Two Questions About Diyba 128GB PAE Patch

[dencorso]

It seems, but I'm still far from sure of it and have no personal experience with the case, that *some* Z390 (Coffee Lake-S and maybe -X) BIOSes map most of the RAM on those boards above 4 GiB, thus exposing just something under 1 GiB for XP, even when populated with 8 or more GiB of RAM. If that turns out to be a fact, it might be an additional reason for using the PAE patch, albeit I would prefer to forgo such boards completely, instead.

[jaclaz]

3 hours ago, caliber said:

@jaclaz Which method are you using ?

NONE.

Read latest dencorso's post to understand why (even if personally I run XP SP2, and - just in case - yes, I know, thanks for telling me that SP3 and unofficial SP4 are available )

3 hours ago, caliber said:

once files are patched do I need to copy paste them in the system32 folder ?

I doubt it will work, unless you have Windows File Protection Service disabled, read the instructions:

https://www.win-raid.com/t4035f45-Windows-XP-Bit-and-Server-Bit-on-Modern-Hardware-242.html#msg96981

jaclaz

[win32]

Creative Labs hardware, as well as Intel graphics drivers and some ATI/AMD graphics are known to be incompatible with PAE with RAM > 4 GB on Windows XP.

But most application software* (and all 16bit stuff in my experience) runs perfectly fine. Multiprocess applications will certainly benefit.

*Video Painter component of Ulead Media Studio Pro 7.3 notwithstanding; a simple rendering operation hogged 2.3 GB before crashing. will try to replicate without PAE. this program dates back to 2003, when most people had 128-512 MB of RAM.

I struggle to find myself using more than 2 GB at any given time though.

However, I don't use NT 5.1 (XP) for these pursuits. Windows 2000 Professional with Extended Kernel + Core also unlocks up to 64 GB of RAM and runs 90% of everything XP runs. The PAE implementation is stable thanks to its direct lineage with its Server counterparts. And it is not as unobtainable as Server 2003 Enterprise Edition.

Edited November 22, 2019 by win32
reliable reports that Intel/AMD graphics work on XP with PAE

[daniel_k]

@caliber and all

I've already posted how to test the patched files without messing with original files:

- Run the WinXPPAE.exe with proper command line switches
- Rename ntkrpamp.exe to ntkrpae.exe
- Rename halmacpi.dll to halpae.dll
- Copy both files to \Windows\System32 folder.

Add this to the end of the [operating systems] section of boot.ini:
multi(0)disk(0)rdisk(0)partition(1)\WINDOWS="Windows XP (PAE Test)" /noexecute=optin /fastdetect /HAL=halpae.dll /KERNEL=ntkrpae.exe

The multi(0)disk(0)rdisk(0)partition(1)\WINDOWS string must match the already existing entry.

Restart and select that boot option.

Edited November 22, 2019 by daniel_k

[Dibya]

Well Win32 my patch never caused any issues with ATI Radeon and Intel HD graphics even for your surprise hibernation works with my patch with so called buggy ATI Radeon .

How it works then ?

Let me explain . XP , 2k and 2k3 all handles pae differently .

XP : keeps double DMA buffer enabled always when 64bit ramdisk is active

2k : Enables dma buffer  when certain kernel calls are called by your driver (on those days high end components only used those API but that not lies today since nt is a major platform now unlike 9x in those days.

2k3 sp2 : has special hal mechanism that detects defective driver and enable DMA buffer when the driver support in separate memory space .

 My new patch will port fullmechanism from win 2k3 unlike my old patch which ported some to prevent just few drivers which troubled me and my user . I am using my new patch for 4months . It is so far God like stable .. I will analyse new patch by Daniel since it enables pae without pae switch

[daniel_k]

@Dibya

This NOT a "best patch" contest.

Do yourself a favor, grab the debug symbols for both XP and Server 2003, find the PAE related code, analyze it, pay attention to the variables.
Plus, all known PAE patches were missing a license check in Kernel.

There is just ONE way to patch HAL and Kernel files, plain and simple.

The most time consuming part of patching the HAL is to find "free space" to add NEW code, ADD relocations and so on.
If you are expecting your "new" patch to work without doing all of this, I can assure you that it WON'T work at all.

I'm done with this discussion.

Edited November 22, 2019 by daniel_k

[caliber]

1 hour ago, daniel_k said:

- Run the WinXPPAE.exe with proper command line switches
- Rename ntkrpamp.exe to ntkrpae.exe
- Rename halmacpi.dll to halpae.dll
- Copy both files to \Windows\System32 folder.

Add this to the end of the [operating systems] section of boot.ini:
multi(0)disk(0)rdisk(0)partition(1)\WINDOWS="Windows XP (PAE Test)" /noexecute=optin /fastdetect /HAL=halpae.dll /KERNEL=ntkrpae.exe

The multi(0)disk(0)rdisk(0)partition(1)\WINDOWS string must match the already existing entry.

Restart and select that boot option.

well... all this was not described this way at the win-raid forum.

https://www.win-raid.com/t4035f45-Windows-XP-Bit-and-Server-Bit-on-Modern-Hardware-242.html#msg96981

this is why I was asking for help.

I had to extract those files from the XP ISO cab folder because they were not available in the system32 folder or just had different sizes

I have found the ntkrpamp file in the windows\driver cahe\i386 and the halmacpi was missing.

the russian patch was tested on another mainboard and it works just fine however it doesn't run on my current computer, hence I wanted to test your patch.

I'm glad to hear @Dibya will be making an executable patch so we don't have to mess with individual files anymore.

the russian patch is simply perfect, anybody with no programming skills is able to install or uninstll it with a single click

 

 

 

 

 

Edited November 22, 2019 by dencorso
flame removed

[daniel_k]

1 hour ago, caliber said:

well... all this was not described this way at the win-raid forum.

https://www.win-raid.com/t4035f45-Windows-XP-Bit-and-Server-Bit-on-Modern-Hardware-242.html#msg96981

this is why I was asking for help.

That's because the patch is intended for advanced users who are used to work with modified system files and understand the possible stability issues that it may cause.

Edited November 22, 2019 by daniel_k

[genieautravail]

I'm using the russian patch 4Gb-Fix (Up to 128Gb) (v. 0.0.0.42 RC2) since one year on one of my main computers (a Zotac zbox id18).

I'm writting this post from it.

The patch is pretty stable but from time to time (several weeks between) I get a BSOD. Especially when I run a browser based on Chromium. 

The CHKDSK command doesn't works with at least one external HDD (depend of the model of controller used in the external case). 

[dencorso]

Just now, genieautravail said:

The patch is pretty stable but from time to time (several weeks between) I get a BSOD. Especially when I run a browser based on Chromium. 
The CHKDSK command doesn't works with at least one external HDD (depend of the model of controller used in the external case). 

Your post illustrates quite nicely my point: one BSOD per year (and none without the patch) is one BSOD too many per year, IMO. And CHKDSK becomes unreliable on top of it? That's dangerous by definition: one cannot rely on one's filesystems if even something so basic as CHKDSK refuses to work. As I said, it's not ready for day-to-day use.

[caliber]

 

1 hour ago, genieautravail said:

The patch is pretty stable but from time to time (several weeks between) I get a BSOD. Especially when I run a browser based on Chromium. 

so the brazilian patch will not support the chinese 360 browser ?

[bphlpt]

@caliber:  The way I interpreted @genieautravail's comment, was that the Russian patch was not 100% stable with Chromium based browsers, I assume that meant the Chinese 360 Extreme, but that it did work with it. The patch has also worked to his overall satisfaction, since he has continued to use it for over a year. He is willing to overlook things like not being able to use CHKDSK with certain drives, and the occasional BSOD. If the software he uses can take advantage of the extra memory the patch provides access to, then he apparently feels that benefit outweighs the infrequent, to him, inconvenience of being forced to restart his machine and other small aggravations. And that's great. Others, like @dencorso, believe that one BSOD is too many.

The PAE patches by @daniel_k (especially when used to access more that 4GB) and @Dibya, of which I believe the Russian patch is one implementation, and probably the future patch by @Dibya and any other PAE patches, all suffer from the same weakness:

the potential occasional instability for some people on some hardware using some software in some circumstances.

Some folks might not have any problems whatsoever, while others can't get their system to last an hour without a BSOD. There is no way to predict with certainty how it will behave for anyone, and no guarantee. That's why MS pulled the feature from XP. They didn't want to deal with the complaints with no way to 100% fix the problem for all users in all situations. (Well, that and they're lazy, and greedy, and... LOL)

Most of us at MSFN are 100% behind the idea that ANY user should have the right to use ANY software, including ANY OS, on THEIR machine if it meets THEIR needs, and we should not criticize them for it. We are here to talk about their, and our, experiences and help others as we can. But I emphasize that we are here to HELP, not do it for them or talk someone through every step, every single time. The general rule is we want someone to give it their best effort to do it themselves, preferably more than once, then come to us and tell us EXACTLY what they are trying to accomplish, EXACTLY why, EXACTLY which steps they took, with EXACTLY which tools, on EXACTLY which hardware, and EXACTLY what the results were compared to EXACTLY what they expected. [ Not just "I tried this program I read about and it didn't work"] At that point someone here might be able to help them thanks to the information they provided. The what they want to do and why are important because we might be able to suggest alternative solutions they hadn't thought of depending on what their need really is. If they are unable, or unwilling, to do the work themselves, then we will probably suggest that they try a different approach, or go do some more research before they try again. After all, as you surmised, and @dencorso and @daniel_k confirmed, much of the software for older OS, and the PAE patches in particular, "was made by geek minds for nerds only....."

If the software you use can really benefit from the extra memory, and IF you are willing to do a lot of the work mostly yourself, and IF you are willing to live with the potential instability, and IF you feel you are capable of getting yourself out of trouble or are willing to just reformat and start over at the worst case, then by all means try the PAE patches and see if one of them works for you. But don't do it just because it's annoying that the OS won't use all of the memory in your machine, as much as I truly understand that motivation. In that case you might find that "using the Gavotte Ramdisk (using RAM above 4 GiB) for a pagefile, TEMP, and the temporary internet files from every browser", as @dencorso suggested, might better meet your needs.

I sincerely wish you the best of luck and hope you are able to get your implementation of XP to work the way you want it to.

Cheers and Regards

Edited November 23, 2019 by bphlpt

[Dibya]

15 hours ago, daniel_k said:

@Dibya
This NOT a "best patch" contest.
[...]
I'm done with this discussion.

No pun intended , I already worked in kernel extention project and I know how to add space to a section , move code in pe file , add import export and add relocation and anything pe32 has to offer. Heck in my early attempt I re wrote many components from assembly .

I am not doing best patch competition here , I am talking why I have wrote my patch despite so many patch exist as none are close to perfect like mine(atleast in case of me and my users). I assure you I am lazy guy and I don't bother about fame so I don't reinvent a wheel for it. I share my stuff so that some nerd can benefit from my work . . I am tagging my friend neverseen here . My patch never makes chrome crash [nor does] it render chkdsk unusable . Patching just kernel will make XP unable to handle memory buffer which will render system into bsod while a application uses more than the physical memory limit . What is the use of showing 6gb ram in system properties if you cannot use it ?   . @neverseen

======

Most patches causes bsod, because hal is not perfect since they can't handle  memory buffer greater than 3.5GB . Back in days when I used fix128,pae64 and other patch shared by dencorso in forum , my system rendered into useless (either freeze or get into bsod ) when it crosses 3.5GB limit , first solution I came up with was using slightly customised windows XP sp1 hal and shared it first time in ryanvm forum . In next iteration , I reimplementation i fixed hal codes which looked different than sp1 hal (12 function change were useless) since sp1 hal broke my Asus xonar sound cars sue to mixing API . In next iteration , I made hybrid of my patch and patch from harkaz . Later after years of updating with new posready files , I released first version which contained all DMA buffer logic from server 2000 and XP sp1 which fixed my usb tethering issue from my phone , fixed ATI Radeon GFX issue (I used to use rx270) also ensured working stand by and hibernation mode in XP .

I am grateful to Rudy for his excellent disassembler and support, wildbill for his petool and blackwingcat for his pemaker (best relocation editor ) .  One day I will show some disassembly code .

This is where Microsoft fixed the bug in 2003 but removed pae code from poor XP through sp2

Here is something for good read for few guys

https://support.microsoft.com/en-us/help/895575/a-process-that-runs-in-the-physical-address-extension-pae-kernel-may-e

Many may notice my patch shows slightly less ram than other patch but that's due another bug present in remote ndis driver (useful for usb tethering)

[genieautravail]

13 hours ago, bphlpt said:

aliber:  The way I interpreted @genieautravail's comment, was that the Russian patch was not 100% stable with Chromium based browsers, I assume that meant the Chinese 360 Extreme, but that it did work with it. The patch has also worked to his overall satisfaction, since he has continued to use it for over a year. He is willing to overlook things like not being able to use CHKDSK with certain drives, and the occasional BSOD. If the software he uses can take advantage of the extra memory the patch provides access to, then he apparently feels that benefit outweighs the infrequent, to him, inconvenience of being forced to restart his machine and other small aggravations. And that's great. Others, like @dencorso, believe that one BSOD is too many.

The PAE patches by @daniel_k (especially when used to access more that 4GB) and @Dibya, of which I believe the Russian patch is one implementation, and probably the future patch by @Dibya and any other PAE patches, all suffer from the same weakness:

the potential occasional instability for some people on some hardware using some software in some circumstances.

The computer is usable!

the BSOD are not very bad. I didn't have lost one file with them.

The problem with the CHKDSK command is only with one 2,5' external drive (one on the five that I have).

BUT, Windows will crash from time to time.

You are warned! 

 

 

The next year, I will only use the Pale Moon fork from Roytam1 and check the frequency of the BSOD. 

[daniel_k]

@Dibya

That's fine, buddy.

From your claims, I can see you really have an in-depth knowledge about the subject.
Looking forward to your perfect patch. It's good to have alternatives.

As long as we respect and don't copy each other's work, it's all good!

Edited November 23, 2019 by daniel_k