Intel CPUs afflicted with simple spec-exec vulnerability

[SenKa]

Just now, dalekphalm said:

That's a whole lot of PR speak. It says "Traditional Windows Desktop" application, but doesn't specifically say x86. Granted, I think it's safe to assume that's what they meant, but I'll believe it when I see it.

"We benchmark the snapdragon 835 running windows in both emulated X86 apps and native UWP apps" - Hardware Unboxed. Your assumption is correct.

 

2 minutes ago, dalekphalm said:

Yes I saw that lawsuit, and it's goddamn stupid. The worst part about it, is that some Judge who doesn't know a thing about CPU's might end up creating a legal definition for "CPU Core". It's a dangerous precedent.

 

The guy was basically mad because he thought "more cores must be faster", when that's not how it works. Cores and Architecture are independent (though related) things. If the architecture regresses, then each core will be slower.

 

So, will he win? Only time will tell - I personally hope he does not win.

Stupid or not, it will create a precedent and it is one we all must follow (at least those of us in the states). Will the ruling be found properly, or even define the question at hand at all? Only time will tell. But, stupid or not it will be the law.

[pas008]

22 minutes ago, dalekphalm said:

 

So, as far as you're concerned, Intel doing something anti-consumer is okay because the consumer is stupid? We'll have to disagree hard here.

Bulldozer had a regression in per core performance. That's why it was worse and couldn't compete. They had a long term strategy that failed horribly.

 

 

what this exists everywhere

from vehicles to food

 

I could go on

 

anti consumer would be saying it has 8 cores when in fact it has 4

or 6 cylinder when its a 4

they didnt dupe anyone into anything they were selling a niche product in a niche market

[SenKa]

1 minute ago, pas008 said:

anti consumer would be saying it has 8 cores when in fact it has 4

or 6 cylinder when its a 4

they didnt dupe anyone

Arguably, thats exactly what AMD did. The "Eight Core" FX chips could be argued to be 4 cores. But, that's a topic for the courts and is in the courts as we speak.

[pas008]

4 minutes ago, SenpaiKaplan said:

Arguably, thats exactly what AMD did. The "Eight Core" FX chips could be argued to be 4 cores. But, that's a topic for the courts and is in the courts as we speak.

didnt it have 8 cores just shared resources?

 

cant remember been awhile since i looked at it

[dalekphalm]

15 minutes ago, SenpaiKaplan said:

"We benchmark the snapdragon 835 running windows in both emulated X86 apps and native UWP apps" - Hardware Unboxed. Your assumption is correct.

 

Stupid or not, it will create a precedent and it is one we all must follow (at least those of us in the states). Will the ruling be found properly, or even define the question at hand at all? Only time will tell. But, stupid or not it will be the law.

Actually, not every case will set precedence. It depends on the specifics of each and every case.

 

Either way, I sincerely hope no definition of CPU is declared. IF we are to make an industry definition, it should be agreed upon by all parties in an Industry Wide Standard.

6 minutes ago, pas008 said:

what this exists everywhere

from vehicles to food

 

I could go on

 

anti consumer would be saying it has 8 cores when in fact it has 4

or 6 cylinder when its a 4

they didnt dupe anyone into anything they were selling a niche product in a niche market

That is anti-consumer, yes, but many other things are also anti-consumer.

4 minutes ago, SenpaiKaplan said:

Arguably, thats exactly what AMD did. The "Eight Core" FX chips could be argued to be 4 cores. But, that's a topic for the courts and is in the courts as we speak.

Disagree - you could argue that their 8 core was a 4 core, but I wouldn't agree with you.

[pas008]

5 minutes ago, dalekphalm said:

Actually, not every case will set precedence. It depends on the specifics of each and every case.

 

Either way, I sincerely hope no definition of CPU is declared. IF we are to make an industry definition, it should be agreed upon by all parties in an Industry Wide Standard.

That is anti-consumer, yes, but many other things are also anti-consumer.

Disagree - you could argue that their 8 core was a 4 core, but I wouldn't agree with you.

niche products in niche markets isnt anti consumer in my book sry

[Stefan Payne]

1 hour ago, pas008 said:

stagnant but developing in many other areas hmmm

The first 8 Core Intel developed was the Beckton in 45nm.

Wich seem to have been like 684mm²

https://en.wikipedia.org/wiki/Nehalem_(microarchitecture)

 

Nehalem was around 300mm² (263mm² for Bloomfield, 296mm² for Lynnfield) or so and sold for as low as ~200€ at the time.

6 Core Westmere was even a bit smaller at 248mm²

And the Ultra High Core Cound Westmere-EX even had 2 Cores more than Beckton while only beeing around 513mm².

And the next Shrink (Ivy Bridfge) was only 133mm² with the smaller GPU (HM-4)

The 6 Core Ivy-E was only around 256,5mm².

Ryzen was around 200mm² when it came out and had 8 cores...

Quote

funny thing is competition is usually what drives innovation

Yes and you know another thing that competition usually does?
It drowns your profit on products. You have to "reasonably price" your products and can't just take whatever you want.

 

Quote

intel released 8 core mainstream on haswell you still be bitching and intel would be stuck with stricker monopoly laws on every cpu sounds like smart business decision

Why so salty?

Ever heard about the i940GML and the "Features" of it? 

Good, that's a bit older...

But right now, try putting a Xeon CPU on a Desktop Board. Oh it doesn't work??
For no reason?

Or the LGA1151 Socket Switch. 

Yeah, there is much to Criticise Intel for bad practices and still people defend the stuff they do, like the Change in TDP Definition, wich is defended by some people because a Company specifies it, it is fine...

Quote

tdp thing has been explained to you many times

So you're saying AMD's ACP was a good idea and Intel was wrong with criticising it??

Oh and there is no Problem in using TDP in a different way than most Engineers understand it. Absolutely...

Quote

10nm is huge problem right now when their supposedly best 14nm products are still ahead in many areas on amds

How can you say such things?!

Just look at all the examples I've given.

 

What you claim is basically that it didn't matter for Intel to have 22nm, when Bulldozer was there.

It didn't make a difference for AMD to have only a not that great 32nm Process for theri CPUs.

 

Quote

plus arent they opening up couple more fabs?

Wich helps in what way when the demand for your product plummets?!

 

And opening more fabs helps in what way with having the inferior process??

I don't get what you're at here...

 

Quote

qualcomm comment lol seriously when they are the biggest monopolistic company on arm cpus kinda funny

Yeah, but right now you are defending one of the worst monopolisitc companys ever in existance.

You know that Intel allowed/licensed their products to a bunch of other companies such as TI, Siemens but also AMD and others.

And when it wasn't convenient for them, they sued the 3rd Party Mamufacturers out of Business. Only AMD and Cyrix Survived at the time.

Cyrics only for a couple of more years until the mid/late 90s and then went bust...

 

And there were other 

Quote

avx, etc is improvements also

Yes and?
When was it introduced?

AVX was Sandy Bridge, wich is what @dalekphalm said, that after 2nd gen Core i-Series they stopped innovating and didn't do too much. 

Well, except for increasing their profit margin, wich is essential for their existance right now, it seems.

 

But with the upcominc competition and the prices we've seen, without any competing Product from Intel, it looks that their profit per chip goes down the toilet and they have to reduce the prices dramatically...

 

Well, with all the People out there that claim they only want to buy Intel they might not need it. And They might be willing to pay the 500€ for allegedly around the same performance than the competition, wich only wants around half the Price...

[Stefan Payne]

32 minutes ago, SenpaiKaplan said:

Arguably, thats exactly what AMD did. The "Eight Core" FX chips could be argued to be 4 cores. But, that's a topic for the courts and is in the courts as we speak.

NO, its 8Cores if you define Core as something that can output Int stuff.

 

I've said it again sometimes. When the FX is a 4 Core CPU, then the Intel 8086-80386 and 80486SX and the x88 Versions are 0 Core CPUs.

Well, they are not, they are 1 Core CPUs.

And they don't have any FPU, Multimedia stuff or other things, just pure Integer excecution processors.

 

Conclusion: AMD FX8k Series is an 8 Core because a Core is not defined and was classically just a thing that does Integer.

FPU was sold seperately (i8087-80387 and even 80487 there was. Also from other companys like WEITEK there were Float Point Processors).

[pas008]

39 minutes ago, Stefan Payne said:

1

The first 8 Core Intel developed was the Beckton in 45nm.

Wich seem to have been like 684mm²

https://en.wikipedia.org/wiki/Nehalem_(microarchitecture)

 

Nehalem was around 300mm² (263mm² for Bloomfield, 296mm² for Lynnfield) or so and sold for as low as ~200€ at the time.

6 Core Westmere was even a bit smaller at 248mm²

And the Ultra High Core Cound Westmere-EX even had 2 Cores more than Beckton while only beeing around 513mm².

And the next Shrink (Ivy Bridfge) was only 133mm² with the smaller GPU (HM-4)

The 6 Core Ivy-E was only around 256,5mm².

Ryzen was around 200mm² when it came out and had 8 cores...

2

Yes and you know another thing that competition usually does?
It drowns your profit on products. You have to "reasonably price" your products and can't just take whatever you want.

 

3

Why so salty?

Ever heard about the i940GML and the "Features" of it? 

Good, that's a bit older...

But right now, try putting a Xeon CPU on a Desktop Board. Oh it doesn't work??
For no reason?

Or the LGA1151 Socket Switch. 

Yeah, there is much to Criticise Intel for bad practices and still people defend the stuff they do, like the Change in TDP Definition, wich is defended by some people because a Company specifies it, it is fine...

So you're saying AMD's ACP was a good idea and Intel was wrong with criticising it??

Oh and there is no Problem in using TDP in a different way than most Engineers understand it. Absolutely...

4

How can you say such things?!

Just look at all the examples I've given.

 

What you claim is basically that it didn't matter for Intel to have 22nm, when Bulldozer was there.

It didn't make a difference for AMD to have only a not that great 32nm Process for theri CPUs.

 

Wich helps in what way when the demand for your product plummets?!

 

And opening more fabs helps in what way with having the inferior process??

I don't get what you're at here...

 

Yeah, but right now you are defending one of the worst monopolisitc companys ever in existance.

You know that Intel allowed/licensed their products to a bunch of other companies such as TI, Siemens but also AMD and others.

And when it wasn't convenient for them, they sued the 3rd Party Mamufacturers out of Business. Only AMD and Cyrix Survived at the time.

Cyrics only for a couple of more years until the mid/late 90s and then went bust...

 

And there were other 

6

Yes and?
When was it introduced?

AVX was Sandy Bridge, wich is what @dalekphalm said, that after 2nd gen Core i-Series they stopped innovating and didn't do too much. 

Well, except for increasing their profit margin, wich is essential for their existance right now, it seems.

 

But with the upcominc competition and the prices we've seen, without any competing Product from Intel, it looks that their profit per chip goes down the toilet and they have to reduce the prices dramatically...

 

Well, with all the People out there that claim they only want to buy Intel they might not need it. And They might be willing to pay the 500€ for allegedly around the same performance than the competition, wich only wants around half the Price...

sorry trying to do this at work on phone 

1

first part I have no clue what you are getting at

yes they developed better instruction sets

 

2

drowns lol intel is drowning in money lol

 

3

then again dont know what you are getting to on that part

 

4

well amd compared their new 7nm to 9900k and just marginally ahead

hopefully way better though

like I said intel hasnt done die shrink yet

 

5

they offer many products besides cpus

and they cant meet demand right now

 

6

and avx and other instructiion sets have got better have they not?

 

like I said I buy best products for me and right now thats intel on the cpu side

 

be nice if they have to lower prices per chip we all win that is what competition does though

 

we dont make products here at my work without having a margin to go by or even pushing it bigger/multiplying it considering who/what/where/when

but if competition quotes it close to us, we drop the price if we have time also at times

meaning why should intel lower its price if they are selling out of all 14nm chips lol

supply and demand, plus people are paying it

its how business works

 

 

[Blademaster91]

2 hours ago, BuckGup said:

Every browser uses javascript. Otherwise almost all sites today would be broken

Yeah but its easy to block, doesn't mean browsers or sites should though, Javascript has too many security holes.

2 hours ago, dalekphalm said:

To be fair, an EXTREME minority of LTT users use stuff like NoScript and block all JavaScript from running by default. They may have an incorrect assumption that everyone does this.

 

But yes, the vast vast vast majority of users have JavaScript enabled - even among us at LTT.

And most users allow ads too, which can infect a system with malware.

[Mister Woof]

Let's do more victim shaming.

 

You are literally blaming Intel for their hardware being victimized.

 

Nothing is 100% secure and all you idiots revelling at intel being compromised to further your AMD fanboyism are basically rooting for the bad guys to keep winning.

 

Even if your personal systems are AMD, I guarantee you that some service you use, including your financial institutions and government, uses Intel and is affected by this. You are not immune and you should not be gloating.

[Dylanc1500]

My goodness, discussion about things is fantastic and should continue. However, the emotion driven conversations happening make having constructive discussion about this topic very hard. Especially when people are discussing and arguing with a basis that is entirely not to do with the topic at hand.

 

On topic though, this is something that will most likely be fixed in the near future. Now, when a fix within the microarchitecture will happen, well that depends on the difficulty and cost (both monetarily and in throughput). Hopefully, again, sooner rather than later. It is of concern but I personally don't feel it is as large as some are making it out to be.

[dalekphalm]

10 minutes ago, Plutosaurus said:

Let's do more victim shaming.

 

You are literally blaming Intel for their hardware being hacked by malicious people.

 

Nothing is 100% secure and all you idiots revelling at intel being compromised to further your AMD fanboyism are basically rooting for the bad guys to keep winning.

Now, I don't personally believe that Intel "knew about it" and chose to not fix the issue - that's some next level tinfoil hat conspiracy stuff.

 

But, one might argue they should do more testing for these kinds of bugs - especially after Spectre and Meltdown shamed Intel pretty badly.

[Mister Woof]

3 minutes ago, dalekphalm said:

Now, I don't personally believe that Intel "knew about it" and chose to not fix the issue - that's some next level tinfoil hat conspiracy stuff.

 

But, one might argue they should do more testing for these kinds of bugs - especially after Spectre and Meltdown shamed Intel pretty badly.

This affects every core generation, so likely this has been around for over 15 years and nothing ever came of it until now.

 

And AMD isn't avoiding this issue because of altruism or some imaginary higher technology. They escaped it out of chance.

 

If they thought up the application of speculative execution that Intel used before Intel did, you better believe they'd have used it too.

[Alan G]

All this doom and gloom must be the real reason why Intel stock is up 14% this year (disclosure:  I have been an Intel shareholder for a lot of years and merrily see the price go up and the dividend checks come in)

[Stefan Payne]

47 minutes ago, pas008 said:

sorry trying to do this at work on phone 

1

first part I have no clue what you are getting at

yes they developed better instruction sets

I'm getting at that Intel has had 8 Core CPUs for more than 10 years.

And with the shrink to even 32nm (Westmere) a 6 core Desktop CPU would have been more than viable as we have two Data sets on that: Westmere with 248mm²

 

Instead they chose to not do it and increase their profit.

47 minutes ago, pas008 said:

2

drowns lol intel is drowning in money lol

right now....

Now imagine the price of processors is halved.

Now imagine the sales of processors goes down 25%

 

And now look at the company form.

And now look at what happens if the reenue, even if it is only the CPU area, drops by more than 50%...

 

47 minutes ago, pas008 said:

3

then again dont know what you are getting to on that part

Show you some anti-comsumer shit that should start a shitstorm but didn't for whatever reason.

 

47 minutes ago, pas008 said:

4

well amd compared their new 7nm to 9900k and just marginally ahead

hopefully way better though

like I said intel hasnt done die shrink yet

AMD was what you called SANDBAGGING, they did show a CPU, one thats on the lower end and could even be called Ryzen 5.

 

Now imagine back in 2016 AMD took an i7-6700K  and took this CPU and demonstrated THAT against the 6700K and would have been slightly faster.

You now get why shit is hitting the fan?! 

There are 12 Cores and 16 Core Versions of the Ryzen 3000 CPU above this!

 

And Why you ignore the power consumption??
Depending on the source/calculations, its between 40% or 50% lower!

So around half the Power COnsumption of the CPU!

That's only marginally better? RLY??

 

And still, its the "low end" 8 Core!

There will be a 12 Core and 16 Core CPU...

They took one of the shittiest Ryzen 3000 CPUs and showed it off against the _FASTEST_ Intel LGA1151 CPU and beat it.

How the heck can you downplay that?!

 

47 minutes ago, pas008 said:

5

they offer many products besides cpus

and they cant meet demand right now

Yes, because they messed up?
What is right now is now, BEFORE AMD really released their 7nm Lineup.

How do you think it will go when someone will offer a product with the same performance, around half the power consumption and half the price??

 

Why don't you mention the Rome demonstration where a SINGLE SOCKET AMD Rome beat a 2 Socket Intel! IIRC also the best they had.

 

You clicked on the "Statista" Link I provided??

They show that CPUs might be around 60% of Intel.

 

[imreloadin]

30 minutes ago, Plutosaurus said:

You are literally blaming Intel for their hardware being victimized.

So if I were to write some pleb code or have shit IT security practices that causes millions of customers to have their personal information breached then I'm not at fault? Do you work for Equifax or something because that train of thought is absolutely ridiculous

[Mister Woof]

Just now, imreloadin said:

So if I were to write some pleb code or have shit IT security practices that causes millions of customers to have their personal information breached then I'm not at fault? Do you work for Equifax or something because that train of thought is absolutely ridiculous

This is not the same thing at all, lol

[imreloadin]

Just now, Plutosaurus said:

This is not the same thing at all, lol

How so? Would I not be a victim in the fact that people would be taking advantage of something I created? That's literally the definition of Victim Blaming...

[Mister Woof]

1 minute ago, imreloadin said:

How so? Would I not be a victim in the fact that people would be taking advantage of something I created? That's literally the definition of Victim Blaming...

"Pleb code and shit security practices" is clearly hyperbole.

 

It affects first gen core CPUs, which have been out for over 10 years, and this is only now coming to light in 2018/19.

 

If it was indeed "pleb code and shit security practices" it wouldn't have taken a decade to exploit.

[Mister Woof]

As the owner of both platforms, and mostly AMD GPUs, whenever I hear about more Intel exploits, my first inclination isn't, "haha Intel you suck, it's all your fault!", but, "damn you society, this is why we can't have nice things."

[imreloadin]

1 minute ago, Plutosaurus said:

"Pleb code and shit security practices" is clearly hyperbole. 

 

It affects first gen core CPUs, which have been out for over 10 years, and this is only now coming to light in 2018/19.

 

If it was indeed "pleb code and shit security practices" it wouldn't have taken a decade to exploit.

I wasn't quoting anything in regards to this particular breach. I was talking about when you said this:

44 minutes ago, Plutosaurus said:

Let's do more victim shaming.

 

You are literally blaming Intel for their hardware being victimized.

If it's victim blaming when it happens to Intel how is it not victim blaming when it happens to others?

Thousands of other companies have had their software and hardware "victimized" by hackers, do you not consider it victim shaming when they get called out on it?

[Mister Woof]

2 minutes ago, imreloadin said:

I wasn't quoting anything in regards to this particular breach. I was talking about when you said this:

If it's victim blaming when it happens to Intel how is it not victim blaming when it happens to others?

Thousands of other companies have had their software and hardware "victimized" by hackers, do you not consider it victim shaming when they get called out on it?

See my above post

 

Also, those companies DO employ shit code and shit security practices, Intel does not.

 

They tried to make abetter faster product and it got exploited in ways they didn't predict, and it happened a decade after the tech was created.

[yian88]

7 hours ago, Dogeystyle said:

Let's all go back to vacuum transistors, punch cards and morse code. Secure, simple, cheap,

I have a better idea leave everything as it is and dont patch anything because there is no vulnerability.

In order to do any of these attacks, attackers and software must be run on that platform along security critical software.

Tell me how on earth does patching meltdown /spectre helps your system if an attacker has gained access to the machine and has privileges to run software/spyware of any kind?

I keep reading these vulnerabilities found by researchers and i fail to understand how can someone take advantage of any of them without access to the machine and have privilege to run software that exploits them.

Someone school me please.

[imreloadin]

Just now, Plutosaurus said:

See my above post

That runs counter to this post however...

15 minutes ago, Plutosaurus said:

This is not the same thing at all, lol

You're the one who stated there is a difference between Intel being subject to "victim shaming" and when it happens to company XYZ. So, again, why are they different things in your opinion?