It's a nice adapter and it's not badly priced either.

I wonder if there will be the possibility of putting a option rom a pci card so that it recognizes an nvme ssd and the OS can be started from it. In this forum (https://www.win-raid.com/t3661f13-Experimenta … Option-ROM.html) they have an option ROM to insert into the motherboard bios. But it seems less dangerous to me, and also very interesting if it could be put on a pci card. I know that there are isa cards for option roms but I have not found any pci other than this (http://www.softworld.es/cu/custodius_lite_2004/) that I don't know if it will be exactly what is necessary, and it could not be easily obtained I also suppose that you could use a network card but it does not seem to me such a clean method.
I am wondering this because I am currently using the usb boot method described here (https://www.win-raid.com/t871f50-HowTo-Get-fu … Systems-with-an -AMI-UEFI-BIOS.html) but for some strange reason, I don't feel totally comfortable with this method.

Sphere478 wrote on 2021-02-05, 14:42:

This may be the first m.2 nvme ever on a pentium 1 era socket 7 motherboard 😂🤣 (it’s the highlighted one.)

Which Linux derivate is that?

Oerg866 wrote on 2021-09-12, 14:37:

Sphere478 wrote on 2021-02-05, 14:42:

This may be the first m.2 nvme ever on a pentium 1 era socket 7 motherboard 😂🤣 (it’s the highlighted one.)

Which Linux derivate is that?

Um, been a while, trying to remember… I think it was debian jessie?

Hoping wrote:

I don't feel totally comfortable with this method.

It's the safest method, because it doesn't change anything in firmware or hardware. And there's usually no problem dedicating one port and some random USB stick for boot loader.

Hoping wrote on 2021-09-12, 13:45:

I wonder if there will be the possibility of putting a option rom a pci card so that it recognizes an nvme ssd and the OS can be started from it. In this forum (https://www.win-raid.com/t3661f13-Experimenta … Option-ROM.html) they have an option ROM to insert into the motherboard bios. But it seems less dangerous to me, and also very interesting if it could be put on a pci card. I know that there are isa cards for option roms but I have not found any pci other than this (http://www.softworld.es/cu/custodius_lite_2004/) that I don't know if it will be exactly what is necessary, and it could not be easily obtained I also suppose that you could use a network card but it does not seem to me such a clean method.
I am wondering this because I am currently using the usb boot method described here (https://www.win-raid.com/t871f50-HowTo-Get-fu … Systems-with-an -AMI-UEFI-BIOS.html) but for some strange reason, I don't feel totally comfortable with this method.

I think I mentioned that the problem is that these seem to run in some kind of pio mode when used on these old systems. I was literally getting kbps out of it, it was dreadfully slow. Totally ruined any thoughts of it’s usefulness.

The fastest thing I have found for old systems is this: promise tx4 sata II 300 pci. You can use that to adapt to a sata nvme I believe though I haven’t tried but pretty sure it would work. I’ve just used it on regular 2.5” it gets really good speeds on my 430tx mobo almost as fast as the ram haha

You can use that to adapt to a sata nvme

It's SATA or NVME. Not both. SATA M.2 can be easily fitted into 2.5 inch adapter.

The Serpent Rider wrote on 2021-10-20, 21:13:

You can use that to adapt to a sata nvme

It's SATA or NVME. Not both. SATA M.2 can be easily fitted into 2.5 inch adapter.

My terms may be confused.

Sata cable to sata stick on a adapter is what I mean 😀

Sphere478 wrote on 2021-10-20, 21:11:

I think I mentioned that the problem is that these seem to run in some kind of pio mode when used on these old systems. I was literally getting kbps out of it, it was dreadfully slow. Totally ruined any thoughts of it’s usefulness.

The fastest thing I have found for old systems is this: promise tx4 sata II 300 pci. You can use that to adapt to a sata nvme I believe though I haven’t tried but pretty sure it would work. I’ve just used it on regular 2.5” it gets really good speeds on my 430tx mobo almost as fast as the ram haha

What I would like to see is a boot rom pci card that recognizes NVME devices and can boot from them. Something like XTIDE that can be inserted into a rom of a network card. I am not targeting computers as old as a Pentium, but I am targeting computers that have native PCI-E, without adapters.
The usb boot method works, but I think a boot rom on a pci card would be intresting.

The HighPoint SSD7505 (4x nvme slots) and SSD7140 8x nvme slots) support booting.

16x PCI-e but they require PCI-e 3.0 or 4.0 and a UEFI BIOS.. and are quite expensive.
https://www.highpoint-tech.com/USA_new/series … ssd7500-faq.htm

So.. I forgot about this option, but you should be able to use CLOVER to pass booting to an nvme drive even if you only have a legacy BIOS setup.
I am pretty sure this should make it possible to boot to even a dumb nvme adapter.

https://www.win-raid.com/t2375f50-Guide-NVMe- … der-method.html

16x PCI-e but they require PCI-e 3.0 or 4.0 and a UEFI BIOS..

So basically it's useless.

Hoping wrote on 2021-10-21, 21:10:

Sphere478 wrote on 2021-10-20, 21:11:

I think I mentioned that the problem is that these seem to run in some kind of pio mode when used on these old systems. I was literally getting kbps out of it, it was dreadfully slow. Totally ruined any thoughts of it’s usefulness.

The fastest thing I have found for old systems is this: promise tx4 sata II 300 pci. You can use that to adapt to a sata nvme I believe though I haven’t tried but pretty sure it would work. I’ve just used it on regular 2.5” it gets really good speeds on my 430tx mobo almost as fast as the ram haha

What I would like to see is a boot rom pci card that recognizes NVME devices and can boot from them. Something like XTIDE that can be inserted into a rom of a network card. I am not targeting computers as old as a Pentium, but I am targeting computers that have native PCI-E, without adapters.
The usb boot method works, but I think a boot rom on a pci card would be intresting.

Ah, with native pcie yeah that may work.

I think if you put grub on a usb drive you could boot to it.

Recently I got myself such a combination to run a large NVMe SSD over PCI for data storage. The adapter itself works and the disk is being correctly recognized. I think it should theoretically use all the PCI bandwidth if possible, though I don't know if it'll affect the operability of other devices on the PCI bus as the bandwidth is shared...

The good part is that unlike USB/SATA, such a disk would not be visible to OSes or boot managers that can't handle it properly, so OSes that I don't intend to use the disk for would not complain.

However, I noticed a caveat when using it with Windows 7, as the PCI-PCIe adapter initially reported Code 10 error and refused to work. Googling about the issue led me to this. After following its instructions the adapter now works correctly, and I'm able to access the disk.

It seems the interoperability affects only Windows 7. On Linux it the adapter and disk work without any issue. Since the OP appears to be running Win10 and there were no mentions of workarounds used, it's very possible that the issue was already fixed on Windows at some point.

LSS10999 wrote on 2021-11-22, 01:21:

Recently I got myself such a combination to run a large NVMe SSD over PCI for data storage. The adapter itself works and the dis […]
Show full quote

Recently I got myself such a combination to run a large NVMe SSD over PCI for data storage. The adapter itself works and the disk is being correctly recognized. I think it should theoretically use all the PCI bandwidth if possible, though I don't know if it'll affect the operability of other devices on the PCI bus as the bandwidth is shared...

The good part is that unlike USB/SATA, such a disk would not be visible to OSes or boot managers that can't handle it properly, so OSes that I don't intend to use the disk for would not complain.

However, I noticed a caveat when using it with Windows 7, as the PCI-PCIe adapter initially reported Code 10 error and refused to work. Googling about the issue led me to this. After following its instructions the adapter now works correctly, and I'm able to access the disk.

It seems the interoperability affects only Windows 7. On Linux it the adapter and disk work without any issue. Since the OP appears to be running Win10 and there were no mentions of workarounds used, it's very possible that the issue was already fixed on Windows at some point.

I encountered extremely bad transfer speeds it seems all the writes and reads had to be translated through the cpu because of the adaptation or something like that. I was literally getting kbps on older systems. And low mbps on newer hardware.

My advice if you want to do this is as follows

Use a promise tx4 sata II 300 pci card and adapt to sata m.2

That should actually come close to pci saturation. And honestly is probably be the absolute fastest configuration you can put on 33mhz 32bit pci. Might be able to go a little faster with a server style card with sd ram cache but not much.

Sphere478 wrote on 2021-11-22, 01:53:

I encountered extremely bad transfer speeds it seems all the writes and reads had to be translated through the cpu because of t […]
Show full quote

LSS10999 wrote on 2021-11-22, 01:21:

Recently I got myself such a combination to run a large NVMe SSD over PCI for data storage. The adapter itself works and the dis […]
Show full quote

Recently I got myself such a combination to run a large NVMe SSD over PCI for data storage. The adapter itself works and the disk is being correctly recognized. I think it should theoretically use all the PCI bandwidth if possible, though I don't know if it'll affect the operability of other devices on the PCI bus as the bandwidth is shared...

The good part is that unlike USB/SATA, such a disk would not be visible to OSes or boot managers that can't handle it properly, so OSes that I don't intend to use the disk for would not complain.

However, I noticed a caveat when using it with Windows 7, as the PCI-PCIe adapter initially reported Code 10 error and refused to work. Googling about the issue led me to this. After following its instructions the adapter now works correctly, and I'm able to access the disk.

It seems the interoperability affects only Windows 7. On Linux it the adapter and disk work without any issue. Since the OP appears to be running Win10 and there were no mentions of workarounds used, it's very possible that the issue was already fixed on Windows at some point.

I encountered extremely bad transfer speeds it seems all the writes and reads had to be translated through the cpu because of the adaptation or something like that. I was literally getting kbps on older systems. And low mbps on newer hardware.

My advice if you want to do this is as follows

Use a promise tx4 sata II 300 pci card and adapt to sata m.2

That should actually come close to pci saturation. And honestly is probably be the absolute fastest configuration you can put on 33mhz 32bit pci. Might be able to go a little faster with a server style card with sd ram cache but not much.

Strange... On my test system (C2Q based) I was able to saturate USB2.0 speeds (~36MB/s) when copying stuffs from a USB stick (over USB 2.0 port) to the NVMe SSD behind the adapter. Maybe what disk you use matters as the NVMe-to-PCIe adapter is passive, so the PCI-to-PCIe bridge is the only thing between between the NVMe disk and the CPU/chipset. (EDIT: That was tested on Linux... not sure if how OS handles things also matters to some extent)

As for Promise TX4 SATAII 300 PCI... I recall having used it on a much older motherboard (Pentium D based) before and it worked great. I could even boot systems from SATA SSDs attached to it and the performance is comparable to that of the native SATA-150 ports, if not better.

Should say that the option ROMs those SATA cards use (for booting support) consume precious upper memory, which can be a problem if you use DOS, but not a problem otherwise. If not using the disks for booting, disabling the option ROM might prevent the UMB consumption as well as preventing incompatible OSes (that I don't intent to use them on) from seeing the disks, though.

LSS10999 wrote on 2021-11-22, 03:25:

Strange... On my test system (C2Q based) I was able to saturate USB2.0 speeds (~36MB/s) when copying stuffs from a USB stick (ov […]
Show full quote

Sphere478 wrote on 2021-11-22, 01:53:

I encountered extremely bad transfer speeds it seems all the writes and reads had to be translated through the cpu because of t […]
Show full quote

LSS10999 wrote on 2021-11-22, 01:21:

Recently I got myself such a combination to run a large NVMe SSD over PCI for data storage. The adapter itself works and the dis […]
Show full quote

Recently I got myself such a combination to run a large NVMe SSD over PCI for data storage. The adapter itself works and the disk is being correctly recognized. I think it should theoretically use all the PCI bandwidth if possible, though I don't know if it'll affect the operability of other devices on the PCI bus as the bandwidth is shared...

The good part is that unlike USB/SATA, such a disk would not be visible to OSes or boot managers that can't handle it properly, so OSes that I don't intend to use the disk for would not complain.

However, I noticed a caveat when using it with Windows 7, as the PCI-PCIe adapter initially reported Code 10 error and refused to work. Googling about the issue led me to this. After following its instructions the adapter now works correctly, and I'm able to access the disk.

It seems the interoperability affects only Windows 7. On Linux it the adapter and disk work without any issue. Since the OP appears to be running Win10 and there were no mentions of workarounds used, it's very possible that the issue was already fixed on Windows at some point.

I encountered extremely bad transfer speeds it seems all the writes and reads had to be translated through the cpu because of the adaptation or something like that. I was literally getting kbps on older systems. And low mbps on newer hardware.

My advice if you want to do this is as follows

Use a promise tx4 sata II 300 pci card and adapt to sata m.2

That should actually come close to pci saturation. And honestly is probably be the absolute fastest configuration you can put on 33mhz 32bit pci. Might be able to go a little faster with a server style card with sd ram cache but not much.

Strange... On my test system (C2Q based) I was able to saturate USB2.0 speeds (~36MB/s) when copying stuffs from a USB stick (over USB 2.0 port) to the NVMe SSD behind the adapter. Maybe what disk you use matters as the NVMe-to-PCIe adapter is passive, so the PCI-to-PCIe bridge is the only thing between between the NVMe disk and the CPU/chipset. (EDIT: That was tested on Linux... not sure if how OS handles things also matters to some extent)

As for Promise TX4 SATAII 300 PCI... I recall having used it on a much older motherboard (Pentium D based) before and it worked great. I could even boot systems from SATA SSDs attached to it and the performance is comparable to that of the native SATA-150 ports, if not better.

Should say that the option ROMs those SATA cards use (for booting support) consume precious upper memory, which can be a problem if you use DOS, but not a problem otherwise. If not using the disks for booting, disabling the option ROM might prevent the UMB consumption as well as preventing incompatible OSes (that I don't intent to use them on) from seeing the disks, though.

Check cpu usage. I think transfer speed is improved by faster cpus

Sphere478 wrote on 2021-11-22, 03:49:

LSS10999 wrote on 2021-11-22, 03:25:

Strange... On my test system (C2Q based) I was able to saturate USB2.0 speeds (~36MB/s) when copying stuffs from a USB stick (ov […]
Show full quote

Sphere478 wrote on 2021-11-22, 01:53:

I encountered extremely bad transfer speeds it seems all the writes and reads had to be translated through the cpu because of t […]
Show full quote

I encountered extremely bad transfer speeds it seems all the writes and reads had to be translated through the cpu because of the adaptation or something like that. I was literally getting kbps on older systems. And low mbps on newer hardware.

My advice if you want to do this is as follows

Use a promise tx4 sata II 300 pci card and adapt to sata m.2

That should actually come close to pci saturation. And honestly is probably be the absolute fastest configuration you can put on 33mhz 32bit pci. Might be able to go a little faster with a server style card with sd ram cache but not much.

Strange... On my test system (C2Q based) I was able to saturate USB2.0 speeds (~36MB/s) when copying stuffs from a USB stick (over USB 2.0 port) to the NVMe SSD behind the adapter. Maybe what disk you use matters as the NVMe-to-PCIe adapter is passive, so the PCI-to-PCIe bridge is the only thing between between the NVMe disk and the CPU/chipset. (EDIT: That was tested on Linux... not sure if how OS handles things also matters to some extent)

As for Promise TX4 SATAII 300 PCI... I recall having used it on a much older motherboard (Pentium D based) before and it worked great. I could even boot systems from SATA SSDs attached to it and the performance is comparable to that of the native SATA-150 ports, if not better.

Should say that the option ROMs those SATA cards use (for booting support) consume precious upper memory, which can be a problem if you use DOS, but not a problem otherwise. If not using the disks for booting, disabling the option ROM might prevent the UMB consumption as well as preventing incompatible OSes (that I don't intent to use them on) from seeing the disks, though.

Check cpu usage. I think transfer speed is improved by faster cpus

shouldnt be, whole point of NVME is its a native PCI bus mastering solution bypassing CPU completely. You setup queues and it pumps data to/from ram without CPU intervention. There is a chance this particular drive had trouble when paired with PCIE x1 1.0 of the PCI-PCIE converter and other brand of controller would work better.

Sorry for a shameless bump after about a year later. I've been experimenting with similar use cases on even older motherboards recently and, well, they all ended up in failure, unfortunately.

Even with such a PCI-PCIe bridge adapter, I can't really make NVMe SSDs work reliably on a chipset that predates PCIe. The NVMe SSD cannot be initialized on Linux as it would not respond to anything (device not ready in dmesg), and while the SSD works with Windows' default drivers to some extent, after some writings the disk would report errors and become inaccessible (recovers after a restart, though).

The bridge itself appears to work fine, and that the SSD itself I used for testing is working just fine on another system using native PCIe, so nothing was really faulty, probably just that chipsets too old can't really handle things like that. I tried several NVMe SSDs of different brands, and they all exhibit similar behaviors.

In the end I can only reliably use NVMe SSDs on PCI slots using such bridges, if the motherboard's chipset already had PCIe support (which was the case for the one that I already made working). In such cases, such bridges would be mostly useful for adding another (slower) storage when there are no more PCIe slots (including x1 ones) to spare.

But again, from previous posts I noticed it was indeed possible for NVMe SSDs to work on older, pre-PCIe chipsets. I wonder if BIOS needs to be aware of such devices somehow...

In my case I was able to get it to work with debian jessie and 430tx motherbord. But it was so slow I didn’t even let setup finish.

The best way to do this is with a sata controler, and a sata m.2 that will work great!

It may be possible to do this on isa by adapting from ide to sata or scsi to ide to sata