New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
@davide I don’t agree.
In your example, you compared accessing a full core with and without hyper threading. That’s not what we’re comparing here. There’s 0 benefits for me as client that the provider is using HT if that means I get to share the physical core as a consequence (if not, it’s even further apart from ”dedicated core”).
Given that a thread with HT can be called a core, they’d still need to offer a dedicated thread to not lie. As in, without HT, they’d have to only provision 1/2 of the CPU allocations on guests, as they’d have to offer a dedicated core instead.
So what can be compared is a dedicated core without HT or a dedicated thread with HT (where someone else will use the cores other thread), and the dedicated core without HT wins.
Tl;dr, as long as there can even remain SOME ambiguity, and they want to keep advertisting dedicated cores, there’s no benefit to me that those are actually threads.
And, if the word usage is valid so long as it carries some form of benefit, which is how I understood your last argument, then it’s valid at Avoro too, the benefit there is that they won’t throttle your CPU usage.
Im inspirational, like deodorant. Speaking of which, yall wanna celebrate friday at a QT?
That is:
A] Right
and
B] Wrong
Both types of threads exist. Don't ask me who came up with the brilliant idea of giving both concepts the same name but someone did.
I posted above the outline of my benchmark to argue that hyper-threading improves the throughput of the machine, and therefore is not a marketing gimmick, as we touched the tangent that hyper-threading itself might be considered fraudulent and not beneficial to the machine's cumulative throughput. Of course VDS customers get worse performance from being assigned hyper-treading cores rather than "real" cores. Let's clarify; the problem here is the definition of core, which is contended between:
If the CPU is considered as a black box, notion #1 is superseded by #2, meaning that whatever number of cores the CPU claims to have is considered the truth, which is how most applicative software count the cores on the system, i.e.
std::thread::hardware_concurrency()(c++11)omp_get_num_procs()(OpenMP)all return the total number of logical cores, not the number of hardware units implemented in the chip. A logical core isn't a thread or the capacity to run a thread continuously without interruption, rather it is a logical construct implemented by the hardware, and is part of the hardware.
Avoro🖕™invented a fourth definition out of their own volition, which is:
This is because a "dedicated core" is always considered a hardware core (logical or not) in the context of VDSs, or at least this is the first time I encounter a phony definition of dedicated core like this.
Hi sir, cores fix next week
well, Avoro @dataforest has put 30 chairs in a room for 60 customers , saying you will get a chair when you need it.
They are saying only 10-12 chairs are occupied and if they see chairs % usage is beyond a certain limit (say 70%), they will redirect customers to another room where there are more chairs.
As long as this background process is functioning , customers will get chair whenever they need.
The issue is, the ordering screen wording is dedicated chair 🪑 and that is lacking clarity or misrepresentation depending on how one sees it.
This whole process is wide open as they decided to accept a large 🪑 order without having room infrastructure in place.
Alright, makes sense, but I find the "true" core count to be the industry standard for CPU's, that correlates with labels on the box. If one can boost performance by say 30 % on average while "2x the amount of cores" (read: by making it seem like 2x by software), I won't agree that you can then claim 2x the cores, and even the manufacturers don't, you don't see Intel or AMD claim that the threads are cores. They don't think so. They always provide "cores" and "threads" in their spec sheet. The fact that it's designed to be treated as 2x the cores for efficiency dosen't make it 2x the cores.
Buying a (what I think we can all agree) 24 core CPU, taking 1/48th of it, and calling it 1 core dosen't make sense to me.
@davide you are very knowledgeable today. Did you eat walnuts for breakfast?
Yep, they accepted an order they couldn't handle.
I'm waiting for someone to start selling t-shirts:
--- I was there! ---
Because marketing needs to use words that will fetch most business.
Photo copier machine with exact same hardware but with different firmware to control number of pages being copied per minute is sold for different price.
Yet, Intel and AMD don't consider it twice the cores. They do marketing as well. If even the manufacturers don't, that's quite the tell.
I won't do this again!
Sweat included or not?
1 cores deliver broken , consistansy error , is guaranteed work 5 year , we waiting for new cores delivered
Regards
m y i m love hype threading
sweat allow, yes
cores is allow , thread is ignore
mom I made it Im in let dedicated core outrage event
Lot of cores
Especially if we count each one twice
my core is broke, we r do decenterlize work on the platform,,, plz fix core asap my bisunes are loosing many money !!!!! or i will charge with the bank!!!
i t i s being fixing , m y i m have 8 month left on thesis on ddos protection repair cores
Have you thought about getting a dotdog and a 48oz soft drink at QT for $3.99?
my im love drama thread