Spectre and Meltdown vulnerabilities...

thomasp

https://www.theregister.co.uk/2018/01/09/meltdown_spectre_slowdown/
the linked article contains some before/after-patch performance figures. not looking good.

right now first of all i'm quite confused as to if and what to install (firmware, drivers, windows + OSX patches for the operating systems). finding all this stuff and ending up with a system that works as well as it has done so far with patches having been squeezed out quickly - the question is if that can be expected at all.

just wondering what the consensus here is considering that we probably are in the user group that will notice the performance hit.

in my situation i have a smooth running windows 7 machine that is dedicated to work that contains no personal data (beyond stored LAN/logins), no games etc and that does not go onto the net (ethernet only) unless it's to fetch a license or access some online help system with most things locked down by firewall and restricted by IP and user-permissions. so far thinking of leaving that machine simply ... unpatched.

the computer will probably be replaced this year anyway (it's a 2014 machine), not sure if and how big a vulnerability this box could be in any case.

Axi5

I don't think performance will be impacted that largely in 90%+ of our tasks.

Certainly opening software, binding GL/DX windows and querying the OS will be impacted. However, real-time or offline rendering, animating, dynamics (excluding caching dynamics), modeling etc. should all be the same as before.

In fact, caching dynamics is where I expect the largest hit to be, dumping data onto disk, constantly querying the OS... yeah can't be good for that. I still don't know if it'll really affect people that much that it would be that bothersome though, 30-35% is the absolute worse case scenario with many reports coming in that it's typically between 5-15% slowdown. Which isn't much worse than having the performance of a previous gen processor, and only affecting certain tasks.

All in all, I'm not that worried!

Dash-POWER

I think the largest impact will be around cloud and database based services where is big amount of I/O. I wouldn't be worry too much about not enterprise sector. There will be measurable differences but not from user perspective visible differences. - If I'm wrong correct me.
PS: Does anyone know if VIA x86 found something similar in their architectures? - I'm just curious. Atmel and their AVR microcontrollers are probably too low for x86 logic.

thomasp

in the last few days i read that a flaw like that was found in ibm's power PC architecture, too. luckily apple has long moved on from those space heaters, so i don't think it would affect us.

there are reports that boot times are supposedly also affected and that some of those patches are leading to system instability. early days i suppose.

Axi5

thomasp said:

in the last few days i read that a flaw like that was found in ibm's power PC architecture, too. luckily apple has long moved on from those space heaters, so i don't think it would affect us.

there are reports that boot times are supposedly also affected and that some of those patches are leading to system instability. early days i suppose.

You might not like to hear this but:
https://www.theguardian.com/technology/2018/jan/05/apple-mac-spectre-meltdown-iphone-ipad-hackers

thomasp

oh i'm well aware that the intel mac is affected, just saying even though power PC seems to be another architecture caught up in this, we do no longer use that one.

Axi5

thomasp said:

oh i'm well aware that the intel mac is affected, just saying even though power PC seems to be another architecture caught up in this, we do no longer use that one.

Ahhh okay, I was under the wrong impression