Computer Specs for Calibre

[ilovejedd]

Just had the (dis)pleasure of using an AMD E-350 Zacate for sending 3,000 ebooks (a whole bunch of fanfics that need to be weeded out) to my PRS-350. I was hoping to log some reading time before going to bed. Alas, the whole process was taking so long that I fell asleep before it finished. Calibre also takes a while to check which books are on the PRS-350 whenever it's initially connected.

There's definitely a processor bottleneck with Zacate as it was pegged at 50% CPU usage meaning it's maxing out one core (not surprising when even a Pentium 4 beats it in terms of single-threaded performance). The question is, how high should I go on the CPU? I don't plan on doing any conversions with this PC. The only use I need it for is:

1. Transfer ebooks to PRS-350.
2. Generate epub catalogs of libraries.

My problem with Zacate is it's too underpowered. The epub catalog generation, while slow, wasn't completely unbearable on the E-350. It's really just transferring ebooks to the reader that's taking forever. The question is how much CPU do I need to speed that up before transfers become bottlenecked by limitations of USB2 or flash memory speed on the reader? Celeron G530 (~$50)? Core i3-2100 (~$110)? Core i5-3450 (~$200)?

[theducks]

RAM as much as the OS supports (the motherboard/processor may support more, but the OS is the limiting factor)
Calibre will use as much CPU (core) as it can get if needed. Keep those fins and fans clean

Each of those 3K books is getting its Metadata updated as it is sent which will eat time.
As you noted: Flash memory is not the fastest horse in the race

[kiwidude]

Personally I would suggest you rethink putting 3,000 items on your device. It will always be your device that is the weakest link in the chain slowing things down - the more books you put on there, the slower it will be every time you connect it to calibre as it has to read the database off it and iterate across its folders to find books that are not in your library etc.

I "limit" myself to around 400 books at a time on my Kindle - still 18 months worth of reading and enough variety to keep me interested, while keeping calibre responsive enough when I plug it in every few weeks to remove read stuff and top it up...

[ilovejedd]

Originally Posted by theducks

RAM as much as the OS supports (the motherboard/processor may support more, but the OS is the limiting factor)
Calibre will use as much CPU (core) as it can get if needed. Keep those fins and fans clean

Each of those 3K books is getting its Metadata updated as it is sent which will eat time.
As you noted: Flash memory is not the fastest horse in the race

Already planning on 8GB RAM for the build. With 8GB kits going for $40, it doesn't make much sense to cheap out even more. Given 32-bit apps are limited to 2GB memory, I think 8GB RAM will be more than plenty. Will be using an SSD so the source drive shouldn't be an issue.

Yes, Calibre will use as much CPU as it can if needed. However, in this case, USB2 and flash memory speed will eventually be the bottleneck. Keep in mind, I won't be using it for conversion and from what I've seen (while I was still awake, anyway), sending books to the device only uses one core.

Originally Posted by kiwidude

Personally I would suggest you rethink putting 3,000 items on your device. It will always be your device that is the weakest link in the chain slowing things down - the more books you put on there, the slower it will be every time you connect it to calibre as it has to read the database off it and iterate across its folders to find books that are not in your library etc.

In this case, I don't think it's the device that's the weakest link yet. From what I can see, the transfer is divided into 3 stages: 1) initial prep? 2) file copying and 3) send metadata

Stages 1 & 3 are the ones taking up so much time. The actual file copying (3,171 files, 328MB) finished fairly quickly.

I'm not going to read all 3,000 ebooks. Reason I'm putting them on the reader is because most of them are probably crap and I plan to mark which ones to keep and which ones to delete on the reader itself.