• Please review our updated Terms and Rules here

New TestFDC Results Registry

glitch

Veteran Member
Joined
Feb 1, 2010
Messages
5,051
Location
Central VA
I'd been trying to reach Dave Dunfield with new TestFDC results since apparently August (forum thread) with no results. So, I wrote a new TestFDC registry into my site:

https://services.theglitchworks.net/ng/testfdc_results

This registry currently includes Dave's last registry update from 2007. There's now a form for entering your results, you can find it as a link from the registry, or here:

https://services.theglitchworks.net/ng/testfdc_results/new

Result submissions have to be manually approved currently so that the registry doesn't get spammed. Text export forthcoming. Any suggestions welcome!

Moderators, if someone wants to sticky this (here or in other forums), I think this would be a valuable resource for anyone wanting to use ImageDisk on non-PC formats.
 
Glitch, it's been my experience that FM-compatible and even 128-byte MFM-capable FDCs on later-model P3->AMD2+ motherboards have proliferated considerably. For example, my FIC KC-19+ P3 motherboard passes all (including 128 byte MFM) tests quite nicely and my Advantech P4 board passes all but the 128-byte MFM test. I've got several later motherboards that I haven't bothered to test, but doubt that they'll flunk the FM test.

Oddly, most of the P1 motherboards that I have flunk the FM test. So older doesn't mean better.

So your list, were it exhaustive, might be very long, indeed.
 
Thanks for creating this new list. I'll make a note to dig up the results of a few additional systems I tested. As machines with real FDCs become harder to find, it will become more valuable to have some idea it the machine will work or not first.

One might expect devices like the Kryoflux or SuperCard Pro to take the place of an FDC, but I have run in to so many problems trying to use these for inter-operation that I think real FDC "tweeners" will need to stick around for a while longer.
 
Glitch, it's been my experience that FM-compatible and even 128-byte MFM-capable FDCs on later-model P3->AMD2+ motherboards have proliferated considerably.

Huh, interesting! I'll have to test a few of mine that have floppy controllers.
 
I'm ready to submit a few new results, but there's a bug in TESTFDC 1.18 that affects my tests.

That's the high-density 128-byte MFM sector test run on a 3.5" drive. All of my tests show "fail" even though I can, using my own code, format write and read back 128 byte sectors at 500Kbps. Is there a different version that I should be using? Interestingly, the 250Kbps MFM 128-byte tests pass. Is perhaps the gap being miscalculated?
 
Interesting. I'm fairly sure I used that same version to test a couple of FDCs a while back that passed all tests. (I have yet to dig those out and re-test). I wonder what is different? Hmmm, if it were incorrectly marking just some, but not all, chips as failed when they really should pass, then nobody would have noticed.

Chuck, does *ImageDisk* read/write/format 128-byte sectors OK on the chip in question? Of course YOUR code would succeed where other failed. :)
 
I've verified the bug. I created a dummy CP/M format with 48x128 byte sectors on each track, two sides, 80 cylinders on a DSHD 3.5 floppy. I copied files from the Win98 /WINDOWS/COMMAND directory--lots of stuff there.

IMD could read the disk and create an image. It could then correctly write the image back to disk (hint: use interleave 2:1 if you don't want to grow old waiting for the process). Here's the image file--it's rar-compressed

Using the same floppy, as target, I get the following results from testfdc 1.18:

Code:
Report on FDC capabilities, issued 18/01/2018 21:48:56:
Single-Density at 250 kbps ............................ Not tested
Single-Density at 300 kbps ............................ Not tested
Single-Density at 500 kbps ............................ Not tested
Double-Density at 250 kbps ............................ Not tested
Double-Density at 300 kbps ............................ Not tested
Double-Density at 500 kbps ............................ Not tested
Double-Density at 250 kbps / 128 byte sectors ......... Passed
Double-Density at 300 kbps / 128 byte sectors ......... Not tested
Double-Density at 500 kbps / 128 byte sectors ......... Failed

So, if you see a result where the 128 MFM test passes at 250 Kbps, but fails at 500 Kbps, disregard the failure.
 
Do we also want results for low-density controllers? I would assume just put "N-not tested" for the 300kbps and 500kbps tests?

I noticed that the low-density controller card with the Tava system I mentioned the other day actually supports FM.

BTW, is 128-byte sector MFM support supposed to work at all on earlier/low density chips? I noticed even an original IBM PC/XT FDC fails that test. Funny thing is since it does not verify what it writes, Imagedisk will happily format and write an image with a few 128-byte MFM 250kbps (low density) sectors, but it can not read it back.
 
The writing bit often doesn't hard fail (as far as the software is concerned) but the actual data written to the disk will be corrupt, that's part of the point in running TestFDC. I've had a bunch of machines that pass on the write section of TestFDC but die on the read/verify bits.
 
Ok, starting to compile a list to send in, and just a few questions and observations:

For "class", the entry page refers to the physical form such as AT/ATX/ISA but the list uses CPU type to indicate motherboards. Is physical form really preferred? And what about proprietary motherboard form factors?

It might be a good idea to add a disclaimer stating that results from a particular product model may vary depending on the FDC chip. This is especially true of the AHA15xx cards that may use either Intel or NSC chips.

For newer motherboards with integrated FDCs, I suspect most just would submit "?" as FDC chip, but after thinking about it, I think ideally this should be the I/O chip (For example, IT8720F), rather than the motherboard's chip set unless perhaps it is built in to the chip set. Usually the only way to get that is to look at the motherboard.
 
That list never got updated anyway. I had submitted some new entries and as I recall the never got added.
 
I get a 404 error. Is your site still up?

Sorry, moved to a new server and didn't update the redirect rules for that one. Here's the new link:

https://services.glitchworks.net/ng/testfdc_results

That list never got updated anyway. I had submitted some new entries and as I recall the never got added.

It got updated and something happened to the DB, I still have all of the submissions, I just need to go back through and approve them/process the notes sections.
 
Ok, great. If adding new stuff will be working in the future, perhaps I'll get off my ass and test a few more things.
 
Just finished integrating a tool that lets me do approvals in a less manual fashion -- I'd been doing it by hand in the DB until now. Better tools should mean faster/more frequent approvals!
 
Hey thanks for bumping this! Noticed my Win98 box's AP5T board was on the list and it motivated me to give it a run, and it supports FM on the built-in FDC like a freaking champ~!

(Also IMD works perfectly inside Win98SE GUI)

EDIT: actually while IMD works just fine, Win 98 REALLY didn't like me accessing the FDC directly without its knowledge. When I went back to access the drive through 98 it made some heinous seeks and bounced off the limiter :evil: (at this point I had a standard MFM DOS formatted disk in there and IMD was no longer running)
 
Last edited:
I've been meaning to share some research I did on good FDCs a while back.

Most enthusiasts know about the good NatSemi FDCs, but there were also good ones made by ALi (Acer Labs, Inc.). Their FDCs appeared in both their Super I/O chips and Southbridge chips. The southbridge chips with good FDC support are the M1535 and M1543, including variants (see below). These southbridges were paired with multiple generations of northbridges, thus it’s potentially possible to find a motherboard up to Socket 478 with good floppy support. In reality, I’ve not been able to find any beyond about Socket 370 however. ALi was overtaken by VIA and SIS after the Socket 370 era and lost market share. ALi’s later southbridges, M1563 and M1567, don’t have floppy support.

ALi also partnered with PC Chips, and several generations of their Aladdin chipsets were re-branded on PC Chips boards. Any of these PC Chips chipsets should have good ALi FDCs in their southbridge: BXcel, HX Pro, Super TX, TX Pro, TX Two, and GFXpro. I own a PC Chips M754LMR+ (GFXpro) and the FDC is great.

It's tedious to sort out all the various labels and generations of their southbridges -- for example, there were M1535, M1535+, M1535D, M1535D+, HT1535+, HT1535D+ -- but from what I can infer, any of them with a "3" or "4" as the 3rd digit should have the good FDC cores. For example, the GFXpro southbridge is HT2543CB1.

The Acer Super I/O chips are less documented (and I need to do more testing), but the M5105 was commonly found on ISA cards in the early ‘90s and has good floppy support. The Diamond Flower MIO-400KF cards I've come across use this chip and are often cheap on eBay.

Hope that helps people sort out their own stock and/or shop for boards with good FDCs.
 
Back
Top