: , but if Commodore was alive in 1996 im pretty sure will make the same chose of Eyetech in 2002 (Xe and Micro was the first AmigaOne)

You mean reselling the 1st and worst eval-board they come across at huge markups and then running away from support ?

Hmm, yeah I can somehow see C= doing that.........

_________________
- We don't need good ideas, we haven't run out on bad ones yet
- blame Canada

@Kronos

It was an example of type of mainboard... Powerpc
i can swap the example with Pegasos too, if you prefer

Last edited by tlosm on 14-Feb-2014 at 11:33 AM.
Last edited by tlosm on 14-Feb-2014 at 11:32 AM.

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

@KimmoK

If you put the ATI GPU in there, why the HECK would you need AGA? Seriously, you have an amazing ability to drag in stuff to overengineer ANY idea.

Regarding shared memory, this is a big hole in my knowledge of PCIe. I would think using the FPGAs ability to do DDR3 (and calling it chip ram) would be more useful than saving 6USD for a DDR3 chip

Since chip ram is such a well understood concept on the Amiga in the first place...

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@tlosm

Actually, the example is valid, but you chose your keyword badly considering your audience.

If you had replaced eyetech with bPlan in the sentence, and cut everything after the comma, even Kronos would have agreed

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@olegil
"amazing ability to drag in stuff to overengineer ANY idea"

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@olegil

Actually no ...

Pegasos1, Pegasos2, TeronCX, TeronTX, TeronMini and A1X1k/whatever are just PCs with an odd choice of CPU.

Efika and the SAMs don't even qualify here through their SoC nature ..... might as well call them "incomplete washing machines"

_________________
- We don't need good ideas, we haven't run out on bad ones yet
- blame Canada

@Kronos

Ok, it's off topic anyway

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@matthey

Quote:

Have you ever considered PowerPC for the high end of the Amiga market and 68k for the low end?

No. I consider 68k as retro market, and PPC as what we're progressing on. I don't think this is the same a low/high ends of the same market. To me, it's a different kind of market split than low/high performance. I see high performance as T4240, and low performance as T10**. I think all new hardware tiers for "us" should be e6500 core (or equivalent 64bit+altivec) based at this point, IMHO, unless "they" make a switch to ARM or x64. IMHO... :)

Minimig (including FPGAarcade etc) are all stuck in OS3.9. No progress there. It's neat hardware, and I like that, and I do want to play with that. But I do not expect to see any significant progress in the OS for it. I have a DE1 to tinker with Minimig, and am now considering my options of Zed, SOCkit or DE-Soc boards to play with beyond the old DE1 board. SOCkit and DE1-Soc both have better VGA than Zed, and SocKit has bigger FPGA than DE1-Soc but is more expensive, and then I have had experience with Xilinx tools, boards and Zed but not with Altera tools myself. Ponder ponder ponder...

For any progressive market, be it OS4 of MOS or AROS, I leave it to them to decide, as I have little choice. OK, AROS in theory gives some choice, but I'm not going to create a new CPU port myself, and it already exists as I understand on any of the three I'd want - x86, ARM and PPC. As neither OS4 nor MOS are going back to 68K, then this is the main reason to look to not consider 68K as a low-end next-gen rig. Second reason is that 68K is not greatly progressing itself, and will likely never see performance of PPC etc.

Last edited by billt on 14-Feb-2014 at 03:08 PM.
Last edited by billt on 14-Feb-2014 at 03:08 PM.

_________________
All glory to the Hypnotoad!

@matthey

Quote:

there are still some powerful high tech PPC processors available. They just have embedded features and optimizations instead of the ones Amiga users want.

In what way is e6500 unsuitable for desktop type uses?

What I see as not what we want are over a dozen ethernet controllers and other IP peripherals that we may not use, or in higher numbers than are useful to "us". So there's a lot of the T4240 that I wouldn't make use of. One or maybe 2 of the 1Gb ethernets. if we get really excited, maybe one 10Gb ethernet port.. But we don't want to have a router on our desk, so I wouldn't hook up all of those. I'm not sure that Interlaken is interesting, or that it is not. SRIO may not be exciting for us. PCIe is. SD is useful, though I'd prefer SDXC to SDHC. I care about capacity there more than speed, so maybe it would work via software. USB and SATA are both interesting to us. The biggest missing bit is audio, so let's do a Nemo and stick a southbridge on one of them PCIe ports. Today, that should also get us more USB2, some USB3, Sata3, more PCIe slots, and some other PC standard goodies.

But I do not see the e6500 as grossly lacking anything. I do see it as having more networking stuff than I'd use in a desktop or laptop. But I'm not sure of anything serious that's missing, that wouldn't be added the same way in an x64 PC.

Granted, my experience with computer architecture is lower than I think you seem to have, so you and a few others may have insights that I don't recognize.

Last edited by billt on 14-Feb-2014 at 03:20 PM.

_________________
All glory to the Hypnotoad!

@olegil

Quote:

@KimmoK If you put the ATI GPU in there, why the HECK would you need AGA? Seriously, you have an amazing ability to drag in stuff to overengineer ANY idea.

KimmoK, he tells me the same thing all the time. :) And my converstaions with Olegil do not include AGA...

_________________
All glory to the Hypnotoad!

@billt

PPC is not a "progressive" market either, in fact I see it as retro as 68k, as a kind of "geek" market. 68k at least has the advantage that it runs on different platforms and there are much more potential buyers for it (at least theoretical, because the commercial market died in about 2001). And it might be that PPC has more performance than emulated 68k but 68k is much more optimized and libraries (f.e. for games) are written in assembler.

Quote:

OlafS25 wrote: @billt PPC is not a "progressive" market either, in fact I see it as retro as 68k, as a kind of "geek" market. 68k at least has the advantage that it runs on different platforms and there are much more potential buyers for it (at least theoretical, because the commercial market died in about 2001). And it might be that PPC has more performance than emulated 68k but 68k is much more optimized and libraries (f.e. for games) are written in assembler.

I'll take a T4240 over a 68060 today. I would take a T10** over an Apollo today.

Understand that I would be interested in a switch from PPC to either ARM64 or to x64. I would not want to see MOS or OS4 switch from PPC to Apollo as the path forward after PPC.

I do want an FPGA Arcade board, as well as some similar boards that have a hardwired ARM inside the FPGA as well. (Zed/SOCkit/DE1-SoC etc)

_________________
All glory to the Hypnotoad!

@billt

I understand that. But it is not very propable regarding AmigaOS, perhaps MorphOS will make a switch someday but as I understand it there is no decision yet if they switch at all and to which platform. And in any case it is years away. AROS is the only realistic option there in the next years.

Regarding 68k, I do not see it as negativ (or "retro") like some here do. I think it offers some opportunities in the future.

@billt

Quote:

The biggest missing bit is audio, so let's do a Nemo and stick a southbridge on one of them PCIe ports.

Quote:

Quote: Quote: @KimmoK [...] Seriously, you have an amazing ability to drag in stuff to overengineer ANY idea. KimmoK, he tells me the same thing all the time. :)

How could that be?

To keep the cost down, why not just keep the board as basic as possible - maybe taking advantage of on-SoC peripherals where that makes sense (see revisions to Nemo ethernet for how such simple things can add cost and time...), but adding as much as possible via PCI/e expansions.

When you're the size of Sony or Apple you can start playing around with the more complex on-chip peripherals. but when you're NRE costs are passed on to each customer in >0.1% chunks, and board revisions cost >$50,000 a pop, it just doesn't make sense for a few watts of power saving, or a few MB/s saved on the PCI bus (imho)*.

* Especially when you consider that drivers for such expansion cards already exist, whereas drivers for on-chip peripherals would have to be written from scratch.

Last edited by Boot_WB on 14-Feb-2014 at 05:35 PM.

_________________
Troll - n., A disenfranchised former potential customer who remains interested enough to stay informed and express critical opinions.
opp., the vast majority who voted silently with their feet.