10 WU = 13Gb+ of RAM. It's good, that i have 32Gb, but not all people have 32Gb.
What is wrong with running a smaller number of tasks if you don't have enough memory to run 10 at once? What is wrong with adding more memory if you really want (and are able) to run more?
I would guess that the analysis of big data may not be able to be done with the same efficiency if memory available to each task was too restricted.
Solarrus wrote:
It will be good, to make WU smaller. 4-6 hours is too long.
What is the difference between running 6 tasks (one after the other) that take 1 hr each as opposed to running one 6 hr task? There is one big difference you should think about. The 6 hr task needs one database record to store the info about it. Each one hr task needs a record so 6 records in total for the same amount of work. The database would need to be 6 times larger and would need more powerful servers to manipulate it with the same speed and efficiency.
It's good to consider what effect your wishes might have on how the project operates.
Do you want to finish your research faster? Or time is not an important value?
1) I have 10 core cpu. I can use them all, or half of them. If u don't care - i can use 5, or 2. I can use 1 core. Is that "effect you wishes"?
2) I have 32GB of RAM. It's not fast. I thought to buy 16Gb of more faster RAM. Many PPL have 16GB of RAM. Why i do not want to buy 32GB of fast RAM, or even 64GB? Because it is 1/4 of my fee. 64Gb will cost 1/2 of my MONTH fee. Not all ppl live at EU or US.
I've already spent my month fee for CPU+MB.
3) I've run boinc from time to time. When i can. As many ppl. Short WU are good, because, you do not need to wait 6 hours while calculation will be finished, if situation will change.
Milkyway@home is good at this case. Short, multy thread WU.
Solarrus wrote:10 WU = 13Gb+
)
What is wrong with running a smaller number of tasks if you don't have enough memory to run 10 at once? What is wrong with adding more memory if you really want (and are able) to run more?
I would guess that the analysis of big data may not be able to be done with the same efficiency if memory available to each task was too restricted.
What is the difference between running 6 tasks (one after the other) that take 1 hr each as opposed to running one 6 hr task? There is one big difference you should think about. The 6 hr task needs one database record to store the info about it. Each one hr task needs a record so 6 records in total for the same amount of work. The database would need to be 6 times larger and would need more powerful servers to manipulate it with the same speed and efficiency.
It's good to consider what effect your wishes might have on how the project operates.
Cheers,
Gary.
Do you want to finish your
)
Do you want to finish your research faster? Or time is not an important value?
1) I have 10 core cpu. I can use them all, or half of them. If u don't care - i can use 5, or 2. I can use 1 core. Is that "effect you wishes"?
2) I have 32GB of RAM. It's not fast. I thought to buy 16Gb of more faster RAM. Many PPL have 16GB of RAM. Why i do not want to buy 32GB of fast RAM, or even 64GB? Because it is 1/4 of my fee. 64Gb will cost 1/2 of my MONTH fee. Not all ppl live at EU or US.
I've already spent my month fee for CPU+MB.
3) I've run boinc from time to time. When i can. As many ppl. Short WU are good, because, you do not need to wait 6 hours while calculation will be finished, if situation will change.
Milkyway@home is good at this case. Short, multy thread WU.