XTIDE Universal BIOS v2.0.0 beta testing thread

[Malc]

Set J1/J2 jumpers to Hi speed mode, See post #488, When using CF originally setup with XUB 1.1.5 and you later re-flash with a recent revision of the XUB (New drive translation code) you will likely have to wipe / partition / format and reinstall, Likewise if you setup the CF card using a recent revision of the XUB and then go back to XUB 1.1.5 you will likely have to wipe and setup again.

So I finally got around to testing this, and did some benchmarks with a 4GB CF card using the 1.1.5 BIOS already on the glitchworks v4 card, and aside from not being able to boot directly off the CF card, it went well. The 1.1.5 BIOS displayed drives 80 and 81 in a menu (there's an MFM drive in the system) and booting from 80 or A: worked, and I could fdisk/format and run disktest and everything checked out. I figured not being able to boot off the CF card was due to LBA support in 1.1.5 so I didn't worry about it at the time.

Then I flashed r602 (from the 6300 itself) and now, on bootup, the v4 card can't see the CF card at all. Both Master and Slave (at 300h) show "not found". I was careful to load IDE_XT.BIN and change the card type to 6300/Olivetti M24, and not much else, and the flash appeared to be successful. Both J1/J2 jumpers on my board are set to compatibility mode (as opposed to hi-speed/chuckmod mode).

What am I missing? The card isn't bricked; the BIOS starts up fine, it just doesn't think there's anything on Master (when the 1.1.5 BIOS could see it just fine).

Also, I never saw a menu for picking the boot hard drive; was that removed between 1.1.5 and 2.0.0 b3, or did I miss another setting?

 

[Trixter]

Whoops, my bad! I was so ingrained with "must not attempt adjacent register reads on M24" that I confused myself into keeping it in compat mode. I'll switch the jumpers tonight and retest.

 

[Trixter]

Okay, full test results follow. The new M24/6300 code is definitely a small improvement, and running the NEC code on an NEC V30 is an additional improvement.

XUB 1.1.5, compat mode, 8086:

Write Speed         : 77.05 KB/s
Read Speed          : 91.04 KB/s
8K random, 70% read : 9.7 IOPS
Sector random read  : 59.0 IOPS

Average access time (includes latency and file system overhead), is 17 ms.

I then flashed it to XUB 2.0.0 b3 (r602) IDE_XT (808x code) and moved jumpers to high-speed/chuckmod mode:

Configuration: 4 MiB test file, 256 IOs in random tests.

Write Speed         : 84.26 KB/s
Read Speed          : 102.71 KB/s
8K random, 70% read : 10.6 IOPS
Sector random read  : 57.5 IOPS

Average access time (includes latency and file system overhead), is 17 ms.

I then replaced the 8086 with an NEC V20:

Write Speed         : 87.52 KB/s
Read Speed          : 103.72 KB/s
8K random, 70% read : 10.8 IOPS
Sector random read  : 63.8 IOPS

Average access time (includes latency and file system overhead), is 16 ms.

Finally, I reflashed with IDE_XTP to get the NEC V20 code:

Write Speed         : 106.83 KB/s
Read Speed          : 142.57 KB/s
8K random, 70% read : 14.0 IOPS
Sector random read  : 69.6 IOPS

Average access time (includes latency and file system overhead), is 14 ms.

Looking solely at read speeds, the improvements were:

1.1.5 to 2.0.0b3: 12% faster (this is what Krille originally wanted to know)
2.0.0b3 8086 code to 2.0.0b3 NEC code: 38% faster

Hope this helps! I broke a 6300 doing this so I'll start a new thread for that one

 

[alecv]

XT IDE Universal BIOS bug?

I've found an old XT-compatible computer with WD-1002 MFM controller and ST-225 HDD.
HDD is formatted with MITAC ADM v1.0 and divided onto two partitions (10Mb+10Mb).
The first partition is bootable but protected with ADM password (I do not know it).
The second partition is open.

After installing XUB (~589) with boot manager module (F2) I see a two "Foreign disk 80h".
"Foreign disk 81h' and my IDE drive as 82h.

IDE drive is bootable (with XUB boot manager), I can see my IDE as C: (80h) and second MFM partition as D: (82h)

Unfortunately, first MFM bartition becames not bootable after XT IDE installation.

 

[Malc]

I've not heard of MITAC ADM v1.0 before, Got a link ?, What version of DOS are you running, What does FDISK say about that partition ?.

 

[alecv]

XUB does not boot disk formatted by ADM.
I've found a bit newer version of ADM (at warez site, sorry)
http://old-dos.ru/files/file_887.html

 

[Krille]

Okay, full test results follow. The new M24/6300 code is definitely a small improvement, and running the NEC code on an NEC V30 is an additional improvement.

Thanks for testing Jim, I hope you get the machine in working order again.

XT IDE Universal BIOS bug?

Nope. Make sure that the MFM controller is not configured for two drives (check jumper settings).

BTW, on a slightly related note, have you tried the new MODULE_MFM_COMPATIBILITY?

 

[pikadroo]

I can't seem to find any mention of a problem I am having with floppy drives, so here it is maybe someone can help.

With the XTIDE bios installed I am unable to format floppies due to a "Invalid media or Track 0 bad - disk unusable" error. My system is probably an extremely rare Cumulus brand 386SX with a built in phoenix bios that has limited IDE support. All version of the XTIDE bios in v2 will boot a CF card just fine into DOS 6.22 on this computer, it works so great I can't even believe it. The floppy format problem only occurs in v2 beta 2 and v2 beta 3 however, I can format floppies when I use v2 beta 1 just fine. If the XTIDE bios is removed from the computer formatting floppies works just fine.

Thanks!
-Drew

 

[maxtherabbit]

I can't seem to find any mention of a problem I am having with floppy drives, so here it is maybe someone can help.

With the XTIDE bios installed I am unable to format floppies due to a "Invalid media or Track 0 bad - disk unusable" error. My system is probably an extremely rare Cumulus brand 386SX with a built in phoenix bios that has limited IDE support. All version of the XTIDE bios in v2 will boot a CF card just fine into DOS 6.22 on this computer, it works so great I can't even believe it. The floppy format problem only occurs in v2 beta 2 and v2 beta 3 however, I can format floppies when I use v2 beta 1 just fine. If the XTIDE bios is removed from the computer formatting floppies works just fine.

Thanks!
-Drew

Funny I ran into the same thing, builds of XTIDE BIOS for AT systems have the issue, but those for XT+ systems do not. I'm looking over the SVN now and it looks like these beta 2 and beta 3 builds are now woefully obsolete. This was probably already fixed but the situation with hosting the code for this project is a hot mess. I'm going to try to build the latest revision of the source available on google code and see what that does.

 

[Twospruces]

Hi,
in the XTIDE hardware thread I have been describing a problem I am having getting a custom XT-CF Version 2 lite card working in a Zenith Z-171.

I think what I need is for XUB to be much more verbose during boot up.
- does it see the CF card?
- does it see the partition?
- does it correctly read the MBR? what is the MBR content?
- does it read the format of the partition correctly?
- can I get full manual control over the boot up process?

Do I need to roll up my sleeves and get ready to compile my own variant of XUB or is there an easy way to see what's going on during XUB boot up?

thx

 

[Trixter]

Hi,
I think what I need is for XUB to be much more verbose during boot up.
- does it see the CF card?

This already happens. You should see two lines that look like this (sample, your values may be different):

Master at 300h: TRANSCEND
Slave  at 300h: Not Found

If you're not seeing this, you have a fundamental issue that needs to be corrected first. Your boot-up screen should at least contain the first seven lines of something like this: http://wiki.sensi.org/images/pcemhd.jpg

- does it correctly read the MBR? what is the MBR content?

If the MBR can't be read, I believe XUB prints a single letter error code. You can look in the source for what the code represents.

 

[Twospruces]

Thanks,
I do in general see XUB report the Cf card presence.
I will try the load that includes the boot loader menu and see if I can get better control over boot up. Thanks

 

[Chuck(G)]

FWIW, I've successfully run the XTIDE with 8 and 32 MB ancient CF cards. I think I've got a 2MB one somewhere around here...

 

[Twospruces]

Hi, I have an interesting update.
First, the major change I made was to use R566 IDE_XTL.BIN so that I have the boot menu available. This has proved very valuable.
Seems that the issue is now apparent as you will see.

Upon initial power up (turn off computer, remove CF card, wait 10 sec, install CF card, power up computer), we see this boot menu.



Here we see the CF card is detected but does NOT show up as a bootable disk option.

CNTL - ALT - DEL, warm restart

on this attempt we see



Now, we see that the CF card is a bootable disk option. In fact I can just select the 80h drive and the machine boots no problem.



As we see, the BIOS messages "Booting C>>C". Great this is how it should work every time!

However,
CNTL - ALT - DEL, warm restart

Now, something different again.



Now we see a new entry - Foreign Hard Disk - with some tiny size allocation - at 80h and the proper disk at 81h.
When I select to boot from the 2nd entry I successfully boot, with the BIOS messages "Booting D>>C".

The pattern repeats. Every time I reboot, another false entry is created and the proper disk moves down by 1 value.
I can still boot the machine.



I can restart the sequence by doing a hard power cycle including removal of the CF card.


OK so - Does anyone understand what is going on, and how to correct this?

THANKS!
(the good news is that it is sort of working..)

 

[Trixter]

That is bizarre. My off-the-cuff guess is that your clone is not clearing memory on a warm restart, and the XUB is getting confused by seeing previous data at startup. That's just a guess though; hopefully Krille can comment.

 

[Twospruces]

Hi, I have an interesting update.
First, the major change I made was to use R566 IDE_XTL.BIN so that I have the boot menu available. This has proved very valuable.

mistake; I'm actually using R580 IDE_XTL.BIN.

* R590 and newer do not work at all.
* R566 seems to behave the same way but I don't have R566 IDE_XTL.BIN (does anyone have it?)
* I have a 16MB Hitachi and 1GB Sandisk, and they are both working equivalently.
* I create the correct MBR on these cards by mounting them (IDE CF adapter) in a Pentium 4 PC running DOS 5. FDISK /MBR
* in the Z171 clone, I can then (using my procedure in prior post) to get the disk visible, then remove and rebuild partition, format C: /s , copy DOS from A to C

Something definitely changes in XUB at some point that makes it incompatible with my machine. I have downloaded the SVN repo; if we can figure out how to get a normal disk detection fix, then R580 could be a baseline revision for a fix.

 

[Trixter]

* R590 and newer do not work at all.

Did you configure them before flashing? They have a slow compatibility mode you likely need to ensure is set before flashing them; they may not work out of the box without configuration.

 

[Eudimorphodon]

Now we see a new entry - Foreign Hard Disk - with some tiny size allocation - at 80h and the proper disk at 81h.
When I select to boot from the 2nd entry I successfully boot, with the BIOS messages "Booting D>>C".

The pattern repeats. Every time I reboot, another false entry is created and the proper disk moves down by 1 value.
I can still boot the machine.

I'm curious, are you using the "Full Operating Mode" setting? Per the docs:

Full mode supports up to 5 IDE controllers (10 drives). Full mode reserves a bit of RAM from top of base memory. This makes possible to use ROM Basic and software that requires top of interrupt vectors where XTIDE Universal BIOS parameters would be stored in lite mode. Lite mode supports only one IDE controller (2 drives) and stores parameters to top of interrupt vectors (30:0h) so no base RAM needs to be reserved. Lite mode cannot be used if some software requires top of interrupt vectors. Usually this is not a problem since only IBM ROM Basic uses them. Tandy 1000 models with 640 kiB or less memory need to use lite mode since top of base RAM gets dynamically reserved by video hardware. This happens only with Tandy integrated video controller, not with expansion graphics cards. It is possible to use full mode if reserving RAM for video memory + what is required for XTIDE Universal BIOS. This would mean 129 kiB but most software should work with 65 kiB reserved.

That it keeps counting up possible "phantom" hard drive numbers past 81h sort of makes it sound like it's going past what would be allocated in "lite" mode?

 

[Twospruces]

Did you configure them before flashing? They have a slow compatibility mode you likely need to ensure is set before flashing them; they may not work out of the box without configuration.

thanks for the tip. I am generally now running XTIDECFG on all raw initial images before I run them, but I missed the compatibility mode. I'll try that.

I'm not using "Full Operating Mode" either. I can give that a try to see if it makes a difference.

This issue seems to be different though. Where does XUB store the detected drive info? On the first try from cold restart, XUB does not even present any HDD for boot up.

 

[Twospruces]

Update re Z171 + XT-CF lite rev2.

* tried R602 again, ran XTIDECFG on raw image, autodetect, turned off block transfers, enabled full mode, saved image.
* installed image into flash
* system now behaves the same as with R580. Same exact problem as described in post 534.

I looked in the config menus, and I re-read the XUB main page. I don't see an option to select "slow mode".

Anyhow, net-net, the latest XUB build shows the problem.