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
What annoys the most is that the intel stock is not going down.
It feels like no one really gives a fuck about that.
@jsg
I asked:
And you replied:
"heinous and disgraceful personal attack"? Really?
You could have simply answered:
"Yes, the "lie and betrayal" account is largely my spin on what happened, but here's a link that partly supports this view."
or
"No, it's not just my personal spin, the "lie and betrayal" account is amply corroborated by others; see e.g. this and that link."
But, no, you didn't answer in one of these two ways. Instead, you spoke of a "heinous and disgraceful personal attack".
(Note, by the way, that all of my other comments were in an exchange with @poisson, not with you.)
You have a documented history here on LET of taking anyone who questions your opinions to be personally attacking you. Everyone who wants to see this can see it, but what has happened over time is that most people have become fearful of questioning your opinions because they're afraid of a ferocious reaction from you. I refuse to be one of these people.
(In the context of personal relationships, this would fall under psychological terror/abuse.)
I can see how a reading of https://mdsattacks.com/#ridl-ng -- a link which you offered only after two attempts by @TimboJones to get independent corroboration of the "lie and betrayal" account -- might lead someone (e.g., you) to read a "lie and betrayal" account into Intel's behavior, but just to note that that web page doesn't mention "lie" or "betrayal" anywhere (as far as I can tell), so it's not obvious to me that the author of that web page would agree about "lie and betrayal". At the same time, the author of that page does criticize Intel's behavior on a number of points.
If Intel's behavior is similar to (e.g.) the Volkswagen emissions scandal a few years ago, which a "lie and betrayal" reading of Intel would suggest, then I imagine that there would be ample grounds for a successful class lawsuit against Intel.
Do Fathers speak like this nowadays?
Fathers do care for their childs and hate companies lying and harming them
@angstrom
I won't respond to your predictable and boring standard reaction re. myself (incl. some ridiculous attempts to paint me as a liar or psycho).
First: No matter how hard you and someone else try to bend it, this is no court room and I need not prove what I say.
Btw. Based on what right exactly do you demand that I provide more information? (hint: that's a rhetorical question but you might profit from considering it).
I suggest you apply a bit of logic. If intel knew in April yet announced one month later how wonderful and immune its new processor generation is then either intel knowingly lied -or- multiple credible researchers conspired to lie about intel and the timeline (which is provably not the case).
Second: A fact is a fact, no matter whether I present it in a way that makes it obvious or not. Similarly a fact is a fact, no matter who (e.g. you) knows about it or not.
The zombieload vulnerability exists, no matter whether you or TimboJones know about it or not and no matter whether you believe me or not. The timeline clearly shows that intel did lie and betray its customers.
If your interest were in the matter you'd simply say "thank you" (for the hint) and, if you happened to be really interested, you'd add "can you provide some links for further research?" - but that's not the case. Your interest is something entirely different.
My interest on the other hand was to inform our community. Anyone feeling a desire to investigate further is free to do so. Funnily so, certain users here did not choose to investigate although it's easy. Rather they chose to complain about me and tried to push me to provide more information - which I even did. And - surprise, surprise - they continue their stupid game .... and Cascade Lake is still vulnerable and intel still provably lied.
Have a nice day.
No, sorry, it doesn't work that way.
You said that I made a "heinous and disgraceful personal attack" against you. You didn't have to say this, no one forced you to say it, but you said it. Since that voluntary negative statement on your part concerns me, I have a right (even an obligation) to react to it. When I react to that statement, you run away. This suggests to me that you have no defense of this statement.
You expressed a strong negative view about Intel's behavior. You didn't have to express this view, no one forced you to express it, but you expressed it. If someone expresses a strong negative view about someone/a company, others have a right (even an obligation) to ask for corroboration of the view. I asked whether there is other corroboration of the "lie and betrayal" view. I didn't ask for or demand "proof". (The original link to ZombieLoad doesn't provide corroboration of the "lie and betrayal" view.)
Again, it's patently obvious by your behavior that you can't stand it if anyone dares to question any strong negative view that you choose to express.
Let's get back to Intel VS AMD, not member vs member.
HaHaHaHaHa! Thanks for amusing me.
And intel processors are still vulnerable and intel still lied and betrayed- provably.
I was right from the beginning. Provably. Have a nice day.
(You don't dictate your rules to me. Try to grasp it already)
Dude, they're not my rules. They're rules of social interaction, not written down anywhere but generally/hopefully learned as part of the process of socialization.
This is precisely what you fail to understand and why you continue to say things like "You don't dictate your rules to me", which reveals that you persist in seeing my question as a "heinous and disgraceful personal attack" against you.
(It's okay, it's probably not your fault. Let's move on.)
Imagine having a twin with the same personality.
I understand disabling Hyperthreading now.
It's unfair, you do not beat the disabled classmate...
At least in the server market, this should result in a significant advantage for AMD (if DCs care enough about things like ZombieLoad).
I don't see the desktop and laptop markets being much affected by ZombieLoad.
Desktop and laptop are also swinging in favour of AMD for different reasons. Cheaper, also power efficient and superb APU integration without much extra cost since AMD has great graphics cards technology too.
I believe you. (I always use older equipment, so I'm not really up-to-date on the current desktop/laptop market.)
But, as you say ("for different reasons"), I tend to doubt that the recent Intel hardware vulnerabilities are the main reason (in the desktop/laptop markets).
let's not debate the topic at hand and get back into member v. member
Holy fuck. You don't know what proof means. I'm asking for it because you didn't provide any. So then it's your word against a public company, subject to stock disclosures and lawsuits based on their public statements. You really don't know the difference between facts and opinions, either. Provide the quote you're specifically saying is a lie and the proof they knew this.
This is why you needed to provide Intel's quote, the link you provided says,
"On July 3, 2019, we finally learned that, to our surprise, the Intel PSIRT team had missed the PoCs from our Sep 29 submission, despite having awarded a bounty for it, p****Intel had failed to address - or even publicly acknowledge - many RIDL-class vulnerabilities on May 14, 2019.
On Oct 15, 2019, we learned that Intel had not found this issue internally and the only other independent finder was the Zombieload team, which disclosed TAA to Intel in April, 2019."
They dropped the ball and didn't know or verify all the flaws. Being wrong and incompetent is different from lying (insert Trump joke here).
You lose credibility when you make allegations and then make people go look for whatever the fuck your point was. Just provide it next time.
My father does
I think that the most decisive player might be the mainboard manufacturers and other large "partners". Reason: It seems that intel lied to them, too.
Same here. Desktops and notebooks are usually run by one(1) entity, so those problems are less threatening for them.
I have been thinking about that for quite a while and I'm less confident than you. Different reasons: the vast majority of desktops isn't about performance but about brand, price, and some other factors.
As for notebooks low power consumption still is a field where intel leads. AMD Ryzen chips for notebooks still are relatively power hungry. I have one of intels J (or N, forgot it) class 4 core, ridiculously low power 4 cores based mini board and it's quite a bit more attractive than AMDs current offers in terms of (low) power consumption and quite respectable performance (good enough for most office desktops).
Funny(?) sidenote: AMDs counterpart, the V series (12xx, 16xx iirc) is considerably more expensive than intels offers.
Also AMD can not use lower price as a main attractor too much unless they are willing (or even targeting) to become the "cheap Joe" again. It seems this they target for "better performance vs price ratio".
Final point: graphics. Yes it's nice to have a by no means high end but reasonably decent graphics capability in an APU - but: intel has learned a lot about graphics capabilities too and what they offer is (a) decent enough for a very significant part of the market, and (b) offers open source drivers.
I think that as of now the 4 decisive points are:
As for us here at the low end hosting market, I don't expect many Ryzens anytime soon. Occasionally maybe, but the large majority of cheap dedis and VPS will stay on 26xx v2 or max v3 or even on 56xx, 24xx, etc. intel processors. Simple reason: plenty available (2nd hand) and at a fraction of the cost of any new (intel or AMD) machine.
@jsg yes, intel still has the lead on power consumption especially in laptops, but AMD is decent now. They just need to design some really power efficient chips (they can sacrifice some raw power for that) and make the selling point the integration with an APU Intel cannot match. Intel's graphics unfortunately has been lagging like crap against Nvidia and AMD and I don't see this changing soon.
Fully agree that the availability of cheap second hand machines will not change Intel's dominance in the short term but they don't make money from second hand machines. That's why they are in trouble. The high send server markets are rapidly moving to AMD, and that's where the money is.
The first one is simple: lower frequency. Which is also the way intel did/does it. But for some reason AMD decided to go for "ours offer much better performance". My take is they did that to escape the former underdog/cheap Joe image from day 1.
As for APUs I think they are over- and underestimated at the same time. Most buyers see an APU basically as "CPU + built in graphics card" and the most important criterion still is the CPU. The graphics part is just a nice plus.
At the same time even rather poor graphics are damn good enough for 95+% of office desktops and notebooks so many large brand name systems come with APUs. It saves some engineering and keeps the BOM a bit lower.
Re quality of graphics I think the vast majority doesn't care that much. Those who do, like Architects, gamers, etc. almost invariably go for a separate graphics card anyway.
I am occasionally using an intel box with some chipset graphics and it's damn good enough(TM). As for crappy integration, hey by far most boxes still run Windows and there it's setup.exe vs. setup.exe ... (for the user).
Misunderstanding. That was remark only for us here at LET. We just won't get a lot of Ryzens and those we get will mostly be gen 1. Simple reason: see my primer, a box that is less than 5 years old is almost a rarity and most nodes (on LET) are even older, 10 yrs is not rare and back then there was only intel (modulo some perverts who went Opteron).
.
That's what I gather from this topic.
Corporations lie. It is to protect their company image.
Scientists lie. It is to earn more money.
Since I cannot verify the issue "yet", I'll just see how it play out. If I want to be safe I would still avoid intel. If I dont care, I would still buy Intel.
There is a good chance the majority if not all of ExtraVM VPS will be migrated to AMD CPUs in 3-6 months.
Thanks for the updates! I have already reflected that you are one of the two I know offer AMD VMs in LEBRE. When you migrate, let me know I will update to offer in all locations instead of just selected locations.
Cheap, higher performing, lower wattage, whats not to love about AMD now?
Amazon had wicked deals on threadripper CPU last BF. Not a big sale on MB's though.With TR3 coming out now, I'm hoping for some big deals on both motherboards and CPU's.
Can you do anything with all those lanes (if purchased on a decent BF sale) or you pretty much stick to Ryzen 39xx for all the cores? Threadripper supports bifurcation, so you can get a lot of NVme ontop of 3 onboard M.2's.
Most of my VPS are on the OVH network still so no. If I did anymore that aren't it would probably be Ryzen though, not Threadripper.
Some I think interesting remarks:
"lanes" - considering that the new Zen generation supports PCIe 4 with double the data rate of gen 3 and plenty lanes at that, I consider the lanes issue settled for good.
"lanes" 2 - An SSD array can easily deliver 2 GB/s to the disk controller and even 4 GB/s is feasible but: that (4 GB/s) can be handled by 4 Gen3 lanes or 2 Gen4 lanes, so disks is not the issue. NVMe on the other hand could be an issue with Gen3 where a fast NVMe could overwhelm 4 Gen3 lanes. As of now 4 Gen4 lanes are plenty for an NVMe drive but if NVMe performance increases those 4/G4 lanes might become a bit tight. And for an NVMe Raid 0 4/G4 lanes already are a bottleneck. But then Zen has plenty of lanes so disks of any kind will not be the problem.
Similarly networking, even 10 Gb/s ports are not a problem. Even with Gen3 one can get 2 "very close to 10 Gb/s" with just 2 lanes. With Gen4 lanes one can even get a "very close to 40 Gb/s" network port.
So, what is the problem solved by (a) Gen4 plus (b) plenty lanes? My take is it's GPUs and the still far off its peak "AI" wave.
Never forget IBM! - or in this case more precisely PowerPC. About 6k$ buys you either a somewhat meager dual 16 core Epyc basic server (no disks, little memory, etc) - or - it buys you a very nice not at all meager dual 18 core PowerPC server with 4(!) Hyperthreads per core and no Spectre/Meltdown/Zombieload worries. As for intel Xeon I won't even mention them; too expensive, too vulnerability ridden.
intel - A propos: An 18 core Xeon v3 went/goes for about 5k$. per processor. The presumably better v4 goes for just about half the price!
So, of course intel did it quietly but they did adapt to the game being increasingly dictated by AMD and their much lower prices. Looking at today's prices I can't but feel that intel follows the line "Yes, we do cut prices considerably but we still ask some premium. After all we are intel".
Btw, the v3 Xeon has a TDP of 145W, the v4 still has a TDP of 120 W, and the 16 core Epyc also has a TDP of 120 W - the 18 core Power9? 90 W, 25% less.
And Sparc? All but being closed down by Oracle. So, would you want to bet on Fujitsu keeping Sparc64 alive and continuing development?
Summary: All, x86-64, Sparc64, and Power9 run most loads - but Sparc isn't even a mid-term option (sigh, I like them) and Power9 does it considerably cheaper than x86-64, both in price/processor and in operating costs (power consumption) and btw it also offers PCIe 4 and plenty other goodies, so you certainly don't get less than with AMD. For Joe and Jane that might be irrelevant and they know only x86-64 anyway but if you run a medium to large company or a big hosting operation it'll probably be a costly error to just ignore IBM. Plus Power9 allows you to claim in big fat letters "free of all the known vulnerabilities!".
I'll hence close with a famous IBM slogan -> Think!
How would you know if they are not under some embargo and just lying they are free from all known vulnerabilities? They have been affected before. It's probably best not to throw stones in the CPU security game.
https://wiki.raptorcs.com/wiki/Speculative_Execution_Vulnerabilities_of_2018
https://www.servethehome.com/ibm-power-confirmed-impacted-security-design-flaws/
https://www.phoronix.com/scan.php?page=article&item=spec-power9-4core&num=1
https://ibmsystemsmag.com/Power-Systems/01/2018/security-vulnerability-power-processors
Let's convict accused people assuming that they deliberately withheld information pointing to their guilt.