Restrictive CPU usage constrains productivity and wastes tons of time and data

The best solution for finding and removing duplicate files.
Post Reply
Callistemon
Posts: 85
Joined: Fri Jun 25, 2021 5:15 am

Restrictive CPU usage constrains productivity and wastes tons of time and data

Post by Callistemon »

Because the Search Indexer in Windows 10 is sloppy enough to index exclusions and prevent external devices from being safely removed, I now use Duplicate Cleaner in a Windows 11 22543 virtual machine on Linux, since the Search Indexer finally was fixed recently. While indexing exclusions, encountering corruption causes Explorer and DWM to fall apart, which cannot be cleared even by restarting those processes. Unstable Windows cannot be installed physically, as although it's just alright at this time, it could become buggy and remain that way for months. It also requires optional diagnostics, which is not something I want enabled on a computer with a rubbish NVIDIA GPU causing blue screens to export all the memory to Microsoft. In this VM, the optional spying is not a problem, as there are no NVIDIA screens to cause the optional spying to send back all the memory.

The slowness of virtualization is not a problem, until Duplicate Cleaner decides to constrict to a fraction of an already fraction. The smaller CPU capacity of the virtual machine would not have an effect with the slower HDDs if not for an even smaller 20% cap. I am already using a virtual machine, so if I wanted to restrict CPU usage, I could restrict the VM itself. The disk usage is only around 30% each and it is expected to take 18 more hours. I am so tired and done with these operations continuing for months, and there is only one solution, which is clear. It is for Duplicate Cleaner to stop being paranoid about the possibility of not being the most deserving program of CPU, when it in fact is. This has occupied both the physical computer and the virtual machine for no benefit of any type. All the hours while the computer is running throughout the night, there is absolutely no need to limit the CPU usage, and the same applies for most of the day.

One of the reasons why I am not using Linux is because buffered writing in Windows is much faster with SMR disks. With all HDDs, less actuation to further destinations and back is quicker, but with SMR, tracks are written more efficiently if further tracks are clear. Without buffered writing, defragmentation during transfer would result in data being written further out, and being encountered later while returning to those areas. When there is data in the next tracks, multiple tracks need to be rewritten at once to prevent tracks from being overwritten. With buffered writing, data that would be written further out and later encountered, instead remains in the memory until the majority of the operation reaches that area. The SMR tracks can be smoothly written like CMR, since there are less used tracks to encounter and work around. This makes the difference between 7 MB/s on Linux, and 80 MB/s on Windows with productive writing. However, it looks as Duplicate Cleaner is cancelling that benefit of quicker writing, from the reading side.

Linux adds files to the Master File Table before finishing the fill, resulting in a good HDD becoming Corrupted and Unreadable multiple times, even when the transfer is cancelled properly and the computer was restarted properly. This requires reformatting away all partitions, which cannot be done when the HDD contains data. I cannot risk corrupting an HDD with data, as backing up the data to another HDD has been made difficult by the slowness of Duplicate Cleaner. If Linux was as desirable, I would be ready to use a competitor duplicating product by now. Enough is enough. Duplicate Cleaner is 80% malfunctioning as of now. I may need to assemble a large desktop computer with multiple server CPUs to use Duplicate Cleaner, even though this AMD Ryzen 7 5700U is plenty when used properly. That would cost more than all the HDDs that I have ever and will ever purchase. There is no way out.
Post Reply