<<<<I'll probably go to 32, since 16 meg DIMMs are quite cheap, and then I also get the slight benefit of memory interleaving>>
Actually, from what I've seen, it's a big benefit. Almost, as good as adding a cache card. Ask Jason, interleaving worked on his machine.>>
I'd have to agree with Vincent here. When I added the correct DIMM to my machine, I went from 143->175 on the MacBench test (about 22% faster). Then I got another 10% or so by adding a 256K L2 cache (up to 199 on the MacBench). When we had the VMUGM meeting, we messed around putting Vince's L2 cache in my machine (before using memory interleaving), and it alone gave about the same increase as interleaving. I think they both give about 20-25% speed boost for the first addition, and then when you add the second, you get another 10-15% boost.
I've read some articles recently that are directly opposed to previous statements from most everybody in the industry in regards to large L2 caches. Apparently, at higher speeds and with more powerful processors (more branch-point and integer units), larger L2 caches DO give an appreciable advantage over smaller ones. For instance, the PowerTower Pro 200 and the PowerMac 9500/200 are basically identical in regards to CPU processing power. However, the PowerTower Pro outperforms the PMac 9500/200 by 15% or so on average. The only difference is that the PowerTower Pro uses a 1Mb cache vs. the 512Kb cache in the 9500.
Another thing: The PowerBase 180 BLOWS AWAY the Performa 6400/200, and the main reason being stated for this is the use of EDO DIMMS and a slightly enhanced memory bridge chip in the PowerBase (which provides the EDO functionality). Apparently many other manufacturers in the Mac clone business (and Apple) will be moving to EDO DIMMS over the next 6 months. All PCI Macs can use the EDO DIMMS, but only the PowerBase machines actually make use of the special features.
