Best Graphics Card With Win98SE Drivers?

[Queue]

If it overheated and stuff melted, you probably had too weak of a powersupply and your card was overdrawing which slowly built up heat until things went bad. This sucks because a hard system freeze is what usually happens instead.

Also, welcome to the BS world of having to plug power directly in to your video card. Since the early Radeon 9800s, it has become more and more common; be glad that model only wants 1 molex plug, and not 2 molex, a 6-pin (roughly 2 molex), 2 6-pin, an 8-pin or even (geh) 2 8-pin power plugs.

Things to know about video cards that need power plugs:

- They need their own independent cable(s) coming from the powersupply, not ones shared with other components (unless you like to risk your hardware). So, don't plug one power cable into your video card AND a hard drive.

- Just because two separate cables come out of the powersupply, doesn't mean they aren't actually sharing their power limits; getting info on which lines of a given powersupply are independent is not usually easy, so good luck.

- If they have multiple power plugs, they need a separate power cable per plug; if a video card has 2 molex plugs, it means you need to plug two independent power cables into it, not two molex plugs on one cable. Yes, this often means you have to find a powersupply that has a buttload of cables coming out of it.

Queue

Edited July 1, 2010 by Queue

[Dave-H]

If you had connected it wrongly (I don't think that's actually possible without use of undue force, because the molex connector has two cut-out corners to prevent it from being plugged the wrong way) the card would have fried right away, not after 1h, IMO. If your card's connector is mounted like the one in the attached picture, the the correct position is as depicted (the yellow cable on the same side as the agp connector). Now, are you sure it was the card and not the power supply that got fried (or maybe both)? It *is* possible that the video card remains good, despite having fried the molex.

Many video cards do connect to a standard 4-pin or 6-pin power supply line. Yours draws a lot of current from the 12V line, from what I just read around. Was it the pin connected to the yellow cable, the one that melted, perchance?

Thanks Den.

Your picture shows exactly what I've got, and yes, it was the yellow cable that melted!

I've checked with a digital multimeter, and that pin on the connector is now reading a direct short with the ground pin next to it, which is probably not a good sign. I fear that the card is indeed damaged, and I'm a bit loathe to try it in the system again in case it damages anything else. I'm terrified of the very expensive motherboard getting damaged of course, as that will leave me with no computer!

If it overheated and stuff melted, you probably had too weak of a powersupply and your card was overdrawing which slowly built up heat until things went bad. This sucks because a hard system freeze is what usually happens instead.

Also, welcome to the BS world of having to plug power directly in to your video card. Since the early Radeon 9800s, it has become more and more common; be glad that model only wants 1 molex plug, and not 2 molex, a 6-pin (roughly 2 molex), 2 6-pin, an 8-pin or even (geh) 2 8-pin power plugs.

Things to know about video cards that need power plugs:

- They need their own independent cable(s) coming from the powersupply, not ones shared with other components (unless you like to risk your hardware). So, don't plug one power cable into your video card AND a hard drive.

- Just because two separate cables come out of the powersupply, doesn't mean they aren't actually sharing their power limits; getting info on which lines of a given powersupply are independent is not usually easy, so good luck.

- If they have multiple power plugs, they need a separate power cable per plug; if a video card has 2 molex plugs, it means you need to plug two independent power cables into it, not two molex plugs on one cable. Yes, this often means you have to find a powersupply that has a buttload of cables coming out of it.

Queue

Thanks Queue.

This is all new to me. As I said, the installation instructions enclosed with the card don't even mention the power connector!

Interesting that you say it should be connected directly to the psu and not share with any other devices, as one of the things they enclose with the card is a power splitter, as if they're encouraging you to do exactly that! This was what fried fortunately, not one of the actual psu cables thank goodness. I had it connected to the same cable that also supplies two hard drives and a fan, from what you say, probably not a good idea!

My psu is a Seasonic S12-600, a 600W model and the most powerful in the range. I need that as I have a dual processor motherboard and a lot of disk drives. I hope the graphics card wasn't the "straw that broke the camel's back"!

I suspect what happened was that the excessive drain from the card on one power line, shared with other things, caused the psu to shut down that line after a while, although it did survive for half an hour or so. It then took a while to reset itself which is why I couldn't get the system to work for a while afterwards. Thankfully it recovered. It didn't shut the whole psu down as some parts of the system were still powering up.

Strangely there seemed to be no sign of overheating damage on any of the other cables, only the splitter that was actually feeding the card, but that may be simply because they are heavier cables.

As I said to Dencorso, I am afraid that the card is now damaged and therefore shouldn't be tried again with different cabling. Do you agree with that? I don't know how I can test it without actually trying it, but the fact that it's showing a short circuit between the pin that the wire melted on and ground doesn't look good.

[dencorso]

As I said to Dencorso, I am afraid that the card is now damaged and therefore shouldn't be tried again with different cabling. Do you agree with that? I don't know how I can test it without actually trying it, but the fact that it's showing a short circuit between the pin that the wire melted on and ground doesn't look good.

Sure. The video card is probably toast, by the looks of it. And I guess you'd need a much stronger PSU, after all, to power another such card. I think Queue's picture about what's happened is probably accurate. Sorry for your new video card. But, then again, I'm glad nothing else was affected. Do browse Corsair's and Huntkey's websites to see what PSU's they're offering nowadays... they're my favorite brands for PSUs.

[Cyker]

Argh, sorry to hear what happened

But at least you got it working for a it!!

Inspired by your success, I had another go! Tried removing my Live!, SATA controller and the Radeon from the board and tried to see if I could get it working.

Tried several different patchmem options and the split8mb utility, different BIOS tweaks that I had written down before and a bunch of other stuff, but still can't get it to boot. Just hangs at "Starting Windos 98..." or "Please wait while setup updates your system files blah blah Completed updating files, continuing to load Windows..."

I had to call it quits again tho' because in my over enthusiasm I have messed up my Win98 install and it just boots into a BSOD in VMM or IOS or just says Windows Protection Error blah blah

I wish I had the tracing tools I have in Linux to try and debug this stuff!

I think there is something fundamentally weird about something on my system which is stopping it from working but I wouldn't even know where to begin!

[Dave-H]

Sure. The video card is probably toast, by the looks of it. And I guess you'd need a much stronger PSU, after all, to power another such card. I think Queue's picture about what's happened is probably accurate. Sorry for your new video card. But, then again, I'm glad nothing else was affected. Do browse Corsair's and Huntkey's websites to see what PSU's they're offering nowadays... they're my favorite brands for PSUs.

Thanks Den.

I reckon it's time to get another card.

I would be really surprised if my psu is inadequate.

It is a 600W model, the most powerful in its range, and I paid good money for it simply on the basis that I wanted it to be running well within its spec, even with my system.

It's the third psu I've had in the system, the other two failed early I suspect (although they both survived quite a long time) because they were being stressed and were running near their limits. The one I've got now is much higher rated than either of its predecessors.

I hope I can make a case that the card was actually faulty, as even if it was stressing the psu, it must have suddenly put a short circuit across the line to burn up the cable like that.

Surely no card (or anything else for that matter) should ever draw enough current to melt its connecting cable!

I think it ran for a while, and then developed a hardware failure, like one of the devices on the card shorted.

That's the case I'll make to the seller anyway, as I don't believe that an inadequate psu could have caused those symptoms. The worst that should have happened surely is the psu overheating and tripping off, or perhaps malfunctioning caused by the supply rail voltage falling.

Unfortunately, I'm sure the seller had only one of them, so they won't be able to replace it. All I can hope for is a refund, and then hope to find another somewhere else.

[Dave-H]

Argh, sorry to hear what happened

But at least you got it working for a it!!

Inspired by your success, I had another go! Tried removing my Live!, SATA controller and the Radeon from the board and tried to see if I could get it working.

Tried several different patchmem options and the split8mb utility, different BIOS tweaks that I had written down before and a bunch of other stuff, but still can't get it to boot. Just hangs at "Starting Windos 98..." or "Please wait while setup updates your system files blah blah Completed updating files, continuing to load Windows..."

I had to call it quits again tho' because in my over enthusiasm I have messed up my Win98 install and it just boots into a BSOD in VMM or IOS or just says Windows Protection Error blah blah

I wish I had the tracing tools I have in Linux to try and debug this stuff!

I think there is something fundamentally weird about something on my system which is stopping it from working but I wouldn't even know where to begin!

Yes, it does sound as if patchmem is perhaps not doing what it should do (I assume that you are using one of the correct versions.) The VMM BSODs you're getting now can happen in my experience if you've got too much RAM and/or a very large registry. Patchmem has cured all that for me.

I'm sure you've tried this, but when it hangs on the restart, have you tried restarting in Safe Mode?

I have known Windows 98 hang on reboots after installs, but Safe Mode starts OK and does the registry updates etc., and it will then boot OK in normal mode.

Worth a try if you haven't tried that.

The only other thing I'd now try before reinstalling Windows is to go to the system\vmm32 folder (in DOS), and restore the backups that should be there of your vcache.vxd and vmm.vxd files. These should have been made by patchmem and have a *.bak extension.

Then boot into Safe Mode with the original graphics card installed and use the system configuration editor to limit the memory to 512MB.

With a bit of luck the system should then boot into normal mode again and you can reinstall patchmem and take it from there.

Edited July 1, 2010 by dencorso

[CoffeeFiend]

It sounds like your card is damaged indeed. There are several possibilities:

-You plugged the power backwards at some point. You say you had it working for a while, but a lot of people are plain unwilling to admit doing this -- although it certainly is possible (I've done it before)

-The power supply passed a very nasty voltage spike damaging components on the video card (also possible, but most likely other components in your computer would have failed too)

-If it was a ghetto PSU I would have been tempted to blame it, but Seasonic makes great units generally speaking (although this is a somewhat older unit, with a weaker 12v rail than I'd wish for, especially with a lot of drives and such -- 432W combined out of 600W; their current models like the SS-600ES and SS-600ET do over 550W). I don't really see it being the source of your problems.

-The card was already damaged somehow and it decided to die just now, heat may have helped a.k.a. bad luck

-As for a power supply being "too strong", it is not an issue. You could have a 1500W PSU if you wanted. That will not damage a card. There's just more power available for various devices, it doesn't make them use more.

-How much power does it need? Not sure, I'd guess around 70W or so (about 6A). They recommend a 400W PSU with 18A on the 12v rail minimum but you already have twice that so I don't really see it being a problem

First of all, I'd have a very good look at the voltages coming out of your power supply. If it's the culprit, you could fry everything in your computer. There's also a possibility that it was damaged or degraded by the video card. If you have a digital multimeter have a look at the 12v rail primarily (between the yellow wire and either of the 2 black ones besides it in a plain old molex). Last resort is checking it in your BIOS.

Once you got that sorted, then you could take pictures of the card, particularly around the power connector, on both sides of the PCB (zoomed in as close as you can get it, and pic has to be sharp or it's useless). This way we can have a look at what happened (what failed or how it failed -- chances are it's a simple TVS diode that failed for instance)

Last thing -- I definitely wouldn't put that card back in my PC if it reads a short and the wires melted on the PSU. That's about as bad of a short as it gets, and the best case scenario is about that -- melting some more wires.

[Dave-H]

It sounds like your card is damaged indeed. There are several possibilities:

-You plugged the power backwards at some point. You say you had it working for a while, but a lot of people are plain unwilling to admit doing this -- although it certainly is possible (I've done it before)

Definitely not!

The connectors aren't reversible anyway, (and I did check that I hadn't forced one in the wrong way round!)

-The power supply passed a very nasty voltage spike damaging components on the video card (also possible, but most likely other components in your computer would have failed too)

-If it was a ghetto PSU I would have been tempted to blame it, but Seasonic makes great units generally speaking (although this is a somewhat older unit, with a weaker 12v rail than I'd wish for, especially with a lot of drives and such -- 432W combined out of 600W; their current models like the SS-600ES and SS-600ET do over 550W). I don't really see it being the source of your problems.

-The card was already damaged somehow and it decided to die just now, heat may have helped a.k.a. bad luck

-As for a power supply being "too strong", it is not an issue. You could have a 1500W PSU if you wanted. That will not damage a card. There's just more power available for various devices, it doesn't make them use more.

-How much power does it need? Not sure, I'd guess around 70W or so (about 6A). They recommend a 400W PSU with 18A on the 12v rail minimum but you already have twice that so I don't really see it being a problem

Thanks, that's very helpful, and confirms my thoughts.

First of all, I'd have a very good look at the voltages coming out of your power supply. If it's the culprit, you could fry everything in your computer. There's also a possibility that it was damaged or degraded by the video card. If you have a digital multimeter have a look at the 12v rail primarily (between the yellow wire and either of the 2 black ones besides it in a plain old molex). Last resort is checking it in your BIOS.

It's a server motherboard with software monitoring of the hardware. All the power rails are shown as being in the centre of their tolerances, as they were before the problem, so it looks as if the psu wasn't damaged.

Once you got that sorted, then you could take pictures of the card, particularly around the power connector, on both sides of the PCB (zoomed in as close as you can get it, and pic has to be sharp or it's useless). This way we can have a look at what happened (what failed or how it failed -- chances are it's a simple TVS diode that failed for instance)

Last thing -- I definitely wouldn't put that card back in my PC if it reads a short and the wires melted on the PSU. That's about as bad of a short as it gets, and the best case scenario is about that -- melting some more wires.

Don't worry, that card isn't going back in the machine!

I can do some pictures if you like, but I've inspected it very closely myself, and can see no physical signs of any problem. No blackening anywhere, no sign of anything open or short circuited.

[rloew]

Argh, sorry to hear what happened

But at least you got it working for a it!!

Inspired by your success, I had another go! Tried removing my Live!, SATA controller and the Radeon from the board and tried to see if I could get it working.

Tried several different patchmem options and the split8mb utility, different BIOS tweaks that I had written down before and a bunch of other stuff, but still can't get it to boot. Just hangs at "Starting Windos 98..." or "Please wait while setup updates your system files blah blah Completed updating files, continuing to load Windows..."

I had to call it quits again tho' because in my over enthusiasm I have messed up my Win98 install and it just boots into a BSOD in VMM or IOS or just says Windows Protection Error blah blah

I wish I had the tracing tools I have in Linux to try and debug this stuff!

I think there is something fundamentally weird about something on my system which is stopping it from working but I wouldn't even know where to begin!

Yes, it does sound as if patchmem is perhaps not doing what it should do (I assume that you are using one of the correct versions.) The VMM BSODs you're getting now can happen in my experience if you've got too much RAM and/or a very large registry. Patchmem has cured all that for me.

I'm sure you've tried this, but when it hangs on the restart, have you tried restarting in Safe Mode?

I have known Windows 98 hang on reboots after installs, but Safe Mode starts OK and does the registry updates etc., and it will then boot OK in normal mode.

Worth a try if you haven't tried that.

The only other thing I'd now try before reinstalling Windows is to go to the system\vmm32 folder (in DOS), and restore the backups that should be there of your vcache.vxd and vmm.vxd files. These should have been made by patchmem and have a *.bak extension.

Then boot into Safe Mode with the original graphics card installed and use the system configuration editor to limit the memory to 512MB.

With a bit of luck the system should then boot into normal mode again and you can reinstall patchmem and take it from there.

Patchmem, and it's /A and /M options, doesn't help or hinder operation of the NVidia Drivers.

The BSOD's and Windows Protection Errors are probably caused by the Driver being incompatable or misconfigured. I have seen this occur during ny experiments with various Cards.

You can uninstall Patchmem simply by running it again. It runs from DOS.

If you uninstall it, I doubt you will be able to run Safe Mode. You will have to limit RAM by setting MaxPhysPage and MaxFileCache in the SYSTEM.INI File from DOS. You will not be able to reach the System Configuration Editor.

I would recommemd not using the /A Option as it probably does not help and may cause unwanted side effects.

[Dave-H]

Patchmem, and it's /A and /M options, doesn't help or hinder operation of the NVidia Drivers.

The BSOD's and Windows Protection Errors are probably caused by the Driver being incompatable or misconfigured. I have seen this occur during ny experiments with various Cards.

You can uninstall Patchmem simply by running it again. It runs from DOS.

If you uninstall it, I doubt you will be able to run Safe Mode. You will have to limit RAM by setting MaxPhysPage and MaxFileCache in the SYSTEM.INI File from DOS. You will not be able to reach the System Configuration Editor.

I would recommemd not using the /A Option as it probably does not help and may cause unwanted side effects.

Rudolph, does that mean that running patchmem with the /A switch was un-necessary in my case?

I do need the /M switch to solve the large registry problem.

Is the /A switch likely to cause harm, and what should I look out for?

The nVidia card was working fine in Windows 98 before it blew up, but I'd already installed patchmem by then so I don't know if it would have worked without it anyway.

The system seems to be working fine at the moment (with the original ATI card back.)

Yes, you're right of course, Cyker won't even get to Safe Mode if he's having the sort of problems I was having before I installed your patch!

[Cyker]

Had a bit of a scare after I re-imaged my '98 partition and it was still VMM/IOS BSOD'ing, and I ended up having to manually scour the registry and inf etc. directories to eradicate driver for all gfx cards (My old 6600GT, X850 and 9200!) then reinstall just the Radeon to get it to boot again and that seems to have done the trick...!

@Dave-H - I think you should be able to get a refund; It's not like you tried to do something weird with the card (Trying to insert a Molex plug the wrong way is actually very difficult! )

TBH, the card is just a no-effort-even-attempted clone of the nVidia reference board, and I bet the QA wasn't great so I'd e more inclined to think it was the card, especially if your card was packed the same way as mine... (Mine was just in a plain box and wrapped in bubblewrap! Seriously... Bubblewrap!? It's only one of the worst things for static buildup!! Wasn't even in a static bag!)

To be fair tho', my card has been pretty solid and quite awesome (Albeit not in Win98! ), even in this heat

As for a replacement, if you don't mind using eBay there are loads of 7950GTs in Hong Kong!

I do wonder whether I should try a 7800GS tho'; I just saw one for about £20 and I know they actually have official Win98 drivers too!

@rloew - Another oddity of my Win98 is that hacking MaxPhysPage and MaxFileCache, even to limit RAM to <512MB, still doesn't allow my Win98 to boot with >1GB RAM (This was my main reason for buying patchmem! )

I think I will take your advice and remove the /A tho' (And maybe the /M and the split8mb.exe too...), as I only put them in for testing. (Haven't done it yet mainly due to laziness )

Edited July 1, 2010 by Cyker

[dencorso]

Don't worry, that card isn't going back in the machine!

I can do some pictures if you like, but I've inspected it very closely myself, and can see no physical signs of any problem. No blackening anywhere, no sign of anything open or short circuited.

Please, do do post pics of it, Dave!

@Cyker: as soon as you have your system stable again, do PM me your machines specs. You do qualify for the list, and I'd gladly add you, provided you send me the required info.

[rloew]

@Dave-H - I think you should be able to get a refund; It's not like you tried to do something weird with the card (Trying to insert a Molex plug the wrong way is actually very difficult! )

Not really, but you get immediate feedback when you do.

@rloew - Another oddity of my Win98 is that hacking MaxPhysPage and MaxFileCache, even to limit RAM to <512MB, still doesn't allow my Win98 to boot with >1GB RAM (This was my main reason for buying patchmem! )

I think I will take your advice and remove the /A tho' (And maybe the /M and the split8mb.exe too...), as I only put them in for testing. (Haven't done it yet mainly due to laziness )

You probably need the /M Option. MaxPhysPage and MaxFileCache cannot fix this issue.

You probably don't need the SPLIT8MB.EXE Program as the /M Option usually is enough.

Rudolph, does that mean that running patchmem with the /A switch was un-necessary in my case?

I do need the /M switch to solve the large registry problem.

Is the /A switch likely to cause harm, and what should I look out for?

The nVidia card was working fine in Windows 98 before it blew up, but I'd already installed patchmem by then so I don't know if it would have worked without it anyway.

The system seems to be working fine at the moment (with the original ATI card back.)

@Dave-H

I added the /A Option to fix a potential problem I saw with ATI Cards.

Since it appears to be of no value with these NVidia Cards I would not use it.

It alters Memory Allocation so it might cause exhaustion of the Shared Memory Arena. It also may cause a problem if it moves a Memory Allocation that must be in the System Area. This is why it is an Option, not a built in feature.

[Dave-H]

@Dave-H

I added the /A Option to fix a potential problem I saw with ATI Cards.

Since it appears to be of no value with these NVidia Cards I would not use it.

It alters Memory Allocation so it might cause exhaustion of the Shared Memory Arena. It also may cause a problem if it moves a Memory Allocation that must be in the System Area. This is why it is an Option, not a built in feature.

Thanks Rudolph.

I will try uninstalling patchmem and then reinstalling it with the /M option only.

When I get a replacement card I'll try it like that and see if it still works.

[Dave-H]

Please, do do post pics of it, Dave!

Your wish is my command!

The first one shows the fried up power connector.

The yellow wire is actually melted onto the adjacent wire!

I managed to break it trying to remove the connector from the card.

I think the connector had welded itself on that pin!