korfhage@CS.UCLA.EDU (09/29/88)
I have been keeping tabs on 68030 upgrades and coprocessors for the Mac,
some information I just got from Dove makes for an interesting comparison
with the recent Apple prices. They are making something to replace the 68020
in a II with a 68030 running at 20 Mhz ($999), 25 Mhz ($1299), or
33 Mhz ($1599). Supposed to ship in mid-October.
What I wonder, though, is just how much performance you can get from the
faster clock speed, given that most macs have 120 ns rams? I asked about this,
and was told that the 25 and 33 Mhz models would run with 1 wait state using
120 ns memory, and 0 wait states with 100 or 80 ns memory. Is there anyone
out there with some 68030 hardware knowledge who can say if this is
reasonable? I though that Mac II's had two wait states built in (it may only
be one). If they do change the wait states for RAM access, does that mean
they have to modify the logic board for the upgrade? Or is it something
simple like a jumper? I don't know mac hardware, and the salesperson did not
seem cognizant of all the details.
Also, I have wondered if anyone sells the PC boards that make up SIMMs?
I'm crazy enough to buy DIPs and solder them in. It may be the only way
to get fast memory at a reasonable price.
Willard Korfhage
ARPA : korfhage@cs.ucla.edu
UUCP : {ucbvax,ihnp4,randvax,trwrb!trwspp,ism780}!ucla-cs!korfhagekorn@eris.berkeley.edu (Peter "Arrgh" Korn) (09/30/88)
In <16327@shemp.CS.UCLA.EDU>, korfhage@CS.UCLA.EDU (Willard Korfhage) said: > > I have been keeping tabs on 68030 upgrades and coprocessors for the Mac, >some information I just got from Dove makes for an interesting comparison >with the recent Apple prices. They are making something to replace the 68020 >in a II with a 68030 running at 20 Mhz ($999), 25 Mhz ($1299), or >33 Mhz ($1599). Supposed to ship in mid-October. > What I wonder, though, is just how much performance you can get from the >faster clock speed, given that most macs have 120 ns rams? I asked about this, >and was told that the 25 and 33 Mhz models would run with 1 wait state using >120 ns memory, and 0 wait states with 100 or 80 ns memory. Is there anyone >out there with some 68030 hardware knowledge who can say if this is >reasonable? I though that Mac II's had two wait states built in (it may only >be one). If they do change the wait states for RAM access, does that mean >they have to modify the logic board for the upgrade?... I believe that the Dove upgrade has space on their 'daughter-board' that slips into the 68020 & 68851 & 68881 sockets for SIMM strips. In this way their processor would talk directly to the RAM over their own bus, rather than the Mac bus. I don't know if they require that you put RAM on this daughter board (and if it's there, whether cards can access it for DMA-related operations). I looked at doing this sort of thing a few months ago, and the conclusion I came to was that the only way to get any real performance increase was w/a daughterboard (the ~15% increase of the IIx isn't a 'real' increase worth writing home about [though a press-release is another matter...]). Anyone from Dove on the net? Peter -- Peter "Arrgh" Korn korn@ucbvax.Berkeley.EDU {decvax,hplabs,sdcsvax,ulysses,usenix}!ucbvax!korn
daveh@cbmvax.UUCP (Dave Haynie) (10/03/88)
in article <16327@shemp.CS.UCLA.EDU>, korfhage@CS.UCLA.EDU says: > > What I wonder, though, is just how much performance you can get from the > faster clock speed, given that most macs have 120 ns rams? I asked about this, > and was told that the 25 and 33 Mhz models would run with 1 wait state using > 120 ns memory, and 0 wait states with 100 or 80 ns memory. Is there anyone > out there with some 68030 hardware knowledge who can say if this is > reasonable? If they're talking about accessing Mac II's motherboard memory, which I assume doesn't do any special tricks, I'd say no, this sounds quite unreasonable. The best you could do with 100ns parts and a 25MHz 68030 is about 2 wait states. If you consider how fast the '030 could really be going at 25MHz in synchronous mode, that's actually 3 wait-states. However, you can get performance up to more like 0 or 1 wait state using various paging and/or interleaving techniques. Lots of the faster '386 machine use these techniques (they, of course, needing all the help they can get). The other solution would, of course, be to supply your fast '030 add-in with an external data and instruction cache, which would then give you maybe an effective rate closer to 0 or 1 wait state. That of course requires that the Mac OS support full caching. > I though that Mac II's had two wait states built in (it may only be one). The plain old Mac II's gonna have 1 wait state for the PMMU, that's built into the PMMU design. Other than that, I guess you need a Mac hardware expert. > Also, I have wondered if anyone sells the PC boards that make up SIMMs? A company called MicroBotics sells pre-made DIP modules that fit a SIMM socket, but I don't know if they sell just the bare boards or not. > Willard Korfhage -- Dave Haynie "The 32 Bit Guy" Commodore-Amiga "The Crew That Never Rests" {ihnp4|uunet|rutgers}!cbmvax!daveh PLINK: D-DAVE H BIX: hazy "I can't relax, 'cause I'm a Boinger!"