Re: how fast were drum memories?

[ben via cctalk]

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:
[...]

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.


YES!  Please send me a PDP-10 computer. PO BOX 1234567 CANADA.

Re: how fast were drum memories?

[Jon Elson via cctalk]

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

Re: how fast were drum memories?

[Jon Elson via cctalk]

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

Re: how fast were drum memories?

[Mike Loewen via cctalk]

   The manual says 125A per phase (3-phase, 208V) starting current.

   I remember spin-down as 20 minutes, but that was a long time ago.  :-)

On Fri, 11 May 2018, Paul Anderson wrote:


The good old days...

I recall the Q-7 drums would take about 15 minutes to come to a stop when
turned off, and would power up in seven seconds or less. Surge current was
around 115 amps?

Paul

On Thu, May 10, 2018 at 3:46 PM, Mike Loewen via cctalk <
cctalk@classiccmp.org> wrote:


On Thu, 10 May 2018, Al Kossow via cctalk wrote:

On 5/10/18 10:37 AM, Al Kossow via cctalk wrote:





On 5/10/18 9:29 AM, Paul Koning via cctalk wrote:

One that comes to mind is the DEC RS04.  It spins at roughly 3600 rpm (a

hair less, so obviously a 2 pole induction motor running off 3-phase 60 Hz
power).


Vermont Research drums (model 1175B) spun at 3450 rpm 3ph 220v, The HP
2773 on the 2000A TSB was from VR so I expect RPM
would be similar for most drums of similar diameter.



Just checked, and the LGP-30 and RPC-4000 drums are both listed as 3600
rpm



   The drums on the SAGE system (12 on each side), are listed as follows:

Diameter:   10.7"
Width:  12.5"
Weight: 105 lbs (cylinder, only), 450 lbs for entire drum assembly
Speed:  2914 rpm
Heads:  Up to 12 R/W bars, with up to 40 heads on each bar, 1
erase bar
6 pairs, one for Compuuter-to-Drum (CD), one for
Other-than-computer-to-Drum (OD)
Head spacing 0.3" apart on each bar
Drum Layout:2048 registers on 33 channels (tracks), 6 fields
Channel spacing is 0.050"
Access Time:Maximum 20ms, average 10ms
Write Current:  110ma

   The R/W bars are arranged in pairs (CD and OD) so that I/O devices can
access the drum independently of the computer.

   More than you ever wanted to know about SAGE drums (thanks, Al!):

http://bitsavers.org/pdf/ibm/sage/3-42-0_Drum_System_Sep58.pdf

   Here's one of the earlier style R/W heads:

http://q7.neurotica.com/Oldtech/SAGE/DrumHead-1L.jpg


Mike Loewen mloe...@cpumagic.scol.pa.us
Old Technology  http://q7.neurotica.com/Oldtech/





Mike Loewen mloe...@cpumagic.scol.pa.us
Old Technology  http://q7.neurotica.com/Oldtech/

Re: how fast were drum memories?

[Peter Corlett via cctalk]

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.

Re: how fast were drum memories?

[Paul Anderson via cctalk]

The good old days...

I recall the Q-7 drums would take about 15 minutes to come to a stop when
turned off, and would power up in seven seconds or less. Surge current was
around 115 amps?

Paul

On Thu, May 10, 2018 at 3:46 PM, Mike Loewen via cctalk <
cctalk@classiccmp.org> wrote:

> On Thu, 10 May 2018, Al Kossow via cctalk wrote:
>
> On 5/10/18 10:37 AM, Al Kossow via cctalk wrote:
>>
>>>
>>>
>>> On 5/10/18 9:29 AM, Paul Koning via cctalk wrote:
>>>
>>> One that comes to mind is the DEC RS04.  It spins at roughly 3600 rpm (a
 hair less, so obviously a 2 pole induction motor running off 3-phase 60 Hz
 power).

>>> Vermont Research drums (model 1175B) spun at 3450 rpm 3ph 220v, The HP
>>> 2773 on the 2000A TSB was from VR so I expect RPM
>>> would be similar for most drums of similar diameter.
>>>
>>
>> Just checked, and the LGP-30 and RPC-4000 drums are both listed as 3600
>> rpm
>>
>
>The drums on the SAGE system (12 on each side), are listed as follows:
>
> Diameter:   10.7"
> Width:  12.5"
> Weight: 105 lbs (cylinder, only), 450 lbs for entire drum assembly
> Speed:  2914 rpm
> Heads:  Up to 12 R/W bars, with up to 40 heads on each bar, 1
> erase bar
> 6 pairs, one for Compuuter-to-Drum (CD), one for
> Other-than-computer-to-Drum (OD)
> Head spacing 0.3" apart on each bar
> Drum Layout:2048 registers on 33 channels (tracks), 6 fields
> Channel spacing is 0.050"
> Access Time:Maximum 20ms, average 10ms
> Write Current:  110ma
>
>The R/W bars are arranged in pairs (CD and OD) so that I/O devices can
> access the drum independently of the computer.
>
>More than you ever wanted to know about SAGE drums (thanks, Al!):
>
> http://bitsavers.org/pdf/ibm/sage/3-42-0_Drum_System_Sep58.pdf
>
>Here's one of the earlier style R/W heads:
>
> http://q7.neurotica.com/Oldtech/SAGE/DrumHead-1L.jpg
>
>
> Mike Loewen mloe...@cpumagic.scol.pa.us
> Old Technology  http://q7.neurotica.com/Oldtech/
>

Re: how fast were drum memories?

[Chuck Guzis via cctalk]

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

Re: how fast were drum memories?

[Mike Loewen via cctalk]

On Thu, 10 May 2018, Al Kossow via cctalk wrote:


On 5/10/18 10:37 AM, Al Kossow via cctalk wrote:



On 5/10/18 9:29 AM, Paul Koning via cctalk wrote:


One that comes to mind is the DEC RS04.  It spins at roughly 3600 rpm (a hair 
less, so obviously a 2 pole induction motor running off 3-phase 60 Hz power).

Vermont Research drums (model 1175B) spun at 3450 rpm 3ph 220v, The HP 2773 on 
the 2000A TSB was from VR so I expect RPM
would be similar for most drums of similar diameter.


Just checked, and the LGP-30 and RPC-4000 drums are both listed as 3600 rpm


   The drums on the SAGE system (12 on each side), are listed as follows:

Diameter:   10.7"
Width:  12.5"
Weight: 105 lbs (cylinder, only), 450 lbs for entire drum assembly
Speed:  2914 rpm
Heads:  Up to 12 R/W bars, with up to 40 heads on each bar, 1 erase bar
6 pairs, one for Compuuter-to-Drum (CD), one for
Other-than-computer-to-Drum (OD)
Head spacing 0.3" apart on each bar
Drum Layout:2048 registers on 33 channels (tracks), 6 fields
Channel spacing is 0.050"
Access Time:Maximum 20ms, average 10ms
Write Current:  110ma

   The R/W bars are arranged in pairs (CD and OD) so that I/O devices can 
access the drum independently of the computer.


   More than you ever wanted to know about SAGE drums (thanks, Al!):

http://bitsavers.org/pdf/ibm/sage/3-42-0_Drum_System_Sep58.pdf

   Here's one of the earlier style R/W heads:

http://q7.neurotica.com/Oldtech/SAGE/DrumHead-1L.jpg


Mike Loewen mloe...@cpumagic.scol.pa.us
Old Technology  http://q7.neurotica.com/Oldtech/

Re: how fast were drum memories?

[Jack Harper via cctalk]

...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:
>
>
> 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


--
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. 

Re: how fast were drum memories?

[Al Kossow via cctalk]

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
> 
> 
> 
> 

Re: how fast were drum memories?

[Al Kossow via cctalk]

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

Re: how fast were drum memories?

[Jack Harper via cctalk]

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. 

Re: how fast were drum memories?

[Chuck Guzis via cctalk]

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

RE: how fast were drum memories?

[Ali via cctalk]

> 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

Re: how fast were drum memories?

[ben via cctalk]

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.

Re: how fast were drum memories?

[Chuck Guzis via cctalk]

"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

Re: how fast were drum memories?

[Paul Berger via cctalk]

On 2018-05-10 4:01 PM, Al Kossow via cctalk wrote:


On 5/10/18 10:37 AM, Al Kossow via cctalk wrote:


On 5/10/18 9:29 AM, Paul Koning via cctalk wrote:


One that comes to mind is the DEC RS04.  It spins at roughly 3600 rpm (a hair 
less, so obviously a 2 pole induction motor running off 3-phase 60 Hz power).

Vermont Research drums (model 1175B) spun at 3450 rpm 3ph 220v, The HP 2773 on 
the 2000A TSB was from VR so I expect RPM
would be similar for most drums of similar diameter.

Just checked, and the LGP-30 and RPC-4000 drums are both listed as 3600 rpm



The 4" drum in a IBM 650 is said to have rotated at 12,500 RPM.

Re: how fast were drum memories?

[Al Kossow via cctalk]

On 5/10/18 10:37 AM, Al Kossow via cctalk wrote:
> 
> 
> On 5/10/18 9:29 AM, Paul Koning via cctalk wrote:
> 
>> One that comes to mind is the DEC RS04.  It spins at roughly 3600 rpm (a 
>> hair less, so obviously a 2 pole induction motor running off 3-phase 60 Hz 
>> power).
> Vermont Research drums (model 1175B) spun at 3450 rpm 3ph 220v, The HP 2773 
> on the 2000A TSB was from VR so I expect RPM
> would be similar for most drums of similar diameter.

Just checked, and the LGP-30 and RPC-4000 drums are both listed as 3600 rpm

RE: how fast were drum memories?

[Rob Jarratt via cctalk]

The fixed head disc used on MU5 had a 20.5ms revolution time and an inner
track data rate of 2.2Mbytes/sec.

Regards

Rob 

> -Original Message-
> From: cctalk [mailto:cctalk-boun...@classiccmp.org] On Behalf Of Paul
> Koning via cctalk
> Sent: 10 May 2018 17:30
> To: Grif <grif...@mindspring.com>
> Cc: General Discussion: On-Topic and Off-Topic Posts
> <cctalk@classiccmp.org>
> Subject: Re: how fast were drum memories?
> 
> 
> 
> > On May 10, 2018, at 10:37 AM, Grif <grif...@mindspring.com> wrote:
> >
> >
> > I wonder how the late generation paging disks (fixed head per track)
like
> DG used in the 80's compared?
> 
> One that comes to mind is the DEC RS04.  It spins at roughly 3600 rpm (a
hair
> less, so obviously a 2 pole induction motor running off 3-phase 60 Hz
power).
> Transfer rate is 0.5 megabytes per second, but still, at that RPM, average
> latency is 8.5 ms.
> 
>   paul

Re: how fast were drum memories?

[Al Kossow via cctalk]

there is a picture of a small 10K RPM drum here
http://museum.ipsj.or.jp/en/computer/device/magnetic_drum/0017.html

On 5/10/18 10:37 AM, Al Kossow via cctalk wrote:
> 
> 
> On 5/10/18 9:29 AM, Paul Koning via cctalk wrote:
> 
>> One that comes to mind is the DEC RS04.  It spins at roughly 3600 rpm (a 
>> hair less, so obviously a 2 pole induction motor running off 3-phase 60 Hz 
>> power).
> Vermont Research drums (model 1175B) spun at 3450 rpm 3ph 220v, The HP 2773 
> on the 2000A TSB was from VR so I expect RPM
> would be similar for most drums of similar diameter.
> 
> from 
> https://ia800809.us.archive.org/27/items/TNM_Drum_memories_-_Vermont_Research_Corp_1966_20170826_0105
> 
> 
> 
> 
> 

Re: how fast were drum memories?

[Al Kossow via cctalk]

On 5/10/18 9:29 AM, Paul Koning via cctalk wrote:

> One that comes to mind is the DEC RS04.  It spins at roughly 3600 rpm (a hair 
> less, so obviously a 2 pole induction motor running off 3-phase 60 Hz power).
Vermont Research drums (model 1175B) spun at 3450 rpm 3ph 220v, The HP 2773 on 
the 2000A TSB was from VR so I expect RPM
would be similar for most drums of similar diameter.

from 
https://ia800809.us.archive.org/27/items/TNM_Drum_memories_-_Vermont_Research_Corp_1966_20170826_0105

Re: how fast were drum memories?

[ANDY HOLT via cctalk]

>>>> from  "Dave Wade via cctalk" <cctalk@classiccmp.org>
To: "General Discussion: On-Topic and Off-Topic Posts" <cctalk@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/gethomas/manchester_drums.html

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_Storage_Sep64_djvu.txt
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

RE: how fast were drum memories?

[Dave Wade via cctalk]

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/gethomas/manchester_drums.html

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-boun...@classiccmp.org> On Behalf Of Ed Sharpe via
> cctalk
> Sent: 10 May 2018 16:51
> To: paulkon...@comcast.net; cctalk@classiccmp.org; cctalk@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  DRUMAS  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@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

Re: how fast were drum memories?

[Paul Koning via cctalk]

> On May 10, 2018, at 11:33 AM, Peter Corlett via cctalk 
>  wrote:
> 
> 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.

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

Re: how fast were drum memories?

[Ed Sharpe via cctalk]

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  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@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

Re: how fast were drum memories?

[Paul Koning via cctalk]

> On May 10, 2018, at 10:37 AM, Grif  wrote:
> 
> 
> I wonder how the late generation paging disks (fixed head per track) like DG 
> used in the 80's compared?

One that comes to mind is the DEC RS04.  It spins at roughly 3600 rpm (a hair 
less, so obviously a 2 pole induction motor running off 3-phase 60 Hz power).  
Transfer rate is 0.5 megabytes per second, but still, at that RPM, average 
latency is 8.5 ms.

paul

Re: how fast were drum memories?

[Jack Harper via cctalk]

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:

> 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


--
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. 

Re: how fast were drum memories?

[Paul Koning via cctalk]

> On May 10, 2018, at 10:57 AM, Chuck Guzis via cctalk  
> wrote:
> 
> 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.

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

Re: how fast were drum memories?

[Jon Elson via cctalk]

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

Re: how fast were drum memories?

[Ed Sharpe via cctalk]

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@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

Re: how fast were drum memories?

[Peter Corlett via cctalk]

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.

Re: how fast were drum memories?

[Chuck Guzis via cctalk]

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

Re: how fast were drum memories?

[Bill Degnan via cctalk]

There are two places I'd check.  The manual for the Royal McBee LGP-30 and
the book Computer Structures: Readings and Examples by G Bell et al.
Bill

On Thu, May 10, 2018 at 10:37 AM, Grif via cctalk <cctalk@classiccmp.org>
wrote:

>
> I wonder how the late generation paging disks (fixed head per track) like
> DG used in the 80's compared?
>
> -Original Message-
> >From: Paul Koning via cctalk <cctalk@classiccmp.org>
> >Sent: May 10, 2018 7:29 AM
> >To: "General Discussion: On-Topic and Off-Topic Posts" <
> cctalk@classiccmp.org>
> >Subject: how fast were drum memories?
> >
> >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
> >
>

Re: how fast were drum memories?

[Grif via cctalk]

I wonder how the late generation paging disks (fixed head per track) like DG 
used in the 80's compared?

-Original Message-
>From: Paul Koning via cctalk <cctalk@classiccmp.org>
>Sent: May 10, 2018 7:29 AM
>To: "General Discussion: On-Topic and Off-Topic Posts" <cctalk@classiccmp.org>
>Subject: how fast were drum memories?
>
>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
>

how fast were drum memories?

[Paul Koning via cctalk]

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