This thread is permanently archived
processor bug
| >thought i could utilise >>here
| >>98253 the article itself is here https://www.nytimes.com/2018/01/03/business/computer-flaws.html
| Howboutdat
| apparently ALL x86, arm and amd64 processors from the last 20 years are fucked for security
| https://arstechnica.com/gadgets/2018/01/whats-behind-the-intel-design-flaw-forcing-numerous-patches/
This is a more in-depth look at the problem. In case this is anyone's cup of tea.
| >>98604
Thanks for the link. It is good to see what is happening under the hood.
It actually seems like Intel processors are the biggest suckers. AMD and basically all ARM processors are somewher in the middle for they use the same design scheme for how the processor adresses memory yet dont implement the cheap ways Intel uses to shortcut certain things and sacrificing speed for proper isolation. Then SPARC and IBM are unaffected, always isolating memory without compromise.
| >>98646
Yeah, it was honestly disappointing and surprising to see that Intel has been doing this for some years-or-so without having their architecture redesigned.
| the worst part is that i think they have known for a while now
| >>98666
They couldnt, if they were to maintain their bussiness advantage on boasting they produce faster processors than the competition. People only see 'wow so fast!', their cashflow doesnt stop, that is all what mattered. Very few people would say Intel hw is *not* the best design but the worst and these always have been considered idiots and laughed at. Redesigning would break backward compatibility with the vastness of other hw/sw products, a disaster, as Itanium demonstrated.
| >>98679
Known they have for they designed the processors that way. But I think also unaware of how such shortcuts and 'unimportant hassles' could spiral down and by cumulative effect become what we are seeing today.
Of all processor architecture, x86 has always been the worst. Why it is dominant is because it is the cheapest.
| >>98699
Oh yeah.. Itanium. Thanks for reminding me of it.
You're absolutely correct here. Sacrificing security in favour of performance, as proven here, in this case here.
Honestly, I think the problem roots deeper than simply this 'bug' alone.
| More bits for the cybdrnetically oriented:
https://www.theregister.co.uk/2018/01/02/intel_cpu_design_flaw/
| To sum it up quoting those who know what they talk about:
an Intel Core i7-3770S GPU will take a 34 percent performance hit, while the new Intel Core i7-6700 will run 29 percent slower.
The Linux fix can be disabled, which you will likely want to do if you use an AMD processor.
| Welp, time to go back to old Pentium Processors
| So is basically everyone completely fucked?
| Meltdown can be fixed in patches, SPECTRE cannot.
| >>101479
that's misleading. neither can be fixed in patches because they're fundamental flaws in the way these hardware systems work. both can be sort of worked around but require massive patching, and things like that aren't likely to be rolled out to every single user program everywhere
| >>101496
People will not care, like they have always done. If they happily publish their lives in social media, surely they wont worry about what they cant see and perceive. And they will get used to the 30% of slowdown increase a lot faster than you think. It is not that we will be screwed, for we have always been that way so nothing will change.
| I doubt this will hurt Intel very much on the consumer end, however a very large chunk of their business is now server equipment, and I doubt Amazon will be quite so forgetful/forgiving.
| Intel could finally lost it's most important monopoly, to both AMD and ARM. I hope the moneyvtgat it brings to this companies will make competition a serious treat in all processor markets.
This thread is permanently archived
| not sure if this should be in /new/ or /tech/ but here goes:
apparently alongside the intel bug, if you've heard of it, there's another bug, which has much greater effects (affects most microprocessors on the market and likely will for an entire hardware cycle).
source is this thread on twitter from an NYT journalist: >>https://twitter.com/nicoleperlroth/status/948684376249962496
thoughts?