DOS ain't dead

Hi,
there's a new faster IDE driver called XIDE from Jack, forwarding the info:

Johnson Lam has posted a new 15-May-2015 DRIVERS.ZIP file, in his "drop
box" website at --

https://dl.dropboxusercontent.com/u/15785527/dos/file/drivers.zip

The new XIDE, XIDE-X and XIDE-P drivers replace the old UIDE/UHDD/UDVD2
drivers. The "X" drivers are "closed source" -- I shall NOT offer any
more source files, after Rugxulo's very UNFRIENDLY post about me and my
work on BTTR, for which he had NO excuse and should be condemned!

XIDE is nearly identical to the old UIDE. XIDE-X is faster, as it has
a buffer to improve cache binary-search speed and also "overlaps" cache
tasks with disk UltraDMA. XIDE-X does require 788 bytes more memory,
all of which goes in "free HMA" by loading with a /H switch. With /H,
both XIDE and XIDE-X use only 880 bytes of upper/DOS memory. Both can
set caches of up to 4-Gigabytes in XMS memory.

XIDE-P is the fastest of all! It uses "Protected Caching", as in UIDE
drivers from 2007-2012, and puts its binary-search table in the HMA, or
in upper/DOS memory. This is faster on protected-mode systems, due to
70% fewer XMS accesses which require a slow Int 15h "trap" in protected
mode. XIDE-P also "overlaps" disk UltraDMA with caching tasks. The
driver permits only 1-Gigabyte caches, due to the memory needed for its
search table (36K total for a 1-GB cache). But XIDE-P gives 2% faster
speed in real-mode, also up to 10% faster speed in protected-mode using
JEMM386/EMM386, etc. Yes, it is 10% faster than XIDE or the old UIDE!

If you use a DOS system that has little "free HMA" (V7.0+ MS-DOS etc.),
XIDE-P still offers a 400-MB cache, using 17,216 bytes of upper memory,
by omitting its /H switch and using a CONFIG.SYS command similar to:

DEVICEHIGH=C:\DRIVERS\XIDE-P.SYS /S400

Larger or smaller caches with XIDE-P do take more or less upper-memory.

If you use FreeDOS, you should be able to set about a 350-MB cache with
XIDE-P for merely 944 bytes of upper-memory, by loading XIDE-P from the
AUTOEXEC.BAT file. This requires "BUFFERS=10" in the CONFIG.SYS file,
using FreeDOS "DEVLOAD", and an AUTOEXEC command similar to:

DEVLOAD /H C:\DRIVERS\XIDE-P.SYS /S350 /H

FreeDOS does not offer "free HMA" before AUTOEXEC runs, so the above is
needed to load XIDE-P (or any of the caching drivers) in its HMA, which
saves memory. I have only V1.0 FreeDOS, from Lucho's old 2008 "boot"
diskette. V1.1 FreeDOS may offer more or less "free HMA" and may need
to be tested, with these drivers.

Do download the latest 15-May-2015 drivers from Johnson Lam's "dropbox"
and try them on your systems. I am very proud of the new XIDE-P, with
its 10% higher speed (I never thought such a gain was possible!), and I
hope the drivers can help you, in your DOS work!

Best wishes,

Jack R. Ellis

---
DOS gives me freedom to unlimited HW access.

> The new XIDE, XIDE-X and XIDE-P drivers replace the old UIDE/UHDD/UDVD2
> drivers. The "X" drivers are "closed source" -- I shall NOT offer any
> more source files, after Rugxulo's very UNFRIENDLY post about me and my
> work on BTTR, for which he had NO excuse and should be condemned!

http://sourceforge.net/p/freedos/mailman/message/34175817/

> Re: [Freedos-user] Getting any CD player to work
> From: John Hupp <freedos@pr...> - 2015-06-04 20:17:05
>
> I correct myself. I was using ide-cd.sys on another machine I was
> working with a couple weeks ago. On this machine I have been using the
> default uide.sys.
>
> But pursuing the driver-as-a-suspect angle anyway, I found a Lite-on
> DOS driver and installed that. CD playing now works. Thank you for
> the idea, Eric!!
>
> If there is a sad note, it is that uide.sys has fallen short on two
> machines in a row. On the other one, I had no CD function at all.
> In this one, I had data CD function, but no audio CD.

Johnson Lam has posted an updated 5-Jun-2015 DRIVERS.ZIP file with a
single XIDE driver, in his "dropbox" at:
https://dl.dropboxusercontent.com/u/15785527/dos/file/drivers.zip

XIDE is now a single driver, using a /P switch to request high-speed
"protected caching", also a /X switch to disable overlap of UltraDMA
disk I-O with caching on an "old" or "odd" mainboard requiring this.
XIDE's overlap of UltraDMA disk input with caching is also improved.

Do try the new XIDE driver, which does everything from only a single
8.5K XIDE.SYS file (12K after UPX unpacking, but it returns unneeded
load-time space back to DOS for other uses).

Also, regarding the following post:
www.bttr-software.de/forum/forum_entry.php?id=14264

Other FD-User posts denote the use of a 15-year-old "Lite-On" CD/DVD
drive, a cheap brand "noted" for trouble only their drivers fix, and
such an old a drive may not be fully ATAPI-compatible. XIDE or its
predecessors are written to support "modern" CD/DVD drives which are
ATAPI-compatible, the "standard" in CD/DVD drives for years. Older
CD/DVD schemes are unsupported.

I tested the 5-Jun-2015 XIDE, using my 2007 Gigabyte CD/DVD which IS
ATAPI-compatible, with "CDP" V1.27 from Simtel playing an "old" Dean
Martin audio CD. No problems, and all "CDP" functions (open/close
tray, rewind, fast-forward, etc.) worked fine. Users of XIDE with
"modern" ATAPI CD/DVD drives should also have no problems playing an
audio CD or reading CD/DVD data files.

Jack R. Ellis

---
DOS gives me freedom to unlimited HW access.

Jack Ellis: http://sourceforge.net/p/freedos/mailman/message/34204671/

Rugxulo: http://sourceforge.net/p/freedos/mailman/message/34205889/

I do not have the ability to ban Jack from freedos-user. However, I am still moderator here at BTTR, and any non-technical (i.e. emotionally rude or directly insulting) comments and posts (from him, even indirectly) will absolutely be deleted!

I did today accidentally notice that I could unban him here, but I will not. He does not deserve any help. He has burned too many bridges. It cannot be tolerated anymore.

I'm really not interested in your personal war but here are some tech. stuff:

I just did testing of the new xide driver (5.6.2015) and first I can tell that there's noticable speed-up with /p switch and under my pmode test program (typical value taken from ~5 runs):

UIDE 2014:
Writting average speed: 58.240 MB/s
Reading average speed: 86.281 MB/s

XIDE /p
Writting average speed: 63.825 MB/s
Reading average speed: 95.086 MB/s

I was a bit surprised that my SSD run slower than HDD (with UIDE even a bit slower):
Writting average speed: 68.518 MB/s
Reading average speed: 70.594 MB/s

Anyway all this speeds are great and some % doesn't matter. Unfortunatelly neither XIDE nor UIDE cannot detect HDD on my Evo N620c notebook that runs pooly ~4,5MB/s that would really need some boost.

Then I was testing analog and digital CDDA playback.
Analog playback (I used DOS Navigator built-in CDP) works fine with both UIDE and XIDE.
Digital playback in MPXPlay and grabbing with DAC doesn't work with XIDE - I got only some staccato noise and MPX UI was busy so I had to eject CD to stop it. With UIDE the digital playback works fine on both my drives (TEAC CDW-524EK and NEC ND-4550A). I sent report to Jack.

---
DOS gives me freedom to unlimited HW access.

> I'm really not interested in your personal war

There is no war, not at all. Despite many hours and years of helping him, he directly blames me for a "very UNFRIENDLY post" here on BTTR, where I only said to someone, "I'm not sure if they work at all anymore". And from that alone, he both closed all sources ("worst betrayal") and insulted me rather heavily in private email.

But I totally ignored him, which just makes him angrier. And despite the fact that this was all related to him accidentally being "locked out" from freedos-user (because of the recent SF.net certificate "bug", thanks to him still using NT 4.0 and Opera 7.54), as of today he has found a way to resubscribe (even though he intentionally avoided ALL of my suggested workarounds and insisted he was "giving up").

But the only thing he did (instead of rejoicing that the problem is solved) was flame me many times and blame me for closing sources. All because of one sentence, which he (erroneously) took the wrong way.

> ... I sent report to Jack.

Keep in mind that I was totally sympathetic and always helping him. Did you not know that he personally requested (many times) that I announce for both FreeDOS and here on BTTR? Why do you think I was always mentioning it? It was no accident, it was what he wanted. But he eventually stopped asking, for no good reason, and he said he was "giving up on FleaDOS" (due to uncontrollable "bug" at SF.net), and then even further took one minor quote of mine out of context, all as a ruse just to be rude and go insane.

There is no war, just him being irrationally angry over an "insult" that didn't even exist. (I wasn't even talking to him! How can I insult a guy who isn't even active on the forum? If I wanted to insult him, I could easily email him privately [like he did to me] or just use many of his own words against him. He has insulted everyone he's ever met! He's the perfect example of rude, worthy of condemnation, but he can't even see it! He still thinks I'm the evil one here, but it's all in his head!)

I tried /x option to check digital CDDA playback and on my NTB but any change...

I tested on another machine at work, a bit newer core i5 750 Nehalem with intel 5x PCH chipset set to IDE mode. The controller and drives was recognized correctly:

XIDE, 6-05-2015. 80-MB Cache. CD/DVD name is MSCD000.
IDE0 Controller at I-O address FF90h, Chip I.D. 80863B20h.
IDE1 Controller at I-O address A080h, Chip I.D. 80863B26h.
IDE0 Primary-master disk is ST3500418AS, ATA-100.
CD0: IDE0 Primary-slave, hp DDVDW TS-H653R, ATA-100.
Kernel: allocated 46 Diskbuffers = 24472 Bytes in HMA

but I didn't get any speed up neither with uide nor xide (both /x and /p tried too). In all cases I got ~15MB/s write and 20MB/s read as well as without the driver.

I don't know what I did wrong before but I retested again with FreeDOS 2040 and MS-DOS 7.10 and got better results:

==================================================

MS-DOS 7.10:

no xide, himem.sys
Writting average speed: 49 MB/s
Reading average speed: 78 MB/s

xide, himem.sys
Writting average speed: 67 MB/s
Reading average speed: 105 MB/s

xide /p, himem.sys
Writting average speed: 67 MB/s
Reading average speed: 110 MB/s

no xide, jemmex.exe
Writting average speed: 52 MB/s
Reading average speed: 81 MB/s

xide, jemmex.exe
Writting average speed: 62 MB/s
Reading average speed: 95 MB/s

xide /p, jemmex.exe
Writting average speed: 61 MB/s
Reading average speed: 95 MB/s

==================================================

FreeDOS 2040:

no xide, himmex.exe
Writting average speed: 48 MB/s
Reading average speed: 49 MB/s

xide, himemx.exe
Writting average speed: 61 MB/s
Reading average speed: 102 MB/s

xide /p, himemx.exe
Writting average speed: 61 MB/s
Reading average speed: 103 MB/s

no xide, jemmex.exe
Writting average speed: 41 MB/s
Reading average speed: 51 MB/s

xide, jemmex.exe
Writting average speed: 54 MB/s
Reading average speed: 85 MB/s

xide /p, jemmex.exe
Writting average speed: 60 MB/s
Reading average speed: 89 MB/s

---
DOS gives me freedom to unlimited HW access.

> I don't know what I did wrong before but I retested again with FreeDOS 2040
> and MS-DOS 7.10 and got better results:
>

Great work RayeR. Any reason to test FD Kernel 2040 instead of 2041 or SVN? FAT16 or FAT32? LFN on or off?

Not quite 3x better with MS-DOS 7. Looks like /p switch is negligible. I also wonder how xide compares vs. Smartdrive.

Great work, again!

> I tested on another machine at work, a bit newer core i5 750 Nehalem with
> intel 5x PCH chipset set to IDE mode. The controller and drives was
> recognized correctly:
>
> XIDE, 6-05-2015. 80-MB Cache. CD/DVD name is MSCD000.
> IDE0 Controller at I-O address FF90h, Chip I.D. 80863B20h.
> IDE1 Controller at I-O address A080h, Chip I.D. 80863B26h.
> IDE0 Primary-master disk is ST3500418AS, ATA-100.
> CD0: IDE0 Primary-slave, hp DDVDW TS-H653R, ATA-100.
> Kernel: allocated 46 Diskbuffers = 24472 Bytes in HMA

BTW, how are you grabbing this IO info? DOS? Linux?

> ==================================================
>
> MS-DOS 7.10:
>
> no xide, himem.sys
> Writting average speed: 49 MB/s
> Reading average speed: 78 MB/s
>
> xide, himem.sys
> Writting average speed: 67 MB/s
> Reading average speed: 105 MB/s
>
> xide /p, himem.sys
> Writting average speed: 67 MB/s
> Reading average speed: 110 MB/s

Dang it. I keep thinking of new things. What's the difference in RAM/UMB/HMA usage between the various configs?

> Great work RayeR. Any reason to test FD Kernel 2040 instead of 2041 or
> SVN? FAT16 or FAT32? LFN on or off?

Simply because I have installed it on that machine some years before and FDK is not much developing during recent years so I didn't check for newer ver...
It was on 62GB FAT32 partition with 32k clusters without LFN driver. Also need to note that there was some variations of speeds so I need to multiple run to get some average or typical value. It's rather information only. The conclusion is that FD is a bit slower than MSDOS and that v86 mode (JEMMEX) also make it a bit slower than with himem. But nothing to be afraid of...

---
DOS gives me freedom to unlimited HW access.

Guys, I have much more serious problem with UIDE and XIDE.
It seems it can provocate an incompatibility with programs using Borland's exteders RTM.EXE and DPMI16BI.OVL

If I have 80MB cache everything is OK.
But if I set another value - (bigger: DEVICE=C:\FREEDOS\XIDE.SYS /S160 /P) but suprisingly also lower (DEVICE=C:\FREEDOS\XIDE.SYS /S40 /P) evil things happen.

If I run any program using RTM & DPMI16BI.OVL like BP.EXE from Borland pascal 7.0 the program immediately collapses with error message: "Ignoring (and losing) VCPI page! - beyond table." This message prints on screen about 500 times and then it ends with another message. In some cases: "JemmEx: exception 06 occured at..." and computer freezes or with exception 0E which is better because computer does not freeze.

Some notes:
1) parameter /P makes no change, other parameters I tried also don't change anything. The same issue had also UIDE.SYS

2) It occurs on my both DOS machines: P4 with 512MB RAM and P2 with 192MB RAM.
Both use JemmeX memory extender in config.sys with A20 parameter.
I was investigating this problem quite deeply few months ago and I was not able to find other possible cause in my setup than UIDE/XIDE

---
DOS-u-akbar!

> If I run any program using RTM & DPMI16BI.OVL like BP.EXE from Borland
> pascal 7.0 the program immediately collapses with error message: "Ignoring

I don't have such problem with BP 7.0 it runs fine even I use 256MB cache. What DOS do you use? I guess it would rather something to do with jemmex options and maybe bios bug/incompat.
Here's my relevant lines:
DEVICE=C:\DOS\JEMMEX.EXE A20METHOD:FAST XMSHANDLES=64 SPLIT SB FASTBOOT VERBOSE
rem UIDE driver have to be loaded after SB live! DOS driver, UIDE cache drives
DEVLOAD /H C:\DOS\PLUS\uide.sys /S256 /D:MSCD000 /H
(I use UIDE.SYS from 19.10.2014 and Jemmex 5.78 from 15.7.2012)

---
DOS gives me freedom to unlimited HW access.

> Guys, I have much more serious problem with UIDE and XIDE.
> It seems it can provocate an incompatibility with programs using Borland's
> exteders RTM.EXE and DPMI16BI.OVL
>
[SNIP]
> Both use JemmeX memory extender in config.sys with A20 parameter.
> I was investigating this problem quite deeply few months ago and I was not
> able to find other possible cause in my setup than UIDE/XIDE

Yeah...take that up with Jack if you haven't already.

I use FreeDOS with kernel 2041.
I am at work now - when I come home I bring here memory info from JEMMEX call from commandline.

¨
> > If I run any program using RTM & DPMI16BI.OVL like BP.EXE from Borland
> > pascal 7.0 the program immediately collapses with error message:
> "Ignoring
>
> I don't have such problem with BP 7.0 it runs fine even I use 256MB cache.
> What DOS do you use? I guess it would rather something to do with jemmex
> options and maybe bios bug/incompat.
> Here's my relevant lines:
> DEVICE=C:\DOS\JEMMEX.EXE A20METHOD:FAST XMSHANDLES=64 SPLIT SB FASTBOOT
> VERBOSE
> rem UIDE driver have to be loaded after SB live! DOS driver, UIDE cache
> drives
> DEVLOAD /H C:\DOS\PLUS\uide.sys /S256 /D:MSCD000 /H
> (I use UIDE.SYS from 19.10.2014 and Jemmex 5.78 from 15.7.2012)

---
DOS-u-akbar!

I solved my problem by changing parameters of DEVICE=JEMMEX.EXE line in config.sys.

I partially fixed the issue by adding parameter MAX=32000 into this line and hopefully finaly fixed by adding also parameter MIN=32000

So now the whole line looks like this:
DEVICE=C:\FREEDOS\BIN\JEMMEX.EXE ALTBOOT MAX=32000 MIN=32000

Now I can any cache size I want.

BTW: Jack started new branch of his drivers. The XIDE is abandoned and 17.6. he released new drivers XHDD and XDVD2.

---
DOS-u-akbar!

> So now the whole line looks like this:
> DEVICE=C:\FREEDOS\BIN\JEMMEX.EXE ALTBOOT MAX=32000 MIN=32000

What it exactly do? It's some kind of VCPI mem limitation, isn't it same as NOVCPI? Or maybe limit it to 32MB. I'm not sure now if BP requires VCPI.

> BTW: Jack started new branch of his drivers. The XIDE is abandoned and
> 17.6. he released new drivers XHDD and XDVD2.

I guess it does the same as xide with disabled cd/dvd support + older udvd2 that was not affected by CDDA digital playback issue.

---
DOS gives me freedom to unlimited HW access.

> I did today accidentally notice that I could unban him here, but I will
> not. He does not deserve any help. He has burned too many bridges.
> It cannot be tolerated anymore.

Hi to all ... I've been recently dead and have not followed any "rants" ... please (whoever is involved is this "war") make peace, some independent admin delete or censor the rude posts if any such still exist.

PS: there exist only 10 types of people: those who know binary and those who don't.

PPSS: there are only 100 DOS'sers around, if anyone starts shooting them away with a six-shooter revolver, then the community will be empty soon, and only 10 cartridges will remain.

---
This is a LOGITECH mouse driver, but some software expect here
the following string:*** This is Copyright 1983 Microsoft ***

When I "solved" my problem with XHDD+JemmEX+borland compatility i later found that I broke the compatilibity with SoundBlaster emulation driver.

When I started SBEINIT.COM file to setup my SB Live as legacy soundblaster I was still getting an error message, something like: "Not loaded, could not allocate memory region bellow 4MB"

So I tried to find configuration to join XHDD+JemmEX+borland+SBEINIT.COM

It looked bad but then I found that I am not using original Borland RTM extender but some unofficial "improved" by Mystic bytes software. I completely forgot it!

When I tried to use the original RTM everything worked like a charm and then was trivial to setup the CONFIG.SYS file.

So this it the reason why you was not able to reproduce my problems. If you want to use the unofficial RTM by Mystic bytes software I uploaded it here: http://www.laaca.borec.cz/newrtm.zip

---
DOS-u-akbar!

> When I started SBEINIT.COM file to setup my SB Live as legacy soundblaster
> I was still getting an error message, something like: "Not loaded, could
> not allocate memory region bellow 4MB"

Do you use SB option in JEMMEX?
SB - SoundBlaster driver compatibility mode
http://help.fdos.org/en/hhstndrd/base/jemmex.htm
As you can see in my config line
DEVICE=C:\DOS\JEMMEX.EXE A20METHOD:FAST XMSHANDLES=64 SPLIT SB FASTBOOT VERBOSE
but currently I don't use this sbemu drivers anymore...

---
DOS gives me freedom to unlimited HW access.

> Do you use SB option in JEMMEX?
> SB - SoundBlaster driver compatibility mode

No, I don't. I don't know either what it exactly doea.
As I remember in the past when I created my config.sys and autoexec.bat files it didn't solve the problem with filling the first 4MB space.

To protect the first 4MB I use a resizeable ramdisk which can be called in config.sys and in command line too.

In config.sys I create 4MB ramdisk, then I load all my device drivers and on end of autoexec.bat I resize the ramsisk to 0. Unfortunetely the disk letter does not disappear from system so then I make JOIN E: C:\freedos\join_e
This workaround works perfectly for me.

---
DOS-u-akbar!

If I remember the SB (or SPLIT?) option do this job for you - reserving some low XMS that driver needs so I didn't need to do this ramdisk tricks. Also I can see that I load UIDE/XIDE in autoexec after SB driver but it could be possible to use /R15 or /R63 option to UIDE that reserve low XMS too.

---
DOS gives me freedom to unlimited HW access.