First, I believe that there is code that limits a machine to two active transfers of any kind at the same time for all projects. This is a change from previous.

Second, I have submitted code for approval that:

1) Can, I believe distinguish between modem/laptops/etc (normally disconnected) and always connected desktops - without any extra settings.

2) Based on #1 either works harder at keeping the queue full (normally disconnected hosts) or deadline safety (always connected hosts).

3) Has a computation deadline that is report deadline - (1 day + project switch time + connect interval) for normally disconnected users or for always connected users the computation deadline is the report deadline - (1 day + project switch time).

4) Entering EDF because a deadline is within 2* the connect interval has been removed.

There are other changes as well. To which project gets the CPU if there is a (several) project(s) with slack time trouble. Projects won't switch as often before the project switch time.

And, no, I don't know when or if the code will be adopted.

[edit]

I forgot

5) Work will be reported if the time now is later than the computation deadline and you have a connection (note that it includes a full queue size in the calculation).

6) The code for determining if the cache is full takes multiple CPUs into account better. Builds an estimate for each CPU if the work is done in EDF order. Any CPU that does not have enough work adds to the global work need. No more of the case on a dual CPU where a CPDN WU fills the queue for both CPUs (it really doesn't, does it).

---

BOINC WIKI

First, I believe that there is code that limits a machine to two active transfers of any kind at the same time for all projects. This is a change from previous.

Second, I have submitted code for approval that:

1) Can, I believe distinguish between modem/laptops/etc (normally disconnected) and always connected desktops - without any extra settings.

2) Based on #1 either works harder at keeping the queue full (normally disconnected hosts) or deadline safety (always connected hosts).

3) Has a computation deadline that is report deadline - (1 day + project switch time + connect interval) for normally disconnected users or for always connected users the computation deadline is the report deadline - (1 day + project switch time).

4) Entering EDF because a deadline is within 2* the connect interval has been removed.

There are other changes as well. To which project gets the CPU if there is a (several) project(s) with slack time trouble. Projects won't switch as often before the project switch time.

And, no, I don't know when or if the code will be adopted.

[edit]

I forgot

5) Work will be reported if the time now is later than the computation deadline and you have a connection (note that it includes a full queue size in the calculation).

6) The code for determining if the cache is full takes multiple CPUs into account better. Builds an estimate for each CPU if the work is done in EDF order. Any CPU that does not have enough work adds to the global work need. No more of the case on a dual CPU where a CPDN WU fills the queue for both CPUs (it really doesn't, does it).

John,
I don't mind having to set a setting that allows me to set up as a modem user (dial up), or broadband.

My experience began with full size 4k handwired core memory, some paper punch tapes, cards, and new devices that looked like audio cassettes, and the tapes sounded really bad on home stereos....

---

John,
I don't mind having to set a setting that allows me to set up as a modem user (dial up), or broadband.

My experience began with full size 4k handwired core memory, some paper punch tapes, cards, and new devices that looked like audio cassettes, and the tapes sounded really bad on home stereos....

I don't really mind having a setting for this either. However, I was told no. In any case, it is not that hard to synthesize.

1) Do you have a network usage schedule set?
2) Do you ask for confirmation before connecting?
3) Was the last time that the network was disconnected within max (queue size, 1 week)?

If any of the above can be answered yes, the host is not always connected.

---

BOINC WIKI

I believe that the sequence was 50, 75, 100, 150, 300, 600, 900, 1200, 2400, 4800, 9600, 19200, 33K, 56K. These are the ones that I have seen, however, I may have missed a few.

Depends on a few things, like the encoding used.

Most of the early systems used more than one stop bit. Early ASCII terminals ran at 110 baud, because it took 11 bits per character (one start bit, 8 data bits, and two stop bits). 110 baud was a nice even 10 characters per second.

The lower numbers generally match up with older teletype networks and 5 bit encoding (Baudot code). 75 baud is about 100 words per minute, and 45 baud (with an odd stop bit) was 60 words per minute.

Just about everything is synchronous serial, not async. so start and stop bits are gone.

... and if we're talking about modem speed, I vaguely remember that everything over 300 baud is really bits per second (the baud rate is a lot slower).

---

John,
I don't mind having to set a setting that allows me to set up as a modem user (dial up), or broadband.

My experience began with full size 4k handwired core memory, some paper punch tapes, cards, and new devices that looked like audio cassettes, and the tapes sounded really bad on home stereos....

I don't really mind having a setting for this either. However, I was told no. In any case, it is not that hard to synthesize.

1) Do you have a network usage schedule set?
2) Do you ask for confirmation before connecting?
3) Was the last time that the network was disconnected within max (queue size, 1 week)?

If any of the above can be answered yes, the host is not always connected.

Yes John, I use Boincview to start and stop communications. (2)If I left it running, it just performs a lot of unneccesary communications on the inhouse network, and that would slow down the kids playing mp3's and movies....

---

The college I was in used keypunches (IBM 26's and 29's)

[edit]

[edit]The good thing was, I was one of the first there to teach myself FORTRAN and wound up helping teach most of the other students (and the Professor) the following year. So I had a program to list out the cards so they could find their mistakes.

yahbut... this was (starting in) 1973... the 026 was just a tad obsolete by then... of course, in my first job outa college, I was using most of IBM's tabulating machines (sorter, reproducer, and a few others I've since forgotten...) and a paper tape to mag tape machine. (non-IBM)

The university I went to in 1977 was still using a set of 026's. The computers were replaced with DEC 20s with terminals after I left.

We were using the 026's until about then... what college? I went to C.C.S.F. ... where we had about (at one time...) 12 026's and 3 029's, later adding more 029's when we expanded into a larger classroom. The noise of 12 026's running at once was incredible, particularly when a number of operators were duplicating...

---

.

Hello, from Albany, CA!...

Or if you were a prankster, scotch majic mend and a couple of dots... LOL

a show of hands, or canes, or walkers. how many here punched cardboard? that's old school computing. god i feel old..............


anyways, be nice if boinc would establish a connection, do it's business, and hangup without having too hold its hand. seems odd they left that out.

Back when data was "real data" and you could see it right there in the holes.

... and I was thinking 45 baud (and Baudot, not ASCII), and "green keys."

Al

---

Please consider a Donation to the Seti Project.

Or if you were a prankster, scotch majic mend and a couple of dots... LOL

a show of hands, or canes, or walkers. how many here punched cardboard? that's old school computing. god i feel old..............


anyways, be nice if boinc would establish a connection, do it's business, and hangup without having too hold its hand. seems odd they left that out.

Back when data was "real data" and you could see it right there in the holes.

... and I was thinking 45 baud (and Baudot, not ASCII), and "green keys."

Al

I remember the old Seti@Home client (a really old version) being able to dial out download / upload WUs and disconnect... This was 1999 though, if I remember correctly. Of course, I could be completely wrong but I do remember that feature.

---

Anyone else used a mechanical hand punch for 80-column cards - they had one plunger for each of the twelve rows. Numbers were easy - a single plunge to make a single hole - but letters needed two fingers, and punctuation three...

You mean like the IBM 010 keypunch. Or the newer IBM 011. They were electric (had an electrical plunger to drive the holes). There was a version of it that was just machanical, that I never really had to use more than a few times. While not that old, I have used the 001, IBM 016 and the model 31 a few times. Even saw a IBM 032 being used that was still in daily use in the early 1970's.

Anyone else used a mechanical hand punch for 80-column cards - they had one plunger for each of the twelve rows. Numbers were easy - a single plunge to make a single hole - but letters needed two fingers, and punctuation three...

You mean like the IBM 010 keypunch. Or the newer IBM 011. They were electric (had an electrical plunger to drive the holes). There was a version of it that was just machanical, that I never really had to use more than a few times. While not that old, I have used the 001, IBM 016 and the model 31 a few times. Even saw a IBM 032 being used that was still in daily use in the early 1970's.

Yes, very like the photo of the 010, but these were pure mechanical - no external power needed (though I think there was also a comparator/validator which must have been powered).

I came across them when I had a vacation job / work experience in the data prep. department of an insurance company c 1970: they were working on almost exclusively numeric data, so the punches were as quick to use as the electric ones.

The regulars taught me to punch with my middle finger, rather than my index finger, to avoid strain - and that's the way I type to this day!

I believe that the sequence was 50, 75, 100, 150, 300, 600, 900, 1200, 2400, 4800, 9600, 19200, 33K, 56K. These are the ones that I have seen, however, I may have missed a few.

You missed 28k8 but not bad attempt, you didn't mention that these speeds are quite often only the download speed, depending on the spec, and with some of the earlier ones the upload speed was still 75 baud. Even today the claimed 56k is only the download speed with variable upload speeds to a max of 33k.
Also not mentioned was the problem, in the days up to 2400 b/s, of communicating across the pond as there were different spec's in USA and Europe and all the tricks that had to be applied so that it didn't time out as you changed frequencies etc.

I believe the max speed for the 56K is actually 53k in the USA... and that the upload speed is proportional (and less than 1:1) to the max line speed of the connection. (max line speed depends on the quality of the phone connection in use, and can vary call to call)

---

.

Hello, from Albany, CA!...

Anyone else used a mechanical hand punch for 80-column cards - they had one plunger for each of the twelve rows. Numbers were easy - a single plunge to make a single hole - but letters needed two fingers, and punctuation three...

You mean like the IBM 010 keypunch. Or the newer IBM 011. They were electric (had an electrical plunger to drive the holes). There was a version of it that was just machanical, that I never really had to use more than a few times. While not that old, I have used the 001, IBM 016 and the model 31 a few times. Even saw a IBM 032 being used that was still in daily use in the early 1970's.

The main problem (from IBM's P.O.V.) was that the 026 was built like a tank... and easily fixable when one did go down. Of course IBM usually required ya to get a maintainence contract on the machines.

---

.

Hello, from Albany, CA!...