Prop 1 mark II with pipelining

IanM

Don't know if this has been discussed before or not so sorry if it has. But wouldn't it be nice if a mark 2 version of the current prop 1 could be designed with instruction pipelining so that each instruction takes 1 clock cycle instead of 4? No other changes (don't know about any changes needed for the hub) just 4 times faster! ie 80 MIPS per prop instead of 20 MIPS.

Would this be possible with the kind of die/process technology changes that would be needed?

The best thing would be if the current 40 pin DIL could still be used as well.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Ian Mitchell
www.research.utas.edu.au

Cluso99

We don't want to slow Chip on the Prop II progress. This would be a complete redesign anyway, so why bother?

The Prop B (64 I/O version) design has been completed, so it is just a matter of getting silicon, if demand warranted it.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:

· Home of the MultiBladeProps: TriBladeProp, SixBladeProp, website (Multiple propeller pcbs)
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: Micros eg Altair, and Terminals eg VT100 (Index)
· Search the Propeller forums (via Google)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm

jazzed

Chip is already implementing the instruction per clock feature. He described how it works in one of the net meetings.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
--Steve


Propalyzer: Propeller PC Logic Analyzer
http://forums.parallax.com/showthread.php?p=788230

virtuPIC

IanM, I don't remember the chip layout exactly but buses take a LOT of area. For PASM youneed at least one instruction read, two operand reads and one operand write. If you do this in on one bus you get your four cycles per instruction. Reducing the cycles per instruction is extremely costly.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Airspace V - international hangar flying!
www.airspace-v.com/ggadgets for tools & toys

Cluso99

virtuPIC:

Chip is doing that in the Prop II. There are 4 simultaneous accesses to hub memory per clock cycle. One instruction fetch, two operand (S & D) fetch and one writeback. The instructions are pipelined, but the net effect is one cycle per instruction. Hub memory has read/write quad long access instructions.

Info on the counters and high speed serial interface between props (48MBps yes MBytes)

Listen to the webinar on the parallax site for the full info

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:

· Home of the MultiBladeProps: TriBladeProp, SixBladeProp, website (Multiple propeller pcbs)
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: Micros eg Altair, and Terminals eg VT100 (Index)
· Search the Propeller forums (via Google)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm

IanM

Perhaps my question wasn't phrased very well. The Prop II effectively executes 1 instruction per clock cycle using an instruction pipeline. Would it be possible, economically and physically, for Parallax to implement that (pipelining only, everything else the same, 32k+32k, 8 cores, 80MHz, 32 IOs, 3.3v etc) in a 40 pin DIL package? Would there be any interest from the community in a Prop 1 that ran at 4x the current speed?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Ian Mitchell
www.research.utas.edu.au

Leon

They simply don't have the resources to make any changes to the design. If you need 4x the performance now, use a different chip.

Leon

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle

Cluso99

It would probably take longer that continuing with the Prop II. And no, we want the Prop II, not a half baked solution.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:

· Home of the MultiBladeProps: TriBladeProp, SixBladeProp, website (Multiple propeller pcbs)
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: Micros eg Altair, and Terminals eg VT100 (Index)
· Search the Propeller forums (via Google)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm

virtuPIC

IanM: I would answer that it is technologically possible but economically not sensible. Prop II will be not much more expensive than prop I. Wouldn't make sense to build something in between.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Airspace V - international hangar flying!
www.airspace-v.com/ggadgets for tools & toys

Rayman

But I would imagine that they will try to create a 40-pin DIP module containing the Prop II...

Microcontrolled

Re-designing the Prop 1 is totally inefficient. You would have to compleatly restructure the chips internal structure. Let's let Parallax get along with the Prop II and not waste time on modifing the Prop I.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Toys are microcontroled.
Robots are microcontroled.
I am microcontrolled.

localroger

What I would like to see is a 40 pin through-hole Prop II that ditches the external address bus, limiting you to the comparable internal RAM as the Prop I, but giving the faster execution. Sort of like the 6510, which was a 6502 with a lot of the lines not brought outside so it would fit in a smaller DIP. This would give the non surface mount guys an upgrade path and probably wouldn't be a lot of extra effort on Parallax's part.

OwenS

From what I understand, you probably can't get DIP packages capable of taking Prop-II sized dies.

I don't remember any mention of the Prop 2 having an external address bus? Though if that is the case, then thats awesome

(PS: The 6502 and 6510 both came in DIP-40)

Cluso99

You should listen to the webinar (March 2009) - follow the links on the parallax home page. A lot of these questions are answered there
Chip talks about perhaps a 44 pin DIP. It is an internal 1.8V part. Parallax do not have the volume for multiple parts, although there will be it seems at least two package types, a DIP (with 32 I/O bonded) and a 100 pin quad of some kind.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:

· Home of the MultiBladeProps: TriBladeProp, SixBladeProp, website (Multiple propeller pcbs)
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: Micros eg Altair, and Terminals eg VT100 (Index)
· Search the Propeller forums (via Google)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm

ImageCraft

In the ideal case, a Prop 1B with 64 IO pins (which should require no significant work) and a Prop 1C with 64 IO and 64K Hub RAM (which requires some work) would be a great step up from the Prop 1, but anything that takes time away from Chip from working on Prop II is a non-starter

Mike Huselton

Why don't we pitch in and pool our tremendous talent to develop this prop variant while Chip finishes the Prop II? Perhaps we also pass the hat around to grow an escrow account to pay the development cost?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
JMH

ImageCraft

You know, I have a great plan for a Mk II Propeller C, so why don't we pass the hat around for that too

localroger

@Cluso99 it sounds again as if Chip is far ahead of me. I was stunned and gratified to realize that there was a 40 pin Prop 1, even though I might rarely use it. (Though I just sold two of them in the form of PropRPM's because they had max232's onboard, but I digress.) I'd be very surprised to hear that the Prop 2 would leave that market totally in the cold. Though given Parallax's record with the BASIC stamps I am very confident Prop I will be around for as long as I need them. And I with every passing day I am thinking I will be needing more of them.