10 May
2018
10 May
'18
7:29 a.m.
Drums were used as main memory in a number of early computers, and as secondary memory for
a while longer. I wonder how fast real ones (actually constructed) managed to be.
What prompted this question is reading an interesting document: https://ir.cwi.nl/pub/9603
(in Dutch), "Principles of electronic calculating machines, course notes February
1948" by Prof. A. van Wijngaarden at the Mathematical Center (now CWI) in Amsterdam.
It's quite a fascinating short introduction into computing technology of that era.
(One comment in the intro: "The field is new. At the moment, the Eniac is the only
working machine..." -- probably not quite accurate given some classified machines,
but not too far wrong.)
The section on main memory describes a bunch of different technoly possibilities, one of
them drum memory. He writes that a drum of 8 cm diameter (a bit over 3 inches) and
"a couple of decimeters height" could hold maybe 100k bits, with a track pitch
of "a few millimeters". So far so good. He goes on to suggest that such a drum
might spin at 1000 revolutions per second, i.e., 60,000 rpm. That seems amazingly high.
I could see it being physically possible for a drum of only 40 mm radius, but it sure
doesn't sound easy. It's a good goal to strive for given that the logic, even in
the days of vacuum tubes, can run at cycle times of just a couple of microseconds. As one
more way to speed things up he suggests having multiple rows of read/write heads, where
the addressed word would be picked up by whichever head sees it soonest. 10 rows and 60k
rpm would give you 50 microseconds average access time which "even for a parallel
computer would be a very attractive number". (Pages 17-18)
I'm wondering what the reality of fast drum memories looked like, and whether anyone
came even close to these numbers. Also, am I right in thinking they are at least in
principle achievable? I know I could run the stress numbers, but haven't done so.
paul
10 May
10 May
7:57 a.m.
On 05/10/2018 07:29 AM, Paul Koning via cctalk wrote:
I'm wondering what the reality of fast drum
memories looked like, and whether anyone came even close to these numbers. Also, am I
right in thinking they are at least in principle achievable? I know I could run the
stress numbers, but haven't done so.
All of the STAR-100 stations, including the paging station used drums.
Jim Thornton and folks at CDC ADL were working on a 100K RPM drum
spinning in vacuo for a paging store, but they couldn't get it to work
reliably. At any rate, STAR was the last system I saw fast drums on
and you can check the figures in the Bitsavers documentation under
cdc/cyber/cyber200. At any rate, a head-per-track drum could be much
faster than a disk.
There were big slow drums, also. Consider the Univac FASTRAND unit.
--Chuck
8:15 a.m.
The Univac FASTRAND *was* slow.
You could stand there watching through the window on the side of the
5,000-pound beast and actually see the enormous drum rotating as it
lumbered along at, what, 14 RPS I think.
Regards to the List -
Jack
At 08:57 AM 5/10/2018, Chuck Guzis via cctalk wrote:
On 05/10/2018 07:29 AM, Paul Koning via cctalk wrote:
----------------------------------------------------------------------------------
Jack Harper, President
Secure Outcomes Inc
2942 Evergreen Parkway, Suite 300
Evergreen, Colorado 80439 USA
303.670.8375
303.670.3750 (fax)
http://www.secureoutcomes.net for Product Info.
I'm wondering what the reality of fast drum
memories looked like,
and whether anyone came even close to these numbers. Also,
am I
right in thinking they are at least in principle achievable? I
know I could run the stress numbers, but haven't done so.
All of the STAR-100 stations, including the paging station used drums.
Jim Thornton and folks at CDC ADL were working on a 100K RPM drum
spinning in vacuo for a paging store, but they couldn't get it to work
reliably. At any rate, STAR was the last system I saw fast drums on
and you can check the figures in the Bitsavers documentation under
cdc/cyber/cyber200. At any rate, a head-per-track drum could be much
faster than a disk.
There were big slow drums, also. Consider the Univac FASTRAND unit.
--Chuck 
12:53 p.m.
On 5/10/2018 9:15 AM, Jack Harper via cctalk wrote:
The Univac FASTRAND *was* slow.
You could stand there watching through the window on the side of the
5,000-pound beast and actually see the enormous drum rotating as it
lumbered along at, what, 14 RPS I think.
Regards to the List -
Jack
Can not be slower than modern computers, It takes 5 minutes for my mail
program to load under windows and grab new mail.
Once core memory became common, fast speeds are relative to other I/O
at the time, of TTY and punch card speeds.
Ben.
1:03 p.m.
On 05/10/2018 12:53 PM, ben via cctalk wrote:
Can not be slower than modern computers, It takes 5
minutes for my mail
program to load under windows and grab new mail.
Once core memory became common, fast speeds are relative to other I/O
at the time, of TTY and punch card speeds.
Actually, the FASTRAND was used on the 1107 machine which used thin-film
memory, so memory was pretty fast already.
Didn't the FASTRAND have "ping" detectors that would register a count
every time a head hit the surface? I recall that the binary-encoded
positioner mechanism was composed of a bunch of solenoids and levers
that could convert a binary input to a head position. (The FASTRAND was
a moveable head, not a head-per-track drum).
--Chuck
1:17 p.m.
Chuck, do you know if the story that the FASTRAND drum was fabricated
from milled/machined sewer pipe is true???
Another apocryphal(?) story is that a FASTRAND unit lies today at the
bottom of Tokyo Bay from when it fell from a freighter unloading crane.
...would have made a bit of a splash :)
Regards to the List -
Jack
Didn't the FASTRAND have "ping" detectors
that would register a count
every time a head hit the surface? I recall that the binary-encoded
positioner mechanism was composed of a bunch of solenoids and levers
that could convert a binary input to a head position. (The FASTRAND was
a moveable head, not a head-per-track drum).
--Chuck
----------------------------------------------------------------------------------
Jack Harper, President
Secure Outcomes Inc
2942 Evergreen Parkway, Suite 300
Evergreen, Colorado 80439 USA
303.670.8375
303.670.3750 (fax)
http://www.secureoutcomes.net for Product Info.
1:25 p.m.
On 5/10/18 1:17 PM, Jack Harper via cctalk wrote:
Chuck, do you know if the story that the FASTRAND drum was fabricated from
milled/machined sewer pipe is true???
it appears one still exists
https://collection.maas.museum/object/261170
the decade is wrong.
and
https://www.youtube.com/watch?v=luPM6XaKZuU
1:26 p.m.
and this says they are 880 rpm
https://www.fourmilab.ch/documents/univac/fastrand.html
On 5/10/18 1:25 PM, Al Kossow via cctalk wrote:
On 5/10/18 1:17 PM, Jack Harper via cctalk wrote:
Chuck, do you know if the story that the FASTRAND drum was fabricated from
milled/machined sewer pipe is true???
it appears one still exists
https://collection.maas.museum/object/261170
the decade is wrong.
and
https://www.youtube.com/watch?v=luPM6XaKZuU
1:44 p.m.
...press POWER ON and watch the lights dim as the helicopter sound spins up :)
Jack
At 02:25 PM 5/10/2018, you wrote:
On 5/10/18 1:17 PM, Jack Harper via cctalk wrote:
----------------------------------------------------------------------------------
Jack Harper, President
Secure Outcomes Inc
2942 Evergreen Parkway, Suite 300
Evergreen, Colorado 80439 USA
303.670.8375
303.670.3750 (fax)
http://www.secureoutcomes.net for Product Info.
Chuck, do you know if the story that the FASTRAND drum was
fabricated from
milled/machined sewer pipe is true???
it appears one still exists
https://collection.maas.museum/object/261170
the decade is wrong.
and
https://www.youtube.com/watch?v=luPM6XaKZuU
2:12 p.m.
On 05/10/2018 01:17 PM, Jack Harper via cctalk wrote:
Chuck, do you know if the story that the FASTRAND drum was fabricated
from milled/machined sewer pipe is true???
I remember hearing the joke that the (2 counter-rotating) drums were
chrome-plated sewer pipe, but I would find that hard to believe.
IIT, I think it was, had one installed on their 1108, which was on the
second floor of a refurbished old building. I heard a story that pretty
early on, the Univac field people were mystified by clusters of "pings"
on the FASTRAND. It turned out that every time a good-sized truck went
down S. Michigan Ave, the pinging would start.
--Chuck
11 May
11 May
4:11 a.m.
On Thu, May 10, 2018 at 01:53:12PM -0600, ben via cctalk wrote:
[...]
Can not be slower than modern computers, It takes 5
minutes for my mail
program to load under windows and grab new mail.
And yet computer performance was perfectly acceptable before you started using
Windows? The cause of your problems and thus the solution seems obvious.
7:34 a.m.
On Thu, May 10, 2018 at 01:53:12PM -0600, ben via
cctalk wrote:
[...]
> Can not be slower than modern computers, It takes 5 minutes for my mail
> program to load under windows and grab new mail.
My mediocre laptop running
linux is fully up and logged in
in 15 seconds! That includes me typing in the password. I
do have a solid state disk.
Jon
7:38 a.m.
On Thu, May
10, 2018 at 01:53:12PM -0600, ben via cctalk
wrote:
[...]
> Can not be slower than modern computers, It takes 5
> minutes for my mail
> program to load under windows and grab new mail.
My mediocre laptop running
linux is fully up and logged in
in 15 seconds! That includes me typing in the password.
I do have a solid state disk.
Jon
12 May
12 May
1:29 p.m.
On 5/11/2018 5:11 AM, Peter Corlett via cctalk wrote:
On Thu, May 10, 2018 at 01:53:12PM -0600, ben via
cctalk wrote:
[...]
YES! Please send me a PDP-10 computer. PO BOX 1234567 CANADA.
Can not be slower than modern computers, It takes
5 minutes for my mail
program to load under windows and grab new mail.
And yet computer performance was perfectly acceptable before you started using
Windows? The cause of your problems and thus the solution seems obvious.
2:31 p.m.
New subject: found what I believe are a few vrc drum pickup heads too bad photos don't post here interested parties,can email
found what I believe are a few vrc drum pickup heads too bad photos don't post here
interested parties,can email
Sent from AOL Mobile Mail
On Saturday, May 12, 2018 ben via cctalk <cctalk at classiccmp.org> wrote:
On 5/11/2018 5:11 AM, Peter Corlett via cctalk wrote:
On Thu, May 10, 2018 at 01:53:12PM -0600, ben via
cctalk wrote:
[...]
YES! Please send me a PDP-10 computer. PO BOX 1234567 CANADA.
Can not be slower than modern computers, It takes
5 minutes for my mail
program to load under windows and grab new mail.
And yet computer performance was perfectly acceptable before you started using
Windows? The cause of your problems and thus the solution seems obvious.
2:51 p.m.
New subject: found what I believe are a few vrc drum pickup heads too bad photos don't post here interested parties,can email
Head-per-track drums were used on airborne radar setups as a
non-computer use. John Meshna used to advertise military surplus ones,
complete with 400Hz spindle motor.
I was tempted at the time to get one, but was warned that the surplus
ones are rarely functional.
--Chuck
2:57 p.m.
New subject: found what I believe are a few vrc drum pickup heads too bad photos don't post here interested parties,can email
these probably get proc control vec drums
Sent from AOL Mobile Mail
On Saturday, May 12, 2018 Chuck Guzis via cctalk <cctalk at classiccmp.org> wrote:
Head-per-track drums were used on airborne radar setups as a
non-computer use. John Meshna used to advertise military surplus ones,
complete with 400Hz spindle motor.
I was tempted at the time to get one, but was warned that the surplus
ones are rarely functional.
--Chuck
3:01 p.m.
New subject: found what I believe are a few vrc drum pickup heads too bad photos don't post here interested parties,can email
vrc get pro. control hpt
Sent from AOL Mobile Mail
On Saturday, May 12, 2018 Chuck Guzis via cctalk <cctalk at classiccmp.org> wrote:
Head-per-track drums were used on airborne radar setups as a
non-computer use. John Meshna used to advertise military surplus ones,
complete with 400Hz spindle motor.
I was tempted at the time to get one, but was warned that the surplus
ones are rarely functional.
--Chuck
3:37 p.m.
New subject: found what I believe are a few vrc drum pickup heads too bad photos don't post here interested parties,can email
My guess was MADRE or NRL:
https://www.wikiwand.com/en/Over-the-horizon_radar
The timing would be right. Acoustic delay lines served the same
purpose, but you couldn't have a mercury delay line in an airplane,
practically speaking.
--Chuck
6:03 p.m.
New subject: More ancient memory
I found this interesting article:
http://pichotjm.free.fr/DisplayDL/DisplayDLus.html
that discusses an ASCII video terminal from about 1970 that employs a
bunch of PAL TV 64 usec. piezo delay devices as a way to reduce cost.
--Chuck
7:36 p.m.
New subject: found what I believe are a few vrc drum pickup heads too bad photos don't post here interested parties,can email
not these heads.....
Sent from AOL Mobile Mail
On Saturday, May 12, 2018 Chuck Guzis via cctalk <cctalk at classiccmp.org> wrote:
My guess was MADRE or NRL:
https://www.wikiwand.com/en/Over-the-horizon_radar
The timing would be right. Acoustic delay lines served the same
purpose, but you couldn't have a mercury delay line in an airplane,
practically speaking.
--Chuck
2:31 p.m.
New subject: found what I believe are a few vrc drum pickup heads too bad photos don't post here interested parties,can email
found what I believe are a few vrc drum pickup heads too bad photos don't post here
interested parties,can email
Sent from AOL Mobile Mail
On Saturday, May 12, 2018 ben via cctalk <cctalk at classiccmp.org> wrote:
On 5/11/2018 5:11 AM, Peter Corlett via cctalk wrote:
On Thu, May 10, 2018 at 01:53:12PM -0600, ben via
cctalk wrote:
[...]
YES! Please send me a PDP-10 computer. PO BOX 1234567 CANADA.
Can not be slower than modern computers, It takes
5 minutes for my mail
program to load under windows and grab new mail.
And yet computer performance was perfectly acceptable before you started using
Windows? The cause of your problems and thus the solution seems obvious.
10 May
10 May
9:12 a.m.
On 05/10/2018 09:57 AM, Chuck Guzis via cctalk wrote:
Jim Thornton and folks at CDC ADL were working on a
100K
RPM drum spinning in vacuo for a paging store, but they
couldn't get it to work reliably.
Anything with high bit density needs to have
the heads very
close to the data surface. With the heads close, you really
need to have aerodynamic flying heads to compensate for
thermal and other variations in the mechanical system. So,
I really have doubts the evacuated drum is going to work.
Jon
9:22 a.m.
On May 10, 2018, at 10:57 AM, Chuck Guzis via cctalk
<cctalk at classiccmp.org> wrote:
On 05/10/2018 07:29 AM, Paul Koning via cctalk wrote:
Faster than a moving head disk, certainly, though head per track disks also existed. DEC
had some fast ones -- RS04 comes to mind.
I looked at the Star peripherals manual. It describes the paging drum as a modified 865
drum, which "rotates at 1800 rpm". So it might have a high transfer rate -- 12
bit words in parallel from 12 heads -- but clearly quite high latency.
paul
I'm wondering what the reality of fast drum
memories looked like, and whether anyone came even close to these numbers. Also, am I
right in thinking they are at least in principle achievable? I know I could run the
stress numbers, but haven't done so.
All of the STAR-100 stations, including the paging station used drums.
Jim Thornton and folks at CDC ADL were working on a 100K RPM drum
spinning in vacuo for a paging store, but they couldn't get it to work
reliably. At any rate, STAR was the last system I saw fast drums on
and you can check the figures in the Bitsavers documentation under
cdc/cyber/cyber200. At any rate, a head-per-track drum could be much
faster than a disk.
12:52 p.m.
"Fast" is a fuzzy term.
The 865A drum that Paul cited did spin at 1800 RPM, but had a transfer
speed of 2MHz per channel. Data was transfered in 12-bit parallel, so
the composite transfer speed was 24Mbit/sec, which isn't too shabby for
1974.
As it was used as a paging drum, transfer speed was probably more
important than rotational latency. A page was either 512 or 65,536
64-bit words (4096 bytes or 512K bytes).
The STAR also used 844 disk drives for data storage.
Each "station" had its own "microdrum" of about 72KB that spun at
3600
RPM, but had a transfer rate of 1MHz. It was used to run code on the
station and also hold the CPU bootstrap. It was normally loaded from a
CE "suitcase".
--Chuck
8:33 a.m.
On Thu, May 10, 2018 at 10:29:06AM -0400, Paul Koning via cctalk wrote:
[...] So far so good. He goes on to suggest that such
a drum might spin at
1000 revolutions per second, i.e., 60,000 rpm. That seems amazingly high. I
could see it being physically possible for a drum of only 40 mm radius, but
it sure doesn't sound easy.
Looking at modern hard disks, I'm unconvinced we could even mass-produce
something like that today.
A 40mm radius is comparable to a 3.5" disk, which are generally 5,400-7,200
RPM. 15,000 RPM is the fastest available, but those tend to be low-capacity and
expensive, and are often 2.5" drives with a huge heatsink. We could perhaps
rotate a very narrow smaller cylinder faster still but then the capacity
suffers further, and the seek time would start to dominate.
9:44 a.m.
On May 10, 2018, at 11:33 AM, Peter Corlett via cctalk
<cctalk at classiccmp.org> wrote:
On Thu, May 10, 2018 at 10:29:06AM -0400, Paul Koning via cctalk wrote:
Drums are head per track devices, so there is no seek. Yes, modern drives do 10k rpm max
on 3.5 inch disks, while 15k rpm disks uses 2.5 inch platters. As I understand it, the
reason is air resistance and the desire to limit drive motor power.
Chuck mentioned a CDC effort to have a drum spin in vacuo. That obviously avoids the air
friction issue, but at the cost of losing the ability to have flying heads. It probably
makes sense to use much reduced pressure, maybe 1% of standard, which still gives you some
lift on the heads.
Note that the document I quoted wasn't talking high density. I'm guessing 3 mm
between tracks, which is easy enough (the RF11 is similar, perhaps somewhat denser if I
remember right). Van Wijngaarden mentions 1000 bits per track (and 100 tracks on a
"few decimeter" length, so about 3 mm per track which is similar to the track
spacing of the DEC RF11). 80 mm diameter means about 250 mm circumference, so we have 4
bits per mm, which is clearly easy enough and doesn't seem to require flying heads.
paul
[...] So far so good. He goes on to suggest that
such a drum might spin at
1000 revolutions per second, i.e., 60,000 rpm. That seems amazingly high. I
could see it being physically possible for a drum of only 40 mm radius, but
it sure doesn't sound easy.
Looking at modern hard disks, I'm unconvinced we could even mass-produce
something like that today.
A 40mm radius is comparable to a 3.5" disk, which are generally 5,400-7,200
RPM. 15,000 RPM is the fastest available, but those tend to be low-capacity and
expensive, and are often 2.5" drives with a huge heatsink. We could perhaps
rotate a very narrow smaller cylinder faster still but then the capacity
suffers further, and the seek time would start to dominate. 
1:03 p.m.
Looking at modern hard disks, I'm unconvinced we
could even mass-
produce
something like that today.
A 40mm radius is comparable to a 3.5" disk, which are generally 5,400-
7,200
RPM. 15,000 RPM is the fastest available, but those tend to be low-
capacity and
expensive, and are often 2.5" drives with a huge heatsink.
I am not sure if that hold true. 15K drives have been around at least for 20
years (unless you are considering that part of today?). I was using Seagate
Cheetah SCSI drives in 9Gb and 18Gb capacities w/ 3.5" platters spinning at
15K RPM back in the 2000s.
They may not be terabyte HDD but 15K RPM was definitely not an issue.
-Ali
8:51 a.m.
SOME OTHER? DRUM? USE? AS? I REMEMBER IT......
?FOR? STORAGE?HP 2000A TIMESHARE SYSTEM? USED? AS? DRUM? ? AS? SYSTEMS? WERE UPGRADED?AND
DISCS ADDED??FOR? 2000C? ETC THRU F? ?SOME KEPT THEIR? DRUM? AS? SWAPPING? MEDIA. AS? ?THE
HEARD PER TRACK?WAS? FASTER THAN? ?MOVING HEAD? FOR? USER? SPACE? AREA? SWAPPING
----? AMD -------AND? SOME OF THE GE-PAC? PROCESS CONTROL STUFF? USED? VERMONT? RESEARCH?
?DRUMS...? ED#?
?
In a message dated 5/10/2018 7:29:19 AM US Mountain Standard Time, cctalk at
classiccmp.org writes:
?
Drums were used as main memory in a number of early computers, and as secondary memory
for a while longer. I wonder how fast real ones (actually constructed) managed to be.
What prompted this question is reading an interesting document: https://ir.cwi.nl/pub/9603
(in Dutch), "Principles of electronic calculating machines, course notes February
1948" by Prof. A. van Wijngaarden at the Mathematical Center (now CWI) in Amsterdam.
It's quite a fascinating short introduction into computing technology of that era.
(One comment in the intro: "The field is new. At the moment, the Eniac is the only
working machine..." -- probably not quite accurate given some classified machines,
but not too far wrong.)
The section on main memory describes a bunch of different technoly possibilities, one of
them drum memory. He writes that a drum of 8 cm diameter (a bit over 3 inches) and "a
couple of decimeters height" could hold maybe 100k bits, with a track pitch of
"a few millimeters". So far so good. He goes on to suggest that such a drum
might spin at 1000 revolutions per second, i.e., 60,000 rpm. That seems amazingly high. I
could see it being physically possible for a drum of only 40 mm radius, but it sure
doesn't sound easy. It's a good goal to strive for given that the logic, even in
the days of vacuum tubes, can run at cycle times of just a couple of microseconds. As one
more way to speed things up he suggests having multiple rows of read/write heads, where
the addressed word would be picked up by whichever head sees it soonest. 10 rows and 60k
rpm would give you 50 microseconds average access time which "even for a parallel
computer would be a very attractive number". (Pages 17-18)
I'm wondering what the reality of fast drum memories looked like, and whether anyone
came even close to these numbers. Also, am I right in thinking they are at least in
principle achievable? I know I could run the stress numbers, but haven't done so.
paul
9:36 a.m.
added hp2000a info when drum was only rotating media it was used for swap and also program
storage..... kids,using a 2000a hp had a bit to tape punching to savesave their
programs... as the drum did not hold a lot
ed# www.smecc.org
Sent from AOL Mobile Mail
On Thursday, May 10, 2018 Ed Sharpe via cctalk <cctalk at classiccmp.org> wrote:
SOME OTHER? DRUM? USE? AS? I REMEMBER IT......
?FOR? STORAGE?HP 2000A TIMESHARE SYSTEM? USED? AS? DRUM? ? AS? SYSTEMS? WERE UPGRADED?AND
DISCS ADDED??FOR? 2000C? ETC THRU F? ?SOME KEPT THEIR? DRUM? AS? SWAPPING? MEDIA. AS? ?THE
HEARD PER TRACK?WAS? FASTER THAN? ?MOVING HEAD? FOR? USER? SPACE? AREA? SWAPPING
----? AMD -------AND? SOME OF THE GE-PAC? PROCESS CONTROL STUFF? USED? VERMONT? RESEARCH?
?DRUMS...? ED#?
?
In a message dated 5/10/2018 7:29:19 AM US Mountain Standard Time, cctalk at
classiccmp.org writes:
?
Drums were used as main memory in a number of early computers, and as secondary memory for
a while longer. I wonder how fast real ones (actually constructed) managed to be.
What prompted this question is reading an interesting document: https://ir.cwi.nl/pub/9603
(in Dutch), "Principles of electronic calculating machines, course notes February
1948" by Prof. A. van Wijngaarden at the Mathematical Center (now CWI) in Amsterdam.
It's quite a fascinating short introduction into computing technology of that era.
(One comment in the intro: "The field is new. At the moment, the Eniac is the only
working machine..." -- probably not quite accurate given some classified machines,
but not too far wrong.)
The section on main memory describes a bunch of different technoly possibilities, one of
them drum memory. He writes that a drum of 8 cm diameter (a bit over 3 inches) and "a
couple of decimeters height" could hold maybe 100k bits, with a track pitch of
"a few millimeters". So far so good. He goes on to suggest that such a drum
might spin at 1000 revolutions per second, i.e., 60,000 rpm. That seems amazingly high. I
could see it being physically possible for a drum of only 40 mm radius, but it sure
doesn't sound easy. It's a good goal to strive for given that the logic, even in
the days of vacuum tubes, can run at cycle times of just a couple of microseconds. As one
more way to speed things up he suggests having multiple rows of read/write heads, where
the addressed word would be picked up by whichever head sees it soonest. 10 rows and 60k
rpm would give you 50 microseconds average access time which "even for a parallel
computer would be a very attractive number". (Pages 17-18)
I'm wondering what the reality of fast drum memories looked like, and whether anyone
came even close to these numbers. Also, am I right in thinking they are at least in
principle achievable? I know I could run the stress numbers, but haven't done so.
paul
9:53 a.m.
I don't think early drums were terribly fast, but this wasn't a problem because
often they were on serial machines, and the data had to sync with the clock speed of the
machine.
I know that the Manchester Mk1 which evolved from the baby had a drum added. The design of
the drum used changed as the machine evolved. There is some info on this evolution here.
http://curation.cs.manchester.ac.uk/computer50/www.computer50.org/mark1/get…
Its interesting to note that the size of the drum was decreased to around 6" as
suggested by others.
The Ferranti Pegasus also had a drum for main storage and delay lines for
"registers".
This was a physically large drum with a capacity of 5120 40 it words.
The Ferranti Pegasus the clock speed was 333Khz and this was derived from clock tracks
written on the drum avoiding any sync problems.
However in order to achieve this transfer rate the designers built the tracks in pairs
with alternate bits coming from different tracks..
The large diameter of the drum gave problems getting consistent flying height for the
heads, which resulted in large changes in signal level.
There are notes on this here:-
http://www.cs.man.ac.uk/CCS/res/res07.htm#h
and here
http://www.computerconservationsociety.org/resurrection/res42.htm#f
If any one is interested I have more pictures of the Pegasus..
Dave
-----Original Message-----
From: cctalk <cctalk-bounces at classiccmp.org> On Behalf Of Ed Sharpe via
cctalk
Sent: 10 May 2018 16:51
To: paulkoning at comcast.net; cctalk at classiccmp.org; cctalk at classiccmp.org
Subject: Re: how fast were drum memories?
SOME OTHER DRUM USE AS I REMEMBER IT......
FOR STORAGE HP 2000A TIMESHARE
SYSTEM USED AS DRUM AS SYSTEMS WERE UPGRADED AND DISCS
ADDED FOR 2000C ETC THRU F SOME KEPT
THEIR DRUM AS SWAPPING MEDIA. AS THE HEARD PER
TRACK WAS FASTER THAN MOVING
HEAD FOR USER SPACE AREA SWAPPING
---- AMD -------AND SOME OF THE GE-PAC PROCESS CONTROL
STUFF USED VERMONT RESEARCH DRUMS... ED#
In a message dated 5/10/2018 7:29:19 AM US Mountain Standard Time,
cctalk at classiccmp.org writes:
Drums were used as main memory in a number of early computers, and as
secondary memory for a while longer. I wonder how fast real ones (actually
constructed) managed to be.
What prompted this question is reading an interesting document:
https://ir.cwi.nl/pub/9603 (in Dutch), "Principles of electronic calculating
machines, course notes February 1948" by Prof. A. van Wijngaarden at the
Mathematical Center (now CWI) in Amsterdam. It's quite a fascinating short
introduction into computing technology of that era. (One comment in the
intro: "The field is new. At the moment, the Eniac is the only working
machine..." -- probably not quite accurate given some classified machines,
but not too far wrong.)
The section on main memory describes a bunch of different technoly
possibilities, one of them drum memory. He writes that a drum of 8 cm
diameter (a bit over 3 inches) and "a couple of decimeters height" could hold
maybe 100k bits, with a track pitch of "a few millimeters". So far so good. He
goes on to suggest that such a drum might spin at 1000 revolutions per
second, i.e., 60,000 rpm. That seems amazingly high. I could see it being
physically possible for a drum of only 40 mm radius, but it sure doesn't sound
easy. It's a good goal to strive for given that the logic, even in the days of
vacuum tubes, can run at cycle times of just a couple of microseconds. As
one more way to speed things up he suggests having multiple rows of
read/write heads, where the addressed word would be picked up by
whichever head sees it soonest. 10 rows and 60k rpm would give you 50
microseconds average access time which "even for a parallel computer would
be a very attractive number". (Pages 17-18)
I'm wondering what the reality of fast drum memories looked like, and
whether anyone came even close to these numbers. Also, am I right in
thinking they are at least in principle achievable? I know I could run the stress
numbers, but haven't done so.
paul
10:14 a.m.
>>> from "Dave Wade via cctalk"
<cctalk at classiccmp.org>
To: "General Discussion: On-Topic and
Off-Topic Posts" <cctalk at classiccmp.org>
Sent: Thursday, 10 May, 2018 5:53:38 PM
Subject: RE: how fast were drum memories?
I don't think early drums were terribly fast, but this wasn't a problem because
often they were on serial machines, and the data had to sync with the clock speed of the
machine.
I know that the Manchester Mk1 which evolved from the baby had a drum added. The design of
the drum used changed as the machine evolved. There is some info on this evolution here.
http://curation.cs.manchester.ac.uk/computer50/www.computer50.org/mark1/get…
Its interesting to note that the size of the drum was decreased to around 6" as
suggested by others.
The Ferranti Pegasus also had a drum for main storage and delay lines for
"registers".
This was a physically large drum with a capacity of 5120 40 it words.
The Ferranti Pegasus the clock speed was 333Khz and this was derived from clock tracks
written on the drum avoiding any sync problems.
However in order to achieve this transfer rate the designers built the tracks in pairs
with alternate bits coming from different tracks..
The large diameter of the drum gave problems getting consistent flying height for the
heads, which resulted in large changes in signal level.
<<<<
See
https://archive.org/stream/bitsavers_icticl1900orageSep64_1140393/1962_Drum…
for the spec on the 1962/3/4 drums for the ICT/ICL 1900 series
When we (City University) acquired our 1905E system (think it was ex Swansea uni.) it came
with a 1964 drum (0.5M 24-bit words) made, I think, by Bryant.
Took up a lot of floor space and was probably the slowest of our peripherals to come up to
"ready" from power on*
we "let it go" when we needed to install more equipment in the computer room -
don't know what happened to it.
* the big CDC at ULCC had a huge drum? - but it had its problems - any power-off, even for
a second or so, meant that it needed
a full hour to be operational again.
? or, perhaps, a special fixed disk.
Andy
2199
days inactive
2202
days old
31 comments
11 participants
participants (11)
-
abuse＠cabal.org.uk -
aek＠bitsavers.org -
andy.holt＠tesco.net -
bfranchuk＠jetnet.ab.ca -
cclist＠sydex.com -
cctalk＠ibm51xx.net -
couryhouse＠aol.com -
dave.g4ugm＠gmail.com -
elson＠pico-systems.com -
harper＠secureoutcomes.net -
paulkoning＠comcast.net