Posts by Allen Paschke

1) Message boards : Random stuff : Amicable Pairs Discovered (Message 1333)
Posted 29 Dec 2019 by Allen Paschke
Post:
About 6 months ago, on the Server Status page, the Current Progress and the ETA at Current Speed were added. This information was a great addition. Is it possible to add the number of Amicable Pairs discovered for this Part and/or during the past week and/or during the past day, on the Server Status page or elsewhere within Amicable Numbers?
2) Message boards : Number crunching : Prime Numbers (Message 1270)
Posted 28 Oct 2019 by Allen Paschke
Post:
You asked a good question. This is hard to estimate because an Amicable Numbers task does not provide any information regarding when it's computing and loading the prime table vs. when it's searching for amicable pairs.

I run Amicable Numbers on two PCs in CPU mode, each using 3 CPUs. Each task requires about 6 hours elapsed time to run, so 8 tasks are run daily across the 2 PCs. I run 8 tasks daily, 56 runs weekly, 240 tasks monthly.

At least for the first hour of each CPU task, if not longer, the prime table is being computed and loaded. During this time period on my PC, my memory utilization is very high and other apps response time is very slow. If the first hour of each task is computing and loading the prime table, then 1 hour could be saved on each task if the prime table were just simply loaded. This would result in a 16.7% reduction in CPU time per task, allowing me to run 288 tasks monthly vs. 240 monthly. If the CPU time to compute and load the prime table is longer than 1 hour, then the savings would be greater.

For a GPU task, it is my understanding that the prime table is loaded using CPU, then the search for amicable pairs is done using GPU. The savings for loading the prime table directly into a GPU task may be more significant.
3) Message boards : Number crunching : Prime Numbers (Message 1267)
Posted 27 Oct 2019 by Allen Paschke
Post:
I understand your concerns regarding corruption/overwriting of a prime table; however, there are steps which can be taken to manage this and minimize the possibility of corruption/overwriting occurring.

- When the prime table is loaded into an Amicable Numbers task, verify that the number of prime numbers is correct and utilize some type of check sum.
If the number of prime numbers or check sum is incorrect, regenerate the prime table.
- Automatically replace and regenerate the prime table on a specific interval, such as when you encounter a prime table that is older than 7, 14 or 28, etc. days.
The date and time of when the prime table was generated can be the first record on the prime table.
- Each Amicable Numbers task is run twice, as a double check.
When the double check task doesn't agree with the first run, run the task a third time.

When I think of how much time my computer spends each month generating the same prime table for each Amicable Numbers task, it is a huge amount of my computer's CPU time. Yes, it is possible that a minimal number of Amicable Pairs may not be found; however, it will really speed up 10^21 and allow getting to 10^22 much faster.
4) Message boards : Number crunching : Prime Numbers (Message 1262)
Posted 26 Oct 2019 by Allen Paschke
Post:
Each time An Amicable Numbers 10^21 task is run, all the prime numbers < 10^11 need to be stored in memory.

Are all the prime numbers < 10^11 computed each time a 10^21 task is run?
If so, is it possible to store all the prime numbers < 10^11 in a file, such as Primes.txt at c:\ProgramData\BOINC\projects\sech.me_boinc_Amicable?

Under Amicable Number preferences, there would be 3 choices
- Amicable Numbers up to 10^21, create Prime Table --- This task would be run only once
- Amicable Numbers up to 10^21 --- Each task would compute all the prime numbers < 10^11
- Amicable Numbers up to 10^21, Prime Table Exists --- The logic would have to be modified to access the Prime Table rather than compute all the prime numbers < 10^11.
5) Message boards : News : The search up to 10^20 (Message 1242)
Posted 24 Oct 2019 by Allen Paschke
Post:
Just had a second task with a "Computation Error", when the task was restarted after BOINC decided to run something else --- 15% completed.
6) Message boards : News : The search up to 10^20 (Message 1237)
Posted 23 Oct 2019 by Allen Paschke
Post:
I am running 10^21 using CPU (3 CPUs). BOINC decided to stop an Amicable Numbers task that was 95% completed to run something else, and when BOINC restarted the Amicable Numbers task, I immediately got an "Error while Computing" abort.

Has anyone else had problems when a partially completed Amicable Numbers CPU task has been restarted by BOINC?
7) Message boards : Number crunching : ETA of (search up to 10^20) (Message 1139)
Posted 19 May 2019 by Allen Paschke
Post:
Once again, Amicable Numbers has slowed down.

Here’s an update regarding the ETA of 10^20, comparing 17 May 2019 vs. 19 April 2019.

On 17 May 2019:
- The project is 89.41% completed
- The task average task runtime is 3.9 hours
- The average number of tasks in progress is 13,400
- Charity Engine is completing 6,625 tasks daily
- 0.03% of the project is being completed daily
- With no slowdown in runtime, the remaining 10.59% of the project, with 0.03% of the project being completed daily,
will be completed in 353 days, or 4 May 2020.

On 19 April 2019:
- The project was 88.52% completed
- The task average task runtime is 3.7 hours
- The average number of tasks in progress was 15,000
- Charity Engine was completing 7,350 tasks daily
- 0.04% of the project was being completed daily
- With no slowdown in runtime, the remaining 11.48% of the project, with 0.04% of the project being completed daily,
would be completed in 287 days, or 31 January 2020.

Over the past 28 days, the expected completion date, as calculated, is now 94 days further into the future, 4 May 2020 vs. 31 January 2020.
8) Message boards : Number crunching : ETA of (search up to 10^20) (Message 1130)
Posted 21 Apr 2019 by Allen Paschke
Post:
Here’s an update regarding the ETA of 10^20, comparing 19 April 2019 vs. 15 February 2019.

On 19 April 2019:
- The project is 88.52% completed
- The task average task runtime is 3.7 hours
- The average number of tasks in progress is 15,000
- Charity Engine is completing 7,350 tasks daily
- 0.04% of the project is being completed daily
- With no slowdown in runtime, the remaining 11.48% of the project, with 0.04% of the project being completed daily,
would be completed in 288 days, or 31 January 2020.

On 15 February 2019:
- The project was 85.59% completed
- The task average task runtime was 3.4 hours
- The average number of tasks in progress was 18,000
- Charity Engine was completing 8,800 tasks daily
- 0.05% of the project was being completed daily
- With no slowdown in runtime, the remaining 14.41% of the project, with 0.05% of the project being completed daily,
will be completed in 288 days, or 30 November 2019.

Over the past 63 days, the expected completion date, as calculated, is now 62 days further into the future, 31 January 2020 vs. 30 November 2019.
9) Message boards : Number crunching : ETA of (search up to 10^20) (Message 1087)
Posted 17 Feb 2019 by Allen Paschke
Post:
I did investigate Charity Engine. Very interesting. 55% of their Work Credits come from Amicable Numbers, with the remaining Work Credits scattered across 20 other Projects. Based on their Recent Average Credit, Charity Engine is completing about 8,800 Amicable Numbers tasks daily.

If you some reason, such as Charity Engine can earn more Grid Coins elsewhere, Charity Engine moves all their Amicable Numbers users to another project, Amicable Numbers would be severely impacted.
10) Message boards : Number crunching : ETA of (search up to 10^20) (Message 1081)
Posted 16 Feb 2019 by Allen Paschke
Post:
On 15 February 2019:
- The project is 85.59% completed
- The task average runtime is 3.4 hours
- 0.05% of the project is being completed daily
- With no slowdown in runtime, the remaining 14.41% of the project, with 0.05% of the project being completed daily,
will be completed in 288 days, or 30 November 2019.

In the last 7 days, the expected completion date, as calculated, is now 49 days further into the future (30 November 2019 vs. 12 October 2019).

I have the following questions and concerns regarding the project:
- How many Amicable Pairs, if any, are being found weekly?
With the 10^20 completion date probably going well into 2020, is it worth all the time and effort for a minimal number of Amicable Pairs?
- With the task runtime slowdown and the escalating completion date, how do you keep people interested in running Amicable Numbers?
A “long time” credit bonus would be beneficial, with a past or future effective date.
The “long time” credit bonus would probably need to be increased several times as Amicable Numbers continues to slow down.
- Since it’s difficult for CPU tasks to complete their run before the 72 hour deadline, how can people who can’t run GPU tasks make a contribution?
Does it make sense to initiate 10^21 tasks for CPU only.
- I’m sure other people have additional questions and concerns.

I would like to hear other people’s thoughts and opinions.
11) Message boards : Number crunching : ETA of (search up to 10^20) (Message 1079)
Posted 9 Feb 2019 by Allen Paschke
Post:
Here’s an update regarding the ETA of 10^20, comparing 18 January 2019 vs. 8 February 2019.

On 18 January 2019:
- The project was 83.59% completed
- The task average runtime was 2.7 hours
- 0.09% of the project was being completed daily
- With no slowdown in runtime, the remaining 16.41% of the project, with 0.09% of the project being completed daily,
would be completed in 182 days, or 19 July 2019.

On 8 February 2019:
- The project is 85.23% completed
- The task average runtime is 3.4 hours
- 0.06% of the project is being completed daily
- With no slowdown in runtime, the remaining 14.77% of the project, with 0.06% of the project being completed daily,
will be completed in 246 days, or 12 October 2019.

Over the past 21 days, the expected completion date, as calculated, is now 85 days further into the future, 12 October 2019 vs. 19 July 2019.
12) Message boards : Number crunching : Amicable Numbers has slowed down (Message 1078)
Posted 5 Feb 2019 by Allen Paschke
Post:
You are correct, Amicable Numbers has slowed down. Over the past week, the task average runtime has increased from 2.7 hours to 3.3 hours.
13) Message boards : Number crunching : ETA of (search up to 10^20) (Message 1061)
Posted 19 Jan 2019 by Allen Paschke
Post:
Here’s an update regarding the ETA of 10^20.

At year end:
- The project was 81.63% completed
- The task average runtime is 2.0 hours
- 0.14% of the project was being completed daily
- With no slowdown in runtime, the remaining 18.37% of the project, with 0.14% of the project being completed daily,
would be completed in 131 days, or 11 May 2019.

Now, on 18 January 2019:
- The project is 83.59% completed
- The task average runtime is 2.7 hours
- 0.09% of the project is being completed daily
- With no slowdown in runtime, the remaining 16.41% of the project, with 0.09% of the project was being completed daily,
would be completed in 182 days, or 19 July 2019.

During the first 18 days of January, the expected completion date, as calculated, is now 69 days further into the future.

With the continued runtime slowdown, 10^20 may not be completed in 2019.
14) Message boards : Number crunching : 10^21 (Message 1050)
Posted 5 Jan 2019 by Allen Paschke
Post:
Since:
- GPU tasks won't run on Intel GPUs
- CPU tasks have become almost impossible to complete within the 72 hour window

For those of us who have Intel GPUs, what are the pros / cons of opening up 10^21 exclusively for CPU tasks, while 10^20 is finished by those who run GPU?

I'd like to make a contribution to Amicable Numbers, but currently I can't run 10^20 CPU or GPU tasks.

Additionally, this would be a good test of the 10^21 CPU tasks.
15) Message boards : Number crunching : ETA of (search up to 10^20) (Message 1046)
Posted 1 Jan 2019 by Allen Paschke
Post:
I would guess late-June / early-July of 2019.

1) The average task runtime has increased from 1.5 hours in mid-December to 2.0 hours currently. --- 0.5 hour increase in 15 days.

2) The task runtime will continue to increase as we slowly spiral closer and closer to 100%.

3) Because of the significant increase in runtimes for CPU tasks, I suspect many people who were running CPU tasks have stopped running them.
(I was running 2 CPU tasks per day, which averaged 15 hours runtime per task. Now that those CPU tasks are taking 40 - 60 hours runtime, I have stopped running them).
16) Message boards : Number crunching : Amicable Numbers has slowed down (Message 1016)
Posted 11 Dec 2018 by Allen Paschke
Post:
At the bottom of the Server Status, it shows the the runtime of the last 100 tasks --- 1.50 hours average, 0.07 hours minimum and 81.97 hours maximum.

Can the runtime be classified by CPU vs. GPU?

Can the runtime be classified even further, such as:
- CPU with 2 CPUs, CPU with 3 CPUs, CPU with 4 CPUs, etc.
- AMD/ATI GPU with 2 CPUs, AMD/ATI GPU with 3 CPUs, etc.
- NVIDIA GPU with 2 CPUs, NVIDIA GPU with 3 CPUs, etc.
17) Message boards : Number crunching : Amicable Numbers has slowed down (Message 1011)
Posted 9 Dec 2018 by Allen Paschke
Post:
I just finished a CPU Task (task 20506407), with 4 CPUs, required 44 hours of CPU time over a 70 hour period to complete. I have been aborting tasks which are 4% - 5% completed after 6 - 8 hours, knowing that they will not get to 100% within 72 hours.

A month ago, CPU tasks, with 3 CPUs, required about 15 hours of CPU time to complete.

At this point, does it make sense to extend the processing window from 3 days to 4 or 5 days?
18) Message boards : Number crunching : Amicable Numbers has slowed down (Message 962)
Posted 31 Oct 2018 by Allen Paschke
Post:
I agree with you, 20 October 2018 sounds about right. Yes, it was before BOINC 7.14.2.
19) Message boards : Number crunching : Amicable Numbers has slowed down (Message 959)
Posted 31 Oct 2018 by Allen Paschke
Post:
I am running Amicable Numbers on 3 different computers, all with Microsoft Windows 10 Professional x64, in CPU Mode with 3 CPUs selected. Each computer has 4 Intel i5 processors. Historically, each task would take 10 - 12 hours to complete. Starting a week or two ago, each task started taking 20+ hours to complete. I did upgrade from BOINC 7.12.1 to BOINC 7.14.1, but I've made no other software changes to any of the 3 PCs. This seems suspicious that all 3 computers would run Amicable Numbers slower.
- Have other people has similar experiences?
- Have there been any changes to Amicable Numbers?
- Are there any parameters which I can control, so Amicable numbers will run faster.
20) Message boards : Random stuff : Social Numbers (Message 770)
Posted 19 Mar 2018 by Allen Paschke
Post:
I would assume that the Amicable Numbers project has a database with the sum of the proper divisors for many numbers.

Are there any plans to use this database, if it exists, for Social Numbers?

Here is an example of Social Numbers for a Sequence of 4:
- The sum of the proper divisors of 1,264,460 is 1,547,860
- The sum of the proper divisors of 1,547,860 is 1,727,636
- The sum of the proper divisors of 1,727,636 is 1,305,184
- The sum of the proper divisors of 1,305,184 is 1,264,460
21) Message boards : News : Credit badges for 100, 200 and 500 thousand total credit (Message 735)
Posted 13 Feb 2018 by Allen Paschke
Post:
Some projects also would give badges for the number of Amicable Pairs found --- 25, 50, 100, 250, 500, 1000, etc.

http://www.primegrid.com/forum_thread.php?id=7805#113989



©2020 Sergei Chernykh