I originally had 1x2Gb (Corsair) in my X-5000 and after reading the recommandation from Kas1e regarding his latest ports I replace it with 2x1Gb of Kingston which were laying around and after running Ragemem, I am getting the expected number for write32/64 (around 1'500)
I did this some years back and even filed a bug report about it (long lost and forgotten, as it seems, probably time to at least file it at AmigaKit).
I currently have 2x4GB modules in bay 0 and 1 (i think), the ones on the right side of the CPU.
My findings from back then were that it doesn't seem to be either of the RAM controllers (both work if used solely) but interleaving between them two. It looks like that interleaving was never really tested when the hardware were designed/developed (the early stages) and when it came to testing with the OS it seems none of the betatesters/developers thought of populating more than 2 of the RAM sockets.
Be it as it may, i tried with 16 GB (4x4GB) back then and the system went into suicide mode...even *IF* i made it into WB it would take only some minutes before it came crashing down frantically (no more warm, no more cold reboot possible, system had to be shut down completely for at least ten minutes before it would ponder about coming back to life).
I don't think it was the RAM modules either (because all four of them were tested in every possible combination, with always the same outcome...populate two differing sockets and prepare to crash)
For me interleaving working at least with two modules. I have modules in sockets DIMM1 and DIMM3 ( first left and first right from CPU ). Only four modules makes troubles.
I tested it few years ago, so I am not sure if it works the same way with sockets DIMM2 and DIMM4 ( far left and far right from CPU ). I will test it and also will check speed without interleaving ( one module, or both modules on one CPU side)
And I have the same modules types, but they not marked like "matched pair". Maybe this can matters too.
AmigaOS3: Amiga 1200 AmigaOS4: Micro A1-C, AmigaOne XE, Pegasos II, Sam440ep, Sam440ep-flex, AmigaOne X1000 MorphOS: Efika 5200b, Pegasos I, Pegasos II, Powerbook, Mac Mini, iMac, Powermac Quad
I "works for you" in the way that you hvae your daily WB running and do a lot of stuff with it without problems, or it "runs whenyou put them in" but not really tested heavily?
I don't think you need "matching pairs", as long as both work in matching ram sockets (0 and 1, 2 and 3) they should work interleaved as well, of not there is something broken or unsupported in the RAM handling.
This CAS 10 thing. Can someone help me because before I buy these modules I want to make sure that I have the correct ones and not miss thr chance to get better ones.
If liberty means anything at all, it means the right to tell people what they do not want to hear. George Orwell.
@rjd324 Why not go that route : just check the models we point out as best tested in thread i quote, and buy then exactly the same with the same markers ?
DRAM: Initializing....using SPD Detected UDIMM AFLD34AN1P There is no rank on CS0 for controller 1. Not enough bank(chip-select) for CS0+CS1 on controller 1, interleaving disabled!
The difference is not that much. In that case, the single AFLD34AN1P that I had was quite good, then.