douglar wrote on 2023-06-28, 14:04:

I found this quote interesting: https://www.reddit.com/r/retrocomputing/comme … o_ide_adaptors/

The problem could be that the cheaper adapters just pass the commands through unchanged while the more expensive adapters bother to emulate ATA commands SATA drives are not mandated to support such as CHS commands. If a CHS-only BIOS is used to access the disk, this makes a difference.

I'm assuming the author was talking about the ubiquitous JMicron stata--pata bridges here when talking about the cheap ones. https://www.jmicron.com/products/list/17

Happen to know if there is a way using Linux hdparm or looking at the ATA IDENTIFY information, etc. whether a drive is one that doesn't support CHS access?
I assume if a drive "doesn't support CHS," that means it still supports PIO mode using LBA addresses?
You remind me that recently I tried a brand new 1TB WD drive through the JMicron converter, and it didn't work. I figured it was different revisions of SATA but maybe this is actually the issue. I went to my old standby drive (Toshiba HDWD110) and it worked just fine.

Has anyone had success with Marvell-based adapters on the PIIX4 southbridge (BX chipset, for example)?

I tried three different motherboards, they all had the same issue, either a read error during Windows setup or other odd system hangs... Replaced with an old '06 datecode JMicron adapter and the problem disappeared completely.

SSD used is SanDisk U110.

mockingbird wrote on 2023-08-03, 06:13:

Has anyone had success with Marvell-based adapters on the PIIX4 southbridge (BX chipset, for example)?

Yes, please see [1].

[1]
Re: Adding XT-IDE option ROM to Asus P3B-F BIOS [Thanks to DenizOezmen, it actually works!!!]

Its shameful Microsoft never added TRIM support to XP.

Ive found the most reliable over long periods of time are the Seagate SSHDs with a SATA to ATA adapter from Startech.
had at least 3 years of daily use and its most than fast enough.

ElectroSoldier wrote on 2023-08-03, 09:38:

Its shameful Microsoft never added TRIM support to XP.

Ive found the most reliable over long periods of time are the Seagate SSHDs with a SATA to ATA adapter from Startech.
had at least 3 years of daily use and its most than fast enough.

An occasional boot from a Linux LiveCD/LiveUSB will allow running TRIM from there, as long as you use a new enough version (mainly kernel) that supports NTFS, trimming NTFS and the disk controller used .

1root@bob-VirtualBox:/tmp# dd if=/dev/urandom of=/tmp/ntfsdrive.raw count=1024 bs=1MB
2
31024+0 records in
4
51024+0 records out
6
71024000000 bytes (1.0 GB, 977 MiB) copied, 9.72184 s, 105 MB/s
8
9root@bob-VirtualBox:/tmp# mkfs.ntfs -F /tmp/ntfsdrive.raw 
10
11/tmp/ntfsdrive.raw is not a block device.
12
13mkntfs forced anyway.
14
15The sector size was not specified for /tmp/ntfsdrive.raw and it could not be obtained automatically.  It has been set to 512 bytes.
16
17The partition start sector was not specified for /tmp/ntfsdrive.raw and it could not be obtained automatically.  It has been set to 0.
18
19The number of sectors per track was not specified for /tmp/ntfsdrive.raw and it could not be obtained automatically.  It has been set to 0.
20
21The number of heads was not specified for /tmp/ntfsdrive.raw and it could not be obtained automatically.  It has been set to 0.
22
23Cluster size has been automatically set to 4096 bytes.
24
25To boot from a device, Windows needs the 'partition start sector', the 'sectors per track' and the 'number of heads' to be set.
26
27Windows will not be able to boot from this device.
28
29Initializing device with zeroes: 100% - Done.
30
31Creating NTFS volume structures.
32
33mkntfs completed successfully. Have a nice day.
34
35root@bob-VirtualBox:/tmp# mount -o loop /tmp/ntfsdrive.raw /tmp/ntfs
36
37root@bob-VirtualBox:/tmp# lsblk -f /dev/loop18
38
39NAME   FSTYPE FSVER LABEL UUID                                 FSAVAIL FSUSE% MOUNTPOINTS
40
41loop18 ntfs               04F2734611D0D9FC                      971.2M     1% /tmp/ntfs
42
43root@bob-VirtualBox:/tmp# fstrim -av
44
45/tmp/ntfs: 971.2 MiB (1018388480 bytes) trimmed on /dev/loop18
46

there is also this https://archive.org/details/trim_20190926

darry wrote on 2023-08-03, 19:00:

An occasional boot from a Linux LiveCD/LiveUSB will allow running TRIM from there, as long as you use a new enough version (main […]
Show full quote

ElectroSoldier wrote on 2023-08-03, 09:38:

Its shameful Microsoft never added TRIM support to XP.

Ive found the most reliable over long periods of time are the Seagate SSHDs with a SATA to ATA adapter from Startech.
had at least 3 years of daily use and its most than fast enough.

An occasional boot from a Linux LiveCD/LiveUSB will allow running TRIM from there, as long as you use a new enough version (mainly kernel) that supports NTFS, trimming NTFS and the disk controller used .

Copy code to clipboard

1root@bob-VirtualBox:/tmp# dd if=/dev/urandom of=/tmp/ntfsdrive.raw count=1024 bs=1MB
2
31024+0 records in
4
51024+0 records out
6
71024000000 bytes (1.0 GB, 977 MiB) copied, 9.72184 s, 105 MB/s
8
9root@bob-VirtualBox:/tmp# mkfs.ntfs -F /tmp/ntfsdrive.raw 
10
11/tmp/ntfsdrive.raw is not a block device.
12
13mkntfs forced anyway.
14
15The sector size was not specified for /tmp/ntfsdrive.raw and it could not be obtained automatically.  It has been set to 512 bytes.
16
17The partition start sector was not specified for /tmp/ntfsdrive.raw and it could not be obtained automatically.  It has been set to 0.
18
19The number of sectors per track was not specified for /tmp/ntfsdrive.raw and it could not be obtained automatically.  It has been set to 0.
20
21The number of heads was not specified for /tmp/ntfsdrive.raw and it could not be obtained automatically.  It has been set to 0.
22
23Cluster size has been automatically set to 4096 bytes.
24
25To boot from a device, Windows needs the 'partition start sector', the 'sectors per track' and the 'number of heads' to be set.
26
27Windows will not be able to boot from this device.
28
29Initializing device with zeroes: 100% - Done.
30
31Creating NTFS volume structures.
32
33mkntfs completed successfully. Have a nice day.
34
35root@bob-VirtualBox:/tmp# mount -o loop /tmp/ntfsdrive.raw /tmp/ntfs
36
37root@bob-VirtualBox:/tmp# lsblk -f /dev/loop18
38
39NAME   FSTYPE FSVER LABEL UUID                                 FSAVAIL FSUSE% MOUNTPOINTS
40
41loop18 ntfs               04F2734611D0D9FC                      971.2M     1% /tmp/ntfs
42
43root@bob-VirtualBox:/tmp# fstrim -av
44
45/tmp/ntfs: 971.2 MiB (1018388480 bytes) trimmed on /dev/loop18
46

Yeah but its a lot of work for so little pay off when you consider the actual performance benefits over a Seagate SSHD.

Page file on an SSD could have some utility though.

darry wrote on 2023-08-03, 06:38:

Yes, please see [1].

[1]
Re: Adding XT-IDE option ROM to Asus P3B-F BIOS [Thanks to DenizOezmen, it actually works!!!]

Thanks, I had no idea that existed! Unfortunately, it halved the write performance (also tested with the Marvell though not shown here):

Default Intel (Microsoft) Driver:

IAA Driver:

I confirmed UDMA was enabled with the IAA app (you also have to get that from the extracted data cab because IAA 2.3 will not install the usual way for the BX chipset).

So I guess I'll just stick with the JMicron for this combo.

douglar wrote on 2023-06-28, 14:04:

I found this quote interesting: https://www.reddit.com/r/retrocomputing/comme … o_ide_adaptors/ […]
Show full quote

I found this quote interesting: https://www.reddit.com/r/retrocomputing/comme … o_ide_adaptors/

The problem could be that the cheaper adapters just pass the commands through unchanged while the more expensive adapters bother to emulate ATA commands SATA drives are not mandated to support such as CHS commands. If a CHS-only BIOS is used to access the disk, this makes a difference.

I'm assuming the author was talking about the ubiquitous JMicron stata--pata bridges here when talking about the cheap ones. https://www.jmicron.com/products/list/17

What do you suppose the "expensive" adapters are?

I have one of these bi directional devices and it doesn't seem to support CHS. https://www.startech.com/en-us/hdd/pata2sata3

What about the Marvell 88SA8052 PATA to SATA Bridge - https://www.marvell.com/content/dam/marvell/e … ief-2015-09.pdf
Or the Genesis Logic GL831 - https://www.arrivalelectronics.co.uk/uploads/ … tasheet_103.pdf

Are there other ones?

Probably depends also on which chipset are they used. So far I never had an issue with JMicron or Marvel.

Comos wrote on 2023-09-18, 11:52:

douglar wrote on 2023-06-28, 14:04:

I found this quote interesting: https://www.reddit.com/r/retrocomputing/comme … o_ide_adaptors/ […]
Show full quote

I found this quote interesting: https://www.reddit.com/r/retrocomputing/comme … o_ide_adaptors/

The problem could be that the cheaper adapters just pass the commands through unchanged while the more expensive adapters bother to emulate ATA commands SATA drives are not mandated to support such as CHS commands. If a CHS-only BIOS is used to access the disk, this makes a difference.

I'm assuming the author was talking about the ubiquitous JMicron stata--pata bridges here when talking about the cheap ones. https://www.jmicron.com/products/list/17

What do you suppose the "expensive" adapters are?

I have one of these bi directional devices and it doesn't seem to support CHS. https://www.startech.com/en-us/hdd/pata2sata3

What about the Marvell 88SA8052 PATA to SATA Bridge - https://www.marvell.com/content/dam/marvell/e … ief-2015-09.pdf
Or the Genesis Logic GL831 - https://www.arrivalelectronics.co.uk/uploads/ … tasheet_103.pdf

Are there other ones?

Probably depends also on which chipset are they used. So far I never had an issue with JMicron or Marvel.

JMicron ones (or at least some chip models) have issues with VIA 686B and Intel ICH4 based IDE ports. (there may be others) Oherwise, they work well.

Mavell ones are reputed to be the most compatible, but there is at least one thread here about a newer chip revision or maybe a counterfeit chip having issues.

Sunplus also makes chips for such adapters. At least one Startech model uses those.

darry wrote on 2023-09-18, 23:10:

Mavell ones are reputed to be the most compatible, but there is at least one thread here about a newer chip revision or maybe a counterfeit chip having issues.

Sunplus also makes chips for such adapters. At least one Startech model uses those.

Yes, I will have to follow up on the newer, unlabeled chips supposedly not being the same as the older Marvell chips with the logo. I tried to transplant one on my mSATA adapter, but I think I received a non-functioning chip. I'm waiting for my next chip to arrive to test.

Sunplus isn't good either.

darry wrote on 2023-09-18, 23:10:

JMicron ones (or at least some chip models) have issues with VIA 686B and Intel ICH4 based IDE ports. (there may be others) Oherwise, they work well.

I've had trouble using the Jmicron bridges with the Promise PDC-20630 based VLB controllers, which was disappointing because it's been the fastest VLB IDE controller that I've looked at, but my fastest IDE devices are Sata based.

douglar wrote on 2023-09-21, 12:05:

darry wrote on 2023-09-18, 23:10:

JMicron ones (or at least some chip models) have issues with VIA 686B and Intel ICH4 based IDE ports. (there may be others) Oherwise, they work well.

I've had trouble using the Jmicron bridges with the Promise PDC-20630 based VLB controllers, which was disappointing because it's been the fastest VLB IDE controller that I've looked at, but my fastest IDE devices are Sata based.

That's interesting. My experience with PCI Promise controllers
(Ultra 100 and 133, AFAICR) and JMicron has been positive .

So, newer Promise ones sarted working whereas newer Intel ones stopped working, 🤣. We need a compatibility table/reference.

douglar wrote on 2023-09-21, 12:05:

I've had trouble using the Jmicron bridges with the Promise PDC-20630 based VLB controllers, which was disappointing because it's been the fastest VLB IDE controller that I've looked at, but my fastest IDE devices are Sata based.

Did you ever get it to work?
I salvaged a 20630 card (EVLSIO-V2-HJ4) from a functional 486 that just got thrown out, but I'm unable to make it work with anything larger than 2GB, be it native IDE or bridged SATA (flash and rust). I don't need any capacity beyond that, but it simply will either not pass POST at all (so the BIOS complains about HDD failure), or never be able to read or write these disks later on. Some are auto-detected at max supported capacity (2GB), some show up as 125MiB/131MB. The EIFDISK tool will show connected drives (EIDE2300 complains about not finding a card...), but adding partitions will also do nothing. Speedsys from PhilsComputerLabs DOSbench pack also displays a Secondary Master with correct device name over a JMicron Bridge, but isn't able to access it.

Bzzz wrote on 2024-10-20, 01:33:

douglar wrote on 2023-09-21, 12:05:

I've had trouble using the Jmicron bridges with the Promise PDC-20630 based VLB controllers, which was disappointing because it's been the fastest VLB IDE controller that I've looked at, but my fastest IDE devices are Sata based.

Did you ever get it to work?
I salvaged a 20630 card (EVLSIO-V2-HJ4) from a functional 486 that just got thrown out, but I'm unable to make it work with anything larger than 2GB, be it native IDE or bridged SATA (flash and rust). I don't need any capacity beyond that, but it simply will either not pass POST at all (so the BIOS complains about HDD failure), or never be able to read or write these disks later on. Some are auto-detected at max supported capacity (2GB), some show up as 125MiB/131MB. The EIFDISK tool will show connected drives (EIDE2300 complains about not finding a card...), but adding partitions will also do nothing. Speedsys from PhilsComputerLabs DOSbench pack also displays a Secondary Master with correct device name over a JMicron Bridge, but isn't able to access it.

I’ve got some new devices to try but recently I’ve been spending my retro time cataloging my stuff on the retro web. I want to get back to this soon.

Are you using your 20630 with the dos driver, a promise bios, or xub?

For people with the J-Micron "Host Bribge" adapter, the fix below will bring you the same speeds as a Startech adapter at less price.
I discovered the fix on the OGXbox forums and have translated it from Spanish.

PcBytes wrote on 2024-10-20, 17:01:

For people with the J-Micron "Host Bribge" adapter, the fix below will bring you the same speeds as a Startech adapter at less price.
I discovered the fix on the OGXbox forums and have translated it from Spanish.

So this fix enables MWDMA? Ill have to try it out.

Based on the schematic, it just connects the PDIAG signal properly.
https://hackaday.io/project/186809-m2-sata-to … rse-engineering

My 2.5" adapter connects pin 46 of the chip directly to the IDE connector with no series resistor or pull-up. It doesn't seem to make any difference because the motherboard I'm using doesn't connect that pin at all (it's completely floating - verified with the schematic), and it still only uses UDMA2.

OK, my adapter certainly matches the unpatched example. I'll have to get on this as soon as I get some 100 Ohm resistors

PDIAG just matters for handling master/slave situation - its only used during reset/poweron and drive detection. Possibility of this being a dual use pin (setup strap) are minimal because it needs to be pulled high in normal use (hence 10K to VCC).
On mine JM20330 pin 46 is directly connected to IDE Pdiag with no weird 100 ohm resistors, and pulled high with 10Khom.

One thing that comes to mind is bad wiring making JM20330 think there is a slave drive unable to finish its init on the cable and going into some slower safety mode.