So when can we sue intel for lost performance?

The class action should begin soon I'd suspect. Server farms especially will be hit hard by this.

So you want a bunch of useless lawyers to make millions and end user end up with worth practically nothing coupon for his next Intel purchase? You really want Intel to waste money on a defense of the case that will deplete them from funds to develop next generation of chips that are safe and fast? Clearly Intel did not did it intentionally. This new attack vector was discovered long after the chips were designed. It is clear that entire industry never understood this attack risk since chips from all architectures are affected. Isn't this pretty stupid in the long term?

Yeah I do. There has to be consequences or else bad behavior propagates. This is basics 101, if you don't know that much you should probably just stop talking and save your reputation. Remember, its better to be thought a fool, than to start typing and remove all doubt.

What bad behavior? Intel will probably prove that it designed the chips in good faith many years ago and offer any affected user a coupon worth $5 for your future Intel purchase. The lawyers will make few millions. You will never see big money. The same lawyers will then turn to AMD and ARM milking them of their money. AMD will be bigger loser than Intel since they are weaker, smaller company and designed their chips far later than Intel when possibly more of this vector of attack was already known. Is that what you want?

Right sure. We should just forget about it and give them a pass. Good point, I totally, 100% unironically agree with you. Completely.

You don't have to agree with me. You have every right to remain wrong about this.

I can understand that you have a backwards interpretation of what the word wrong means. And thus can agree by that interpretation that you are "right".

Worrying about someone else "backwards" interpretation is the last thing you should worry about.

Oh, and what's the first thing?

First thing? Probably get better educated. Never hurts

But then how would I descend to your level in conversation? Its difficult enough as is...

Yes, that might be an issue. I did not specified what you should study. Certainly if you pick woman studies or gender studies we might have an issue with communication. If you pick something more logical. You know, science, mathematics, physics, then maybe you could grasp some of what I'm saying.

I said I wanted to get down to your level, not to confuse you. But seeing as how you had things backwards up there, I can understand why you, also again here, say the opposite of the truth.

I hate to break it to you, but there is absolutely no risk that you can confuse me. Not even close.

The fact that you think that proves that I already have.

Just made my day, cheers mate!

Yes of course I do because Intel basically did that to AMD and caused them many billions of dollars lost profit which was one of the main reasons AMD stagnated.
It's time Intel get's put in it's place.

@Jonathan can you further explain this please? What did Intel do to AMD in the past? Thanks.

That's a really long question.

Hrm. https://arstechnica.com/inf...

Should get you started.

See my reply to ja_1410 above

Did what? Intel won fair and square. Produced for years better and faster chips after Athlon forced them to compete. Your logic also fails to notice that AMD chips are subject to the similar flaws and exposed to the same lawsuit. The speculative execution attack vector is brand new a family of attacks many of which are yet undiscovered. It might very well be that future discoveries will affect Ryzen far more than Intel chips. Then what? Be careful what you wish for.

No they certainly did not win fair and square. Do a bit of research on Intel's anti-competitive behaviour and the many many antitrust lawsuits world wide in US, Europe, Japan, South Korea. Intel's modus operandi has been to try and block AMD in any way possible, even if illegal, because they know that any lawsuit will take 10 years to resolve. Ten years is a loooong time in tech and if youn ca block your competition from any significant amount of sales for 10 years it will massively damage their ability to fund new research and improvements in their CPUs. It did cost them $1.3 billion dollars fine in European antitrust lawsuit but by that time it was too late and the damage was done.
In summary
- Broke cross licensing agreements with AMD that were supposed to last 10 years, just 2 years after the agreement.
- Blocking AMD from selling to major OEMs by giving hidden rebates or even direct payments on condition the OEMs did not use AMD. OEMs including Dell, HP, NEC, Lenovo & Acer.
- Making fake benchmarking software that falsely shows better performance on Intel than AMD
- Bribery
More info:
https://youtu.be/osSMJRyxG0k
There a list of links in the description of the video if you want to read further.

You are entitled to any opinion. The fact remains, Intel chips were since 2006 (Core 2 architecture) far better than AMD and remained on top for over a decade. Only recently AMD managed to produce chip that comes close enough to force Intel to look for serious redesign. I believe Intel already paid numerous fines for noncompetitive behavior. You should also notice that you are proposing to use lawsuit alleging some flaw to punish Intel for unrelated alleged wrong doing. This is pretty naive bastardization of the law. Isn't accused has the right to defend itself against accusations?

Likewise you're entitled to your opinion even if it's based on ignorance.
Whether you like it or not, fact remains over 3-4 decades Intel were acting unethically and often illegally to stifle competition from AMD. This caused AMD to lose out on sales and run out of funds to keep developing good CPUs which was major factor in bulldozer disaster, they also had to spin off their entire manufacturing process just to survive. The fines Intel received were miniscule in comparison to the damage they caused AMD and the financial benefit they got in the long run. Sure AMD did some things wrong as a business, but sooo much of their problems were actually either a direct or indirect result of Intel unethical ans illegal actions.
I'm not suggesting Intel should continue to be punished for unrelated wrong doings. What I'm saying is they have a history going back to the very earliest of days in PCs of doing the wrong thing on purpose and often breaking laws.
Karma is a bitch. They haven't changed there business culture and bad practices much so they should continue to face lawsuits, including over security flaws until they start doing the right things. The trouble is they created a scenario of being almost a monopoly at one point not by having better products but by stifling competition. I think the trouble they have now with their security flaws is 110% deserved. In fact they deserve to fall behind AMD for many years so AMD has time to build itself up as a company so it can survive in the long term. Consider that Intel is over 10x larger now than AMD and can throw unlimited funds and new products. So they are now behind AMD for the first time but will come back fighting hard with new products. AMD needs a few years leading in both technology and sales to be prepared for when Intel hits back around 2021-2023, yes it does take about 5 years to develop a CPU architecture and that's how long it will take for Intel to have a chance to catch up.

But feel free to ignore the facts and blindly try and validate your feelings about a company that doesn't even care for you. I bet you didn't even watch the video or research the links contained in the description so still remain ignorant on what Intel did wrong.

Like I said karma is a bitch and it's time AMD get the success they deserve and Intel suffer for at least a few years for AMD to build up their war chest to fight Intel hard.

Ideal situation is for both companies have equal 50% market share each and are equally innovative and competitive in their own right without facing any anticompetitive behaviour from the other. This is the only way that will be a good result for everyone is fair equal compete which will force both companies to keep developing better products for everyone.

AMD lost revenue because it was unable to compete. They created couple brilliant CPU in early 2000 and then stagnated. Karma is a bitch. If you are going to use legal system to "punish" Intel, keep in mind that Spectre like flaws affect AMD also. The lawyers will not stop on Intel.

Inform yourself first before arguing. Are you aware of Intel anticompetitive and antitrust cases? I've already provided a link to check, you can easily also google more details.
AMD is significantly more secure and only vulnerable to 1 or 2 flaws. The fix had no noticeable affect on performance.
Intel on the other hand has had about a dozen in the last year and has a significant performance hit when you try to patch existing systems.

What Intel some alleged and some proven and fined noncompetitive past behavior has to do with safety of the speculative execution? Can you, please explain me unless you are driven by pure hatred of Intel like some kind of AMD fanboy?
No, AMD is not "significantly" more secure. AMD Ryzen chips just like ARM and Intel x86 are built around speculative execution with caching memory areas that are forbidden in the process and can be then then decoded by side effects. The Spectre type of flaws affects entire industry and we are only seeing the tip of the iceberg since it is not a single flaw but the whole chip structure is built around flawed idea with some effects almost impossible to fix by patches and requires full chip redesign. No, Intel chips do not have "significant" performance hit in general when mix of various tasks is considered typical for home PC. The flaw only shows in singular specific tasks. It is true that Ryzen has lesser impact with currently discovered attack methods but we only see tip of the iceberg on the vector of attack that was simply not understood by the entire industry. In a few years we might see flaws that affect Ryzen far more than Intel. If you make the case for Intel to be fined for loss of performance and security exposure then AMD and various ARM manufacturers are guilty by the same ruling and will end up paying fines too.

Intel boys and apple boys are very similar in reasoning, dont waste your time...

intel only had faster chips due to their "superior" prefetching. Now we know why it was "Superior". AMD had a huge ipc advantage over intel in the early days for years until the release of the "now affected" cpu lines we have today. Guess why? I have no brand loyalty but I have been around and seen just how far Intel goes to "WIN". They cannot be allowed to behave in that manner and deserve a record breaking class action suit at least.

The fact remains Core 2 outperformed K8 core easily. Why it would matter for the end user how it was achieved? If superior prefetching was the enabler what stopped AMD from designing its own superior prefetching for over a decade my friend? The CPU is a combination of memory access, prefetching, clock speed, heat dissipation, cache management, microcode efficiency, power consumption, instructions per cycle etc. Good design balances all of them such that the final product executes code faster than competitor at similar price point. Intel "deserves" punishment only in your mind. We have laws against unfair competition and any company felt it was hurt by alleged actions they could have had their day in the court. Using legal system to punish someone for some action using something else is plainly wrong. We need people trust in legal system, not to actively propose to undermine it.

And you are wrong of course on the "only prefetching". Intel Conroe was capable of more IPC than Athlon at the time.

Core was not better at IPC. Only their prefetching due to far less checks or authentications than that of Amd who bad the foresight when they implemented theirs. They only had process and prefetching and faster cache but core was slow as Fack. Only sandy had IPC improvements over whatever Amd had.

Yes it was.
https://external-preview.re...

U use a cb graph? Their cutting corners in the prefetching is what gave them the performance edge so that graph does not argue anything. I worked with IBM at the time and we did tests in various ways depending on what we were looking for from platforms which included tests/benches with/without cache and at various clocks etc to determine which part would provide utmost performance for whatever application / solution. Prior to core, Amd had an edge with their k7 and apart from Intel's process advantage later which meant better clocks, there was little and even worst IPC vs AMD. Sandy is the CPU which set them apart from everything else on the planet. X86 ARM Power etc. Nothing was better And their scale allowed them to position it in low power up to HPC in a way never seen before. In any case, knowing at the time how their prefetcher worked, I can tell u this alone gives them a major edge. Sorry I can't give more dynamic details on what we worked on but Dont overlook the fact that it plays a major part in their performance lead. Look at the performance impact when patched, they are similar in percentage to gains Intel chips get gen over gen... That's how significant of a performance gain they get from cutting those corners. As far as I am concerned. Intel and everyone else including Amd should have had the foresight and came up with a better solution allot earlier. They should all be fined or put under some serious presure.

Nice try, except in practically every benchmark Core outperformed K7 at the same clock speed. No, it is not "cutting corners in prefetching". You can prefetch as much data as you want but you still need to process it in the CPU pipeline and that shows up in time spent - IPC. How do you define IPC if not a speed of CPU at the same clock? Could you, please provide a link to generally recognized benchmarks that show Conroe lagging behind K7 at the same clock speed? No? I though so.
Yes, I know Athlon was outperforming Pentium before Core. That was the time when Intel was trying high clock speeds over IPC and failed. Core was the Intel answer and it did outperformed Athlon by large margin. Practiaclly in every software category Conroe was faster or much faster than K7 at the same clock speeds. That lasted with every CPU released until this day. Only Ryzen comes close to Intel CPU IPC but still fails to match.

Here is more than "cb" graph and multiple benchmarks:

https://www.anandtech.com/s...
Notice that Intel clocks its CPU slower than your K7.

I never said k7 was competing with conroe. Nice try. K7 was socket 462(socket a) in the period with Intel Pentium 4. Conroe FP vs 939 and AM2 was not superior except for higher clocks and faster cache and shortcuts in prefetching at the cost of security. Also Intel paid lots to have those benchmarks favor them in that period as well so most of these benchmarks a useless. I'm talking about real world server/workstation application use cases here. Amd was highly competitive until sandy. Intel was highly unti competitive before sandy. Nice try bud.

And I clearly said Intel core, not Pentium, had major IPC advantage. You keep coming back comparing Ahtlon to Pentium which I never said that was a good CPU.

And just to clarify your false beliefs in AM2 superiority:

https://www.anandtech.com/s...
https://www.extremetech.com...

Punitive damages, look it up.

The law was supposed to be about fairness. It degenerated to a game of get rich quickly for those who can't do anything useful for society. Why don't you design and produce a secure, fast chip that can compete with Intel instead?

Absolutely yes.

Why? Do you like to destroy good people and award bunch of human hyenas just for fun?

Except they DID intentionally do it. They intentionally made a less secure chip in exchange for performance. many of these vulnerabilities were found to be bad practice design in the name of gaining extra performance at the expense of security. They should be held accountable for allowing those design flaws to go forward.

Literally 20 years ago.... They didn't design this on purpose it is a flaw in their speculative execution. That means the chip tries to guess what you're going to use next and loads it there's a flaw with that. ALL moders cpus do it.

I remember the time when they had "a flaw" in their compiler as well. Guess what, it severely impacted AMD performance and was completely intentional.

How do you know so? The researcher from German university that discovered this vector of attack said in interview that this was very new approach to the security that nobody thought of before. Intel chips were designed many years earlier. So claiming that Intel did it intentionally is plain wrong. Can you point to any article describing this type of security flaw from 4 -5 years ago?

This was back when spectre was first discovered, I don't remember the exact report that said it, but there was a research firm that looked at the architecture and basically said that Intel did a design decision to sacrifice security for speed, since it would be difficult to exploit it. It's touched on in this article about the class action lawsuit, but I can't find the article I originally saw, that these allegations are likely based on. https://globaldatareview.co...

You missing the fact that Spectre affects practically every speculative execution CPU from every manufacturer simply because no expert in the field ever thought about this vector of attack. The article you listed is clearly lawyer speak in preparation to extort money. You can rest assured that if they win against Intel they will in turn sue all the rest. By your logic Intel, ARM, IBM, AMD all should be then sued for this because they all ignored this vector of attack. Who do you think will be the winner in this? Are you naive to think that customer will benefit from this?

How about their responsibility?

The problem is that they did not did it on purpose but rather it was discovered just recently as never thought vector of attack.
I know US legal system society and propaganda loves to blame past generations for "crimes" committed applying modern ethics or recent scientific knowledge. Famous lawsuits against asbestos are a good example. In most cases it is used to make millions for a few lawyers and society has nothing good out of it.

You can't, there is no 'lost performance', most security mitigations you apply at all levels will cost some performance. You can gain full performance by disabling these mitigations, and in fact there are additional features you could turn off to gain yet more perf.

It's a cost/benefit/workload decision, anyone operating at any scale should be doing a security architecture and threat model and determining which of these mitigations makes sense for their scenario.