skafen wrote on 2026-06-08, 11:14:

Congratulations that's great news. So all you had to do was bridge R103, R109, R102, R88, R86, R108 then? Did you use resistors?

What are your jumper positions for 512K on J14 and 15?

Thanks.

Yes, only bridge R103, R109, R102, R88, R86, R108. I use 22Ω resistors to bridge it. My SRAM chips' latency is 15ns, it cannot run 2-1-1-1 at 50MHz, do you think it is normal?
J14 pin3 goes to cache A16 and HA18, pin2 goes to tag A14, pin1 goes to GND. J15 pin3 goes to cache A15 and HA17, pin2 goes to tag a13, pin1 goes to VCC. So the jumpers are useful only when you have tag chip bigger than needed. So, 512K is J14: 2-3 J15:2-3 256K is J14: 1-2 J15: 2-3 128K is J14: 1-2 J15:1-2
(the side with triangle mark is pin1)
It looks like J8 only decides whether pci clock is passing through an extra buffer. That's very strange to me, do you have any idea about that?

Maybe you need the 373 chips for signal integrity at 50MHz but looking at this post:- 486 Cache size and speed questions seems to indicate that 2-1-1-1 timing is only guaranteed up to 33MHz with 15ns so you might be unlucky with your particular cache chips not able to go any faster.

Not sure on J8 at all.

Thanks for the jumper info. I'm hoping to get mine tomorrow or Wednesday and try just bridging the resistors as I don't have any spare SMT resistors.

skafen wrote on 2026-06-08, 11:59:

Maybe you need the 373 chips for signal integrity at 50MHz but looking at this post:- 486 Cache size and speed questions seems to indicate that 2-1-1-1 timing is only guaranteed up to 33MHz with 15ns so you might be unlucky with your particular cache chips not able to go any faster.

Not sure on J8 at all.

Thanks for the jumper info. I'm hoping to get mine tomorrow or Wednesday and try just bridging the resistors as I don't have any spare SMT resistors.

Let's see if your board can run the cache at 2-1-1-1 at 50MHz. I'm too lazy to desolder these resistor networks. According to this post, it seems fair that my 15ns chip can reach 2-1-1-1 at maximum 40MHz bus speed.

Actually I'm wondering how you figured out it is not the BIOS that disabled L2 cache, and if BIOS did that, what can we do?

Ah I have experience in that matter... PCChips M912 V1.7 Fake to Real Cache and new cache not recognised. I Suspect Bios version. Check out my post here with a PCChips M912 - I've done a couple of these now with Fake to Real Cache. Now the post refers to the bios disabling L2 cache specifically. But on this Amaquest the J446 bios I'm using does NOT disable the cache at all will attempt to use it as long as you don't disable it yourself in the bios, hence the crashing when the chips are in the board without the resistors.

I think only PCChips were mean enough to attempt to actually disable it completely and other manufacturers just have a routing in the bios that says always display 256K cache. My other PCChips is like that and also an Azza made board. If you put real cache in e.g. 128k then they read correctly.

I managed to get access to mine today and fiddle with it to no success. I bridged the six resistors fine but the board still either crashes or doesn't recognise the cache.

I've got a set of 32K chips for 128K and a set of 64K chips for 256K and a couple of extra 32k for tags. Neither set will work they crash just after the hard disk detection screen or on the post summary screen depending on what I push/wiggle. So I am not 100% happy with the cheap socket for the tag at least if not the cache too. Although it does correctly detect if 128 or 256k is installed and show it correctly on the summary screen.

The cache chips are tested too both in another board and my TL866-III Eeprom programmer.

I've got no more sockets in currently and actually no desire to meddle with it further at the moment so think it can just go back in to storage for now.

Winnt5, I am glad that you got yours going though. Have fun with it.

Sad to hear that. Maybe try testing if all the signals are connected well and if 74f373 is working? Here are some pictures of my ap8548 board with 512K cache.

Very nice. Here's the figures from my own X5-133 chip on an Asus PVI-486SP3. Looks like yours is tuned very well.

SpeedSYS CPU Score 49.80
SpeedSYS L1 Cache Throughput (MB/s) 117.27
SpeedSYS L2 Cache Throughput (MB/s) 47.78
SpeedSYS RAM Throughput (MB/s) 35.73
SpeedSYS GPU RAM Throughput (KB/s) 15171