kerl@xenon.UUCP (Dan Kerl) (10/21/88)
According to the 17-October issue of EET, the NuBus bandwidth of the NeXT
system is 25 Megabytes per second, not that NeXT's implementation of NuBus
has a 25 Megahertz clock. This bandwidth is consistent with the NuBus clock
rate of 10 MHz, since it is a 32-bit bus.
With a 10 Mhz clock, the data rates on NuBus are as follows:
word burst length rate (in Megabytes/sec)
------------------------------------------------------------------------------
1 (4 bytes) 13 1/3 (competing with another master)
1 (4 bytes) 20 (parked on bus - no competition)
2 (8 bytes) 26 2/3
4 (16 bytes) 32
8 (32 bytes) 35 5/9
16 (64 bytes) 37 11/17
------------------------------------------------------------------------------
Note that the arbitration time for subsequent masters of NuBus can be hidden
if the burst length is 2 words or longer. There are three bus cycles required
for arbitration; the first being a cycle in which the request line is
inactive, followed by 2 arbitration cycles.
With respect to the Mac II implementation of NuBus, the biggest problem I
have with it is their lack of support for bus retries (ACK=try again later)
I don't know if NeXT has this problem or not...
Dan Kerl
Intergraph Corporation
uunet.ingr.b11.xenon.kerlbcase@cup.portal.com (Brian bcase Case) (10/23/88)
>According to the 17-October issue of EET, the NuBus bandwidth of the NeXT >system is 25 Megabytes per second, not that NeXT's implementation of NuBus >has a 25 Megahertz clock. This bandwidth is consistent with the NuBus clock >rate of 10 MHz, since it is a 32-bit bus. Well, the EET article is, as is so typical, simply inaccurate. The NeXT NuBus runs at 25 MHz.
pf@diab.se (Per Fogelstr|m) (10/27/88)
In article <10360@cup.portal.com> bcase@cup.portal.com (Brian bcase Case) writes: >Well, the EET article is, as is so typical, simply inaccurate. The NeXT >NuBus runs at 25 MHz. One thing that strikes me: How many slots can NuBus have when it uses this high clock rate ?? Multibus II have 20 at 10Mhz and the clock distrubution in the middle. (However having it there or somewhere else dosen't really matter). As i understand it NeXT NuBus uses CMOS swing on the signals wich also complicates the bus driving. Any comments ??
daveh@cbmvax.UUCP (Dave Haynie) (11/01/88)
in article <461@ma.diab.se>, pf@diab.se (Per Fogelstr|m) says: > > One thing that strikes me: How many slots can NuBus have when it uses this > high clock rate ?? From the BYTE photos, the NeXT layout looks like this: +---+---+----------+---+---+ | | | | N | | | | | Disks | e | | | | | | X | | | | | | T | | | | | | | | | | | | | | | | | Power | | | | | | | | | +-1-+-2-+----------+-3-+-4-+ Having the disks in the center might be prettier, but you've got to pay a penalty in line lengths going to slots 1 and 2. Of course, this forces you to deal with a foot long backplane even if it doesn't have that many real slots. The limit on NuBus anyway is 16 slots, though there's a pretty good chance you's have a hard time running a 16 slot NeXTBus machine, especially given that goofy NuBus asymmetric clock. -- Dave Haynie "The 32 Bit Guy" Commodore-Amiga "The Crew That Never Rests" {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: D-DAVE H BIX: hazy Amiga -- It's not just a job, it's an obsession
aglew@urbsdc.Urbana.Gould.COM (11/06/88)
>From the BYTE photos, the NeXT layout looks like this: > > +---+---+----------+---+---+ > | | | | N | | > | | | Disks | e | | > | | | | X | | > | | | | T | | > | | | | | | > | | | | | | > | | | Power | | | > | | | | | | > +-1-+-2-+----------+-3-+-4-+ > >Having the disks in the center might be prettier, but you've got to pay a penalty >in line lengths going to slots 1 and 2. Of course, this forces you to deal with >a foot long backplane even if it doesn't have that many real slots. The limit >on NuBus anyway is 16 slots, though there's a pretty good chance you's have a >hard time running a 16 slot NeXTBus machine, especially given that goofy NuBus >asymmetric clock. > >Dave Haynie "The 32 Bit Guy" Commodore-Amiga "The Crew That Never Rests" > {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: D-DAVE H BIX: hazy > Amiga -- It's not just a job, it's an obsession > Of course, it's possible that they have a lot of cables going into the top and bottom surfaces of the boards in slots 3 and 4, rather than going into connectors on the board edge; but what I've seen of the cube makes me doubt that.