home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Amiga Collections: MegaDisc
/
MegaDisc 27 (1992-03)(MegaDisc Digital Publishing)(AU)(Disk 1 of 2)[WB].zip
/
MegaDisc 27 (1992-03)(MegaDisc Digital Publishing)(AU)(Disk 1 of 2)[WB].adf
/
Articles
/
Not_So_Slow_Memory
/
Not_So_Slow_Memory
Wrap
Text File
|
1992-03-30
|
3KB
|
79 lines
WHEN SLOW RAM IS FAST RAM
by Peter Flynn
(( 27 )) (( 27 )) (( 27 )) (( 27 )) (( 27 )) (( 27 )) (( 27 )) (( 27 ))
While mucking around with my Phoenix Board I found that sometimes slow ram
is faster than fast ram.
Some Puzzles with Slow Ram and Fast Ram.
Machine used for tests: A1000 with Phoenix Board (1 meg chip, 1 meg slow)
and XEL external memory expansion (1 meg fast). Prior to getting the
Phoenix Motherboard replacement I had been using my old A1000 with the XEL
memory expansion board for a total memory of 1.5 megabytes. Sometimes I
had noticed that it seemed to run slower than the A500 that I had before
picking up a second hand A1000 and at other times it ran at the same speed.
I once had the two machines side by side and noticed a difference while
running the F/A 18 demo flight.
When I installed the Phoenix board I had 3 megabytes memory in total and
was surprised that when I ran SysInfo V2.40 by Nic Wilson it ran faster
without the fast ram expansion than it did with it. Results 1.00 without
and .82 with (these are speeds of an A500 so .82 means an A500 completes
the test in 82 % of the time of the A 1000 with external memory). Other
speed tests were made and generally, but not always, the computer was
faster without "fast" ram. I normally use "fastmemfirst" in
startup-sequence but found it made no difference to the speed if it was
present or not.
In going through the Phoenix Board documents I noticed the memory could be
configured in a number of ways. It was an easy matter to switch off the
second megabyte of internal memory - the "slow" ram.
A test was arranged using an A500 revision 5A (1.3 kickstart) with and
without internal memory expansion and with and without XEL memory board and
my Phoenix board A1000 in similar combinations. The program used was the
F/A 18 Interceptor demo flight. Times are accurate to a second or two.
Time is for a completed flight.
Computer Time Configuration
A500 6.48 1/2 chip, 1/2 slow, 0 fast
A500 6.48 1/2 chip, 1/2 slow, 1 fast
A500 8.07 1/2 chip, 0 slow, 1 fast
Phoenix 6.48 1 chip, 1 slow, 0 fast
Phoenix 6.48 1 chip, 1 slow, 1 fast
Phoenix 9.00 1 chip, 0 slow, 1 fast
Phoenix 6.48 1 chip, 0 slow, 0 fast
The above times are in minutes and seconds.
These results suggest some feature of the F/A 18 program itself affects
the results obtained. One thing that I find odd is why the absence of slow
ram slows the program if fast ram is present - perhaps someone can
enlighten me? Another finding is that the two computers do normally run at
the same speed.
I have tried different memory chips in the 80 - 120 nanosecond range with
no change to results.
Peter Flynn
Berkeley N.S.W.
(( 27 )) (( 27 )) (( 27 )) (( 27 )) (( 27 )) (( 27 )) (( 27 )) (( 27 ))