@ella: Yes I'm using 2 SSDs in parallel in the server. The SSD can do, I think, 
40k IOPS. The algorithm just makes a fairly big number of random accesses for 
every search. Some optimization can be done there, but it will always be 
painful in some cases. So putting it all in RAM seems like the way to go.

I have gotten the index for the 72 GB of data down to ~100 GB, just need to fix 
the last bugs for the new format.

I can't get AWS for lack of a credit card, but I think I don't need to. Regular 
dedicated servers are probably better for this application anyway.

@immortal.discoveries: Are you asking this as a killer question or an actual 
question? I don't respond to killer questions :)

@James: What good is a Google search for a bot? Google charges for every 
automated access. Also it's not the full search I want.
------------------------------------------
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T6322565b7d29a2a0-M64291be25f5dddb81cacb392
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Reply via email to