-
Hello MLAers! We've re-enabled auto-approval for accounts. If you are still waiting on account approval, please check this thread for more information.
AE TW2340 (TW1340) Accelerators - Maximum RAM
- Thread starter superjer2000
- Start date
Thanks @zigzagjoe. I ordered a few of those 030's and should have them in a couple of days.
That being said, the purpose of this exercise was to see if I could coerce my TW2340 accelerator to show more than 4MB of RAM. I have now found somebody with the same accelerator (well, the SE version) that does support > 4MB, and his processor is still an 68EC030. He sent me pictures of his card and I compared it to a card that is only recognizing 4MB, and the only difference appear to be:
* A CPLD with updated programming (i.e. an AE v1.2 sticker versus a v1.1 sticker); and
* What appears to be a rev B. ROM.
The CPLD is a Lattice pLSI 1024-60LJ
The EEPROM is 27C256
This makes sense to me as back in the day, it seems AE shipped out some of these cards with 4MB support but then had owners mail the cards back for an update to support >4MB.
68kmla accelerator gurus @zigzagjoe and @Bolle, if I am able to get a card with an updated CPLD/EEPROM, what are the chances I would be able to clone them to upgrade my current card? I'm assuming the EEPROM would be no issue, but my guess is AE would have set the security fuse for the CPLD. If that's the case, is it pretty much impossible to make a clone (assuming I could find a suitable CPLD programmer)?
Thanks!
You have the right of it. You would require an attack/backdoor to bypass the security in order to dump the code. WIthout that, you'd be left trying to poke at inputs and see what comes out... with something this complex you may as well be designing it anew.
That is interesting. I'm looking into options to see if I can buy an updated TW1340 in hopes that I can clone the CPLD and EEPROM.
In other news, I received the full 68030 chips (thanks for the source @zigzagjoe. I am a pretty proficient solderer but wow, that QFP package was a pain.
I need to do some further inspection on my soldering as, while the computer booted fine there were a few things that were odd:
* It is running much slower than with the 40mhz EC030: Speedometer shows about 7x whereas before it was about 12x versus a Classic. More noticeable is graphics performance is WAY slower. Down to about 2x vs 6x before
* Speedometer and the AE control panel are not showing an 030 MMU, but rather an 020 PMMU. Virtual no longer complains about the lack of an MMU but the computer crashes on boot, with a garbled screen and a weird sound playing through the speakers when Virtual is turned on (it boots fine with Virtual turned off)
* The FPU is no longer showing as available. I am wondering if this is a soldering issue and also if this could be what is driving the lower Speedometer scores (especially for graphics) if possibly patching some of the SANE routines through then FPU would increase graphics performance? Just a guess here. Note that the graphics performance is really noticeable with scrolling, with the accelerator unaltered windows would scroll insanely fast. Now it's maybe 2x faster than a stock Classic.
It looks like the reduced speed was related to the FPU not being recognized. I went back over my soldering and the FPU is now being recognized and the accelerator is working and benchmarking at full speed. I can only assume the driver is making use of the FPU to speed up regular operations, including scrolling etc.
The MMU is still a no-go (not recognized generally and Virtual crashes) - I think Bolle is right that the MMU isn't being set up. My guess is this accelerator is designed to support extra memory through some dark magic that doesn't rely on Virtual or an MMU which is aligned with the marketing materials. I'm trying to see if I can buy an accelerator that IS supporting the extra RAM. I am aware of a couple in the wild - one I received pictures of still had the EC MMU-less processor.
The MMU is still a no-go (not recognized generally and Virtual crashes) - I think Bolle is right that the MMU isn't being set up. My guess is this accelerator is designed to support extra memory through some dark magic that doesn't rely on Virtual or an MMU which is aligned with the marketing materials. I'm trying to see if I can buy an accelerator that IS supporting the extra RAM. I am aware of a couple in the wild - one I received pictures of still had the EC MMU-less processor.
Similar threads
