Alternate Kernel Fails to Start (WinXP)

NTDEV
1

Hi All
(this is my first post to NTDEV, please forgive any faux pas I may commit in asking for help)

I have been searching for a solution many weeks, and was referred to this forum from the stackoverflow forum. With great respect, I turn to you members asking something you would never consider under “normal conditions”.

Q: Is it possible to start Windows XP using the NT 3.1 kernel ?

Without knowing the context, this could easily be dismissed as a foolish, misdirected effort (it sounds mad to me), but please let me explain WHY.

The goal is to optimize the performance of a particular open-source music player, originally coded for WinXP. We are driven by the love of Music - and it’s high-quality playback. “We” are a small group of very determined DIY “audiophiles”, who’ve found extreme removal of unnecessary OS components radically improves the sound of the player. We’ve stripped XP down to a single purpose: running 2 programs - a shell and the music file player. All other functionality is gone.

Recently the idea of using a simplified kernel came up, and one of the team suggested “/kernel=” in the boot.ini file would do the job. This has not worked out for us. None of the group are developers, so the implications and technical requirements to achieve this are known only by trial-and-error. We do a lot of internet searching, but the idea (ie: running an antiquated kernel in support of a gutted, later-model OS) is nowhere to be found.

(sorry for the rambling description, but I think it’s integral to understanding the Why)

If somebody can tell me definitively “That Won’t Work !”, then it’s a valuable answer I can take away (and get on to other aspects of the project). We’ve checked the basics over and over (got the syntax and file locations correct), yet when asked to boot using the alternate kernel (christened “Neptune” for reasons unknown to me), the system halts with the (familiar) message:

“Windows could not start because the following file is missing or corrupt:
\system32\ntoskrnl.exe
Please re-install a copy of the above file.”

Of course it’s right where it belongs in system32, so I obviously don’t know what is in control at the time this goes down. Rebooting the same partition and specifying the original XP-Kernel on the boot menu loads fine, and the player functions normally.

If there is some reference I can be directed toward, I’d be so grateful. I’m also prepared to give more detail on the test conditions, but prefer not to fill up the forum with unsolicited info.

Thank you all for reading this far, I hope somebody can make sense of our situation and guide us accordingly.

Cheers,
Grant

ps: Yes, the result is worth the trouble and effort - it is a very special thing we’ve built for ourselves. The software is detailed at -> http://cicsmemoryplayer.com/

2

Negative. This would not likely work with any version mixing and matching,
but as 3.1 predates the move of some subsystem stuff to win32k, this will
never work.

Good luck,

Mm
On Apr 26, 2012 12:24 AM, wrote:

> Hi All
> (this is my first post to NTDEV, please forgive any faux pas I may commit
> in asking for help)
>
> I have been searching for a solution many weeks, and was referred to this
> forum from the stackoverflow forum. With great respect, I turn to you
> members asking something you would never consider under “normal conditions”.
>
> Q: Is it possible to start Windows XP using the NT 3.1 kernel ?
>
> Without knowing the context, this could easily be dismissed as a foolish,
> misdirected effort (it sounds mad to me), but please let me explain WHY.
>
> The goal is to optimize the performance of a particular open-source music
> player, originally coded for WinXP. We are driven by the love of Music -
> and it’s high-quality playback. “We” are a small group of very determined
> DIY “audiophiles”, who’ve found extreme removal of unnecessary OS
> components radically improves the sound of the player. We’ve stripped XP
> down to a single purpose: running 2 programs - a shell and the music file
> player. All other functionality is gone.
>
> Recently the idea of using a simplified kernel came up, and one of the
> team suggested “/kernel=” in the boot.ini file would do the job. This has
> not worked out for us. None of the group are developers, so the
> implications and technical requirements to achieve this are known only by
> trial-and-error. We do a lot of internet searching, but the idea (ie:
> running an antiquated kernel in support of a gutted, later-model OS) is
> nowhere to be found.
>
> (sorry for the rambling description, but I think it’s integral to
> understanding the Why)
>
> If somebody can tell me definitively “That Won’t Work !”, then it’s a
> valuable answer I can take away (and get on to other aspects of the
> project). We’ve checked the basics over and over (got the syntax and file
> locations correct), yet when asked to boot using the alternate kernel
> (christened “Neptune” for reasons unknown to me), the system halts with the
> (familiar) message:
>
> “Windows could not start because the following file is missing or corrupt:
> \system32\ntoskrnl.exe
> Please re-install a copy of the above file.”
>
> Of course it’s right where it belongs in system32, so I obviously don’t
> know what is in control at the time this goes down. Rebooting the same
> partition and specifying the original XP-Kernel on the boot menu loads
> fine, and the player functions normally.
>
> If there is some reference I can be directed toward, I’d be so grateful.
> I’m also prepared to give more detail on the test conditions, but prefer
> not to fill up the forum with unsolicited info.
>
> Thank you all for reading this far, I hope somebody can make sense of our
> situation and guide us accordingly.
>
> Cheers,
> Grant
>
> ps: Yes, the result is worth the trouble and effort - it is a very special
> thing we’ve built for ourselves. The software is detailed at ->
> http://cicsmemoryplayer.com/
>
>
>
> —
> NTDEV is sponsored by OSR
>
> For our schedule of WDF, WDM, debugging and other seminars visit:
> http://www.osr.com/seminars
>
> To unsubscribe, visit the List Server section of OSR Online at
> http://www.osronline.com/page.cfm?name=ListServer
>

3

That Won’t Work !

The kernel itself is the os. The dlls in the os that run inuser mode depend on it, sometimes at a binary layout level. You can’t boot a 3.1 kernel into XP, just like you can’t use a Linux kernel to boot mac OS X. None of the XP drivers could load, let alone boot.

d

debt from my phone

From: xxxxx@hotmail.com
Sent: 4/26/2012 12:24 AM
To: Windows System Software Devs Interest List
Subject: [ntdev] Alternate Kernel Fails to Start (WinXP)

Hi All
(this is my first post to NTDEV, please forgive any faux pas I may commit in asking for help)

I have been searching for a solution many weeks, and was referred to this forum from the stackoverflow forum. With great respect, I turn to you members asking something you would never consider under “normal conditions”.

Q: Is it possible to start Windows XP using the NT 3.1 kernel ?

Without knowing the context, this could easily be dismissed as a foolish, misdirected effort (it sounds mad to me), but please let me explain WHY.

The goal is to optimize the performance of a particular open-source music player, originally coded for WinXP. We are driven by the love of Music - and it’s high-quality playback. “We” are a small group of very determined DIY “audiophiles”, who’ve found extreme removal of unnecessary OS components radically improves the sound of the player. We’ve stripped XP down to a single purpose: running 2 programs - a shell and the music file player. All other functionality is gone.

Recently the idea of using a simplified kernel came up, and one of the team suggested “/kernel=” in the boot.ini file would do the job. This has not worked out for us. None of the group are developers, so the implications and technical requirements to achieve this are known only by trial-and-error. We do a lot of internet searching, but the idea (ie: running an antiquated kernel in support of a gutted, later-model OS) is nowhere to be found.

(sorry for the rambling description, but I think it’s integral to understanding the Why)

If somebody can tell me definitively “That Won’t Work !”, then it’s a valuable answer I can take away (and get on to other aspects of the project). We’ve checked the basics over and over (got the syntax and file locations correct), yet when asked to boot using the alternate kernel (christened “Neptune” for reasons unknown to me), the system halts with the (familiar) message:

“Windows could not start because the following file is missing or corrupt:
\system32\ntoskrnl.exe
Please re-install a copy of the above file.”

Of course it’s right where it belongs in system32, so I obviously don’t know what is in control at the time this goes down. Rebooting the same partition and specifying the original XP-Kernel on the boot menu loads fine, and the player functions normally.

If there is some reference I can be directed toward, I’d be so grateful. I’m also prepared to give more detail on the test conditions, but prefer not to fill up the forum with unsolicited info.

Thank you all for reading this far, I hope somebody can make sense of our situation and guide us accordingly.

Cheers,
Grant

ps: Yes, the result is worth the trouble and effort - it is a very special thing we’ve built for ourselves. The software is detailed at -> http://cicsmemoryplayer.com/

—
NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at http://www.osronline.com/page.cfm?name=ListServer

4

> Without knowing the context, this could easily be dismissed as a foolish, misdirected effort (it sounds

mad to me), but please let me explain WHY.

Lots of pain for no gain.

very determined DIY “audiophiles”, who’ve found extreme removal of unnecessary OS components
radically improves the sound of the player.

Do you have an idea to remove “unnecessary” parts of your car so it will consume lesser fuel? :slight_smile:

Yes, there are people doing this, but then all car vendor’s bets are off and the car will run by mere luck.

The usual way for media players is to a) use much more RAM b) disable the devices which are the interrupt sources. If you do not need network - disable all network adapters in Device Manager, disable Intel’s onboard audio if you have better professional audio hardware, disable all USB hosts, probably switch from onboard to better (like Promise) SATA controller, use SSDs and so on.

Booting XP with the older kernel will not work, since user-mode essentials of the OS rely on the new kernel stuff. Can you put the 30-years old carbutator pushrod engine to your car in a hope it will consume lesser fuel, or will have better dynamics? no, you can’t.

Surely there are shops which will do the major remake of your car to fit non-standard engine and gearbox there, but sorry, Windows is not suitable for such remakes. Consider Linux which is open source.

–
Maxim S. Shatskih
Windows DDK MVP
xxxxx@storagecraft.com
http://www.storagecraft.com

5

>

The goal is to optimize the performance of a particular open-source music
player, originally coded for WinXP. We are driven by the love of Music - and
it’s high-quality playback. “We” are a small group of very determined DIY
“audiophiles”, who’ve found extreme removal of unnecessary OS
components radically improves the sound of the player. We’ve stripped XP
down to a single purpose: running 2 programs - a shell and the music file
player. All other functionality is gone.

As others have mentioned you can’t do what you are asking.

If it’s Open Source and runs on a minimal system (eg no GUI) then are there any major blockers to porting the thing to Linux which you can configure in almost any way you can imagine? The drawback that you need to support it yourself, but you’re already at that point wanting to muck around with NT kernels.

I guess the other thing you might encounter is a lack of (quality) driver support for your preferred hardware under Linux…

James

6

This would seem like the best idea to me too.

mm

-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of James Harper
Sent: Thursday, April 26, 2012 2:32 AM
To: Windows System Software Devs Interest List
Subject: RE: [ntdev] Alternate Kernel Fails to Start (WinXP)

The goal is to optimize the performance of a particular open-source
music player, originally coded for WinXP. We are driven by the love of
Music - and it’s high-quality playback. “We” are a small group of very
determined DIY “audiophiles”, who’ve found extreme removal of
unnecessary OS components radically improves the sound of the player.
We’ve stripped XP down to a single purpose: running 2 programs - a
shell and the music file player. All other functionality is gone.

As others have mentioned you can’t do what you are asking.

If it’s Open Source and runs on a minimal system (eg no GUI) then are there
any major blockers to porting the thing to Linux which you can configure in
almost any way you can imagine? The drawback that you need to support it
yourself, but you’re already at that point wanting to muck around with NT
kernels.

I guess the other thing you might encounter is a lack of (quality) driver
support for your preferred hardware under Linux…

James

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer

7

Also, you seem like you enjoy this sort of hacking. If so, you’ll probably
enjoy Linux hacking as well.

mm
-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of James Harper
Sent: Thursday, April 26, 2012 2:32 AM
To: Windows System Software Devs Interest List
Subject: RE: [ntdev] Alternate Kernel Fails to Start (WinXP)

The goal is to optimize the performance of a particular open-source
music player, originally coded for WinXP. We are driven by the love of
Music - and it’s high-quality playback. “We” are a small group of very
determined DIY “audiophiles”, who’ve found extreme removal of
unnecessary OS components radically improves the sound of the player.
We’ve stripped XP down to a single purpose: running 2 programs - a
shell and the music file player. All other functionality is gone.

As others have mentioned you can’t do what you are asking.

If it’s Open Source and runs on a minimal system (eg no GUI) then are there
any major blockers to porting the thing to Linux which you can configure in
almost any way you can imagine? The drawback that you need to support it
yourself, but you’re already at that point wanting to muck around with NT
kernels.

I guess the other thing you might encounter is a lack of (quality) driver
support for your preferred hardware under Linux…

James

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer

8

Another point is that best possible audio performance (which I assume the
audiophile seeks) requires a realtime operating system and/or dedicated
hardware. All that is required is to deliver PCM audio (or DSD if you
prefer) to the chosen DA converter without data corruption and with high
precision timing. I don’t think that any Windows system, no matter how
stripped down, could guarantee this. Perhaps the OP should transcode all the
source files to PCM and play them on suitable dedicated hardware. Mike.

>>>>>>>>>----- Original Message -----
From: Martin O’Brien
To: Windows System Software Devs Interest List
Sent: Thursday, April 26, 2012 10:36 AM
Subject: RE: [ntdev] Alternate Kernel Fails to Start (WinXP)

Also, you seem like you enjoy this sort of hacking. If so, you’ll probably
enjoy Linux hacking as well.

mm
-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of James Harper
Sent: Thursday, April 26, 2012 2:32 AM
To: Windows System Software Devs Interest List
Subject: RE: [ntdev] Alternate Kernel Fails to Start (WinXP)

The goal is to optimize the performance of a particular open-source
music player, originally coded for WinXP. We are driven by the love of
Music - and it’s high-quality playback. “We” are a small group of very
determined DIY “audiophiles”, who’ve found extreme removal of
unnecessary OS components radically improves the sound of the player.
We’ve stripped XP down to a single purpose: running 2 programs - a
shell and the music file player. All other functionality is gone.

As others have mentioned you can’t do what you are asking.

If it’s Open Source and runs on a minimal system (eg no GUI) then are there
any major blockers to porting the thing to Linux which you can configure in
almost any way you can imagine? The drawback that you need to support it
yourself, but you’re already at that point wanting to muck around with NT
kernels.

I guess the other thing you might encounter is a lack of (quality) driver
support for your preferred hardware under Linux…

James

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer

9

Well, everything depends on how far the OP wants to go…

The most obvious way to go is to build a specialized RT distro with his target app running as a kernel module that, apart from other things, takes over Linux scheduling and interrupt handling the way RTLinux does in order to achieve real-time performance - a “default” scheduling and interrupt handling that are provided by
plain vanilla Linux kernel are obviously inadequate for the true RT performance that the OP, apparently,
needs, regardless of configuration options used to build the kernel

Anton Bassov

10

I think you’re barking on the wrong moon.

If you want more responsiveness (less latencies) you need to carefully choose thread priorities in your application, amount of buffering, and also make sure your file reading functions don’t cause the file cache to get bloated at a cost of code page discard. Windows XP is *notorously* *bad* in that regard. Use FILE_FLAG_NO_BUFFERING when reading large files.

You may want to make sure you have SP3 installed on your XP. XP Gold has extremely bad USB stack which could spend unbelivably long times in DPC, like a constipated granpa.

11

OK, I checked your player page.

I’m afraid you’re addressing the problem from the wrong side. As it’s often happens with audiophiles, you’re chasing your phantom problems. As you rejects double blind tests (“Procedures such as blind testing and focus groups are of little help”), a reasonable person would stop right there and ask you: how you can be sure you’re actually hearing differences?

A proper test setup would make sure the output levels are matched within at most 1% (about 0.1 dB); there is no clipping, and the frequency responses of different DACs are linear within <1dB (at least matched). It’s been reported that little differences (as little as 1dB) in the output level often create an appearance of different coloration of the sound. This is just artifact of human hearing. This is why it’s important to match output levels for testing.

Jitter is just group delay noise. You want to know how much such noise you can notice? Create a test file, and introduce artificial group delay noise to it. You can do it by using a sliding filter with fractional delay, similar to the technique used for resampling. Also, you can measure actual jitter by using a white noise test sample, a quality ADC and then applying some not-too-difficult math to the result.

You want quality output? Don’t use the onboard DAC. But for what it’s worth, I’ve measured pretty good result for some of my computers’ onboard sound chips.

There is no voodoo or Force, young padawan. Even though you might find my lack of faith disturbing, use instruments for measurement, and run double blind tests to find out what makes difference.

12

Small correction:

For jitter measurement a pink noise sample at about 1/2 of the band (for example, at 11 kHz) would be more appropriate. That would simplify the math.

13

Hello Gentlemen…

I appreciate the rapid response. Thank you all, for your clarification and comments.

@Martin, @Doron, @Maxim:
Thank you for pointing out why 3.1 CAN’T work in this context (ie: subsystem stuff moved to win32k, user-mode essentials).

We were hoping (…the Hope of the Uninformed) that there is enough commonality between kernels, such that the very few things still running would be able to function. I presume not, but is it possible an intermediate version (Win2K ?) meets this need ?

OK, if I may, let me comment on some on the points made in the replies…

“you enjoy this sort of hacking”:
Indeed, HACKING is the only word for it - there is not a single programmer/coder among us (well, I did do CICS, COBOL, VSAM etc. back in the Dark Ages, but nothing that is “PC programming”), so quite literally we have reduced our XP footprint through painful trial-and-error. The miracle is - IT WORKS - and gloriously so.

Why Windows (or, Why Not Linux ?):
The software is written for XP, and we are not coders (…porting to Linux cannot be done by THIS volunteer army). Describing the software as “Open Source” may be a bit misleading… it was written by one individual and offered Free-To-All through the Audio Asylum PC Audio forum ( http://www.audioasylum.com/cgi/vt.mpl?f=pcaudio&m=31286 ).

The last update to the player was 2010-Oct-10 (cicsPlay.exe, v2-b39), and the author has not been heard from since 2011-Oct-10. The website is under his control, and has not been updated with any of our Slimming efforts. This is the end of cPlay’s development, and it is only by chance (and mad tinkering) that it’s performance has been further improved through shrinking the OS.

BTW, though the original forum post declares “…Source code (4.1MB) is available via email.”, requests for such are unanswered, and there are no known copies available to us. We’re stuck with the software as-is.

Another Linux blocker is the requirement for an ASIO driver for our preferred soundcard.

Hardware Optimization:
There is a suggested hardware complement from which to get maximum results with the cMP software. The author also suggests Services and Devices that can be eliminated from a vanilla XP install, in an effort to reduce activity at both the motherboard & OS level. We under-clock and under-volt the system to promote a calmer EMI/RFI environment. Much of what Maxim and Alex suggest is addressed in this recipe.

The ESI Juli@ PCI soundcard has been proven a solid, affordable performer, and the choice of most who build a dedicated machine.

A subset of the community have gone to great lengths in providing clean power to their setups (ie: DIY linear supplies and/or batteries).

@Alex:
Thank you for taking time to look into the cMP/cPlay documentation… and for your considered response (I sense you’re quite familiar with issues in high-end audio). I agree that measurements are important as a basis for understanding the road to improvement. Why do you say “I’m afraid you’re addressing the problem from the wrong side.” ?

Our experience has been “Less Leaves More”. The more we remove from the OS, the more detail we hear from the same old familiar music files. There’s no doubt the changes manifest at the speakers. Perhaps it IS simply decreased jitter. The root cause may be unknown, but the direction is clear. We want “more of that, please”.

Thank you again to everyone for their input. I appreciate the depth of knowledge represented here in the NTDEV List.

Cheers,
Grant

BONUS: A sidebar on Jitter:
My converter (since 2003) is the Benchmark DAC1 - a high quality, revealing unit - which today is one of the common yardsticks referenced in equipment reviews. The designers at Benchmark are truly “engineers”, holding that the whole truth is expressed in the very detailed and impressive measurements they publish. The ad copy proclaims “Benchmark’s proprietary UltraLock clock system achieves an unprecedented immunity to jitter.” Immunity, mind you. Their position: if the DAC1 can lock to the signal at all (ie: tell the difference between positive- and negative-going voltages), it will extract the data and the audio performance CANNOT be improved or degraded by the effects of jitter.

Bollocks of course… else my CD, DVD and computer sources would all sound the same when playing the same disc via the DAC1. They don’t. How could I possibly detect something as ephemeral as less “software-induced jitter” through this unit if it is “immune” to the artifact. The DAC1 never fails to report a change made upstream - unless I have made an error.

My good friends at Benchmark are Blinded by their Sliderules… but I love them for the exceptional product they make, which apparently is far more capable than they know !

14

“The more we remove from the OS, the
more detail we hear from the same old familiar music files.” Me too, I tend to notice more and more details as I listen to the same file over and over again. In the same setup.

OK, I’m afraid again that source-synchronized DACs are solving a problem that should not exist in the first place. And jitter is just caused by them trying to solve that. An USB-connected DAC with some nominal amount of internal buffer and its own clock generator makes this problem non-existent.

Anyway, get a good ADC (you can have a USB-connected one supporting 24/96000 for under $100) and run measurements.

And DON’T TRUST YOUR EARS. DON’T TRUST YOUR MEMORY. DO DOUBLE BLIND TESTS. Make two boot disks, with stripped-down and with regular OS. Make that stripped down to different grade. That could be different boot partitions, so you don’t have to get too many disks. For example, have two disks with partitions of different grade. Make sure you run the test while not knowing what is booted. Have an aide select that for you.

Good luck. Don’t be disappointed when you find out the truth.

15

Have you looked at Reactos? It is something you may tweak.
– pa

16

> A subset of the community have gone to great lengths in providing clean power to their setups (ie:

DIY linear supplies and/or batteries).

Do you remember the buck converters which power the CPU? they can introduce audible high frequency noise.

The hard disk can introduce it too.

–
Maxim S. Shatskih
Windows DDK MVP
xxxxx@storagecraft.com
http://www.storagecraft.com

17

>

> A subset of the community have gone to great lengths in providing clean
power to their setups (ie:
>DIY linear supplies and/or batteries).

Do you remember the buck converters which power the CPU? they can
introduce audible high frequency noise.

The hard disk can introduce it too.

Sounds like the best plan would be completely custom hardware (eg FPGA based) in a heavily shielded box driven by a PC and connected via fibre :slight_smile:

James

18

>Do you remember the buck converters which power the CPU? they can introduce
audible high frequency noise.

The hard disk can introduce it too.

The irony of all that is that with a computer-based player, I have to go to great lengths to mask the mechanical noises (disks and fans), before I can hear any circuit noise. Unless it’s a second tier brand motherboard with onboard audio, that lets me hear mouse cursor movements and screen redraws. I’ve had one of those.

19

Note that if you are living in the present instead of the past, Vista has
a new feature to support very high thread priorities for ordinary apps.
Look at the Multimedia Class Scheduling Service (MMCSS).
joe

I think you’re barking on the wrong moon.

If you want more responsiveness (less latencies) you need to carefully
choose thread priorities in your application, amount of buffering, and
also make sure your file reading functions don’t cause the file cache to
get bloated at a cost of code page discard. Windows XP is *notorously*
*bad* in that regard. Use FILE_FLAG_NO_BUFFERING when reading large files.

You may want to make sure you have SP3 installed on your XP. XP Gold has
extremely bad USB stack which could spend unbelivably long times in DPC,
like a constipated granpa.

NTDEV is sponsored by OSR

For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars

To unsubscribe, visit the List Server section of OSR Online at
http://www.osronline.com/page.cfm?name=ListServer

20

Arrrgh… it’s so hard to keep this one on topic :wink:

@Pavel
Thank you for pointing out ReactOS. I just recently became aware of the project - actually a month ago we began using their drmk.sys and winspool.drv modules (smaller = better), and last week managed to remove drmk.sys completely (by substituting portcls.sys from Win2K).

I’ve not tried installing ReactOS yet. Do you know if they can support ASIO drivers ? (No ASIO = No Juli@ = No cMP system)

@Maxim
I am hardly Leader of The Pack in P/S matters. Presently I use a slightly modified Antec Earthwatts EA 430, connected only to P4 & P24 of the motherboard. SSD (OS partitions and “local” music files) and HDD (music library) are externally powered by simple 5V/12V power bricks, providing a modicum of isolation.

@James
Ha-HA ! Even better plan would be to have funding available to attempt such a setup ! Good one.

@Alex
“Me too, I tend to notice more and more details as I listen to the same file over and over again. In the same setup.”
Excellent… then you too are an attentive listener. I have this experience quite often, for I will focus on a particular aspect/instrument in a song, and discover some new relationship I’d never noticed before.

Perhaps I should not have started discussing DACS, as they are downstream from the process I’m trying to improve (ie: the creation of an S/PDIF bitstream).

“And DON’T TRUST YOUR EARS. DON’T TRUST YOUR MEMORY. DO DOUBLE BLIND TESTS.”
I don’t quite know how to respond to this advice. It won’t help me next time I mix a record, where SOUND -> EARS -> PERCEPTION -> JUDGEMENT is the only way to arrive at the right blend of all constituent elements.

I’m SURE I hear sounds differently from those people not engaged in a life-long obsession with music and the technology of sound, and I’m SURE I hear “better sound” as I reduce the complexity of my OS. The trend is real and unmistakable.

Perhaps we shouldn’t bore list readers with this tired old debate (we’re both right, aren’t we ?)

@Joe
Thanks for mentioning Vista & MMCSS. The reason most of us are “living in the past” with XP is, that’s the environment it was written for. The change log repeatedly mentions low level optimizations, memory handling and such. I don’t know how much of this would break on later platforms. Over the years a few individuals have tried running on Vista and Win 7… with mixed results.

I don’t know the details, but I’ll bet XP is easier to deconstruct than any recent OS. In fact calling ours XP is probably misleading - because forum members “know” what XP is. Our present slimmed OS is more The Creature That Came From XP (cue B-movie soundtrack !).

With a “Windows” folder weighing in at 16 MB, and teeny, tiny registry hives [default - 8 kb/ sam - 8 kb/ security - 8 kb/ software - 12 kb/ system - 28 kb] it’s an XP in name only. If only there was a mini-kernel to get our limited functionality off the ground…

Wish us Luck, and thanks for reading.
Grant

ps: I Double Dog Dare you to try a software-only implementation, to get a taste of what this fabulous player is capable of doing. FAT32 partition, Windows XP Home-SP1 (and forgo USB devices), ASIO soundcard (or in a pinch ASIO4ALL > your audio device) and a competent Hi-Fi.