Volatile USB RAM disk?

[GiGaBiTe]

You could try creating a block device on video RAM:

https://wiki.archlinux.org/index.php/swap_on_video_ram

There are a few gimmick video cards out there with an absurdly large amount of video memory that will never be able to use all of it practically. I've done this with several PCIe and PCI video cards. You aren't limited to swap, you can use any file system and store regular files on it. I got some pretty impressive speeds on PCIe video cards, usually 800 MB/s+ on both reads and writes. Speeds depend on the specific GPU though since you have to go through the memory controller on the GPU.

 

[Chuck(G)]

I'm not at all sanguine about data in video RAM surviving rebooting into different operating systems. Are you?

 

[KC9UDX]

The Amiga using DOS 2.0 or higher used the RAD: device.

You beat me to it. Another one of those things that is trivial on Amiga yet impossible on a PC.

But I wonder if it is possible on the PC. Is there some way to fool the BIOS into mis-reporting the amount of RAM to the booting operating system (does the OS even get this information from the BIOS?), and then increase the amount of RAM after boot?

 

[Chuck(G)]

I'm going with the Pi Zero approach; it's cheap and not-OS dependent.

What's curious is that I've found more posts on the Web over the last few years that were looking for exactly this sort of thing--a volatile RAM disk. Apparently, my customer isn't the only paranoid one. (FWIW, the customer is not a private party; just that the information content is unknown and could potentially be sensitive).

 

[KC9UDX]

So there won't be a new market for core or bubble memory.

 

[hjalfi]

(Tangent: I once added a Hercules card framebuffer as a Linux swap device. It worked fine, and you could tell by the way the screen flickered with gibberish what it was doing, but the 8-bit ISA bus slowed the computer down so painfully it wasn't worth using.)

 

[lowen]

I've got a bit of an unusual problem.

I need a USB (or other) RAMdrive that will persist across boots (e.g. rebooting operating systems) but not across power cycles.
...

Going back to the top here.... is such a thing even possible? I know at least one of my laptops power cycles the USB ports when doing even a soft reboot. I haven't tested any of my desktops to see (I have a USB power meter that makes it very easy to see a power cycle; the voltage will dip below the normal 5V for at least a brief blip).

I'm not at all sure if firewire cycles power on a reboot or not.

 

[Chuck(G)]

In my case, it doesn't matter, the Pi Zero can be powered by an external wall wart, so no big deal. If powered by USB, it's probable that a schottky diode and a big cap will carry the device through the power blip.