WildBill

Check out this article on OSNews: http://www.osnews.com/comments/21713 If only they knew... Edit: I looked at the log and I'm testing version 1.0.0 to see if some changes help. In the meantime, has anyone tried skinning yet? I'm curious to see how it works for you.

It definitely looks like an OpenGL/driver issue. Are you using onboard video or a card? Are there better drivers out there? For my card I have to use hacked XP drivers from www.omegadrivers.net, for instance, because ATI dropped Win2k support.

Here's a screenshot from a fresh install I just made: http://www.mediafire.com/download.php?rmjmrinazqf This has the latest Doom3 1.3.1 patch applied, though the program ran even from a clean install from the CD's.

That's weird...just to be safe I reloaded the download link from MediaFire and updated it. Now it should be correct.

Aha!!!! :thumbup :thumbup I learned something tonight. Delphi 6's memory manager is NOT thread-safe by default -- and won't be unless you set IsMultiThread to True when your program starts. I've had Delphi 6 since it came out in 2001 and I can't believe I never knew that. Version 0.9.9 is posted, and it's working great on my dual-core setup. I also fixed a couple of other bugs, and getting skinning should now work for people (the problem was that the skin .INI file setting wasn't being saved to the registry).

Intel DX48BT2 (it's also called BOXDX48BT2). I can't say it was easy, but I have it working. Here's a post where I describe what I went through to get both cores working (E8600 Core 2 Duo CPU): http://www.msfn.org/board/index.php?showto...st&p=864917 Win2k Pro works, I can't speak for Win2k server. For reference, in case it helps: Motherboard: Intel BOXDX48BT2 CPU: Intel E8600 3.33GHz Core 2 Duo OS: Windows 2000 Professional (installed with SP4 slipstreamed) RAM: Crucial 4Gb (2 2Gb DDR3 Ballistix PC3 - 10600 CL6, 1.8V - BL2KIT25664BA1336) Video: ATI X1800XT 512Mb (Using www.omegadrivers.net XP/2K driver 4.8.442). ATI XP/2K driver link Getting sound working was tricky. Scroll up to post 291 for a link to the sound driver that worked for me.

That's really strange. I've been playing Doom 3 on Win2k for years. I've never had a problem with it. I'm actually looking forward to reinstalling it with the new motherboard+hard drive to see how it plays. For other games, you might like these posts http://community.codemasters.com/forum/sho...ad.php?t=258793 http://community.codemasters.com/forum/sho...ad.php?t=359988

Yup, Win2k is working with the ACPI Multiprocessor HAL and both cores active on the Intel board. To say it was an unbelievable PITA is an understatement, though. First, in the Win2k textmode setup, it couldn't get through the formatting portion of my 1TB hard drive, even though I was willing to let it format to only 128Gb at first. I would up having to do the following: 1. Use a parallel IDE-to-laptop kit to connect a spare 30Gb laptop hard drive, and install on that instead. 2. Fully patch Win2k and install a bunch of drivers. 3. Install Norton SystemWorks 2003, which includes Norton Ghost 2003. 4. Coax Symantec LiveUpdate to fully patch Ghost 2003 to the newest revision so it could handle SATA drives. 5. Enable 48-bit LBA support. 5. Ghost from the 30Gb laptop drive to the 1TB drive. 6. Use PartitionMagic to resize the partition to fill the 1TB drive. 7. Copy a ton of files from my old C: drive to the new one. 8. Reinstall a lot of applications (and I'm nowhere near finished). There was a lot of trial and error in each of these steps, hence why I haven't slept since Friday night.

I had to install with both cores active to get the right HAL to load. Anyhow, it took all night (I haven't slept), but I'm almost back to normal now, with both cores working. I still have some data to copy from my old hard drive and a lot of programs to reinstall, but things are working. The only problem I'm having is with my keyboard. It's a first-gen Microsoft Natural Keyboard, and is by far my favorite keyboard ever. Unfortunately sometimes the PC forgets it's connected. I'm going to have to search the net to see if I can find a better PS/2-USB converter, because I really don't want to have to get a different keyboard. Aside from that, the computer is working really well, and if I can fix this keyboard issue then I might not need that Asus motherboard after all. I'll just have to find a use for it somehow (or maybe sell it). I tried SmoothText, and now I see what you're talking about. Explorer immediately crashed with it. At least now I can recreate the problem, which is the first step toward fixing it. At this moment, though, I'm worn out. I'm just going to relax and inch my way back toward full operation tonight.

Just for kicks I went and bought a 1TB SATA hard drive and decided to experiment. I tried a clean install of Win2kPro with SP4 slipstreamed, and once the installation finished I tried turning on both cores. The PC actually booted into Windows, but Task Manager still thinks there is only one CPU (it will show only one CPU graph). Now I'm patching from Windows update to see if that works. We shall see... As an aside, I did some searching for hotfixes, and I ran across a couple of Windows 2000 USB hotfixes: KB822568 and KB836662. You have to request download, but they emailed them to me in a matter of minutes (I guess a bot does it). I don't know if I need them, but I can't help but wonder: are there any other hotfixes that might help that I don't know about? Is there a master list somewhere of all Windows 2000 hotfixes? (in case you're wondering, I'm posting this from my Win2k laptop...) Edit: GRR. I had it fully patched, had most drivers loaded, and was copying files and found that it was using the ACPI Uniprocessor HAL. I need to reinstall using the multiprocessor HAL and see if that works...

I tried the hotfix, but it didn't help. It's still pretty amazing that you even found that. It was at least worth a try. I still can't get Intel's "high definition audio" working, and my keyboard occasionally dies which makes me have to unplug and plug-in my PS/2-to-USB adapter (the motherboard doesn't have any PS/2 ports). The motherboard also doesn't have a floppy port, and it took me HOURS this afternoon to find a way on the net to make a bootable USB stick so I could flash to the newest BIOS. I'm rapidly becoming disgusted to the point that I'm considering getting another motherboard from a different vendor. I've exclusively used Intel motherboards for years, but I really feel let down. If I could only verify that any other one I tried would play nice with Windows 2000... I'm about 95% likely at this point to drop some money on another motherboard tonight, but this time I'll make sure it meets my needs. Edit: we'll see if the Asus Rampage Extreme works better for me. If nothing else, it has more SATA ports and a PS/2 keyboard port. Apparently it's also a lot more configurable at the BIOS and should take the same type of RAM I bought for the Intel motherboard. (I do have two legal copies of XP Pro sitting on the shelf...I just don't want to have to reinstall everything and I really HATE the thought of being at the mercy of Microsoft's activation, store-bought copies or not. So I plan to stick with 2k right up to July 13, 2010.) Edit #2: Well, I did eventually get sound working, after maybe the sixth or seventh driver I tried...this one finally did the trick: http://www.x-drivers.com/catalog/drivers/s...c9221/6245.html So if anyone else is trying to get an Intel DX48BT2 working on Windows 2000, at least that might help. I'm still going to an Asus motherboard, though. I've HAD IT with unplugging my PS/2-USB adapter and plugging it back in every few minutes because my keyboard froze up. :realmad:

Well, I'm certainly *trying* to figure out the dual-core problem. I wen and bought a dual-core CPU, new motherboard, and RAM. Specifically, I got an Intel Core 2 Duo 3.33GHz CPU, Intel DX48BT2 motherboard, and 4Gb ram. Unfortunately, Windows 2000 absolutely refuses to run (or even install) unless one of the cores is disabled. It looks like it's an ACPI issue that is only fixed in an XP SP2 hotfix. So unless I switch to XP I'm stuck at single core for now (and switching to XP is something I absolutely refuse to do at this point). That's not to say that this PC isn't noticeably faster than the 3.4GHz Pentium-4 (Prescott) I had before, even though that had 4Gb RAM, too -- the bus speed is faster and this CPU has a much better architecture (and the chipset is better). I still don't have my onboard sound working yet, though, since I'm having a hard time tracking down a Win2k driver for it. I installed a Windows hotfix that I hope will help, but I have to reboot first to see if it does. At least I tried. I only wish it were possible to port whatever ACPI fix was made for XP back to 2000, or at least get 2000 to not barf with a STOP 0x0000009F (DRIVER_POWER_STATE_FAILURE) when both cores are running. That's simply beyond me, though. In the meantime, I've released version 0.9.8. It has a number of improvements that should make it more stable, though I can't call it bulletproof yet. Suffice to say that it's better in the sense that some potential bugs have been found and squashed. I'm hoping that the new system will still help me further track down bugs, but only time will tell. I know that version "1.0.0" is approaching. It won't mean anything different than 0.9.9 or 0.9.8, but I suppose it's a measure of how far SmoothText has come...

Thanks for the tip, Tihiy. I posted 0.9.7 with calls to FlushInstructionCache. Are the newer versions not working for you?

I posted version 0.9.6, which I *hope* will help, but admittedly it's a guess. I think the problem might be when overriding Windows API calls, as according to Microsoft memory accesses aren't atomic when there is more then one CPU (or CPU core). So when overriding the API (or restoring it back to the original), I first set the process affinity to use only one CPU/core (whichever is the first one specified in the mask that Windows returns). Then I restore it when finished. Processes will still use all cores, but not in that critical period when overriding or restoring API calls. At least, this is what I hope will happen, but I don't have a means to test it. All I can say is that it's had no adverse effect on my single-processor system.

I downloaded vmware player, but I have no idea how to create a virtual Win2k installation with it. How would I set up a VM that simulates two cpu's? Edit: never mind. I found some instructions, but apparently vmware can't simulate two cpu's on a single-processor system. That means that I won't be able to do any multiprocessor testing to find out how to fix the problem. I have a hunch as to where to look, but no way to do the testing I would have to perform to make it work.

I've only tested it on single-core machines. The machine I develop it on has a 3.4 GHz Pentium 4.

I looked at the logs, but so far I haven't been able to figure out where the problem is. In the meantime, I made a lot of changes that should improve matters. Version 0.9.5 is up, which hopefully will be even more stable.

Yup, it's coded in Delphi 6 Professional (you don't actually need the Pro version, but that's the version I always buy). I was going through the code tonight preparing to make some more improvements (planning to simplify things even more), and I found that the hash code that I thought was thread-safe wasn't thread-safe after all. I made a bunch of small changes, and it should be thread-safe now. I've posted a 0.9.4 update, and I really think this will improve reliability.

It took a while, but I've managed to track down several problems for version 0.9.3, and I hope fixing them will make the program more stable (please see the top post for the link, as usual). I'm certainly seeing less problems so far, but if you wouldn't mind testing it I'd appreciate it. Enjoy...

The DLL has all the debugging information in it so I can see where a memory leak takes place. The problem, though, is if there is an exception there is always a memory leak, so it's of limited benefit. I occasionally see Firefox crash, but it never seems to leave a log when it does. I've also seen some weirdness when Override Standard Controls is enabled. In the interest of narrowing it down, would you mind trying it with everything except text smoothing turned off? That is, turn off skinning, overriding standard controls, translucent taskbar, etc. Basically I need to know what subsystem is causing the problems so I can start looking closely at it. In the meantime I've expanded skinning in 0.9.3 to include status bar controls, but until we narrow down the remaining source of instability I've been sitting on it.

I've posted 0.9.2, which should work a lot better. I haven't updated the NT4 version, that will come later after I've gotten the Win2k version nailed down. I still see occasional weirdness, but it's a lot better than before. Any bug reports are always greatly appreciated, of course

SmoothText should be able to use just about any skin, though you have to configure the skin .INI files by hand at this point. Just a little update: it looks like the cause of all the exceptions was in my hashing code. I completely redesigned it from top to bottom and I think I might have finally beaten the exceptions. I need to do some more testing, but I think that I might be able to release version 0.9.2 very shortly.

I'm using IE6, my system is fully patched. Are you using the newest DLL? (see my last post). It should cause your FastMM logs to be complete (no blank entries). I suspect that the problem might be in my hash code. I'm redesigning it so it will be much simpler and easier to debug. I don't have the redesign completed yet, but I'll keep you posted.

Okay, here's another DLL. I didn't properly include all the debug information and that's why the stack trace is incomplete. This will definitely give better FastMM stack traces. http://www.savefile.com/files/2087947 I also found several memory leaks with it, and those are fixed, too.

It shouldn't, but you never know. If it helps, I'm running: Norton Internet Security 2002 Norton SystemWorks 2003 (including AntiVirus 2003)