|
Post by Syniphas on Mar 26, 2010 11:25:35 GMT -5
Still nothing on my 486, I suspect it's the motherboard...
|
|
|
Post by Tom Maneiro on Mar 26, 2010 12:33:33 GMT -5
Maybe a fried CPU or rotten RAM... There is a law on those old motherboards regarding RAM: if it uses 72-pin SIMMs, it's very difficult to get the memory placed correctly at the first try. Actually, you may need a few tries, a cleaning brush, and some TNT. Also, 486 don't even bother to boot if you're using EDO RAM. Stick to FPM modules. Heh, i should try those spare 32M FPM's on my 486DX board... but there is no fun on a 486DX-50 Gah, i should have lifted some DX4-100's from the university stash
|
|
|
Post by Syniphas on Apr 4, 2010 19:51:13 GMT -5
I've tested 3 different CPUs, and two of them are supposedly working (a friend tested, said they were working but I didn't get to see it)
This computer was working before my aunt upgraded to a Pentium, then I got the motherboard from her and bought a SB16 to go with it. Now I can't get it to work at all. Maybe you guys could help me as I have little to no knowledge with 486 motherboard, just modern ones...
|
|
|
Post by Tom Maneiro on Apr 4, 2010 23:22:43 GMT -5
PROTIPS WHEN DEALING WITH ANCIENT BOARDS: - Check the jumpers - older boards are damn picky!. No-name boards tend to have a gazillon of jumpers to set the CPU type and speed (SX, DX, multipliers, RAM speed, blah blah blah). Kustom big-brand boards (read: IBM/HP evil crap) have those jumpers... often unlabeled on the board, but on the PC case (do you still have the original case, right?), or some obscure page inside the manufacturer website/FTP. - Give the board a good cleanup (use your brush army, be careful but don't be too gentle), and some washing with alcohol - RAM... As i've said earlier, avoid big and EDO sticks. If the stick is 16 MB or bigger, chances are that it's EDO, and those won't work on 386. Even if you get fat FPM sticks, some boards may fail to boot. Use 4 or 8MB for being in the safe side - Winblows 95 is happy with 32MB on 4x8MB setups - Check for loose IC/jumpers. - Try another PSU... - Do your board use one of those dreaded DS128(8)7 bricks? It may have a dead battery, and some boards fail to boot if the RTC has bogus data due to a dead battery, so you will need to replace the module. If your board uses an standard button-cell battery, try replacing it.
|
|
dave
Moldy Popcorn
You can't beat the raw processing power of a Tualatin PIII!
Posts: 26
|
Post by dave on Apr 5, 2010 19:19:59 GMT -5
It's been ages since i've been here! Hope everyone had a good easter! Another thing with 486 boards (I know it sounds rather silly, but i've done it before and the CPU worked after putting it in the right spot) check pin 1 is in the right spot, because these old things aren't keyed (only god knows why )
|
|
|
Post by Syniphas on Apr 10, 2010 16:36:18 GMT -5
Tom, the processors I was using were the ones that were working with the board before it was taken apart. All jumpers are set up to a 486 DX2, which is also the Cyrix processor I have right now.
Other jumpers are confusing and most of the time unlabeled, such as the speaker jumper as well (which I haven't been able to get a beep out of at all).
My setup is basically an unbranded 486 board, a Cyrix 486 DX2, 16mb (2x8mb, I have no idea what kind of RAM it is, the computer ran on these so they should work), a SB16, 1mb Trident video card and a nondescript serial card for COM1 (mouse), LPT1, HDD and floppy (none of which are onboard)
|
|
|
Post by Tom Maneiro on Apr 10, 2010 23:46:33 GMT -5
Your sticks are 8MB, and you said that the board used to work with these, so they're FPM. However, try with other RAM modules - your current sticks could be fried (yes, i've experienced that crap - noname RAM modules are sometimes pure crap).
Some of the other parts could be dead too - try booting without the I/O port card. And sometimes, motherboards are plain funky, and refuse to work for no good reason...
|
|
|
Post by Syniphas on Apr 11, 2010 16:34:13 GMT -5
Yeah I don't have any other memory sticks, these were working just fine until my Aunt upgraded her computer and this was disassembled. I just can't seem to put it back together in a working state...
I probably fried the processor already with the confusing socket plus jumpers and whatnot but I can't be suuuuuuure
|
|
|
Post by GiGaBiTe on May 2, 2010 4:15:24 GMT -5
DX4-100's from the university stash I have 3 x Am5x86 P75 133 MHz processors, I could let go of one if you really needed it. You'd have to pay shipping, etc. though.
|
|
|
Post by Tom Maneiro on Jul 30, 2010 17:48:29 GMT -5
Duke Nukem 3D testing in progress.... Unfortunately i had to repair the damn Wintendo Registry first - the joys of a Win9x korrupt registry that refuses to load... But this is 2010, we have USB and Internet. This MS KB article saved my ass, again: support.microsoft.com/kb/131431Now, wish me some luck... UPDATE: Experiment result: Total expected failure Before booting this time, Windows 95 decided to corrupt the registry, and i was stuck in a weird loop of "Not enough memory for continue booting Windows" and "Windows found a error while doing a registry backup, can't backup your 140KB USER.DAT because i think that 85MB != enough free space, enjoy your plain green Windows desktop" error messages. Finally, after reading that MS KB article, here is how you repair a fuxxored registry on Win95: Having healed Marika's registry, let's advance with the Duke Nukem 3D testing (i'm using the shareware version for space reasons). The manual says that it should run (slowly) on anything below a 486DX-66, but 8MB of RAM and VGA graphics are a must. However, in Win95, you could fool memory checks thanks to swap. But... "invalid opcode"?! Looks like 3DRealms is using some 486+ instruction This happens under pure DOS - no swap, no RAM cheats: If someday i get 4MB SIMMs, i will perform the test again, to confirm if Duke is really using some 486+ instruction - if that's true, we're DOOMed
|
|
|
Post by theelf on Aug 3, 2010 15:50:30 GMT -5
Hi tom! i have a 386DX40 with 32mb (4mb simm x8), a SVGA ATI 1MB I can´t run Duke3D in msdos/win9x But in Win9x, i can boot xDuke a windows port vision.gel.ulaval.ca/~klein/duke3d/, very slow, and the new version hangs... NOTE: I have a old 386DX laptop, without sound. One day i want to play doom with music.... then i make a primitive Parallel Sound Card, and in Linux, using the LPT sound card in kernel, i play Doom with music using timidity++.... Slow like hell... but works.... Maybe you cna play duke in linux.... ?¿ In any case.... Doom forever... ;D
|
|
|
Post by GiGaBiTe on Aug 7, 2010 5:04:01 GMT -5
Also, 486 don't even bother to boot if you're using EDO RAM. Stick to FPM modules. Every x86 Intel CPU up until the i3/i5/i7 series has had the memory controller integrated into the north bridge, processors are irrelevant to the type of memory. All that matters is if the memory controller on the NB supports FPM or EDO memory. "invalid opcode"?! Looks like 3DRealms is using some 486+ instruction This happens under pure DOS - no swap, no RAM cheats: If someday i get 4MB SIMMs, i will perform the test again, to confirm if Duke is really using some 486+ instruction - if that's true, we're DOOMed The primary purpose of pagefile is to swap unused chunks of main memory out to disk, in order to free main memory for other applications. If the machine doesn't have enough physical memory to fit the entirety of X application, swapfile won't help. If an application loads part of itself in main memory and part of itself in swap, then proceeds to try and access a virtual memory location in swap, it will result in a segmentation fault due to not being able to load the virtual address into main memory (which is what D3D is doing.) If you install at least 8-16MB in the machine, duke should run, just really slowly. The 80486 only added a few additional instructions over the 80386, and I really doubt 3drealms used any of them. The D3D engine code was written on a 386 in Watcom and was probably compiled as such, and shouldn't have a problem.
|
|
|
Post by Syniphas on Aug 7, 2010 9:54:53 GMT -5
Man, now I have all kinds of old RAM since that trip to that computer store with piles of old stuff (I think I have about 32mb in 4mb sticks), wish I could get Tom some.
|
|
|
Post by Tom Maneiro on Aug 7, 2010 12:56:53 GMT -5
Also, 486 don't even bother to boot if you're using EDO RAM. Stick to FPM modules. Every x86 Intel CPU up until the i3/i5/i7 series has had the memory controller integrated into the north bridge, processors are irrelevant to the type of memory. All that matters is if the memory controller on the NB supports FPM or EDO memory. And here is the detail - i don't know about any consumer 486 motherboard with a EDO-capable northbridge (perharps some server motherboard does EDO?), so any normal user assume that "486 isn't compatible with EDO RAM". Back then there were like a trillon of chipset brands and models, most of them with zero documentation. And for the swap and D3D... Well, i guess so, but that "invalid opcode" error is a bit misleading, i would have expected it to say "segmentation fault" or something. But you're right, swap is not RAM
|
|
|
Post by GiGaBiTe on Aug 7, 2010 16:23:43 GMT -5
And here is the detail - i don't know about any consumer 486 motherboard with a EDO-capable northbridge (perharps some server motherboard does EDO?), so any normal user assume that "486 isn't compatible with EDO RAM". Back then there were like a trillon of chipset brands and models, most of them with zero documentation. And for the swap and D3D... Well, i guess so, but that "invalid opcode" error is a bit misleading, i would have expected it to say "segmentation fault" or something. But you're right, swap is not RAM Earlier 486 motherboards probably don't support EDO memory, but the later ones definitely do. I have some 64 MB 72 pin EDO SIMMs and they've worked in several 486 boards I've had in the past fine. And like I said in one of my other posts, I have some Am5x86 133's and 4 x 4MB 30 pin SIMMs if you'd like to buy them off me (I'd only charge what it costs to ship them.)
|
|