monohouse
Content Type
Profiles
Forums
Events
Posts posted by monohouse
-
-
is it in any way possible to allow execution of processes that utilize SSE2/3/3.5/4/4.1/4.2/5 instructions to run ?
is it possible to add to win2k the functionality ? I think this will require adding to the kernel the ability to save and restore SSE2/3/3.5/4/4.1/4.2/5 registers for processes, and with the recent modifications that allow some NT6 softweare to run then maybe also AVX/AVX2 instructions ?
0 -
had a similar problem after manually integrating the DirectX 11 update for Vista, the solution is to boot again from the disc and at the windows setup press Shift+F10 for command prompt then chdir to C:\windows\winsxs and delete the files cleanup.xml and pending.xml, if your windows drive is not mounted use the mountvol command
0 -
YE ! my windows 2000 boat is AFLOAT - THANK YOU ALL and blackwingcat ! (and this message is being written on windows 2000 !)
I didn't have onboard serial, it turns out that the nLite integration of the drivers was messed up because it was done after the HFSLIP, so I started from scratch and performed the INF and AHCI driver integration before the HFSLIP process - and it worked !
so what is up ? ACPI-PC and ACPI-MP cause the freezing of the setup at "setup is starting windows 2000", MPS-MP causes setup BSOD, after installation changing from Standard-PC to ACPI-MP causes BSOD at boot - all that no surprises and no revelations (and this cannot be fixed in safe mode - it doesn't allow you to enter safe mode at all).
so I advice everyone who is planning on building a windows 2000 boat - make sure to integrate storage and INF drivers in nLite before starting the HFSLIP process
the good news:
- Radeon HD 7950 - thanks to blackwingcat's 13.4 driver works properly (out of the box - as in via device manager manual installation)
- network card driver - works properly (out of the box) - but make sure to set your network parameters before the first reboot otherwise you will get "stuckings"
- Realtek audio - works properly (out of the box)
- hardware accelerated DirectX games work properly (out of the box)
- MSI Afterburner 2.3.1 works properly (out of the box)
- RivaTuner 2.24 works properly (out of the box)
- PAE works and I can access all 32 GB of RAM (but the question remains if the drivers will cooperate, another question is how mutch "golden" memory is used to the "windowing games")
- SMB works and I can connect to my samba linux server (out of the box)
- CPU temperature is good (no high idle temperatures bug) (and hwinfo 32 works)
- hidusbf USB filter driver works and allows to go all the way to 1000 Hz with USB xHCI
- need for speed most wanted works very well with all the mods and there are no "stuckings"
- and it's FAST, despite the 2 facts that my windows XP x64 SP2 was nLited and only one CPU core is used and this windows 2000 installation was not - this one is MUCH faster
a few bugs:
- Left 4 Dead works at 85 FPS (when in windows XP x64 SP2 it's at 150 FPS) (maybe for the same reason Race Driver GRID exits - no SSE2 support in the OS)
- all websites in firefox are black when display is at 16/32 bit, but works properly when display on 256 colors
- Starcraft 2 is stuck at loading screen and eventually crashes
- Race Driver GRID says the processor does not support SSE2 instructions and exits
0 -
is the Radeon HD 7950 driver works ?
I see you have 780 Ti so I guess it work, I will have 780 Ti later, too
with ACPI-PC system is stuck in "Setup is starting windows 2000", same with standard-PC and ACPI-MP and MPS-MP
I tried all the options in the F5 menu including not using the F5 menu, they all make it stuck at "Starting windows 2000"
I have mobo gigabyte Z87X-OC Force with bios 10b
I tried to integrate INF and AHCI drivers with nLite but it didn't help, I don't know if the AHCI drivers I am using are effective because my BIOS is configured to use IDE mode and the southbridge is a PCH on this mobo, so the 8.9 driver has no effect on it because it only supports up to ICH10R
0 -
blackwingcat, I have been glancing at your work on windows 2000 and I notice how much you invested in it, such dedicated work on supporting windows 2000 makes me want to go nostalgic again. I have a Haswell system and I do like older versions of windows. (right now I am running XP x64) but I am wondering, why would you and the community that is focused around the project would invest in such a great undertaking ? what are the reasons and motivations to support and develop this platform ? what are the benefits from running windows 2000 ? why isn't it easier to run windows 2003 ? with native x64 and multi-processor support out of the box ? what makes windows 2000 so special that justifies all the work that was invested in it ?
I am willing (and able) to find out for myself (I don't use USB3 and have my own added network card already), is there a big patch I can download with all the modifications you and people that worked on it with you and apply it onto an ISO of windows 2000 ?
0 -
not using and never have used sysprep, I would like to know how, but the instructions that were provided for SP1 integration weren't clear enough for me, I simply gave up on service packs and updates to get vlite working, i don't need them
how small did you manage to get your installation ? you have any suggestions for me, maybe some batch I could try to remove more ?
0 -
I've successfully combined four deletion procedures on 7 Ultimate RTM x64, reduced the size of the installation to 1.73 GB:
0. create a virtual machine with a standard 7 ultimate x64 RTM install (I recommend virtualbox), install the vista WAIK on it (6001.18000.080118-1840-kb3aikl_en.iso) and vlite 1.2 (install 1.1.6 as well if you want to save the wimfltr)
1. Guest(VM): run 7xtreme64, do the DISM removal (as per the instructions in the xtreme64 manual!): "mount.bat", "removal.bat", "commit.bat" (don't forget to copy the sources\setup.exe to sources\setup.exe at destination afterwards)
2. Guest(VM): save the boot.wim separately < Updated: this may not be necessary if you choose rebuild all images with rt7lite
3. HOST(PM): use the removal.bat image as input to rt7lite 2.6.0 beta, make sure not to remove winsxs in rt7lite 2.6.0 beta, otherwise anything goes, I will attach a config (inf) to get you started
4. HOST(PM): copy the boot.wim that was previously saved and replace the old < Updated: this may not be necessary if you choose rebuild all images with rt7lite
5. Guest(VM): use the output from rt7lite 2.6.0 beta (with the original boot.wim < (Updated: this may not be necessary if you choose rebuild all images with rt7lite)) as input to vlite 1.2 (vlite 1.2 causes problems with the boot.wim that rt7lite 2.6.0 beta creates, that's why it is necessary to make sure that before vlite 1.2 process is started, the correct boot.wim is present
6. before you make ISO in vlite 1.2, make sure you delete everything from sources except setup.exe, boot.wim and install.wim
7. VM: optionally do steps 14 through 17 of the 7xtreme64 guide (cleanup.bat) to save an extra 0.03 GB (it will have many errors but it will install fine)
8. VM: optionally apply dead0's bat (removex64.bat) by: performing 7xtreme64's mount, copy the bat into the mountdir (preferably in an "Administrator:" cmd prompt, run it from the mountdir (also in an "Administrator:" cmd prompt, like: "D:\s2>copy vboxsrv\\share\\removex64.bat .")
9. VM: optionally apply dead0's SP1 bat (batch7x64sp1.bat) (copy into the mountdir), when finished delete the bat and do "commit.bat", load the directory in vlite 1.2 and create an ISO
10. if you get installation errors "windows setup could not configure windows to run on this computer's hardware" or "Windows could not complete the installation, to install windows on this computer, restart the installation" do this:
10.1 press shift+F10
10.2 cd oobe
10.3 msoobe <- wait for it to finish
10.4 audit
10.5 it's safe to press ok on all queries, it should reboot and you should get your system booted at last
a simple rule to help you know on which system to run the operation is: just remember to perform vista WAIK operations on the VM (vlite and 7xtreme64) where the vista WAIK is installed, and the rt7lite 2.6.0 beta operations on the host machine where a different version of imaging driver is installed
the flow of images should look something like this:
7xtreme64's removal up to step 11 (including)>rt7lite 2.6.0 beta ([almost]whatever you want except winsxs)>vlite 1.2>ISO
there are 3 problems with this preset for rt7lite:
sysdm.cpl not available solved by not removing System Properties in System category of rt7lite
network connections not accessible < solved:
enable netman (Network Connections) without depending on nsi (Network Store Interface Service):
navigate the registry editor to: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\services\Netman
open the DependOnService value and delete all entries except RpcSs
reboot the system
if you are using any programs that require the DHCP Client service (such as Tunngle), this method cannot be applied to it, the DHCP Client service will complain about no more endpoints
can't load the driver of Plug and Play software device enumerator in device manager, preventing sound card drivers from being installed < solved by not removing Drivers/Media Managers/Microsoft Media Manager in rt7lite 2.6.0 beta, the Drivers\System\High Definition Audio Bus Driver and Drivers\Media Managers\Audio\WDM Audio Drivers may also be necessary for audio to function
also, you can disable User Account Control without using the GUI: (because the vlite preset deleted access to the UAC GUI)
Windows Registry Editor Version 5.00
[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Policies\System]
"EnableLUA"=dword:00000000
"ConsentPromptBehaviorAdmin"=dword:00000000now I am removing drivers to reduce size of kernel code, this is a list of drivers from system32\drivers:
NOTE: deleting the drivers is a manual process, those files cannot be seen by cmd.exe, you may have to take ownership and give full access to the "drivers" directory
required for FAT32 access ? fastfat.sys
required for usb storage access ? usbstor.sys
required for USB ? and/or USB mouse ? hidclass.sys hidparse.sys hidusb.sys mouclass.sys mouhid.sys usbehci.sys usbohci.sys usbd.sys
required for disk drives ? disk.sys
required for cpu idle ? processr.sys
required for keyboard ? kbdclass.sys
required for IDE hard drives ? pciidex.sys pciide.sys
required for USB keyboard ? kbdhid.sys
required for intel ICH10 ? iaStorV.sys
intelppm.sys <- required for Intel processor driver to work ? <- I tested this and seen no change in my cpu temperatures or frequencies
amdppm.sys <- required for AMD processor driver to work ?
required for Audio and Audio drivers:
swenum.sys
drmk.sys
drmkaud.sys
portcls.sys
ks.sys
ksthunk.sys
required for networking:
afd.sys
tdi.sys
nsiproxy.sys <- may not be needed if using a proxy
required for NVIDIA Driver:
dxgmms1.sys
required for boot (BSOD):
Wdf01000.sys
cng.sys
Classpnp.sys
npfs.sys
watchdog.sys
msfs.sys
required for boot (error message):
netio.sys
msrpc.sys
volsnap.sys
fltmgr.sys
fwpkclnt.sys
ksecpkg.sys
ksecdd.sys
mountmgr.sys
partmgr.sys
volmgr.sys
wdfldr.sys
vdrvroot.sys
battc.sys
msahci.sys
compbatt.sys
spldr.sys
wmilib.sys
msisadrv.sys
ndis.sys
pci.sys
forcibly removing from registry (Start = 4) causes BSOD:
fvevol.sys
rdyboost.sys
can be deleted:
hcw85cir.sys
agilevpn.sys
bxvbda.sys
evbda.sys - ebdrv
ql2300.sys -
http.sys
PEAuth.sys
spsys.sys
rdbss.sys
brserid.sys
mrxsmb.sys
mrxsmb10.sys
mrxsmb20.sys
b57nd60a.sys
netbt.sys
vmbus.sys
exfat.sys
storport.sys
scsiport.sys
nvstor.sys
ndiswan.sys
rmcast.sys
rasl2tp.sys
tunnel.sys
nv_agp.sys
raspptp.sys
amdsata.sys
sbp2port.sys
hidbth.sys
usbcir.sys
tdx.sys
usbccgp.sys
parport.sys
bridge.sys
serial.sys
smb.sys
raspppoe.sys
browser.sys
wanarp.sys
ipfltdrv.sys
ohci1394.sys
bthmodem.sys
fileinfo.sys
stream.sys
gagp30kx.sys
uliagpkx.sys
amdk8.sys
uagp35.sys
termdd.sys
appid.sys
agp440.sys
mup.sys
agilevpn.sys
vwififlt.sys
ndproxy.sys
ndisuio.sys
fsdepends.sys
dumpfve.sys
winhv.sys
pcw.sys
brserwdm.sys
vmstorfl.sys
hidir.sys
netbios.sys
tcpipreg.sys
rndismp.sys
discache.sys
ndiscap.sys
filetrace.sys
kbdhid.sys
usbcamd2.sys
usbuhci.sys <- may be needed for AMD USB
monitor.sys
tape.sys
amdxata.sys
wacompen.sys
hidbatt.sys
sermouse.sys
vwifibus.sys
crcdisk.sys
ndistapi.sys
serenum.sys
tdtcp.sys
asyncmac.sys
wimmount.sys
mcd.sys
vmbushid.sys
ws2ifsl.sys
isapnp.sys
usb8023.sys
brfiltlo.sys
vwifimp.sys
cmbatt.sys
cmdide.sys
viaide.sys
intelide.sys
tdpipe.sys
aliide.sys
amdide.sys
MTconfig.sys
brusbmdm.sys
brusbser.sys
hwpolicy.sys
sffdisk.sys
sffp_sd.sys
wmiacpi.sys
sffp_mmc.sys
wfplwf.sys
pciide.sys
mskssrv.sys
errdev.sys
brfiltup.sys
mshidkmdf.sys
mstee.sys
mspclock.sys
mspqm.sys
vms3cap.sys
null.sys
secdrv.sysNOTE: driver list is still being updated, be careful with it
gross weight of kernel code removed: 13,123,096 bytes in 120 drivers
p.s. if you find more/less drivers to delete, post here or msg me so I could maintain the list for everyone else that may/not need a driver
0 -
nice tool, thanks, not all that big when it comes to removal but it leaves a fully functional image even with everything removed, it's an extra weapon in the arsenal
I have made some progress making a lite RTM version, got it down to about 4GB so far, vlite 1.2 doesn't work for me so I am using 1.1.6 which is much more functional and stable, with the help of dead0's bat's the winsxs is down to about 1 GB
0 -
I am also looking into making a lite 7 x64 sp1, I've already spent a couple of days trying to figure out what method to use and which guide to follow, so far no go, you may have better luck if you give up SP1 and use RTM(SP0), it seems to be better supported in vlite and there appear to be more batch-based removal procedures for RTM than for SP1, they seem to rely on vlite nonetheless, I think the best would be to combine all three extreme batch removal topics (if that is even possible) to make a super-lite 7, I am willing to give up SP1 (and all future SPs) for a super-lite edition, if only I knew for sure that it would work (I suppose I should start with getting at least one of these projects to function xD)
0 -
trala I didn't fully understand your process, but if you managed a 250 mb install.wim, that probably puts you in first place, I am on the extreme removal side of things and so far my efforts are only able to produce about 1.3 GB install.wim 7 x64 SP1, that's after a whole process of rt7lite and then vlite 1.2 with almost everything removed in both of them, there can't be that big a difference between SP0 and SP1 can there ? if I give up SP1 and decide to use only SP0 and use both rt7lite and vlite 1.2 on it, will I get 250 MB install.wim ?
0 -
I can't see/find the option "component cache (winsxs)" in vlite 1.2, where is it supposed to be ? Isn't it in components menu ?
0 -
from the beginning: (is that right ?)
1. install WAIK 1.1 for vista SP1
2. use GimageX as done on post #105
3. use myselfidem's instructions as on post #166
but what components are not removable in vlite ? and how much can be deleted from the 7 installation ? I missed the part that deals with the actual deletion
0 -
I think that problem happened because I was using an x86 platform to operate on an x64 vista, when I used an x64 platform it worked fine without making the mentioned directory
0 -
the processing takes place and then when it reaches this point (after the counting is finished):
the vLite window disappears and that's it, I've tried vLite 1.2 and vLite 1.1.6
0 -
thanks, I don't have an AMD, but from my experience all those power management features are bad, I was playing quake4 the other day and noticed that I had strange sudden drops from 60 to 30 fps I couldn't figure out why or how they were occuring and then I was trying to figure out how to define the proper thresholds for the speedstep, when I found the software I accidentally removed the speedstep, and then when I loaded the game again it was going 60 fps all the way all the time, wasn't long before I went to the BIOS and disabled all power management features
it may be possible that the video card is doing the same thing, it may be worth investigating by using a non-ACPI HAL/kernel combination and see if it has any effect on performance, power management is good, but it seems speedstep is not it, hope AMD have done better, I think sometimes the best power management is to turn off the system
0 -
yhe that worked
0 -
I was trying to patch XP with service pack 1, what files ? I was using update/update.exe
nLite is no longer compatible with extracted service packs ?
0 -
while on the subject of adding features, there is one thing: is it possible that the TMP and TEMP directories of the environment variables could be "System Variales" and not "User Variables" ? because the setting does not change %TEMP% and %TMP% of the system only the user's
0 -
I agree, NOTHING is completely true when running something that you did not write yourself.
as for waiting/looking for files - that is where Deleted Edition's anti-dll-loading technology comes in
But an optimized system may very well beat a system that is faster on paper.thell me about it, I got a P3 500 mhz that smokes the newest systems, like my core 2 3880 mhz, of corse not when it comes to performance tests, but GUI and boot, is faster, just until I get to delete it as much as possible
0 -
emulators will not give you accurate boot time indications, nor will they provide any accurate performance measures, and boot time DOES NOT depend on your hardware, not even your hard disk speed, or cpu speed for that matter, more importantly, boot time does not represent system performance, but one thing is for sure - the more files you delete, the faster is BOTH your system AND your boot times
0 -
you didn't understand, I mean that you trust what the "list of fixes" say, not whether you trust microsuck itself, but "In a way" "is basically saying you trust MS"
let's be clear: I don't trust anyone, not you, not MS, not linux, not bush - only me and only the changes that I make to the system - and that is why Deleted Edition exists, along with the speed improvements, but then "In a way" "is basically saying you trust me" but not entirely - because all the changes that I make are fully documented, and concept is simple - deleted as much as you can - if it can be deleted - delete it, and when you delete microsuck's files - it means that you are not "basically saying you trust MS", because the more microsuck files you delete, the less you "trust MS", also I only trust my own examinations and tests, and you don't so "is basically saying you trust MS"
also, the only reason that I allow myself to be making threads in a forum that is based on a domain that starts with "ms" is because I trust nuhi and his work on nLite and that I whant to contribute what I can to make sure that some of these bugs can be fixed and because I believe that nlite is worth that, if it was ANYTHING else I would never allow myself to type a site that starts with "ms", so "basically saying you trust MS" ? maybe, I guess it is hard to clearly define that, maybe everyone that runs that microcrap sooner or later "basically saying" they "trust MS", I can only be sure of one thing, that im trying hard not to, and maybe with nLite and Deleted Edition, that can be achieved
as for netlogon - that is all I know for now, since it is obvious that 2003 is slower, there is not much need to test it any further, just check if that maximum number of connections option is availabe when integrating sp1, and whether or not it should be there in the first place, just strange that microsuck would limit connections on server
0 -
- "same settings" - what settings ?
- "running the same" - what tests did you do to verify that ? could you show me ?
- "You say it is not broken", and I have proof !
- "the list of sp1 and sp2 fixes say otherwise" - do you trust microsuck ? or facts ? or yourself ?
- "little faster" - im sorry, but 1000 MB/s of memory speed is 10-20 times faster than the L1 cache of a pentium 200mmx, and faster than the L2 cache speed of a pentium 3 500 mhz, see here: http://manoa.flnet.org/server.html, you can't call that a "little faster"
- "software advances" ? it seems to me more like software fallbacks, just how do you define that ?
- "in it for the performance and size" - if the "software advancements" are not that, then what are they for ?
- "but I like it" - I find it hard to understand why would anyone like to run their software slower
I should get to testing SP0 and SP1 to see what differences are there if any
also not to offend but when you stated a little faster and running the same - any chance you were testing in software emulators ? just to clear any doubts all tests were running on HARDWARE, no emulators
I thought machines are faster so that you could put their speed to use, not to waste
0 -
p(1) 2
patching error occured
check if one of the addons is overwriting one of these files with non-original file(already patched outside nLite)
uxtheme.dll
syssetup.dll
tcpip.sys
usbpost.sys
don't use already patched files, use Patches page instead.
btw was testing SP1 for 2003, got this error after integrating service pack 1 into 2003, wonder if it's going to work, I have added the tcpip patch because it was selectable, which seems strange that microsuck created connection limit for 2003 server....but maybe a nlite bug, but only occurs when service pack 1 integrated
apparently it works, here are tests results:
XPPro SP3 | 3SEE SP0 | 3SEE SP1 | 2000 SP0 | 2000 SP3 |
Memory Read - 9404 - 8503 - 8497 - 8527 - 8475
Memory Write - 8855 - 8853 - 8852 - 8854 - 8858
Memory Copy - 8586 - 8322 - 8256 - 8357 - 8088
Memory Latency - 52.9 - 53.0 - 52.7 - 52.7 - 52.8
CPU Queen - 7879 - 7880 - 7880 - 7878 - 7878
CPU PhotoWorxx - 29127 - 29299 - 29097 - 29365 - 29205
CPU ZLib - 51769 - 51568 - 51532 - 51745 - 51733
CPU AES - 14281 - 14268 - 13834 - 14303 - 14286
FPU Julia - 14844 - 14724 - 14722 - 14720 - 14725
FPU Mandel - 7186 - 7187 - 7185 - 7186 - 7145
FPU SinJulia - 10813 - 10812 - 10811 - 10810 - 10810
SuperPI 1M - 13.109 - 13.156 - 13.172 - 13.188 - 13.187
SM2 Primordia Ag: 166 - 169 - 167.93520 - 174.71199 -
SM2 Mem BW MBs: 8102.50 - 7986.46 - 7984.72 - 8013 -
SM2 AES 7.3 - 7.43 - 7.51 - 7.35496 -to summarize those results:
#1 problem of 2003 Server is memory performance, which is not fixed in SP1
#2 a definite performance degradation with increasing service packs is proven, even worse memory performance
#3 the "good" news: other performance factors are less affected, but eventually are affected from the memory speed
the system on which tests were run is in signature, the system configuration that was used is #5 (maximum) with 2.0V on RAM (instead of 1.95), the system uses water cooling, temperature does not affect test results (throttling is also disabled in bios)
maybe I should test xp/2003/2008 x64 editions and 2000 also - too bad x64's cannot be compared to x86
as to your question - I hope these results answer it, I should add XP SP0, XP1 later, maybe 2008, but im not sure, since 2008 is based on vista it is highly unlikely that it's performance could approach that of 2003 or 2000, let alone XP, even with the fact that 2008 is up to 17% faster than Vista SP1, it seems to me that the best win32 is XP, at least as far as software performance is concerned, but what else is there anyway ?
0 -
yhea, I get confused all the time, so many " last sessions" that I have no time to read all of them which one was what - manual save/load into directories outside nlite is the only way to keep it orderly and know what is what
manually deleting the presets directory every time before it loads is not convinient, on the border of needing to write a batch script to "properly" load nlite.exe (deleting the directory before it starts every time)
regardless, good version 1.4.5 beta 2 is
0
Windows 2000 on Haswell?
in Windows 2000/2003/NT4
Posted · Edited by monohouse
I got it to install on my Gigabyte Z87X-OC Force mobo with a 4790K in ACPI-MP mode by disabling the serial port in the BIOS but there is another problem: the mobo has a PLX chip that is used for all the red PCI-E slots, the (re)installation of the chipset drivers doesn't fix the error which is found in device manager: the device cannot start because it cannot find resources for it to use, have to disable other device to enable it
this error is shown on the "Intel(R) Xeon E3-....PCI Express x16..." "device"
but it doesn't say what other device - so I search manually the IRQ/MEM/IO addresses of each of the devices in the system category in device manager: found only one device that has overlapping addresses with it is the "PCI Bus", so I try to disable "PCI Bus", try to install the "Intel(R) Xeon E3-....PCI Express x16..." driver on the "PCI bus" device, try to use alternate driver on "Intel(R) Xeon E3-....PCI Express x16..." device
later I find that "Intel(R) Xeon E3-....PCI Express x16..." is sitting on top of "PCI Bus" so the addresses are supposed to overlap ?
my video card and network card are on the PLX chip and I don't see them in device manager, it's funny that there is no "VGA compatible comtroller" in device manager and the system works fine otherwise
--------------------------------------------------------------------------------------------------------------------------------------------
I looked into my XP x64 installation and found out I have conflict between the PCI-E PLX chip and VgaSave driver
here is what I tried:
1. switch to IGPU as display device while the cards are connected and use the IGPU as boot display device
2. install drivers (nvw2k35306.cab) for both other cards (successful)
(tried also nvw2k35906.zip, nvw2k35560.cab - they both give error "cannot load driver for device")
3. switch back to the card as display device (780 Ti) and use it as boot display device
4. it's still conflicts and all the devices connected to the PLX chip are gone from device manager
any way to dodge this problem ?
I tried to look into the registry see if I could change those conflicting addresses, could not find what I was hoping for (load VgaSave into a different MMIO range).
I tried to uninstall VgaSave a couple of times, it screwed 2 installations by BSOD with a followed reboot and no safe mode, I guess no point trying it again :x
--------------------------------------------------------------------------------------------------------------------------------------------
I tried everything, moving the card to another slot, replacing vga.sys with version from windows xp SP0,SP3, a version from windows 2003 SP0, a version from 2000 SP0, tried to replace HKLM\SYSTEM\CurrentControlSet\Services\nv to VgaSave and VgaSave renamed with VgaCompatible set to 0 and 1, tried to change VgaSave Start mode from 1 to 2,3 and 0, tried to enable more devices in BIOS to force it to move MMIO address to another range with no luck here either, tried updated VGA driver videoprt.sys still no luck
--------------------------------------------------------------------------------------------------------------------------------------------
ok so the story of this problem is that when the system is installed as Standard-PC the conflict does not exist and the system detects the devices but if installed as any of the ACPI options the conflict exists, I tried to change it after installation but this consistently causes INACCESSIBLE_BOOT_DEVICE 0x0000007B BSOD, however before switching from Standard-PC to ACPI-MP and backup the hal.dll of the Standard-PC configuration and rename it to for example hal-std.dll then add a boot.ini option for it as /HAL=hal-std.dll - it becomes possible to enter the system in safe mode, in addition it turns out the problem is not with the PLX chip - the conflict exists also when the cards are connected directly to PCI-E bus that is in the CPU, I tried many different combinations of /HAL and /KERNEL to boot the ACPI-MP mode but all of them failed mostly on the 7B BSOD, I also tried to uninstall and change the "fancy" IDE controller drivers to windows standard drivers but that failed as well, also tried to install the system without integrated chipset drivers but the problem was the same only the name of the conflicting device changed (instead of PCI-E x16 controller it's called PCI-PCI bridge)