Arleson wrote: ↑08 Sep 2020, 00:50
The problem seems to have been solved with the boot in copy2ram. What would be the explanation for that? I am booting from a simple USB stick.
Unfortunately I cannot answer you that.
But I can tell you my performance experience on my 8 core Toshiba Satellite with 4 GB of RAM.
Due to keeping the pre-installed Windoze7 I have partitions in none-consecutive order. "Partition table entries are not in disk order" is how fdisk calls it.
For many years I had space allocated for a Linux swap file but that could not be created.
Then a recent version of GParted could create the swap partition.
Here is what fdisk -l tells me, I replaced some geometry starting and ending numbers with symbols where
a is the lowest number
and g the highest.
and also replaced the actual ending numbers of the Sectors sizes with some zeros.
Same with overall values of bytes & sectors & Disk model, also² the Disk identifier is fake
Code: Select all
root@porteus:/# fdisk -l /dev/sda
Disk /dev/sda: 465.00 GiB, 500107000000 bytes, 976700000 sectors
Disk model: Hitachi HTS00000
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: dos
Disk identifier: 0xBlaBla
Device Boot Start End Sectors Size Id Type
/dev/sda1 * 2048 a 3000000 1.5G 7 HPFS/NTFS/exFAT
/dev/sda2 a+1 b 899000000 429G 7 HPFS/NTFS/exFAT
/dev/sda3 b+1 f 54000000 26.1G f W95 Ext'd (LBA)
/dev/sda4 f+1 g 19200000 9.2G 7 HPFS/NTFS/exFAT
/dev/sda5 e f 45700000 21.8G 83 Linux
/dev/sda6 c d 9000000 4.3G 82 Linux swap
Partition table entries are not in disk order.
Now for years I had a
swap file on /dev/sda5 (because the swap on /dev/sda6 could not be created) - and when I failed to killall my browser
aka killall palemoon when the swapping got too severe, the system would end more than once in a non responsive state, swapping itself to death.
Since I have the
swap partition that improved greatly.
Code: Select all
root@porteus:/# type sx
sx is a function
sx ()
{
echo $(date +%d.%m.%Y\ %H:%M:%S) ____________________________________________________________;
{
read firstLine;
echo "$firstLine";
while read f t s u p; do
let "s2 = $s / 1024";
let "u2 = $u / 1024";
printf '%-40s%-16s%-8s%-8s%-8s\n' $f $t $s2 $u2 $p;
done
} < /proc/swaps
}
root@porteus:/# sx
27.12.2020 02:33:55 ____________________________________________________________
Filename Type Size Used Priority
/dev/sda6 partition 4435 1376 -2
/dev/zram0 partition 774 624 100
So, things I learned:
First, I repurposed the Windoze7 either hibernate or swapfile on /dev/sda2 as Linux swap file.
Performance was abysmal as soon as disk swap had to be used. Because NTFS is no native Linux partition.
Therefore I replaced that by creating a ~ 4GB swapfile on /dev/sda5 which is of type ext3. That
improved the performance somewhat. But with the issue of system freeze when my main browser palemoon got too greedy.
Then all of a sudden I could use /dev/sda6 as swap partition.
Now I do not have to monitor beginning heavy disk swap usage any more. Even when palemoon gets too greedy, the GUI of XFCE would freeze, but for only less than 10 seconds, often only for 1 or 2 seconds.
So the overall conclusion for me
swapfile on NTFS - worst performance
swapfile on ext3 - okay performance but system completely freezes when browser too greedy
native Linux swap partition - best performance, not a single system freeze, only short freezes even when browser too greedy
So my tip: if possible create a Linux swap partition using a recent GParted.