The cache chips are in the top-right corner which is fake. You have to replace them with functional SRAM chips.

https://theretroweb.com/motherboards/s/octek- … -1-486-rev.-1.0

dominusprog wrote on 2024-09-12, 13:25:

The cache chips are in the top-right corner which is fake. You have to replace them with functional SRAM chips.

https://theretroweb.com/motherboards/s/octek- … -1-486-rev.-1.0

Yeah, I already got to that point. 😀

Are you sure those are cache chips? Because this board has none at all, you need to supply them via RAM, and they can only be 15ns.
Could probably try making a RAM PCB that accepts 15ns memory chips, been diving into PCB making recently, so it's worth a shot.

Czeky wrote on 2024-09-12, 08:59:

Apparently this board can use external cache only thru RAM memory, and it can only be 15ns, it has no cache sockets on board at all.

The manual only mentions "Dynamic Cache Architecture", which is a write buffer and not a proper cache (it does nothing for reads, for instance).

Czeky wrote on 2024-09-12, 08:59:

I've also been wondering if it would be possible to give any 486 system external cache via the ISA / VLB bus? It might be slow on ISA, but what about VLB, and is it even possible to make an external cache card (even if it requires a separate driver in DOS)? If so, then many people who have 486 boards with no or fake cache, and have no soldering experience could benefit from this.

No, the ISA/VLB buses essentially extended the CPU's address and data lines to the expansion slots. You can add extra memory that way (this was in fact required on the 5150 PC as it only supported 64KB or 256KB on the motherboard) but not cache. The memory controller has to support caching and manage cache memory (looking up memory reads in cache first, retiring cache lines, etc.) and the datasheet for Headland HTK340 chipset used on this board does not mention support for a cache, it only mentions the write buffer.

It does mention a Headland HT44 secondary cache and that is mentioned in this thread about a Dell 486 that does have cache sockets for a board with that cache controller. The Octek board doesn't appear to have this cache socket I think you're out of luck.

Octek DCA, there is a thread about it Octek DCA2 Dynamic Cache Architecture 2, 486 Boards it requires special snowflake unobtanium EDRAM with only source being rare Mylex DAC960 RAID controllers.

rasz_pl wrote on 2024-09-13, 06:52:

Octek DCA, there is a thread about it Octek DCA2 Dynamic Cache Architecture 2, 486 Boards it requires special snowflake unobtanium EDRAM with only source being rare Mylex DAC960 RAID controllers.

Thanks for the thread, that PCB image with the desoldered memory chips is super useful, though I might ask him if he has any more images of that RAM stick.

Even if my board does not support these 15ns RAMs, it should still be useful for ppl who have DCA2 boards. Just need to find alternative memory chips to those RAMTRON branded ones...

rasz_pl wrote on 2024-09-13, 06:52:

Octek DCA, there is a thread about it Octek DCA2 Dynamic Cache Architecture 2, 486 Boards it requires special snowflake unobtanium EDRAM with only source being rare Mylex DAC960 RAID controllers.

The other source are a very specific, equally unobtanium Transputer TRAM module produced by Transtech. These contained an Inmos transputer CPU and a variable amount of "ramtron" branded edram chips effectively giving the normally cache-less transputer CPU a cache.

I have exactly one of these modules. They are very rare.