On 28/09/17 12:32, Zrubi wrote:
Hash: SHA256

On 09/21/2017 08:47 PM, Teqleez wrote:
Hi all.

I assume this is relevant not only for considerations related to
the practical limit/benefit of increasing RAM in existing
computers, but also regarding the specifications when buying new

- At the following amounts of RAM; 8GB -> 16GB -> 24GB -> 32GB ->
48GB -> 64GB, how much does the requirements for the graphics card
memory increase with each step, if we assume the
normal/daily/"permanent" usage to be 70-80% of the available RAM?

This obviously depends on how many different VM's one has open, and
how much video memory each one needs, but can we make an assumption
on some sort of average number for this, to see if it is possible
to find some kind of rule-of-thumb figures?

For example, if we assume a "normal"(?!?) Qubes-OS user whose sole
reason for increasing memory is in fact to be able to run more
Qubes simultaneously; how many more Qubes can he/she expect to be
running per extra 8gb of RAM, and how much more will each such step
"typically" require of the graphics card memory, if we assume a
"linear" growth in the number of concurrently open Qubes?

I am guessing that with the capacity to run (increasingly) many
Qubes, the amount of applications running in each one will be
lower, to the point where we most often choose to run
"one-app-qubes". For the sake of this example we could also assume
that this example user is a "lazy" person who will not bother
configuring minimal templates, but only use the default shipped
fedora-24(++) template for each one, just for the sake of
simplicity and playing around with these numbers a bit.

IMHO there is no real connection between RAM used by the OS and apps,
a and the RAM ins your VGA. (VRAM)

VRAM is used for:
- - game (3D) data rendering, which is out of scope in case of Qubes
(maybe if you using Compiz with fancy desktop effects)

- - display rendering.

The later is still need more and more VRAM because of the bigger and
bigger common resolutions Like:
800x600 -> 1024x768 -> 1920x1080 -> 2560x1080 -> 4K -> 8K -> etc.

So you still need more and more VRAM to render all your pixels in your
screens. These can be multiplied if you start using multi-monitor setups

But these needs are not growing as fast as your conventional RAM needs.
I'm using 3xHD screens, and the 512Mb VRAM used by the integrated VGA
is still enough for that,  while my OS has 16Gb RAM.

- -- Zrubi
Version: GnuPG v2


Ref. "(maybe if you using Compiz with fancy desktop effects) "

For me, the desktop environment affects my well-being and mood very much, so yes some of such "effects" are important, both for efficiency and for the general good feeling when working etc.,, not the fancy part.

I prefer to have 12+ viewports/workspaces, each with its own name and background image, so I can associate specific tasks/applications to each one, plus with a wrap-around function to navigate beyond the borders. I use Compiz for this on Ubuntu, and fortunately all of the above is possible in Qubes without any extra tools.

Compiz provides customized window transparency so that when I write, I arrange for good (text++) contrast towards the nice background image on each desktop, on a per-window basis. I use Alt+[numeric"+"] and Alt-[numeric"-"] keyboard shortcuts to increase/decrease the transparency on the active window. (Not sure if this is possible in Qubes out-of-the-box?)

With this as the background (all done outside of the VMs), along with the initially mentioned points about increasing use of one-app-vms, are you saying that we can basically just keep upping the RAM without worrying about the potential need for increased display memory?

I would like to know what are the minimum requirements for the graphics card in this respect, for "both ends of the spectre", so to speak;

a) ~16GB RAM, which is getting increasingly possible with sub-€1000 laptops (just for the perspective on the limits).
b) >32GB RAM, for the high-end computers.

Will it be possible to use a (compatible) laptop in the sub-€1000 range as long as it has enough RAM, or should one also verify that the display card meet some kind of minimum specifications (for the scenario where we are going to run a LOT of VMs, only limited by the available RAM)?

+ How is this different with Qubes 4.x compared to 3.2?



You received this message because you are subscribed to the Google Groups 
"qubes-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to qubes-users+unsubscr...@googlegroups.com.
To post to this group, send email to qubes-users@googlegroups.com.
To view this discussion on the web visit 
For more options, visit https://groups.google.com/d/optout.

Reply via email to