Couldn't help but reanimate an old thread. Incidentally, this is why I was asking about programming and debugging multiple props with a single PC.
I thought I would look at more numbers and p'raps add to the list of big numbers in the Collatz Wikipedia article. So I built a mini-Big Brain. Nine MiniProp slaves looking at about 1.3 billion seeds per hour. Each slave uses 6 PAM cogs to examine numbers (total of 54 cogs in parallel), one PASM cog to issue numbers, and one spin cog to talk to the master prop. The master propellor issues groups of numbers to the slaves and keeps track of the seeds that require the most iterations. Each time a bigger one is found, it is displayed on the LCD.
Couldn't help but reanimate an old thread. Incidentally, this is why I was asking about programming and debugging multiple props with a single PC.
I thought I would look at more numbers and p'raps add to the list of big numbers in the Collatz Wikipedia article. So I built a mini-Big Brain. Nine MiniProp slaves looking at about 1.3 billion seeds per hour. Each slave uses 6 PAM cogs to examine numbers (total of 54 cogs in parallel), one PASM cog to issue numbers, and one spin cog to talk to the master prop. The master propellor issues groups of numbers to the slaves and keeps track of the seeds that require the most iterations. Each time a bigger one is found, it is displayed on the LCD.
I think that's great, can't wait to see updates on this, couldn't duplicate because of the number of Propeller Mini's. I bet it does have a feeling of great power. Is your program nine Propeller Mini's specific?
I think that's great, can't wait to see updates on this, couldn't duplicate because of the number of Propeller Mini's. I bet it does have a feeling of great power. Is your program nine Propeller Mini's specific?
Great Job taking this to the extreme.
The nine mini's are running the exact same program. I used a spi interface; each has a unique chip select.
The nine mini's are running the exact same program. I used a spi interface; each has a unique chip select.
"Anything worth doing is worth overdoing."
Well you went and done it now, from now on any posts here by you will require a shot of your LCD and current progression.
How long do you expect this could run? Does it have BU power in case of an outage? Hate to see this giant number cruncher loose its place. This is on the level of the P2 with all the cores involved.
One trillion I understand, trying to fathom al the processing involved to get there in almost 38 days is something else.
Using 9 +1 Propeller's in parallel processing is cool in it self. But as Phil said, you have useful, meaningful output. Did you also log data? Probably not, that would require an additional memory device or a PC running 24/7.
What is the lllonnggg.. . term on this, will it overflow? Or left alone, go from one trillion to the next. And eventually years from now reach a 1 zillion seed.
I forgot what comes after 999 trillion, will have to look at the national debt and maybe get an idea.
Because of the progression: million, billion, trillion, quadrillion, quintillion, sextillion, etc...
Thanks Seaith for that brain refreshing.
I knew that at one time, school days maybe. It didn't sink in because I couldn't relate to it at the time. No means except pencil and paper, nothing exciting there. Had to be pushed to "+,-,×,÷. And then came fractions / decimal's, girl's. Well you know, "Pay Attention In School" and don't waste the teachers time.
Hi Tom, It has been awhile, is your big ten Propeller mini-brain still collatz conjecturing? I always look forward to your updates.
Mike, I got to 1,444,338,092,271 (1408 cycles) after 53 days and realized that the next interesting one would be another 30 days and the next after that another 40 days and so on exponentially. "Soon" they would come hundreds of days apart and then thousands. So my mini-BigBrain is sitting on the shelf waiting for another computational chore.
What's a matter Heater, bet you wouldn't leave your Linux machine running 24/7 over something like this. If it isn't done under a few seconds you get board, am I right?
Au contraire, Linux systems I built 10 years ago, OS and software not hardware, have been up and running for ten years now. Many hundreds of them in 5 different countries. New ones coming to the USA as we speak.
Well, OK, they are not doing "something like this". They control things that people think are important.
Well, other OK, at home nothing stays up very long. I think the longest was a Raspberry Pi and camera pushing images to the internet for half a year. It finally failed when it's SD card went on fire. No really.
You perhaps have a point though, unless someone is paying me to be dedicated to a task my interests wonder around pretty quickly.
Anyway, my "good grief" was about the fact that in 53 days we get: 53 * 24 * 60 * 60 * 1000000 * 20 = 91584000000000 Propeller PASM instructions executed in a single COG.
This Collatz run got up to 1,444,338,092,271
So I make that 91584000000000 / 1444338092271 = 63.4
Instructions per test.
Well, that does not seem anything like enough. I'm wondering how many Props were working on this.
Comments
I thought I would look at more numbers and p'raps add to the list of big numbers in the Collatz Wikipedia article. So I built a mini-Big Brain. Nine MiniProp slaves looking at about 1.3 billion seeds per hour. Each slave uses 6 PAM cogs to examine numbers (total of 54 cogs in parallel), one PASM cog to issue numbers, and one spin cog to talk to the master prop. The master propellor issues groups of numbers to the slaves and keeps track of the seeds that require the most iterations. Each time a bigger one is found, it is displayed on the LCD.
Nicely done.
-Phil
LOL
Excellent.
Whatever happened to the "Big Brain" guy?
http://humanoidolabs.blogspot.com/search?updated-min=2016-01-01T00:00:00-08:00&updated-max=2017-01-01T00:00:00-08:00&max-results=1
He must be busy building a new lab in a place that does not have internet, (the moon)?
I think that's great, can't wait to see updates on this, couldn't duplicate because of the number of Propeller Mini's. I bet it does have a feeling of great power. Is your program nine Propeller Mini's specific?
Great Job taking this to the extreme.
The nine mini's are running the exact same program. I used a spi interface; each has a unique chip select.
"Anything worth doing is worth overdoing."
Well you went and done it now, from now on any posts here by you will require a shot of your LCD and current progression.
How long do you expect this could run? Does it have BU power in case of an outage? Hate to see this giant number cruncher loose its place. This is on the level of the P2 with all the cores involved.
Top Line is the seed
2nd Line is slave number, cog number, and iteration count
3rd line is elapsed time
4th line is beginning of last batch issued
Using 9 +1 Propeller's in parallel processing is cool in it self. But as Phil said, you have useful, meaningful output. Did you also log data? Probably not, that would require an additional memory device or a PC running 24/7.
What is the lllonnggg.. . term on this, will it overflow? Or left alone, go from one trillion to the next. And eventually years from now reach a 1 zillion seed.
I forgot what comes after 999 trillion, will have to look at the national debt and maybe get an idea.
Funny how many people think it's a zillion.
I wonder how quad got in there.
Tom has ten Propeller's tied up in this, who will give in first? Propeller overflow or Tom having to re-purpose some of his hardware.
Because of the progression: million, billion, trillion, quadrillion, quintillion, sextillion, etc...
Thanks Seaith for that brain refreshing.
I knew that at one time, school days maybe. It didn't sink in because I couldn't relate to it at the time. No means except pencil and paper, nothing exciting there. Had to be pushed to "+,-,×,÷. And then came fractions / decimal's, girl's. Well you know, "Pay Attention In School" and don't waste the teachers time.
-Phil
Thanks Phil, I saved a screenshot of your post for reference.
-Phil
Hah. Maybe it's enough that a Billiard is 1e15. I'm definitely storing this one away for pub trivia...
Mike, I got to 1,444,338,092,271 (1408 cycles) after 53 days and realized that the next interesting one would be another 30 days and the next after that another 40 days and so on exponentially. "Soon" they would come hundreds of days apart and then thousands. So my mini-BigBrain is sitting on the shelf waiting for another computational chore.
Wow, they have been working very hard.
That's 17 character's on one line of the 20 characters per line LCD, if your including separator's.
You must have the mother-load of mini's. This project deserves a cabinet of it's own.
Do you have the power supply on a BUs?
That's a lot of computation progress to loose over a mains glitch.
What's a matter Heater, bet you wouldn't leave your Linux machine running 24/7 over something like this. If it isn't done under a few seconds you get board, am I right?
Well, OK, they are not doing "something like this". They control things that people think are important.
Well, other OK, at home nothing stays up very long. I think the longest was a Raspberry Pi and camera pushing images to the internet for half a year. It finally failed when it's SD card went on fire. No really.
You perhaps have a point though, unless someone is paying me to be dedicated to a task my interests wonder around pretty quickly.
This Collatz run got up to 1,444,338,092,271
So I make that 91584000000000 / 1444338092271 = 63.4
Instructions per test.
Well, that does not seem anything like enough. I'm wondering how many Props were working on this.