4K low level format compatible/issue discussion

[syscl]

@gujiangjiang @wmchris @bozma88 @R1MnTnA ... Hi, I've created a new topic for you to discuss about the 4K low level format issue related to NVMe. Let #26 related to HWP + corrupted data issue. :)

Thank you, syscl

[gujiangjiang]

thanks

[krim404]

@gujiangjiang to be honest: i dont think that the switch didnt work, i'm pretty sure these are different problems which are not related to each other.

Its possible your toshiba ssd is defective. So i would suggest doing an intensive SSD checkup with bad sector search - just in case... If the SSD is really defective i would suggest swapping it as soon as possible. Such a problem can (and will) easily extend itself to something you can not repair yourself...

If the device is not defect: i would suggest doing the 4k switch without any bias in mind. it's possible your dell gives a false "non existing" message albeit the disk is available and works well. Also it's still possible your drive suffers from something like the "dirty shutdown" bug, which should be no problem in daily usage after the OS is installed correctly.

Some of your screenshots from linux could also result from a bad NVMe configuration, auto discovery problem because of no existing partition table and much more.

So you see: there is still hope ;-)

[darkhandz]

HackrNVMe.kext + 512 -> ubuntu USB live -> 4K -> USB Clover without HackrNVMe.kext -> TimeMachine -> IONVMeFamily.kext

XG3 info:

[krim404]

@darkhandz thank you. So it works for you, too. Unless someone else is reporting the same problem i would say this issue is not replicateable.

[shixuev5]

I follow @wmchris 4k tutorial, ubuntu USB live -> 4K -> refresh install 12.4 without any IONVMeFamily patch in clover. use computer two days, no problem.

[nos1609]

@syscl & @gujiangjiang there is a new firmware issued by dell for xg3 to fix this. The cause is not the 4k format, but the "unsafe shutdown" counter instead. My xg3 also ended this way and got replaced by a hunix drive. But why for are you performing a full 8hour low level format? It's not only incredibly increasing the wear level, but also os dangerous for toshiba ssd wich is too hot for the tiny xps13 case and thermal config. If you check the rd400-a setup, it contains a thermal pad to dissipate the heat from the controller. Also take a look at 7370 and 9365 teardown, especially the copper heatspreader used over the ssd to prevent overheat. Just update the ssd firmware and your toshiba gonna stay alive. There is nothing special at converting it to 4k lba, it is exact the same hardware as the rd400. Only the firmware is not issued by ocz, so it lacks some optimization and need updates for bugfixes. I've performed the conversion on the drive two times 4k-512-4k and it died two weeks after. It started disappearing every two reboots and the counter of unsafe shutdowns was over 1.5k. 2012 I had an ocz vector. It also died after a forced power-button shutdown. After that 3 revisions now are free of that issue (vt150, vt180 and vx500) even if they are the direct successors of the original vector and two of them even contain the same controller. But the firmware is different. And this makes the major difference in stability.

[krim404]

@nos1609 he already upgraded the patch. the tutorial also informs about this patch. Low level format does not wear the drive and does not take 8 hours.

there are no known thermal problems for the SSD. do you have a source for this claim?

[nos1609]

@wmchris this one as an example, some more googling "rd400 throttling". Same issue as the 950Pro. In 960Pro the issue was solved with a copper-layered sticker. According to 8 hours I agree that maybe I misunderstood the mentionned duration. But can you prove somehow that low-level format is defferent from the usual one, consisting of writing zeroes to the whole capacity of the ssd? Cause with such approach this counts as a rewrite and results in the TBW counter increase. If so, it is a degradation called wear-levelling. ;)

[krim404]

low level format does not require zeroing. you switch to 4k, then the whole data will be unreadable, so it "feels" brand new. you just recreate the partition table.

[gujiangjiang]

@darkhandz 你的能通过ePSA检测么？

[krim404]

do not start ePSA, it fried my Toshiba SSD and i got a SAMSUNG as replacement from Dell... its possible its just a coincidence, but its also possible that ePSA increased the voltage because the 4k drive was "not readable"

samsung has 1/3 speed of the toshiba. toshiba is not available anymore and acording to the dell hotline will never be.

[gujiangjiang]

@wmchris

I have replaced a new XG3 from dell and the speed is well than the old one.

So i want to know if the xg3 or if just the 4k sector was safe now?

[bozma88]

@wmchris , I've already run ePSA multiple times with my OEM 256GB Toshiba NVME THNSN5256GPUK SSD (XPS 13 9360) formatted to 4K. No issues. As a side note, XPS line comes with only 2 o 4 PCI-e lanes active, so faster SSDs like Samsung 960 will run much slower than their declared max speed. Anyway, my Toshiba real-world performance is mediocre, not faster than my old SATA Samsung 950PRO in everyday usage patterns. But that's ok for me...

[krim404]

See http://www.insanelymac.com/forum/topic/319764-guide-dell-xps-15-9550-sierra-10122-quick-installation/?p=2420478

Samsung is much slower :/ but not the theoretical 30%, only 50%.

[gujiangjiang]

GPUK is xg4 wih tlc not xg3 with mlc. Xg3 is GPU7