That's the later revision, D and E should be able to only cache up to 128mb. The G revision should do 512mb.

I wonder if it works with k6-2+ and k6-3+ mobiles. It should work but can you test this?

meljor wrote:

That's the later revision, D and E should be able to only cache up to 128mb. The G revision should do 512mb.

I wonder if it works with k6-2+ and k6-3+ mobiles. It should work but can you test this?

That's correct, rev. G has almost all bugs fixed. There should be no tag SRAM chip installed on such mainboards as the internal tag SRAM of M1542 works properly. Therefore a 512Mb caching limit and better overclocking.

You can get a K6-2+ or K6-3+ running after flashing the latest 3.0 BIOS. Although it probably needs some better capacitors as well.

http://web.inter.nl.net/hcc/J.Steunebrink/k6plus.htm

All Capacitors have already replaced 😀 K6-2+ works on this Mainboard.

But SD-Ram combatiblity is poor.
I tested many many RAM Modules until i found a Pair that works. A 256Mb Double Side MDT PC133 and a 64Mb PC100 Single Sided.

Current View:

Configuration:
K6-2 450Mhz
320MB Ram
Riva 128ZX 8Mb (For testing purposes)
Intel Ethernet 100 Pro
3ware 9500 SATA Controller - PCI 64bit @ 32bit with 128Mb PC133 SO-DIMM installed.
40Gb Intel 320 Series SSD
ARGUS Prototype Nr. 14 ISA
Tandy Soundcard Prototype r1 ISA

OS:
Slackware 14.2 with Xorg and Fluxbox
Windows 98 / MS-DOS

Stupid.. weird problems again.

L2 Cache disabled, and memtest runs ok..

But is L2 Cache defective or memory controller ?
someway to figure this out ?

What comes to my S7AX, I have replaced the big caps near the CPU socket and have no issues with memory compatibility. There were 4 no-name STONE 1000uF caps and one empty place. Junk pretty much. Put in 5 Rubycon MBZ 2200uF/6.3V. That's 3 of them on Vcore and 2 on +3.5V which goes to Vdd, chipset, cache chips, memory and the AGP slot.

Commate screwed up by implementing +3.5V using an LDO rather than PWM and not putting a large heat spreader on the regulating MOSFET located near the IDE connectors. That's 70% efficiency only, so a lot of heat goes straight into a PCB. Two double sided SDRAM modules, then an AGP video card with no aux power supply and this MOSFET will burn through the PCB. I think I'll replace it with a DC-DC PWM converter capable of 85% to 90% efficiency.

Got it fixed, the ALI M1542 BGA had broken solder joints.. i used a lot of flux and a heatgun to reconnect it.. hopefully it will work forever 😉
And Memory Modules are running now 😀

All my Capacitors are new already.

Yeah the Voltage Regulator is really hot, there is a heatsink on them and Air from CPU Fan is cooling it down.

Does yours accept 512Mb Modules ?

matze79 wrote:

Does yours accept 512Mb Modules ?

I haven't tried it because the cacheable range is 512Mb using internal tag and valid/dirty bits, no external tag bits and a single 512Kb PBSRAM cache chip. The memory controller itself supports up to 1Gb per slot according to the spec, so 512Mb double ranked SDRAM modules with eight 256Mbit chips per rank should work.

My mb [chaintech ct-rsa2] also do tricks for example Radeon 9200se can reset pc, fx 5500/5700 irq problem [code12] geforce 440mx black screen after driver install, matrox g550 seams to work well but i cant run one 3d test from pc mark 04 to get a score, only Rade 128 works fine but its a bit to slow 😒

ph4nt0m wrote on 2018-10-09, 22:26:

matze79 wrote:

Does yours accept 512Mb Modules ?

I haven't tried it because the cacheable range is 512Mb using internal tag and valid/dirty bits, no external tag bits and a single 512Kb PBSRAM cache chip. The memory controller itself supports up to 1Gb per slot according to the spec, so 512Mb double ranked SDRAM modules with eight 256Mbit chips per rank should work.

Few questions left unanswered in the last posts of this thread that should be noted

You can indeed install up to 1.5gb on this chipset using three slots though some ram types may not work

1gb sticks while technically on paper should be supported don’t seem to work, possibly due to a bios glitch? It hangs at ram counting or aborts count at 512mb

If it aborts count at 512mb you can put that stick in slot 3 and install 256 mb sticks in slots 1 and two and the bios will count 1gb but if you put the 1gb in slot 1 it only counts 512mb if that makes sense. But this is only on motherboards that abort at 512mb the ones that count 1024 or even up to 2048 seem to just sit there after the count bot knowing what to do. So max is 1.5gb for now unless someone can find a patch. I assume 512, 512, 1024 would count to 1.5gb

As for cachable area. The use of a k6-3, k6-2+ or k6-3+ cpu of any model will circumvent this limitation even allowing caching up to the max supported ram of 1.5gb theoretically 3gb also if someone got it to work.

I’ve been having the crashing issue with a 9800 pro I removed the card though and will begin using a quadro2 soon but if I try it again I will try the riva tuner trick. Thanks for this post!

L2 Cache was also defective on my Board.

But Videocard problems still remained after Repair.
So i gave up on ALI Alladin and replaced it with HOT591P.

The BGA Fix didnt last long too.