Rebirther:
I am a CPU only cruncher of this project.
I was close to breaching the top 20.
Due to the addition of GPU project, I will never be represented fairly in my efforts.
Please correct this by modifying credits given for CPU efforts.
I know GPU's will always compute better than a CPU. But, we cannot ignore the efforts of CPU's in the pursuit of Primes.
My machines are busy recomputing Base 31. They take almost a week to complete these computations. Please consider modifying the credits to allow for Time Spent.
Best,
Prescott
/ 
Reply
Quote
rebirtherVolunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 2 Jan 13
Posts: 7479
Credit: 43,654,617
RAC: 40,497
Rebirther:
I am a CPU only cruncher of this project.
I was close to breaching the top 20.
Due to the addition of GPU project, I will never be represented fairly in my efforts.
Please correct this by modifying credits given for CPU efforts.
I know GPU's will always compute better than a CPU. But, we cannot ignore the efforts of CPU's in the pursuit of Primes.
My machines are busy recomputing Base 31. They take almost a week to complete these computations. Please consider modifying the credits to allow for Time Spent.
Best,
Prescott
Sorry to hear that. I already thought about this after starting the new subproject but I cant change it in script. You could use the stats per application. The total credits arent useful but per badge. And of course solving any Base has more honor.
/ Reply
Quote
Rebirther:
You could increase the credits per workunit per CPU job to match the Time Spent per GPU to make it a fair distribution.
I guess this would cause confusion for Actual Work Done. Since GPU's are much more powerful than a CPU.
Just a suggestion, I will continue to search for Primes.
Please let me know if I find anymore.
Best,
Prescott
/ Reply
Quote
rebirtherVolunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 2 Jan 13
Posts: 7479
Credit: 43,654,617
RAC: 40,497
Rebirther:
You could increase the credits per workunit per CPU job to match the Time Spent per GPU to make it a fair distribution.
I guess this would cause confusion for Actual Work Done. Since GPU's are much more powerful than a CPU.
Just a suggestion, I will continue to search for Primes.
Please let me know if I find anymore.
Best,
Prescott
I cant. Its always different from non-AVX to AVX, AVX to AVX2, AVX2 to AVX512 and all to GPU.
/ Reply
Quote
Thank you for letting me know.
Best,
Prescott
/ Reply
Quote
I believe Reb was very conservative when you look at Collatz credits of 30k per only 120 seconds. Examples: https://boinc.thesonntags.com/collatz/results.php?hostid=818876&offset=0&show_names=0&state=4&appid=
/ Reply
Quote
How much lower is the GPU credit going to be reduced to?
Impossible to work out the average credit for any GPU because it reduces every couple of days :(
Only 300 per WU now.
____________
/ Reply
Quote
rebirtherVolunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 2 Jan 13
Posts: 7479
Credit: 43,654,617
RAC: 40,497
How much lower is the GPU credit going to be reduced to?
Impossible to work out the average credit for any GPU because it reduces every couple of days :(
Only 300 per WU now.
The credits for TF are around 10k per hour, the last bunch of WUs for the 71-72 range were setup so the next batch would be longer as we starting 72-73 range soon, a news will follow.
/ Reply
Quote
Thank you for the information.
I could see the processing times decreasing, but couldn't see why an 8 second gain was causing 100 credit loss :)
____________
/ Reply
Quote
How much lower is the GPU credit going to be reduced to?
Impossible to work out the average credit for any GPU because it reduces every couple of days :(
Only 300 per WU now.
The credits for TF are around 10k per hour, the last bunch of WUs for the 71-72 range were setup so the next batch would be longer as we starting 72-73 range soon, a news will follow.
I understand the theory behind reducing credits as the GPU tasks get shorter, but it does not work out to getting 10k per hour all the time. If that were the case the the PPD of a GPU would stay the same every day...but it doesn't. Just saying :)
/ Reply
Quote
rebirtherVolunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 2 Jan 13
Posts: 7479
Credit: 43,654,617
RAC: 40,497
How much lower is the GPU credit going to be reduced to?
Impossible to work out the average credit for any GPU because it reduces every couple of days :(
Only 300 per WU now.
The credits for TF are around 10k per hour, the last bunch of WUs for the 71-72 range were setup so the next batch would be longer as we starting 72-73 range soon, a news will follow.
I understand the theory behind reducing credits as the GPU tasks get shorter, but it does not work out to getting 10k per hour all the time. If that were the case the the PPD of a GPU would stay the same every day...but it doesn't. Just saying :)
yes, its only estimated with 10000cr*xmin/60min on a RX 5500 XT
/ Reply
Quote
Before jumping to the 72-73 bits range here’s an interesting sub project;
-for range lower than 99M clean all up to 73 bits, I believe this can be done in one week.
https://www.mersenne.ca/status/tf/0/0/2/0
/ Reply
Quote
rebirtherVolunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 2 Jan 13
Posts: 7479
Credit: 43,654,617
RAC: 40,497
Before jumping to the 72-73 bits range here’s an interesting sub project;
-for range lower than 99M clean all up to 73 bits, I believe this can be done in one week.
https://www.mersenne.ca/status/tf/0/0/2/0
yes, but this is reserved by the GPU72 project? I will ask the admin if these are available but will be working on later in the future.
/ Reply
Quote
Before jumping to the 72-73 bits range here’s an interesting sub project;
-for range lower than 99M clean all up to 73 bits, I believe this can be done in one week.
https://www.mersenne.ca/status/tf/0/0/2/0
yes, but this is reserved by the GPU72 project? I will ask the admin if these are available but will be working on later in the future.
Please ask him. In that range you have a mix of fast and slow wus but most it will please the community in trying to find known factors for that range almost PRP’ed.
/ Reply
Quote
rebirtherVolunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 2 Jan 13
Posts: 7479
Credit: 43,654,617
RAC: 40,497
Before jumping to the 72-73 bits range here’s an interesting sub project;
-for range lower than 99M clean all up to 73 bits, I believe this can be done in one week.
https://www.mersenne.ca/status/tf/0/0/2/0
yes, but this is reserved by the GPU72 project? I will ask the admin if these are available but will be working on later in the future.
Please ask him. In that range you have a mix of fast and slow wus but most it will please the community in trying to find known factors for that range almost PRP’ed.
The answer is reserved by the GPU72 project.
/ Reply
Quote
3,900 credits per WU now and much less time wasted on stop/start and delivery.
Thanks :)
____________
/ Reply
Quote