what makes you think that someone soldered on the wrong ram. The card looks the same as one on vga museum. http://www.vgamuseum.info/index.php/cpu/item/ … logic-cl-gd5429

Surely thats not a coincidence.

Looks like the correct ram to me and it , according to you OP, is functioning as it should.

You've done similar posts like this in the past. Whiy?

Okay, I found this info on the current EDO DRAM installed.
It reads that the access time is from -25ns to -40ns so I think it should be okay for this video card.
http://pdf.datasheetcatalog.com/datasheets2/38/384008_1.pdf

Yeah, it's fine. The soldered on looks like 45ns and the socketed is 25ns so no problem there.

When it comes to latency, lower is better. 25ns is faster than 45ns. It's not listed on the VGA data sheet because those speeds didn't exist when the chip was released.Someone upgraded this card with much newer SOJs.

"correct" ram can differ on each card as well.
I once had two identical Trident 9000 ISA cards, same type number, same PCB and same components. Even serial numbers were not too far behind.

Both cards had 512kB in four 44256 chips, but that's where the similarity between ram ended. Brand was different, even timings were different (one had 80ns rated ram, other 60ns rated). Both were equally slow in benchmarks and I believe the 60ns chips were terribly underperforming. No harm in that though...

To me it always seems the card builders search for fairly cheap ram that matches the minimum requirements (or quality) and from there is just whatever they can get.

waterbeesje wrote on 2021-06-13, 13:51:

"correct" ram can differ on each card as well. I once had two identical Trident 9000 ISA cards, same type number, same PCB and s […]
Show full quote

"correct" ram can differ on each card as well.
I once had two identical Trident 9000 ISA cards, same type number, same PCB and same components. Even serial numbers were not too far behind.

Both cards had 512kB in four 44256 chips, but that's where the similarity between ram ended. Brand was different, even timings were different (one had 80ns rated ram, other 60ns rated). Both were equally slow in benchmarks and I believe the 60ns chips were terribly underperforming. No harm in that though...

To me it always seems the card builders search for fairly cheap ram that matches the minimum requirements (or quality) and from there is just whatever they can get.

Yes, that is why I was skeptical about the additional DRAM chips used on this card. But it turns out it is okay for this card but most likely will not work on all video cards. I would try to match the soldered on DRAM chips for compatibility.

Intel486dx33 wrote on 2021-06-13, 14:11:

[...]

Yes, that is why I was skeptical about the additional DRAM chips used on this card. But it turns out it is okay for this card but most likely will not work on all video cards. I would try to match the soldered on DRAM chips for compatibility.

Once again, lower = better.

"45" means the chips can take it if they get consecutive accesses 45ns apart. "25" means the chips can take it if they get consecutive accesses 25ns apart. 25 can take everything designed for 45ns and then some. The other way round would not work, putting 45ns chips in an application requiring 25ns.

Maybe you're more familiar with wiring terminology. You can use AWG24 for everything designed for AWG26, but not vice versa.

"Compatibility" 🤣