• Please review our updated Terms and Rules here

ISA Video card strange sparkly artifacts

Yeah I tried swapping the two socketted ram chips first.

Just tried all 4 combinations of jumper settings, no effect.
 
maxtherabbit said:
Sorry, I meant "replacing" them not just reversing their position which you said you tried already
Click to expand...
I guess I can try that next, I spent some time yesterday desoldering and socketting the two ram chips in my OAK card so I can try popping those in. If that doesn't change anything I guess the next thing would be to desolder and socket the other two chips in the et4000
 
Replacing the two socketted chips didn't affect anything. I'm trying to think of anything else I can try before desoldering the other two chips.
 
I suspect you are seeing DAC snow. Some early VGA cards had snow if the palette was changed without waiting for vertical retrace. I don't think there is any hardware fix.
 
Plasma said:
I suspect you are seeing DAC snow. Some early VGA cards had snow if the palette was changed without waiting for vertical retrace. I don't think there is any hardware fix.
Click to expand...

I second that it's either this or some chipset-level issue with handling memory contention that's causing an effect similar to CGA snow, I doubt changing out the RAM chips is going to do anything. (Since the RAM is apparently passing a memory test if it is memory contention it's in favor of CPU access vs. screen refresh, so from the computer's standpoint the card isn't "broken". A problem like this would not be the RAM chip's fault.)

Since it apparently happens when "a scene is loading" the DAC snow theory sounds right to me, since a lot of games did the thing where they'd load a scene and fade it in by hammering on the DAC.
 
Hmm okay. It did seem highly unlikely that vram was the cause especially after passing ram tests.

What about 16 bit bus access? Could my xt-ide bios be hurting performance?
I guess I could take another hard drive, partition it at 500mb, put something like doom on it and see if it runs faster without the xt-ide bios installed.
 
Divarin said:
What about 16 bit bus access? Could my xt-ide bios be hurting performance?
Click to expand...

The only thing that should be affected by having the XT-IDE BIOS on the bus is access to the VGA BIOS, and if you have video shadowing enabled then even that should be a non-issue.

(You could try benchmarking the system with BIOS shadowing enabled and disabled to make sure the shadowing is working, and then if you're curious you could disable shadowing and run the same benchmark with and without the NIC card installed, but the only things that really make much use of the BIOS are simple DOS text programs. Graphics programs might use the BIOS for mode switching, but after that practically nothing else.)
 
Also if it works ok with the 0WAIT jumper installed, then leave that on since it should be faster.
 
I'm highly skeptical of the VGA snow explanation. Yes the viewpoint is an early ET4k card but I've never seen this behaviour before
 
Well I could desolder and socket the 2 chips and try swapping them out just so I can prove without a doubt that it's not the ram. If so then I could ask the seller if he'd exchange it for another since he has more of the same.
Of course there's always a danger when desoldering, I could overheat the chips or something
 
Well desoldering a PLCC DAC with a hot air tool is arguably more dangerous than zipping out 2 DIPs. I still suspect memory
 
I doubt the seller will take it back once you've modified it. If you have the option of exchange (or return) I would go that route.
 
maxtherabbit said:
I still suspect memory
Click to expand...

The $50,000 question is if those artifacts stick where they show up. If they look like "snow", IE, transient flickers with only the duration of a frame then RAM being bad seems really unlikely to me. Is the theory that data is getting corrupted when it's written to the card (which would imply the artifacts would be stable), or it's intermittently corrupting data read back out, but only when it's the video hardware doing it (because it apparently passed a RAM test)?
 
maxtherabbit said:
I'm highly skeptical of the VGA snow explanation. Yes the viewpoint is an early ET4k card but I've never seen this behaviour before
Click to expand...

I have seen VGA DAC snow a few times but I don't remember the specific cards. Some references:

https://www.phatcode.net/res/224/files/html/ch34/34-01.html
http://www.osdever.net/FreeVGA/vga/vgadac.htm#flicker
http://qzx.com/pc-gpe/vgaregs.txt
http://www.nextcomputers.org/NeXTfiles/Docs/Software/OPENSTEP/vbe20.pdf - page 9
http://www.bitsavers.org/components/s3/DB019-B_ViRGE_Integrated_3D_Accelerator_Aug1996.pdf - page 178

It would be easy enough to test by reading/writing the DAC in a continuous loop without waiting for retrace.
 
Eudimorphodon said:
The $50,000 question is if those artifacts stick where they show up. If they look like "snow", IE, transient flickers with only the duration of a frame then RAM being bad seems really unlikely to me. Is the theory that data is getting corrupted when it's written to the card (which would imply the artifacts would be stable), or it's intermittently corrupting data read back out, but only when it's the video hardware doing it (because it apparently passed a RAM test)?
Click to expand...
I've seen almost this exact behaviour with an IBM EGA that had bad 4x16 chips soldered to it. It also passed the checkit VRAM test
 
You must log in or register to reply here.
Back
Top