rmariani@watnot.UUCP (Rico Mariani) (01/27/86)
In article <550@well.UUCP> farren@well.UUCP (Mike Farren) writes: >In article <11403@watnot.UUCP>, rmariani@watnot.UUCP (Rico Mariani) writes: >> matter?) with no parity checking and no wait states. Well actually >> that's not *quite* true, if you try to access the same memory that is >> being refreshed then the CPU may see as many as 3 wait states. From >> what I understand, this happens very infrequently. As for expandability, > > It shouldn't happen infrequently, it should happen not at all. The design >of the 68000 allows refresh to be interleaved invisibly with the memory >accesses (look at how the coprocessors in the AMIGA work), so you should >NEVER have to resort to wait states for refresh! If there ARE wait states, >I wonder about the design... I'm not sure exactly why this was necessary, but I'll try to find out and I'll get back to you on that... >> Now for the bad news, the price. It is going to be somewhere in the >> $1000-$1500 range (not $800 as mentioned before). > > Let's see... 2Meg = 72 256K rams @ $2.50 each - 180.00 > Support chips (liberal) - 40.00 > PC board and case - 25.00 > for a total of $245 manufacturing cost. Given a 3 to 1 markup (not >atypical) that's a cost of about $750 that I would expect to see. $1500 >for 2Meg with current chip costs is a rip-off! There are some problems with your pricing... a) you haven't added anything to do with Labour... i) assembly ii) testing b) your pricing on the PC board is a bit on the low side c) cost of memory just went up by 20% (apparently Japanese sources bumped their prices, so US sources followed suit, I could be wrong on this...) d) You're thinking in U.S. Dollars, my prices are all in Canadian Cost with labour in $USA is closer to $350 Markup (* 3 if you like) $1050 What's the Canadian Dollar worth? Well add 40% anyway ... $1470 Does that look more reasonable to you? I hope so... By the way, I didn't say the board sells for $1500, I said it would be in the $1000 to $1500 ballpark. However, looking at these figures my hopes of a cheap board are diminishing... Also, I am told that the circuitry to do the automatic system configuration is relatively complicated, this could add to the cost of the support chips, that also makes this price seem much nicer. I'll try to post more details later... Lo Song -Rico {utzoo|allegra|decvax|ihnp4|linus}!watmath!watnot!rmariani
mykes@3comvax.UUCP (Mike Schwartz) (02/01/86)
When I heard that Tecmar was selling a 1MB board for $1000, I honestly considered going into the ram board business, myself. Does anyone out there read InfoWorld (heaven forbid)? I see lots of mail-order places carrying 1.5 MB ram boards for the AT for $300-$500. I had one of the hardware designers here at 3Com do a rough estimate of what a 2MB ram board would cost in small quantities, and he came up with less than $300, including RAM, sockets, pc board, and all the other parts. Labor was not included. Figure the manufacturer will mark up 100% then the retailer would do likewise and you get $900. However, I quickly saw that by NOT INCLUDING THE RAM, the price falls to under $300 for an unpopulated board. When you buy a board with RAM, you are paying $4 for each RAM chip, plus $8 per chip in markup. I was prepared to invest the money to fund the development of the board, but two things stopped me. First of all, in Appendix G of the Amiga Hardware Technical Reference Manual, it says that the exacthardware and software protocols for how boxes and boards on the extended bus are supposed to behave. I did not want to make something that would not work with other 3rd party boards. Second, I found that a company called MicroForge on the East coast would sell me a 2MB board populated for $700 (I am a legit developer). They list March 1 as First Customer Ship, so I will believe it when I see it. In the mean time, MicroForge is supposedly selling 20MB hard disks with the minimum other stuff (1 slot expander and power supply) to developers for $1275 (if my prices are still correct). Although I would very much like to have a bunch more RAM, I think that disk capacity is critical with just 2 floppies online at a time. Plus the increase in speed gained by hard disk vs. floppy access will probably be much more appreciable to human beings than the extra gained by having a Ram disk instead. Plus, Ram disk gets lost too easily. I am wishing constantly that someone would make a simple 256K or 512K ram board that just plugs in the side (SCREW future compatibility!!!!!). 'C' is incredibly verbose, as far as code generated goes, so I imagine that the effective amount of storage for a 20MB hard disk (in terms of comparable number of comparable function programs) will probably be the same as 10MB on an IBM PC. I make this observation based on how it seems like I get about the same amount of stuff on these huge 880K floppies as I get on my PC's puny 360K ones. I ordered a copy of MANX 'C' for $400 (ouch), and I am happy to pay so much for their compiler for several reasons: 1. Manx supports the Amiga! 2. Manx is 2x faster and 2x smaller than Lattice on the PC in just about every category (compile time, code size, code speed). 3. Manx is HIGHLY rated for the MacIntosh. 4. $400 is tax deductable for legit developers 5. $400 is not too much to pay for a real good program which I will be using 99% of the time. I will review the compiler in detail when I am familiar with it. I didn't want to buy their big package (I spent $200 bucks more than for their next smaller one), but it was the only way I could also get library source. In the big package they give me a whole bunch of Unix trash (grep, vi, etc.) that I wouldn't pay $.02 for (I prefer to write my own tools). I feel the other biggest reason I wanted to pay the extra $200 was reason #1 above.
jec@iuvax.UUCP (02/04/86)
$300 sounds kind of low to me, but if he could get a functional
board working, he could probably make good enough money selling the
schematic to a magazine or something. I'm sure AmigaWorld would like
to receive at least one technical article before the end of 1986. I
don't know their rates, but if their payroll department as a loose with
money as AmigaWorld is with 4-color printing, you could probably retire
on it.
Seriously, though. I think a 2MB kit would be a nice project
for someone. There are already a lot of IBM PC types already doing
similar things.
James E. Conley Usenet: {ihnp4,pur-ee,purdue}!iuvax!jec
I.U. Dept. of Linguistics Phone: (812) 335-8702 [Work]
401 Lindley Hall (812) 333-6702 [Home]
Bloomington, IN. 47405