Jump to content

About AMD Drivers on Vista


greenhillmaniac

Recommended Posts

I've decided to start a little investigation on recent AMD drivers and Windows Vista (mainly the ones that work with the Rx 300 series and newer), because those drivers won't install, giving error 31 when trying to do it on Device Manager (even with the appropiate INF modding), and I love Vista too much, to let my GPU be the only reason I can't have it installed :wub:.

The earliest drivers I found, supporting the Rx 300 series, were located on the Gigabyte website, dated from April 2015, version 15.20 (link, in case you're curious: http://download.gigabyte.eu/FileList/Driver/vga_driver_amd_w7_w8_15.4.exe). I do believe these are prior to the big driver revision of 15.7 (all of the drivers until 15.6 are revisions of the Omega drivers 14.12. I know this because in the INF file they all say version 14.50x). With Catalyst 15.7, the driver version bumped to 15.30. Starting with versions 15.20 and 15.30, a simple INF mod would not suffice to make the drivers work with Vista.

As such, I started to dig around the .sys and .dll files of recent drivers with Dependency Walker on my Vista laptop, to see what kind of dependencies were introduced. I started by first comparing the 14.12 .sys files, with the drivers provided on Microsoft Update Catalog, dated 19-04-2016, version 16.150, and here's the difference (since it's too many images, I've compressed all the results to my Mega account): https://mega.nz/#F!ApB1hQZL!dsRHCNz9HPSyJUOYI2AZZA
Basically, atimdag.sys introduced two dependencies on NTOSKRNL.EXE that Vista doesn't have, namely sprintf_s and RtlDowncaseUnicodeChar. All the other .sys files have the same group of missing dependencies, which I'm not sure matter, since, like I said, 14.12 installs just fine on Vista.

In terms of .dlls, both versions seem to have the same kind of missing dependencies, that is, missing IESHIMS.DLL and a missing function in MFPLAT.DLL, MFCreateMFByteStreamOnStream (there are A LOT of dlls on AMD's drivers, so I only looked at the main ones, like amd_opencl64.dll, amdh264enc64.dll,...).

Now I'm kind of stumped. I don't know where to go from here. What do I need to do? Mod the drivers? Mod the system files? How do I add calls to DLLs? I feel like there might not be too much work involved in making these drivers work on Vista, because certain basic parts of it work. What do I mean by this? Well, when trying to install the drivers through Device Manager and getting Error Code 31, if I go to change the display resolution, my monitor's resolutions appears, where it otherwise only listed 600x480, which means basic display detection might be working.

So, can anyone help me/guide me to making the Rx 300 (and perhaps Rx 400) series work on Vista? Summoning @blackwingcat and others :thumbup

Link to comment
Share on other sites


I'm glad that someone is starting to take interest in making modern drivers work on Windows Vista. Unfortunately, even NVIDIA recently stopped supporting Windows Vista as of driver version 365.19, and the new GTX 1000 series cards have no Vista drivers at all. I'm wondering if an INF mod to the Windows 7 RTM drivers would be sufficient to make the drivers work. Even if driver updates eventually put a stop to the INF modding method, at least there would be a working driver for the GTX 1000 series. 
I don't have any AMD cards to test drivers, but I do have a GTX 760. I may try the latest drivers on Vista when I get a chance, and report back with results, so at least NVIDIA users would also know where Vista compatibility stands on that boat. Unfortunately, even if modern GPUs had Vista drivers, you're going to be stuck with using either Ivy Bridge or FX series processors, because Intel Haswell and later just simply won't accept Vista in any shape or form, due to a bug that causes the boot process to fail (although not every time, it happens in about 1 in 5 reboots from my testing). I'm not sure exactly what happens, but somehow, the services just don't seem to start up in the correct order, and it causes the whole boot process to fail, which throws up the "Interactive logon process initialization has failed" error message. Changing numerous BIOS settings did nothing to fix this, and installing all the latest updates also didn't fix it. Hopefully someone on here will look into this issue and find out what is exactly causing the problem, because as it stands now, Vista is pretty much unusable on Haswell because of this.

Edited by 2008WindowsVista
Link to comment
Share on other sites

4790k/Z97 user here. 2008 is correct about Ivy+ being spoiled with Vista. As far as I can tell though, only significant repercussion is the timing bug that occasionally occurs upon reboot (aka. Interactive logon process initialization has failed) It can be hair pulling, but it is pretty doable :w00t:

I've been out of the Radeon loop for quite a while, so can't really comment too much. User smeezekitty did various patches that went as far as Catalyst 15.6, I think? Personally used 15.4 on my 8570d and R7 240 setups. Beyond that got quite murky. Nvidia I'd like to believe will be more lenient towards mods. It always has been in the past. Hoping somebody is willing to step forward in making some patches to various graphics drivers. Most significant drawback using this OS anymore.

Link to comment
Share on other sites

5 hours ago, 11ryanc said:

I've been out of the Radeon loop for quite a while, so can't really comment too much. User smeezekitty did various patches that went as far as Catalyst 15.6, I think? Personally used 15.4 on my 8570d and R7 240 setups. Beyond that got quite murky. Nvidia I'd like to believe will be more lenient towards mods. It always has been in the past. Hoping somebody is willing to step forward in making some patches to various graphics drivers. Most significant drawback using this OS anymore.

Actually, the patch @smeezekitty did for the 14.12 drivers worked for all releases until the 15.6 drivers (the last ones to belong to the 14.500 branch). He did an extra patch for the 15.3 drivers, I think, but that's it.

If you'd like to know, the Rx 300 series (released June last year) do not work on Windows Vista, because the only drivers working with the cards are from the 15.20 branch or newer, which introduced some extra dependencies that the OS doesn't have. No INF modding works this time. Some extra work is needed. That's why I oppened this thread.

Edited by greenhillmaniac
Link to comment
Share on other sites

30 minutes ago, 2008WindowsVista said:

Smeezekitty might be able to tell if the missing functions can be stubbed, and he may be able to stub them. Sadly, it doesn't seem anyone else cares about Vista :(

There are dozens of us! We might not be many, but what we lack in numbers we make up for in passion.

Does @smeezekitty go online often? I still think that atimdag.sys is the main file preventing the installation of the basic drivers (even if stubbing the missing fuctions is possible, other driver functionality might still be semi-incompatible, and require aditional dll files, like power management, or hardware encoding).

Link to comment
Share on other sites

OK, this will be somewhat off topic, but I want to say it: how come everyone hates Vista so much (save for those few passionate users who found it perfect)? It received a lot of bad press during it's initial release, I'm sure because of MS' mishandling of the "Designed for Windows Vista" thing, but every time I've used it, I've found it to be very decent, and in many ways actually quite reminiscent of XP, and with the classic theme and classic start menu enabled, one can barely tell the difference unless they look really closely.

The same is true of Windows ME. Anyone remember that? Whenever I've used it, I've found it also quite decent as OSes go, and when configured properly, it's even a bit better than 98 (which, along with 95, is my one of my favorite OSes in the 9x series).

OK, got that out of my system now.

Carry on... :)

c

Link to comment
Share on other sites

4 hours ago, cc333 said:

OK, this will be somewhat off topic, but I want to say it: how come everyone hates Vista so much (save for those few passionate users who found it perfect)? It received a lot of bad press during it's initial release, I'm sure because of MS' mishandling of the "Designed for Windows Vista" thing, but every time I've used it, I've found it to be very decent, and in many ways actually quite reminiscent of XP, and with the classic theme and classic start menu enabled, one can barely tell the difference unless they look really closely.

The same is true of Windows ME. Anyone remember that? Whenever I've used it, I've found it also quite decent as OSes go, and when configured properly, it's even a bit better than 98 (which, along with 95, is my one of my favorite OSes in the 9x series).

OK, got that out of my system now.

Carry on... :)

c

A combination of Windows XP having 6 years to prosper and gain a huge base of diehard users and therefore becoming the OS that users and developers alike became accustomed to, Vista requiring much more powerful hardware to run than XP, some vendors (namely NVIDIA) not properly testing their drivers on Vista before releasing them causing several unnecessary issues and blaming Microsoft/Vista for all of it, Vista initially suffering from a few bugs of its own (albeit fixed with later patches and SP1/2), Microsoft not properly stating the system requirements (512 MB of RAM for Vista? Yeah, right), the fact that Vista was essentially an almost complete rewrite coming from XP which caused compatibility issues with legacy software (albeit mostly fixed later on), and most importantly, once one group said Vista was bad, it became a chain reaction until pretty much everyone started knocking Vista- some that more than likely never even used the OS for themselves- which ensured that Vista failed. 
You're kidding yourself if you think you're going to be able to run any NT 6 OS out of the box optimally on anything less than 1 GB of RAM (and even that is pushing it), and in 2007, that was asking a lot. Although, if Microsoft had made 1 GB the minimum, I would bet that would've made Vista a bit of a better (in terms of performance) experience for users. Although, from my experience, service packs (and perhaps prior patches) eased the experience on lower end hardware, and brought performance up to par with Windows 7, and in some cases, performance has been even better on Vista SP2 than 7 for me and other users alike.

Although, by the time Microsoft worked out all the kinks and developers got their act together by releasing Vista-ready drivers and software (or at least updated their software to support Vista), the damage had already been done, and Vista was forever known as a failure. Then, Windows 7 comes out, and everyone loves it, even though they don't realize that it's basically Windows Vista with a new name, new UI, and a few features to make it look like a completely different, new OS compared to Vista, when in reality, it's essentially the same thing- especially once Vista is fully patched. 

Anyway, I won't get too deep into the whole Longhorn/Vista story (since it's a HUGE one to tell), so time to get back on topic.

Smeezekitty hasn't been online in over a month, although lately he's been coming online less frequently than usual. Perhaps if/when he sees this thread, he could solve these issues by stubbing the calls. I guess I should've been more specific; anyone that has the experience/knowledge to make Vista work with modern software/drivers doesn't care enough about it to make it happen, with Smeezekitty being one exception, although he might not be as experienced as Blackwingcat who seems to care more about Windows 2000 than anything (not saying there's anything wrong with that either, it's a great OS). At this point, we can only hope, you have to remember, Vista wasn't exactly universally loved by many, so it won't have as many enthusiasts that are willing to help it out as Windows 2000 or XP for example. Although, the fact that Windows ME has an unofficial service pack should warrant that Vista is bound to get something of that sort sooner or later. ;)

Edited by 2008WindowsVista
Link to comment
Share on other sites

  • 2 weeks later...
On Saturday, July 16, 2016 at 9:11 PM, greenhillmaniac said:

There are dozens of us! We might not be many, but what we lack in numbers we make up for in passion.

Does @smeezekitty go online often? I still think that atimdag.sys is the main file preventing the installation of the basic drivers (even if stubbing the missing fuctions is possible, other driver functionality might still be semi-incompatible, and require aditional dll files, like power management, or hardware encoding).

I have fixed newer win 7 crimson graphics for XP , but there is toooooo many crashes .

I hope that if i succeed to get read of these problem then the driver can be applied to both xp/vista

Link to comment
Share on other sites

You're very right. I haven't been on much. And I guess I've sold out since I have installed Windows 7.

I still have a dual boot partition with Vista so I can do testing and development on Vista.

It is relatively easy to figure out the missing function and create a wrapper for usermode code. Things get trickier with kernel mode code (including the .sys) that is loaded in device manager.

Unfortunately, my experience with driver development is a little limited and the meaning of the errors isn't clear to me. A patch is probably possible but will require research

Link to comment
Share on other sites

8 hours ago, Dibya said:

I hope that if i succeed to get read of these problem then the driver can be applied to both xp/vista

Don't forget that XP and Vista are two almost entirely different beasts. What might be crashing a lot on XP and require a lot of reverse engineering of system and driver files, will probably be a lot easier on Vista (given it's basically Windows 7, but a lot prettier and lighter :)). NT 6.0 and NT 6.1 are a lot closer than NT 5.1 and NT 6.1.

8 hours ago, smeezekitty said:

You're very right. I haven't been on much. And I guess I've sold out since I have installed Windows 7.

Who can blame ya! Vista is suffering a worse fate than XP, with drivers being made only for XP and 7+, and programs that abandon both XP and Vista, without even considering the latter.

8 hours ago, smeezekitty said:

It is relatively easy to figure out the missing function and create a wrapper for usermode code. Things get trickier with kernel mode code (including the .sys) that is loaded in device manager.

Unfortunately, my experience with driver development is a little limited and the meaning of the errors isn't clear to me. A patch is probably possible but will require research

Unfortunatly I am even more illiterate when it comes to this kind of thing. I can't promise much help, but I can give a test system (since my card needs the latest drivers, I think it could be a perfect test subject :yes:).

Also, did some more research, and here are the driver versions AMD has developed up to this day, which might be useful, if you intend to do your research (in chronological order):

- 14.50x - AMD Catalyst 14.12 up until Catalyst 15.6 beta. Based on the Catalyst Omega release, these drivers don't support the RX 300 series, but are the last known ones to support Vista.

- 15.20 - only found these on the Gigabyte website, which leads me to believe these are an early version of the Catalyst 15.7? They are dated from April 2015, according to the files inside the exe, so they seem older than the AMD released 15.15, which are dated June 2015... (link: http://download.gigabyte.eu/FileList/Driver/vga_driver_amd_w7_w8_15.4.exe)

- 15.15 - just found out about these on the AMD website. These are the first official drivers to support the RX 300 series, but seem to be newer than the 15.20 that I found on the Gigabyte website (which is weird, going by the build number). Haven't tested these on Vista, but I'm assuming they won't work, like all 15.xx drivers and later (link: http://support.amd.com/en-us/kb-articles/Pages/AMD-Radeon-300-Series.aspx)

- 15.20 - the first 15.xx drivers supporting all other cards from AMD's lineup, and made for Windows 10. Starts with Catalyst 15.7 up to Catalyst 15.11.1 beta

- 15.30 - introduced with the Crimson UI overhaul of the drivers. Starts with Crimson 15.12 all the way to Crimson 16.2.1 beta

- 16.101 - discovered by the good folks at Guru3D forums. The only drivers on the 16.xx line to support the non-GCN GPUs (link: http://forums.guru3d.com/showthread.php?t=408883)

- 16.15 - Vulkan support and a few new AMD features. The first official 16.xx release. Starts on Crimson 16.3 and ends on Crimson 16.5.1 Hotfix

- 16.20 - honestly, I don't know what these drivers brought to the table... Begins on Crimson 16.5.2 Hotfix and ended on Crimson 16.7.2 Hotfix

- 16.30 - recently released with the Crimson 16.7.3, made for Windows 10 RS1. Still haven't tested these. Apparently this driver version reduces DX11 driver overhead, according to the Guru3D forum guys (great forum for AMD drivers BTW)

I think this is everything. If you find anything new, tell me. This info should provide some ground work to get started.

EDIT 1 - Corrected some info and added the missing driver entries, up until the most recent

EDIT 2 - Added info on the 16.101 drivers.

Edited by greenhillmaniac
Corrected and added some info
Link to comment
Share on other sites

  • 4 weeks later...

15.6 R9 380.JPG

Yup, there is a way to make the Rx 300 series work on Vista! But it's super unstable, and shouldn't be used for a day to day basis.

Here's how I did it: Got the Catalyst 15.6 beta drivers, added my HW ID to the INF file and moddified it to install on Vista, and voila!

A few things to note: for some reason transitions, fades to black, UAC calls and screen resolution changes are super slow... Probably because the driver was meant to work on a R9 285 and is running on a R9 380... This means the R9 Fury series probably don't work with this mod. The Windows Experience Index Score crashed the system when I tried to run the test. Like I said, this is a super unstable way to get the drivers working. Also, AMD External Events kept crashing at startup, don't know why. I applied all the fixes @smeezekitty made, so I dunno.

error.JPG

Link to comment
Share on other sites

  • 3 weeks later...

(Seems like I'm the only one keeping this thread alive :wacko:)

New driver version from AMD!

16.40 - released with Crimson 16.9.1. It introduces even less driver overhead to DX11, and Vulkan API version 1.0.21. Haven't tested Vista compatibility, but I assume it's the same as previous versions (that is, non functional :()...

Link to comment
Share on other sites

  • 4 weeks later...
On 9/16/2016 at 0:20 PM, greenhillmaniac said:

(Seems like I'm the only one keeping this thread alive :wacko:)

New driver version from AMD!

16.40 - released with Crimson 16.9.1. It introduces even less driver overhead to DX11, and Vulkan API version 1.0.21. Haven't tested Vista compatibility, but I assume it's the same as previous versions (that is, non functional :()...

Well, I hate to say it, but Windows Vista continues to die at an alarming rate, and still next to no one really cares :(
Mozilla now recently announced that they will cut off XP/Vista support after Firefox v51.x mainline, and 52.x ESR. Yes, Vista is closer to 7 than XP and should be on the same support level as 7 rather than XP, but that fact didn't seem to matter to AMD, Google, NVIDIA, and now Mozilla. Although there's still a few months left until this occurs, it will be here and gone before you know it.
As I mentioned a while back, NVIDIA finally quit developing Vista drivers with version 365.19, and XP was developed on slightly longer but eventually they discontinued that too, which was another instance of Vista being unfairly left out. However, it may be possible to feasibly mod the latest Windows 7 drivers and successfully install them on Vista since Windows 7 RTM seems to still be supported, but I haven't tried it, and it will more than likely become a dead end soon, just like the AMD drivers. Also, Intel Haswell/Skylake compatibility is still ridiculously broken on Windows Vista to the point of near unusability, so if you had any plans to try and run Vista on that hardware, you might as well forget it. 

I know I have a sad, hopeless outlook on the future of Vista at this point, but it's true... Windows Vista is dying and will continue to do so, and while Vista does have its fans and admirers (I'm one of them!), the fact of the matter is and as I've said before, pretty much no Vista admirers have the knowledge or experience to make Windows Vista work with this new software and hardware, and anyone that does doesn't care a thing in the world about Vista at all, and they most likely hated it and have probably forgotten about it by now. 
I don't want to "rain on your parade" (for want of a better term) or make you lose interest or your enthusiasm for Vista, but the point I'm trying to make here is that I wouldn't get my hopes high for Vista, because it looks to me like it's going to die a quick, quiet death alongside Windows XP, and no matter how much we express our frustrations about this whole situation, it might as well be falling on deaf ears. Efforts to keep XP alive are currently being conducted by @Dibya and @Blackwingcat, but they don't seem to care about Vista, unfortunately. :(

Edited by 2008WindowsVista
Link to comment
Share on other sites

5 hours ago, 2008WindowsVista said:

...Mozilla now recently announced that they will cut off XP/Vista support after Firefox v51.x mainline...

:o That's not good at all. 

As a strange side note to that, something happened today...I booted up my Vista drive (haven't done that in a while) to run a program I only keep on Vista, and decided to update Firefox (I had v46). I ran the internal updater and I could only get to as current as v47.1. I tried the updater multiple times but no luck getting a prompt for v49.1. I didn't attempt to download and run the full 49.1 installer, though, so I can't say for certain there is an issue regarding Vista/Mozilla.

EDIT: Vista/XP support ending for Firefox http://www.ghacks.net/2016/09/27/firefox-53-no-support-for-windows-xp-or-vista/

Edited by Luxman
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...