Hi Ariel!
Yea, I suspected as much, but I tried to see what they do in io.jl and 
*thought* they too had a for loop… Not so sure now.

On Thursday, August 13, 2015 at 7:23:05 PM UTC+10, Ariel Keselman wrote:

in the first version you read a single large chunk while in the second you 
> read many small chunks.
>
> I think reading many small chunks is much slower due to how disk IO works. 
> First it has to look for the data, and second, read only a small chunk 
> while it could read a larger chunk using the same time. 
>
> I mean, disks (I think SSDs too) are optimized for larger reads. Also 
> reading one chunk would spare the many lookup delays.
>
> Seems this IO penalty is large compared to the cost of allocating the data 
> variable...
>
​

Reply via email to