3 ms·
That's interesting. It's also a concept that I thought of when I was an intern at Imprimis in 1989 (back before they got bought by Seagate). I was told that thi
by bradknowles 4y ago
That's interesting. It's also a concept that I thought of when I was an intern at Imprimis in 1989 (back before they got bought by Seagate). I was told that this concept was already in use in mainframe grade drives that Control Data built, but was too expensive to put in consumer 5.25" drives.
I'd be interested to see what has changed with this technology to make it more cost effective now.
- bradknowles 4y agoAnother idea I had come up with and mentioned at the time I was at Imprimis, was the idea of reading & writing from/to all heads simultaneously, as opposed to doing so in serial. Again, I was told that this concept was already in use on mainframe drives. I wonder if we will ever see it in consumer drives?
- rbanffy 4y agoI always assumed that, since all heads of an assembly are on the same point over the disk, they would transfer data from all the surfaces simultaneously. If drives don't do that I'd say they are cutting costs (because a n-platter drive would need (n-1)x the signal processing power of a single platter one. This drive behaves, and presents itself, as two drives and the head assemblies rotate on the same axis and each has access to half the platters. I would love if they managed to cram two full sets of 10 heads on opposite sides of the platters, but that would most likely confuse computers unless it's completely abstracted away by firmware.
- rbanffy 4y agoThis isn't the same stuff we saw on mainframe drives - those drives had two heads on the same platter, making it possible to read and write on different parts of the same cylinder at any given time. IBM made some of their HDA (head-disk assemblies) in a factory in Brazil not too far from where I lived.