Re: Is Parallel Programming Hard, And, If So, What Can You Do About It?

On 5/21/2025 12:39 AM, Lawrence D'Oliveiro wrote:

On Tue, 20 May 2025 20:54:08 -0700, Stephen Fuld wrote:
 

And furthermore, you saw earlier that the host can turn on or off,
either for all writes or a particular write, the completion return
before the data is on the disk.

 The host can *request* such a thing, nothing more. And of course the drive
can answer in the affirmative.
 Whether a suitable action actually takes place is another matter entirely.

No need to FUD the HDD manufacturers.
In this case, they would stand to gain little from ignoring these requests, but ignoring these requests would make the drives much less attractive on the market.
Cutting this corner is unlikely to go unnoticed, as people would likely notice if the drives were prone to irrecoverable data corruption to the filesystem journals and similar in the case of power disruptions and similar.
Well, nevermind possibly Seagate, whose seeming market strategy is to make drives that have crappy reliability, occasional issues with data corruption, and that are prone to fail after a few years.
Guess (as some executive rubs their hands together):
"Man, these drives last too long, we need to make them less reliable..."
"Then people will buy new drives..."
"And they will choose us because the drives are cheaper...".
Then me a few times, "Well, maybe Seagate drives have stopped being crap? Surely they can't stay in business if their drives are still crap." Drive then fails after a few years, so, still crap...
They will be like, "Hey... Look at this drive, it is both big and cheap..." people will buy them, but them the power of Seagate and the power of SMR combine to make reliability suck (at least assuming the drive doesn't just randomly decide it doesn't want to work anymore and becomes entirely unresponsive after 2 or 3 years).
Once again leaving one to wonder, "If the drives are this crap, why do people still buy from them?...".
Like, it can't really just be that I get unlucky every time I consider to try giving them another chance...
Meanwhile, WD's CMR drives seem to be fairly reliable.
Though, SMR is more of a mixed bag, more TB/$, worse performance and reliability.
Seemingly SanDisk is better for SSDs though.
For HDDs, have mostly been having good results with WD Red drives (my PC has several of them).