jueves, 17 de marzo de 2016

SSD usage in Linux

This are some essential points to consider if you want to use an SSD hard drive in a Linux system.


1. Get enough RAM
2. Avoid using SWAP

If you want to use the option of hibernating, then set the swapines to zero. But try to avoid hibernation.

echo -e "vm.swappiness=0" | sudo tee -a /etc/sysctl.conf

3. Disable acces time logging

in your /etc/fstab file add the option "noatime" in every partition of your SSD drive. For example, change every “errors=remount-ro” to “noatime,errors=remount-ro”

4. Run fstrim every day

It will prevent your SSD from slowing down. You can use cron to run this program automatically in the background every day.

echo -e "#\x21/bin/sh\\nfstrim -v /" | sudo tee /etc/cron.daily/trim
sudo chmod +x /etc/cron.daily/trim

5. Monitor your SSD using S.M.A.R.T.

Check once in a while that the Media_Wearout_Indicator value of your SSD is not lower than 10 (it starts with a value of 100).

sudo smartctl -data -A /dev/sda

source: https://www.leaseweb.com/labs/2013/07/5-crucial-optimizations-for-ssd-usage-in-ubuntu-linux/

martes, 15 de marzo de 2016

DeepDream through all the layers of GoogleNet

Video depicting all the layers of a deep nerual network. The Convolutional Neural Network has been pretrained with ImageNet. The first input image is a picture of myself and at every step the image is zoomed with a ratio of 0.05. At every step the actual input image (frame) is forwarded to the actual hidden layer. The error is set to be the same representation in order to maximize all its activations. Then, a backward pass is computed to modify the input image. After 100 iterations of zooming in one layer, the next layer is used. See more in my Aalto personal web-page