30% performance hit? I'm sure that heavily depends on the workload... and I'm also sure you lose performance when HT is on, depending on the workload as well.
That would make sense, my understanding is that with a 100% pegged CPU hyperthreading won't be super beneficial as they aren't real cores, just smarter scheduling. You can't really schedule 100% load better, however for applications that are latency specific it would make more sense, as you don't have the CPU pegged, you just want a faster response.
Sure you can. You can do math while another HT is waiting for memory. Sometimes you can even multiplex use of multiple ALUs or one HT can do integer and another can do floating point.
It's actually under high multithreaded load that HT shines, especially if that load is heterogenous or memory latency bound.
I too was once under the misapprehension that HT was "just smarter scheduling", until I took a university course in microarchitecture that explained how Simultaneous Multithreading actually works in terms of maximising utilisation of various types of execution units. I wonder why "smarter scheduling" became a common understanding.
Not in this case. Using "The frog croaks 'ribbit'." as an example sentence [0] with contextual clues, I only got translations that, when translated back, yielded either "rib", "rabbit" or "ribbon".
But after reading the thread I see the intended public
That seems an oddly teleological argument; judging the intention of a speaker by extrapolating from the response of a single person who replies to them enables drawing almost any conclusion you want.
Given that you've commented to my post, isn't that like saying you are my intended audience - despite my having no way to know you were going to create a HN account yesterday?
The problem free speech message boards will always have is that the only people who seem to be genuinely motivated to post on free speech platforms are people who can't get on normal platforms. As a result free speech platforms almost always end up dominated by extremists and racists. Partly because they're the ones with the biggest motivation to use the platforms, but partly because their use of the platform drives away more moderate users. Plenty of people believe you have a right to free speech but also believe in their own right not to hang around and listen to you.
'Unsavory' is an interesting way to describe speech that instructs, encourages, or incites the commission of murder, which has a terminal effect on the speech of the people who die and and a chilling effect on the speech of those around them.
Logically, it would appear that murder is the ultimate form of censorship, by not merely constraining speech but eliminating its very possibility. Perhaps you could articulate how you resolve this philosophical conundrum?
This is not a problem if you are behind a CG-NAT. If you are not (that should be the default) then ask your ISP to put you behind one. If they don't offer that service, then it's time to shop around.
Asking your ISP to cripple your connection like that is a horrible "solution", and usually isn't a change they're prepared to make by request. If you have the option of shopping around for ISPs, the one that doesn't do CG-NAT is usually the best choice.
I disagree with you. The majority of users don't care about being behind a CG-NAT (what you call "crippling"), and CG-NAT offers a very big layer of protection that avoids problems like the one on this article.
NAT adds latency to a number of applications (VoIP, video conferencing, gaming). Not so much in the translate IP/ports and keep some state, but in connection establishment. CG-NAT only makes this worse (not to mention it's becoming impossible to troubleshoot when issues arise).
Users don't explicitely care because they don't know. It doesn't make much difference when viewing YouTube videos, but there's more to the Internet than cat videos.
NAT is not a security layer. It's possible through techniques like STUN and such to discover and reach hosts behind a NAT.
CG-NAT is crippling because I want to receive incoming connections like anyone else who has a connection to the Internet should be able to. Router manufacturers can do better. The world does not have to consist solely of cloud-based middle-men who take full advantage of the fact that all your data has to pass through them, and that you have to trust them.
What's somewhat ironic to this discussion is that some Linksys routers modify STUN responses, which breaks legitimate functionality if the router is used with dual-NAT or CG-NAT:
This is very unlikely. Any ISP with some experience will make sure customers are not able to interact any more than they could from outside the network. That's network 101.
Having worked for large ISP's for around a decade, cheap comes first, customer safety comes last.
Also, QUIT BREAKING THE INTERNET. CGNAT is complete crap that breaks the internet peer model. Even ISPs not using CGNAT are pushing complete crap on users. For example last week I had a user that VOIP stopped working. They received a new integrated cable modem router from their ISP. If you rebooted the unit VOIP worked about an hour, after that it would stop passing VOIP packets (ran tcpdump on the server and watched them stop). If we ran VOIP over another port it would work (but had different issues related to changing around 50 phones so we only used it for testing). There were no options to disable SIP_NAT, nor any other settings that would fix the problem. Since this ISP also provided their own phone service I found the whole debacle rather anti-competitive. They simply have no interest in contacting the modem vendor and having them fix the problem.
We ended up supplying our own modem and router in this case and the problem was resolved.
