Cloning NVME PCIe3 to NVME PCIe4

nekrosoft13

[H]ard|Gawd
Joined
Jan 4, 2005
Messages
1,581
so I'm trying to clone from Intel 660p 1tb (NVME gen3) to Gigabyte Aorus 1tb (NVME Gen4)

I'm using Macrium Reflect, and everything I tried so far, every single time I get inaccessible boot device.

Using the same software and same clone disk option from intel 660p to Intel 660p zero problems, or going from 660p to 760p no problems...

only when I try to clone from Gen3 to Gen4 it doesn't work.

Anyone seen anything like this?
 
There's a chance that maybe the drive is formatted with a different block structure? I've noticed that issue before when trying to clone an ADATA drive to a sabrent drive, it just wouldn't let me do it.
 
so I'm trying to clone from Intel 660p 1tb (NVME gen3) to Gigabyte Aorus 1tb (NVME Gen4)

I'm using Macrium Reflect, and everything I tried so far, every single time I get inaccessible boot device.

Using the same software and same clone disk option from intel 660p to Intel 660p zero problems, or going from 660p to 760p no problems...

only when I try to clone from Gen3 to Gen4 it doesn't work.

Anyone seen anything like this?
Have you tried not using the "smart" copy and instead do an exact copy?
 
My Sabrent gen4 drive it came with a format utility for 512bytes or 4k sectors.
 
try cleaning the drive with diskpart and dont initialize it or anything, then clone. ive had issues where im reusing a ssd from a dead macbook, i clone a windows drive to it and it wont boot. but if i clean the drive first and let the clone tool set it up it works every time. ive had it happen a few times as we have a tonne of ssds from dead/decommissioned macbooks.
 
The generation of the nvme protocol does not matter as long as you are using it on the proper working ports. You have another issue going on that does t have anything to do with it being gen 3 vs gen4.

this is of course assuming you know for sure the board is compatible and the drives are properly configured.
 
I am not saying this will work in your case: https://www.newegg.com/riitop-model...ler-card/p/17Z-0061-00052?Item=9SIA6V87HX5336

But that's what I used before and it worked, although I used Corsair MP NVMe drives.

I didn't want to slap my NVMe drive in an SSD slot so I opted for an adapter and I've also used ASUS adapter as well and it worked on both adapters.
I have one of these that's been holding a 1tb nvme (gen 3) ssd for almost a year now in my system. It's a surprisingly cheap and effective solution that actually cools better than sticking them against the motherboard.

That said, it didn't let me clone between the different brands for whatever reason when I tried. 🤷‍♂️
 
The generation of the nvme protocol does not matter as long as you are using it on the proper working ports. You have another issue going on that does t have anything to do with it being gen 3 vs gen4.

this is of course assuming you know for sure the board is compatible and the drives are properly configured.

what I don't understand is I did it couple dozen times with m.2 gen3 drives (Intel, crucial, kingston) all worked fine, now i'm doing same exact way, even took out a motherboard that I used few times in past succesfully and nothing.
the gen4 drive i started with was brand new when I started, i wiped it few times to entire nothing crazy was going on.
last week I purchased a usb3.2 to nvme adapter to try it outside, and to no surprise it didn't work, then I tried it with a gen3 drive and it worked fine.

when it doesn't work, it goes though entire process, says data has been copied successfully, visually all partitions look fine, then i swap the drives and upon boot it says inaccessible boot device
 
if you have a copy of hirens, try lazesoft imaging, it too has never failed to clone for me if other software fails.
 
"inaccessible boot device" sounds like a Windows message (are you trying to boot Windows?). Like it doesn't have the correct driver installed...or it does, but for some reason it's picking a wrong one.
 
"inaccessible boot device" sounds like a Windows message (are you trying to boot Windows?). Like it doesn't have the correct driver installed...or it does, but for some reason it's picking a wrong one.
since it supposed to be excact copy, all drivers are there. there is no reason why driver would be affecting it.
 
I had the same exact problem the OP is having when I was trying to clone the Adata SX8200 Nvme to the WD Black SN850. Identical : Inaccessible boot device error.

Trying 3 different cloning proggies didn't help. It was a fluster cluck. I was forced to throw in the towel and do a full install the old way.
 
Sometimes the clone software can give a rogue drive letter which stops it booting. Try removing any drive letters before you disconnect the new drive. Especially if the hidden partitions have been given a drive letter.

I use Partition Wizard myself. Great bit of software.
 
I'll be trying to migrate my OS from a gen3 to gen4 NVME this weekend but I am going to use this enclosure.
I plan on using Paragon's Migrate OS to SSD which does a migrate vs a clone. Been using this software for the past decade hundreds of times with no issues till recently when trying to clone a non-UEFI system to be used in an UEFI machine.
Got one to work by manually creating the UEFI partition on the destination drive after the migration for my friends upgrade, but when I tried to do it for my upgrade, I was getting errors when trying to create the UEFI partition, so I ended up doing a clean install on my system.

IMG_7065.JPEG
 
All you guys having problems going from gen 3->4 have you tried going to in to device manager and changing the driver to "standard nvme express driver" vs. the manufacturer specific driver?
 
The issue is that you-basically-cannot clone SSD to SSD. This is due to the fact that not all firmware's and controllers are the same. There is a lot of translation that happens, and is similar to much of the LBA and 4k sectors issues that previously was with HDD's. But that was mainly in the BIOS/UEFI, not with the host interface, nor with the controllers and software. That defacto-standard was at a particular level of equal functionality and allowed a more fluid bit-to-bit transfer. One notable issue with SSD's you can actually buy two of the same storage devices, but depending on time-of-manufacturing you may have different firmware, a different controller, and maybe no volatile cache when it is supposed to have some!

So, SSD 101: it is all about the tech inside, mainly firmware and controller. These two control the mappings and interaction of the translation of commands and such to the interfaces and to the flash and then back up. When you clone bit-to-bit the layers do not translate from that host flash to the target flash mappings. You can with some software due to data migration, but cannot with docking and some software solutions due bit-to-bit. The PCIe adaptor will work properly because communication to the device is translated between the flash and interface as it normally would. This is how I usually will do my SSD's is when I upgrade I just copy the data to a large local HDD, shutdown and remove SSD drive, install new SSD, boot, then setup partition, and finally copy data over using it as like it was the drive before (why install software?). If it is a NVMe I rather do a new full install if it is a system drive. Although, you can copy/migrate the data and make the drive bootable or let software do some of the work. Basically, as noted by Zepher, this is data migration. Copying the data from one end to another.

Drivers only participate in particular functions of the device, or not; NVMe was meant to kind of be driverless and agnostic. So, Windows has a driver, Intel and AMD have a software suite for controlling PCIe and AHCI devices with their drivers/software (that was that one issues with AMD and their PnP ID recently), and you can have a provided driver through the vendor such as Samsung. Samsung's driver is better than Windows, and some functions are not exposed when using Window's or the storage suites software such as Intel's Rapid Storage Tech. It can make it simplistic, but when you do not install the best way you will likely have something pop up later, such as the AMD one mention previously.

Anyways, I come from the time in which not many and pretty much mostly all did not know what a compatible mode controller and a native mode controller was when SATA interface was introduced. They would say install the "native" SATA driver and not know what they even meant nor what it actually was when the device was actually never in native controller mode. This created even more confusion to those that didn't know what to set the BIOS options to when it came to legacy IDE, native IDE, and AHCI settings because the never learned and much of the legacy compatibility can also exacerbate everything.
 
The issue is that you-basically-cannot clone SSD to SSD. This is due to the fact that not all firmware's and controllers are the same. There is a lot of translation that happens, and is similar to much of the LBA and 4k sectors issues that previously was with HDD's. But that was mainly in the BIOS/UEFI, not with the host interface, nor with the controllers and software. That defacto-standard was at a particular level of equal functionality and allowed a more fluid bit-to-bit transfer. One notable issue with SSD's you can actually buy two of the same storage devices, but depending on time-of-manufacturing you may have different firmware, a different controller, and maybe no volatile cache when it is supposed to have some!

So, SSD 101: it is all about the tech inside, mainly firmware and controller. These two control the mappings and interaction of the translation of commands and such to the interfaces and to the flash and then back up. When you clone bit-to-bit the layers do not translate from that host flash to the target flash mappings. You can with some software due to data migration, but cannot with docking and some software solutions due bit-to-bit. The PCIe adaptor will work properly because communication to the device is translated between the flash and interface as it normally would. This is how I usually will do my SSD's is when I upgrade I just copy the data to a large local HDD, shutdown and remove SSD drive, install new SSD, boot, then setup partition, and finally copy data over using it as like it was the drive before (why install software?). If it is a NVMe I rather do a new full install if it is a system drive. Although, you can copy/migrate the data and make the drive bootable or let software do some of the work. Basically, as noted by Zepher, this is data migration. Copying the data from one end to another.

Drivers only participate in particular functions of the device, or not; NVMe was meant to kind of be driverless and agnostic. So, Windows has a driver, Intel and AMD have a software suite for controlling PCIe and AHCI devices with their drivers/software (that was that one issues with AMD and their PnP ID recently), and you can have a provided driver through the vendor such as Samsung. Samsung's driver is better than Windows, and some functions are not exposed when using Window's or the storage suites software such as Intel's Rapid Storage Tech. It can make it simplistic, but when you do not install the best way you will likely have something pop up later, such as the AMD one mention previously.

Anyways, I come from the time in which not many and pretty much mostly all did not know what a compatible mode controller and a native mode controller was when SATA interface was introduced. They would say install the "native" SATA driver and not know what they even meant nor what it actually was when the device was actually never in native controller mode. This created even more confusion to those that didn't know what to set the BIOS options to when it came to legacy IDE, native IDE, and AHCI settings because the never learned and much of the legacy compatibility can also exacerbate everything.
So have I managed to deal with all those customer SSD upgrade clones (SSD to SSD and SSD to NVMe) the past 5+ years? I must have got lucky I guess?? :)

Totally possible to clone SSD to SSD or NVMe. I often do a initial build on a old Intel 60GB SATA SSD while waiting for the 500/1TB SSD/NVMe to come in and then clone it over when it arrives. I have all sorts of M.2/mSATA/NVMe USB/SATA adapters that help.

Usually just slap it in my cloning dock and press the button.
 
I used Paragon to migrate from gen3 to gen4 and had no issues.
New 5600X/Dark Hero/980Pro system booted up fine with the clone from my 5800X/Crosshair VIII/ WD-SN750 .
 
Last edited:
Yeah I just did it with no problems last week. I think the trick is what I posted above (for nvme) you need to uninstall the manufacturer specific driver and go to generic before cloning.
This is the issue. Windows is trying to boot with the Intel RST driver, whereas the new drive uses the generic Microsoft driver. Uninstall the Intel controller from Device Manager and choose to delete the driver software. Uninstalling the RST software is not enough as it won't uninstall the driver. Alternatively you can set Windows to boot in safe mode which should automatically switch the driver.
 
So have I managed to deal with all those customer SSD upgrade clones (SSD to SSD and SSD to NVMe) the past 5+ years? I must have got lucky I guess?? :)

Totally possible to clone SSD to SSD or NVMe. I often do a initial build on a old Intel 60GB SATA SSD while waiting for the 500/1TB SSD/NVMe to come in and then clone it over when it arrives. I have all sorts of M.2/mSATA/NVMe USB/SATA adapters that help.

Usually just slap it in my cloning dock and press the button.
Note, that I said -basically- cannot. If you are familiar with "shirnkflation" you can also understand the latter comment on firmware, controller, and possible lacking of volatile cache. If the drives are "exactly" exact, then you most likely will be able to clone bit-to-bit without issue. Remember there is a difference to copying data from end to end, and cloning a drive-which is why I say bit-to-bit for this is format, file system, and everything written and noted in the journal etc. However, the lacking of understanding of SSD's is still prevalent in the latter comments with the RST configuration and such as I noted in the post. Many have not learn from the ground up the core logics and into the SSD logics. Be one with hardware.....or you will find that the CPU you purchased lesser of cost doesn't support an instruction or two, a chipset that doesn't support a function or two, etc (caveat emptor). Still you should note all the conversations in the forums about not having the right format, sectors, performance issues, etc (I wonder why?).......this is a lot of unknowing in a technical forum and for something that has been around for a good while now.

Like I mentioned, you need to know what you have, and one of the issues is the manufacturers are swapping parts which has spawned much of what I note. But even if it is not "shrinkflation" SSD 1 controller 0 is Rev A and SSD 2 controller 1 is Rev B, or C, etc I would not clone due to different mapping, wear leveling , and translations done. So, the question is you think it works, but is it correct/proper (rhetorical)?
https://www.tomshardware.com/news/adata-and-other-ssd-makers-swapping-parts

This is the domain of the firmware and controller. This picture explains the border well in which PCIe, SATA, and commands from the host are translated into flash commands, positions, channels, and mappings then accessed, written, etc. This is a top view, but not an internal view-meaning they are not the same with all SSD's. But hey, it is your hardware and your possible issue for not knowing-wasn't that half the battle?

51b2a39e418253fb8b5b5f34e7cc7b204d959f25.png
 
thats a lot of rambling "tech talk" to still be wrong. cloning ssd to ssd/nvme works, it CAN be done.
 
thats a lot of rambling "tech talk" to still be wrong.
So funny, and yet, so true. :ROFLMAO:
Cloned many a SSD and yet to have a problem. Last week cloned HP gen 3 SDD to Samsung gen 4 SSD with Macrium Reflect. Zero issues. New drive running at full speed. Am I doing it wrong all this time?
 
Look into Acronis True Image.

Awesome program for backing up, cloning and restoring.
 
I used Partition Wizard Technician Edition to clone my last drives. Fast as fuck, and it just... works.
 
The issue is that you-basically-cannot clone SSD to SSD. This is due to the fact that not all firmware's and controllers are the same.
Incorrect. Cloning a drive works at the logical bit level, which is above all the hardware and firmware abstractions. There's no way to clone a drive at the NAND chip bit level with any of the software mentioned here - this obviously wouldn't work on anything but identical hardware / firmware.
 
Not Windows .. but on my TrueNAS server I was using a USB to SATA cable to power/run a 120gb SSD for a boot drive .. changed motherboards to one that had an M.2 slot and installed a 120GB NVMe SSD. Mirrored the USB to SATA connected SSD to the M.2 NVMe SSD and then removed the USB connected SSD from the configuration and booted solely with the M.2 drive without issue...

..not Windows ... but kind of related ... kinda ;)
 
Note, that I said -basically- cannot. If you are familiar with "shirnkflation" you can also understand the latter comment on firmware, controller, and possible lacking of volatile cache. If the drives are "exactly" exact, then you most likely will be able to clone bit-to-bit without issue. Remember there is a difference to copying data from end to end, and cloning a drive-which is why I say bit-to-bit for this is format, file system, and everything written and noted in the journal etc. However, the lacking of understanding of SSD's is still prevalent in the latter comments with the RST configuration and such as I noted in the post. Many have not learn from the ground up the core logics and into the SSD logics. Be one with hardware.....or you will find that the CPU you purchased lesser of cost doesn't support an instruction or two, a chipset that doesn't support a function or two, etc (caveat emptor). Still you should note all the conversations in the forums about not having the right format, sectors, performance issues, etc (I wonder why?).......this is a lot of unknowing in a technical forum and for something that has been around for a good while now.

Like I mentioned, you need to know what you have, and one of the issues is the manufacturers are swapping parts which has spawned much of what I note. But even if it is not "shrinkflation" SSD 1 controller 0 is Rev A and SSD 2 controller 1 is Rev B, or C, etc I would not clone due to different mapping, wear leveling , and translations done. So, the question is you think it works, but is it correct/proper (rhetorical)?
https://www.tomshardware.com/news/adata-and-other-ssd-makers-swapping-parts

This is the domain of the firmware and controller. This picture explains the border well in which PCIe, SATA, and commands from the host are translated into flash commands, positions, channels, and mappings then accessed, written, etc. This is a top view, but not an internal view-meaning they are not the same with all SSD's. But hey, it is your hardware and your possible issue for not knowing-wasn't that half the battle?

View attachment 358862


TLDR - You can clone SSD/NVMe perfectly fine. You are wrong on this one. Reality says different. Move on.
 
I've used this for years now. One of the best bits of software around.

Certainly an upgrade from the days of Partition Magic though it owes a lot to that once great partition manager. I have Partition Wizard on the Ventoy boot disks I make as one of my invaluable PC tools. Some of the content in this topic is a bit puzzling to me since I'm pretty sure I used it as recently as a few months ago to clone one of my SSDs to another.
 
Are you getting Inaccessible_Boot_Device on a blue screen with the sad icon, or just the white text on a black background When you reboot?
 
So I clone SSDs a lot for my reviews, using Reflect, and I've learned the key to avoiding inaccessible boot device errors is making sure the two drives never coexist during a boot cycle after they've been cloned.

In other words, I've got drive A, which already has Windows on it, and drive B which I want to clone drive A's contents over to. Drive B *must* be blank. And by blank, I mean it needs to be completely uninitialized; I use an external USB adapter to temporarily connect the drive to a (different) booted system, fire up diskpart, and run the "clean" command on the drive.

Once drive B is blank, I connect both drives via the onboard m.2 slots on my test bench and boot them up. Then I use Reflect to clone the drive, A->B. As soon as Reflect is done cloning, I shut down the system and remove drive A and move drive B into its slot.

This procedure works 100% of the time, regardless of which SSD I am moving to or from.

HOWEVER, if after drive B has received a clone, if I reboot the system with both drives A and B connected, something goes awry; I find myself permanently unable to use either of those drives as a source for cloning new drives in the future. Drive A continues to boot up if it is alone in the system, but if I use it as a source for cloning a new drive (let's say drive C) that new drive will give an inaccessible boot device error. The only way to 'fix' this issue is to run a diskpart clean on the drives and start fresh, or clone to them from some other drive that never 'coexisted' with another drive.
 
So I clone SSDs a lot for my reviews, using Reflect, and I've learned the key to avoiding inaccessible boot device errors is making sure the two drives never coexist during a boot cycle after they've been cloned.

In other words, I've got drive A, which already has Windows on it, and drive B which I want to clone drive A's contents over to. Drive B *must* be blank. And by blank, I mean it needs to be completely uninitialized; I use an external USB adapter to temporarily connect the drive to a (different) booted system, fire up diskpart, and run the "clean" command on the drive.

Once drive B is blank, I connect both drives via the onboard m.2 slots on my test bench and boot them up. Then I use Reflect to clone the drive, A->B. As soon as Reflect is done cloning, I shut down the system and remove drive A and move drive B into its slot.

This procedure works 100% of the time, regardless of which SSD I am moving to or from.

HOWEVER, if after drive B has received a clone, if I reboot the system with both drives A and B connected, something goes awry; I find myself permanently unable to use either of those drives as a source for cloning new drives in the future. Drive A continues to boot up if it is alone in the system, but if I use it as a source for cloning a new drive (let's say drive C) that new drive will give an inaccessible boot device error. The only way to 'fix' this issue is to run a diskpart clean on the drives and start fresh, or clone to them from some other drive that never 'coexisted' with another drive.

It might have something to do with disk IDs. A straight clone will produce a disk with the same ID, which could cause problems. Alternatively, the cloning software could change the disk ID, but that would prevent booting unless it does some repairs.
 
It might have something to do with disk IDs. A straight clone will produce a disk with the same ID, which could cause problems. Alternatively, the cloning software could change the disk ID, but that would prevent booting unless it does some repairs.
I always assumed that was the case, but it never bothered me enough to investigate. Once I figured out the procedure, which was just to make sure they didn't coexist during a boot, it became an easy problem to avoid and never interfered again. I have done 40+ SSD->SSD clones over the past year just for the review work alone and it's been easy as can be.
 
Slight bump to an old topic, but I *just* cloned my ADATA SX8200 that was getting a bit old to a WD Black SN850* (gen4) via USB-C to M.2 enclosure using the WD version of Acronis True Image and it worked perfectly. Had the settings on automatic, put the new drive in the computer as the OS drive and it booted up immediately and everything is exactly as it was, except now my OS drive is 3x faster and with 4x the capacity. Did a crystaldiskmark bench and I'm getting the rated 7000MB/S read speed, even while being actively used as the OS drive.

*I know there's a firmware issue with the SN850 and it not giving full performance if connected to the ryzen chipset, however, i knew this was going to be connected to the processor directly so this doesn't affect me.
 
Back
Top