[comp.sys.mac.hardware] LC Prediction

hpoppe@ncar.ucar.edu (Herb Poppe) (11/02/90)

In article <2730569E.19505@orion.oac.uci.edu> ajauch@orion.oac.uci.edu 
(Alex Jauch) writes:
> I have a Mac II (about the speed of a LC I presume) ...

Although the LC has a 68020 running at 16MHz (as does the Mac II), it is 
configured with a 16-bit data path to memory instead of the 32-bit data 
path of the Mac II. Since two bus cycles are required to read 32-bits from 
memory on the LC (where only one bus cycle is required on the Mac II), the 
LC is only half as fast as the Mac II, all other things being equal.

When pre-announcement rumors about the LC intimated it would have a 68020, 
I was disappointed. Since this was to be a low cost machine, it would be 
unlikely that they would offer the machine with a PMMU. That has turned 
out to be the case. So, no virtual memory in System 7.0 for the LC. 
However, when the machine was announced with a 16-bit data path instead of 
a 32-bit data path, my disappointment evaporated.

Why is that, you say?

Prediction:

The reason that the LC is not going into full scale production until 
January '91 is because the LC will not be built with a 68020! Instead, it 
will sport a Motorola 68025. As I recall from Motorola's announcement 
early last year, the 68025 is basically a 68030 with a 16-bit data bus. 
That is, it has on-chip PMMU. At the announcement, Motorola said that they 
and "a major computer manufacturer" would make a significant announcement 
in the summer ('90). To my knowledge, no such announcement was made. My 
guess is that the "major manufacturer" was to have been Apple with news of 
a 68025 LC, but the chip wasn't far enough along for the announcement to 
take place. Hence the delay in availability of the LC until January.

Of course, its possible that the 68025 will never see the light of day, or 
that it will be an LCx that has this chip. This chip is the equivalent of 
the 80386SX, and like that chip, could play an important role in low-cost 
desktop and portable Macs.

Herb Poppe             hpoppe@ncar.ucar.edu
NCAR                      (303) 497-1296
1850 Table Mesa Dr.
Boulder, CO  80307-3000

wayne@helix.nih.gov (Wayne Rasband) (11/03/90)

In article <9037@ncar.ucar.edu> hpoppe@ncar.ucar.edu (Herb Poppe) writes:
> Although the LC has a 68020 running at 16MHz (as does the Mac II), it is 
> configured with a 16-bit data path to memory instead of the 32-bit data 
> path of the Mac II. Since two bus cycles are required to read 32-bits 
> from memory on the LC (where only one bus cycle is required on the Mac II), 
> the LC is only half as fast as the Mac II, all other things being equal.

The LC isn't as slow as you might expect. I  run some benchmarks on it the 
other day, and, due to its built-in video,  it actually out performed the 
Mac II on a few graphics intensive benchmarks.  The 16-bit data path does 
make the LC somewhat slower for CPU intensive tasks, but notice how It 
runs small integer benchmarks, such as the Sieve, as fast as the Mac II. 
This is probably because the code fits entirely inside the 68020's tiny 
256 byte instruction cash. It is clear from these results that you 
wouldn't want to do number crunching on the LC without adding a third 
party FPU.


Radius Benchmark V1.0b6

               SE        LC      Mac II    Mac IIfx

Float        44.7         10        0.5        .2
Trig          452         86          1        .3
Butterfly     149         29        7.6       1.6
Ripples      1095        238       27.4         9
Sieve        41.4        8.8        8.5       3.1
Moire          28        8.2       13.2       3.8        
Circles      25.7         19         21      17.5
Arcs         15.8        9.8        10.9        8


Speedometer 2.5

                 SE       LC      Mac II     Mac IIfx

Whetstones       7.2     35.6      51.5        199
Dhrystones       916     2104      2982      11321
Sieve           34.2      7.9       7.9        2.3
Savage           395       70      47.3       11.8

CPU              .99     3.23      3.39       12.4
Math             .98     4.76      6.72       26.2


"Sharpen" filter in Image(pixels/sec)

                 SE        LC      Mac II    Mac IIfx

                 NA        57K      60K        172K

planting@cs.pitt.edu (Professor Harry Plantinga) (11/07/90)

>Although the LC has a 68020 running at 16MHz (as does the Mac II), it is 
>configured with a 16-bit data path to memory instead of the 32-bit data 
>path of the Mac II. Since two bus cycles are required to read 32-bits  
>from memory on the LC (where only one bus cycle is required on the Mac II),  
>the LC is only half as fast as the Mac II, all other things being equal.

16-bit data path, yes; half the speed, not necessarily.

If I remember correctly, a data access on a Mac II has two wait states.  
That means that a memory access takes 3 cycles.  If memory access on a Mac 
LC has 2 wait states, it could be that the two transfers occur within 4 
cycles.  That would make the LC 75% the speed of the II.

Notice that I am not saying that this is the case.  I don't know anything 
about the Mac LC.  I'm just pointing out that the "half as fast" comment 
need not be true.  In fact, MacUser benchmarks make it much more than half 
as fast.




----------
Harry Plantinga
planting@cs.pitt.edu

jfr@locus.com (Jon Rosen) (11/09/90)

In article <9087@pitt.UUCP> planting@cs.pitt.edu (Professor Harry Plantinga) writes:
>>Although the LC has a 68020 running at 16MHz (as does the Mac II), it is 
>>configured with a 16-bit data path to memory instead of the 32-bit data 
>>path of the Mac II. Since two bus cycles are required to read 32-bits  
>>from memory on the LC (where only one bus cycle is required on the Mac II),  
>>the LC is only half as fast as the Mac II, all other things being equal.
>
>16-bit data path, yes; half the speed, not necessarily.
>
>If I remember correctly, a data access on a Mac II has two wait states.  
>That means that a memory access takes 3 cycles.  If memory access on a Mac 
>LC has 2 wait states, it could be that the two transfers occur within 4 
>cycles.  That would make the LC 75% the speed of the II.
>
You have also overlooked the fact that the data path only effects memory
operations... Many instruction mix analyses that I have seen indicate
that only 25-35% of all instructions in a normal application load are
memory instructions (hence the effectiveness of RISC architectures)...
Most instructions are reg-to-reg arithmetic, bit twiddling, compares,
branches and the like... Since the 68020 is pipelined (8 byte pipe
I believe), its instruction queue will be filled most times to that
the smaller data path will not have that much effect on instruction 
fetches (they will be overlapped with other operations)... My guess
is that a 16-bit data path 020 would only see a 20-25% loss in speed
vs a 32-bit data path 020 given equal clock rates and no wait states.
If the above issue regarding wait states is also taken into account,
the difference might only be 15% slower...
 
Jon Rosen

gillies@m.cs.uiuc.edu (11/09/90)

I can't believe they saved more than $2 by crippling the LC with a
16-bit datapath.  Does anyone know for sure if this was a money -
saving move (if so, how much)?

Why did they do this?  Was it the marketing guys complaining about too
good a price:performance ratio?