A faster IDE controller will do nothing for you as long as the data bus feeding the controller is limited to 133MHz (or 66MHz in even older machines.) The slowest link controls the speed of transfer, period. In those days we sped things up by using techniques like DMA to offload tasks from the processor onto the controllers. Even DMA is limited to the data rate of the bus, though.
As for old solid state stuff, the RAM for the 16 Pentium 3 systems I have here is much cheaper than newer RAM modules required by newer machines. Sure, it doesn't work in those newer machines, why should it? But in the machines for which it was designed, it works just fine. The price of 512MB of RAM for these machines is less than the original price of the 64MB that came in the machines back when they were made. I don't see that as bad, but rather as an advantage. Memory always gets cheaper over time. Back in about 1982 I paid $100 for 16KB of RAM and it was a good price. (Yes, KB, not MB.) I worked on an IBM mainframe system that had only about 128KB of real memory itself, but had very impressive virtual memory capabilities and speed so that any of the hundred or more processes it was running at a given time might have access to up to 2GB, which was amazingly huge back then.
Nowadays that $100 will buy a GB or more of RAM, but the performance you get for it is all eaten up by wasteful code that spends most of its time painting pretty pictures rather than delivering the useful data that is really needed.
no subject
Date: 2009-10-22 04:12 pm (UTC)As for old solid state stuff, the RAM for the 16 Pentium 3 systems I have here is much cheaper than newer RAM modules required by newer machines. Sure, it doesn't work in those newer machines, why should it? But in the machines for which it was designed, it works just fine. The price of 512MB of RAM for these machines is less than the original price of the 64MB that came in the machines back when they were made. I don't see that as bad, but rather as an advantage. Memory always gets cheaper over time. Back in about 1982 I paid $100 for 16KB of RAM and it was a good price. (Yes, KB, not MB.) I worked on an IBM mainframe system that had only about 128KB of real memory itself, but had very impressive virtual memory capabilities and speed so that any of the hundred or more processes it was running at a given time might have access to up to 2GB, which was amazingly huge back then.
Nowadays that $100 will buy a GB or more of RAM, but the performance you get for it is all eaten up by wasteful code that spends most of its time painting pretty pictures rather than delivering the useful data that is really needed.