Slow resume from hibernation

[ese002]

Hibernate is reasonably quick. A minute or two tops. Initial resume is also pretty quick. However, between initial resume and when the system responds as it should can be be half an hour. In between, the disk churns and churns after every little action. swapoff -a makes the system responsive sooner at the expensive of being more painfully slow during the transition.

It is reasonably clear what is happening. Resume only loads critical pages. Everything else is left to page in as needed. With 6GB typically paging into 16GB of RAM in haphazard fashion, that is a lot of random reads. Ideally, the whole image would be loaded into RAM in one sequential burst. Are there any knobs to accomplish this or something with similar effect?

For what it's worth, here is what I run.

Swap is a 32GB RAID1 partition on a pair of 3TB SATA drives.
NAME TYPE SIZE USED PRIO
/dev/md23 partition 32G 0B -1

eric@crab ~ $ uname -a
Linux crab 4.1.12-gentoo #1 SMP Sun Dec 13 11:40:13 PST 2015 x86_64 Intel(R) Core(TM) i5-3570 CPU @ 3.40GHz GenuineIntel GNU/Linux

eric@crab ~ $ sudo swapon
NAME TYPE SIZE USED PRIO
/dev/md23 partition 32G 0B -1

md23 is raid1 on 3GB of spinning rust

md23 : active raid1 sdb3[0] sdc3[1]
33537920 blocks super 1.2 [2/2] [UU]

The rest of the system is up to the moment Stable with a handful of ~AMD64 exceptions