search

madMAx43v3r / chia-gigahorse

221 stars 32 forks source link

Problem with plots #51

Open zxc5558769 opened 1 year ago

zxc5558769 commented 1 year ago

Question1： When making compressed plots，there is a problem.The first polt always goes well,since it is finished,the process start to copy the plot from tmpdir to finaldir,and at the same time the program start to making the secend plot.When the process comes to [P3]Table 5 LPSK,the process stopped until the copy process of the first plot is finished.It is a waste of time. Since the performance of SSD is good enough in both reading and writing performance for the process,how could I make the process not stopped when it comes to [P3]Table 5 LPSK.

zxc5558769 commented 1 year ago

Question2： In P1,the speed of my SSD is not stabile in each process, some times avg 1.8G/s and some times avg 0.6G/s,the model of SSD is sumsung PM1733 3.84t.Technically it should not be like this.

zxc5558769 commented 1 year ago

Question3： In some case, the problem in Question1 did not occured. For example,while the first plot is in the final step that copy from tmpdir to finaldir,the second plot is complete as well.After the data copy into the tmpdir in the second plot,the third plot process will stopped at [P1] Table 3"waited for desk flush".Actually the flush is already finished,and all the data is in tmpdir.Why dont the tmpdir copy plots to different finaldir at the same time, instead of waiting the copy of first plot to complete.

madMAx43v3r commented 1 year ago

There is no explicit waiting in the code. This behavior has been observed in Windows before.

zxc5558769 commented 1 year ago

There is no explicit waiting in the code. This behavior has been observed in Windows before.

OK.May I ask you why the first plot process runs faster than each process after it.For example,in P1 of the first process the -2 ssd write avg 1.8g/m but in the second one the speed droped to avg 0.8g/m which cost two more minutes in p1. Ps:the temperature of sad is 45 degrees

FarmingKid commented 1 year ago

I had the same problem. I had in use a Crucial 500GB NvME. the first plot goes well times around 20sec per Table. the plots after that much slower around 100sec per table. The Active Time on the NvME was everytime at 100% at use and the speed slows down to 100mb/s. I changed the nvme to an Crucial 1TB NvME and after that no problems. Maybe you give a better look at your ssd again