General discussions, chit-chat
August 31st, 2026, 18:04
Some Reddit user has been spamming the Internet with claims that his 64GB SATA SanDisk P4 SSD has reported 9PB of host writes. This appears to be the thread that launched it all:
https://old.reddit.com/r/HDD/comments/1um1lee/ssd_firmware_crashes_and_controller_failures_are/This is his latest thread with his most recent results:
https://www.reddit.com/r/HDD/comments/1w05xux/9_petabytes_and_still_going_endurance_16_year_old/This is my latest summary:
https://old.reddit.com/r/HDD/comments/1w0zt0e/9_petabytes_and_still_peddling_nonsense/He has blocked me from responding to his threads.
My theory is that his SSD is counting TRIM commands as host writes. Does anyone have an alternative explanation?
August 31st, 2026, 18:39
SSD TRIM is an operating system command that tells your solid-state drive which data blocks are no longer in use so it can erase them ahead of time for faster future writes. With trimming, the SSD can efficiently clear unneeded data that would otherwise slow future write operations.
August 31st, 2026, 19:36
It seems sending TRIM commands is all he's doing unless Windows defrag does some other shit too.
I'm going ro send some TRIM commands and see if host writes goes up ..
August 31st, 2026, 19:52
Hardcore Games wrote:SSD TRIM is an operating system command that tells your solid-state drive which data blocks are no longer in use so it can erase them ahead of time for faster future writes. With trimming, the SSD can efficiently clear unneeded data that would otherwise slow future write operations.
we know this, einstein
August 31st, 2026, 20:10
A painless test would be a quick format under Windows (NTFS). The host writes should increase by a few megabytes, or 64GB if TRIM-ed sectors are counted as writes.
August 31st, 2026, 20:26
If I tell it to TRIM an external SATA SSD, host writes increases by 1 (LBA I suppose).
August 31st, 2026, 21:14
Arch Stanton wrote:If I tell it to TRIM an external SATA SSD, host writes increases by 1 (LBA I suppose).
Then that explains it. It seems that his reference to cache was a red herring. He must know that his exercise is phony, so what is his purpose? Maybe he's just looking for notoriety and more clicks for Youtube.
September 1st, 2026, 5:03
That I am wondering too. He's basically abusing SMART to tell us SMART tools are flawed.. As if we weren't having this debate already 20 years ago.. He seems very keen on pointing out on what tech websites and magazines his experiment is mentioned.. I wonder what a psychologist would make of those hand written notes.
September 1st, 2026, 10:59
Just to be clear, if you send 1 TRIM command to unmap 100 sectors, does the count of host writes increment by 1 or 100?
September 1st, 2026, 13:36
I ask Windows to TRIM all free space. Many more than 100 sectors, I see SMART counter go up by one. I only have 1 SATA SSD I can try this on.
September 1st, 2026, 16:25
i think it is a bit hard to control how many sectors the OS trims, so you'd be better off with sending the commands yourself and checking the smart afterwards...
September 1st, 2026, 17:50
I do it same way as the guy in the experiment. I just wanted to see correlation TRIM <> Host writes SMART parameter. Furthermore I was doing it on drive I am using.
September 2nd, 2026, 9:59
This is the thread that started him on his adventure:
https://www.techpowerup.com/forums/threads/i-forced-a-16-year-old-64gb-ssd-to-write-1-petabyte-and-it-didnt-die.350145/He has since spammed several other forums with updated results that have no reference to the original thread. It seems that his intro is basically just technobabble. His intention seems to be to prove that SSD controllers are robust. To me, he's just proving what everyone already knows. Moreover, if these writes truly are "ghost data" writes or "virtual cache operations" which don't arrive at the NAND flash, then why bother sending TRIM commands at all?
Maybe there is a more detailed explanation in his hand-written scrawl, but my eyes hurt when I try to read it.
He writes "if max is 18gb per trim", which suggests that he doesn't have control over the TRIM operations. AIUI, the DATA SET MANAGEMENT ATA command requires that the host specifies the LBA ranges that are to be unmapped, so who/what is in control of TRIM-ing?
By exploiting a massive storage telemetry loophole in Windows 11 using a high velocity 5 second macro loop, I managed to trick the system into aggressively logging massive "ghost data" writes directly to the drive's odometer. Because these are virtual cache operations, the actual data vanishes into thin air before it can physically destroy the legacy MLC NAND flash cells.
my project completely debunks standard factory TBW limits as a hardware kill-switch and shows just how incredibly resilient legacy, over-engineered storage controllers really are.
I officially obliterated the 32-bit integer limit weeks ago. Next Testing the limits of a 48-bit register!
Remember its a telemetry and firmware test, not a physical destruction test.
The purpose isn't to burn the physical NAND to ashes; it's to see how a budget 2010 storage brain handles modern, enterprise-level digital stress under Windows 11. The fact that it crossed a Petabyte without a single firmware exception or interface crash is the real victory here!"
Also I set a auto click for every 5 seconds to trim it, my research is found amazing things so far, but all I can say sata 2 technically is durable much so as today's ssds,
Place bets where it may roll back to 0 or max out, I'll let u know
Approx Max per trim is 18gb that every 5 seconds
Minutes in a day: 60 minutes multiplied by 24 hours equals 1 440 minutes.Total daily data: 216 GB per minute multiplied by 1 440 minutes equals 311 040 GB. That is if max is 18gb per trim.
September 2nd, 2026, 10:18
so who/what is in control of TRIM-ing?
He's using an auto-click tool to spam Optimize button in Windows Optimize tool. He's doing something super simple, looks at changes in CDI and fabricated a puffed up story around it.
September 2nd, 2026, 10:28
the actual data vanishes into thin air
Why is he referring to "actual data" if he is not writing anything? What is his "macro loop" during for the remainder of those 5 seconds? If it's doing nothing, then that contradicts his claim that he is thrashing the drive's controller.
I could understand writing data at 100 MB/s into the drive's pseudo-SLC cache and then following it up with a TRIM command that unmaps the drive's entire free space. However, the SMART result would only make sense if the drive were falsely counting TRIM-ed LBAs as host writes. That is, there must be a bug in SMART.
September 2nd, 2026, 10:59
Arch Stanton wrote:so who/what is in control of TRIM-ing?
He's using an auto-click tool to spam Optimize button in Windows Optimize tool. He's doing something super simple, looks at changes in CDI and fabricated a puffed up story around it.
Is it really that simple? If so, then it's a massive con job.
September 2nd, 2026, 11:04
fzabkar wrote:Arch Stanton wrote:so who/what is in control of TRIM-ing?
He's using an auto-click tool to spam Optimize button in Windows Optimize tool. He's doing something super simple, looks at changes in CDI and fabricated a puffed up story around it.
Is it really that simple? If so, then it's a massive con job.
https://youtu.be/yXJVJnJotLU
September 2nd, 2026, 11:22
Arch Stanton wrote:fzabkar wrote:https://youtu.be/yXJVJnJotLU
In his first thread at Reddit, I suggested to him that SMART was counting TRIM-ed sectors. Instead, he said he would add my hypothesis to his to-do list, and would attend to it after the endurance test finished. I don't think he has any interest in doing so.
September 3rd, 2026, 8:53
September 3rd, 2026, 9:17
And he blocked me ..
Powered by phpBB © phpBB Group.