More Intel leaks.. this one is not good though

[Stefan Payne]

1 minute ago, mr moose said:

2. AMD are effected, (ignore at your own peril).

Why do you lie and not mention that only Spectre one may (or may not) affect AMD??

Spectre 2 and Meltdown isn't a Problem for Ryzen.

[leadeater]

11 minutes ago, Stefan Payne said:

Spectre 2 and Meltdown isn't a Problem for Ryzen.

Until someone can predict with reasonable reliability what the branch predictor is going to do in advance and know which memory address is going to be used then Spectre 2, or what ever new name is used if found and named separately, is a problem. This is why the researchers said AMD processors are not off the table for both Spectre 1 and 2, they could not demonstrate 2 on any AMD processors but that isn't the same thing as being not susceptible to it.

 

I'd be rather cautious of outright saying AMD processors are not effected by Spectre 2 when the primary source of information regarding these security issues are not saying so.

[leadeater]

16 minutes ago, JoostinOnline said:

-snip-

But... sometimes I like a good pointless argument . Fine I'll go find something better to do.

[pas008]

23 minutes ago, Stefan Payne said:

Yes because all that matters right now is Spectre 2, not 1.

But it seems like you want to defend Intel and say that everyone is affected wich is lying by omission.

 

@pas008

Read the Arstechnica Article I posted.

https://arstechnica.com/gadgets/2018/01/heres-how-and-why-the-spectre-and-meltdown-patches-will-hurt-performance/

The "Zen escapes (again)" part. 

There you know that AMD isn't affected by most of it. 

Because they weren't targeted on a theory to crack them

 

Do you realize how many Windows patches microcode patches for security have gone on without our knowledge?

I don't know how long you been dealing with computers this one is only big because of mass social media

 

These large security firms are getting paid better now than being small hacker groups like the old days which are criminal so now they have more minds finding ways to break shit

 

Logic?

Quit being a fanboy or anti fanboy

 

[ItsMitch]

15 minutes ago, Stefan Payne said:

Why do you lie and not mention that only Spectre one may (or may not) affect AMD??

Spectre 2 and Meltdown isn't a Problem for Ryzen.

Oh my god we've lost him

[Zodiark1593]

3 minutes ago, leadeater said:

But... sometimes I like a good pointless argument . Fine I'll go find something better to do.

I used to like pointless arguments too. It just gets too predictable however. Even a surefire victory provides little satisfaction.

[Stefan Payne]

11 minutes ago, leadeater said:

Until someone can predict with reasonable reliability what the branch predictor is going to do in advance and know which memory address is going to be used then Spectre 2, or what ever new name is used if found and named separately, is a problem. This is why the researchers said AMD processors are not off the table for both Spectre 1 and 2, they could not demonstrate 2 on any AMD processors but that isn't the same thing as being not susceptible to it.

 

I'd be rather cautious of outright saying AMD processors are not effected by Spectre 2 when the primary source of information regarding these security issues are not saying so.

Might need to use Google Translate, but here is what someone I'd trust says:

https://www.forum-3dcenter.org/vbulletin/showpost.php?p=11602856&postcount=1355

 

So yes, it is highly unlikely to explit this on Samsung Exynos or AMD Ryzen processors because of the way the speculative excecution works. Because you can only access a certain area of memory, so it leaks way less than on Intel processors.

 

But since AMD has a huge advantage in this are we have to doctor the language so that it sounds that its as bad as on Intel, so that nobody thinks AMD is any safer.

That's what it leads to, doesn't it?

 

And to exploit it on AMD seems rather impractical, don't you agree?

[pas008]

3 minutes ago, Stefan Payne said:

Might need to use Google Translate, but here is what someone I'd trust says:

https://www.forum-3dcenter.org/vbulletin/showpost.php?p=11602856&postcount=1355

 

So yes, it is highly unlikely to explit this on Samsung Exynos or AMD Ryzen processors because of the way the speculative excecution works. Because you can only access a certain area of memory, so it leaks way less than on Intel processors.

 

But since AMD has a huge advantage in this are we have to doctor the language so that it sounds that its as bad as on Intel, so that nobody thinks AMD is any safer.

That's what it leads to, doesn't it?

 

And to exploit it on AMD seems rather impractical, don't you agree?

Then why do ryzen pro cpus exist?

Read the difference between then first

[mr moose]

16 minutes ago, leadeater said:

Until someone can predict with reasonable reliability what the branch predictor is going to do in advance and know which memory address is going to be used then Spectre 2, or what ever new name is used if found and named separately, is a problem. This is why the researchers said AMD processors are not off the table for both Spectre 1 and 2, they could not demonstrate 2 on any AMD processors but that isn't the same thing as being not susceptible to it.

 

I'd be rather cautious of outright saying AMD processors are not effected by Spectre 2 when the primary source of information regarding these security issues are not saying so.

 

14 minutes ago, JoostinOnline said:

Don't try to reason with him.  It won't get you anywhere.

 

11 minutes ago, pas008 said:

Quit being a fanboy or anti fanboy

 

 

11 minutes ago, SC2Mitch said:

Oh my god we've lost him

 

10 minutes ago, JoostinOnline said:

Spend 5 minutes looking through his posts.  I'm not sure why he hasn't been banned yet, but he trolls the forum like every day.  I try to ignore him, but sometimes his advice is so terrible that it's going to end up costing people money and I can't ignore it.

 

Yes it appears I have fallen victim to a troll.   I guess - "there is no difference between trolling, ignorance and stupidity.  They all flow from the same mindset".

 

 

 

 

[leadeater]

2 minutes ago, Stefan Payne said:

Might need to use Google Translate, but here is what someone I'd trust says:

https://www.forum-3dcenter.org/vbulletin/showpost.php?p=11602856&postcount=1355

 

So yes, it is highly unlikely to explit this on Samsung Exynos or AMD Ryzen processors because of the way the speculative excecution works. Because you can only access a certain area of memory, so it leaks way less than on Intel processors.

 

But since AMD has a huge advantage in this are we have to doctor the language so that it sounds that its as bad as on Intel, so that nobody thinks AMD is any safer.

That's what it leads to, doesn't it?

 

And to exploit it on AMD seems rather impractical, don't you agree?

Impractical or not yet know how isn't not possible though, it doesn't matter what sources you go find they all reference the research paper and the people who conducted it and it is their opinion after doing their research that Spectre 2 could be exploited on AMD processors however they were not able to figure out how. They could be wrong and it might not be possible or made any more practical but I wouldn't go round saying Spectre two outright is not an issue for AMD processors.

[Stefan Payne]

2 minutes ago, leadeater said:

Impractical or not yet know how isn't not possible though, it doesn't matter what sources you go find they all reference the research paper and the people who conducted it and it is their opinion after doing their research that Spectre 2 could be exploited on AMD processors however they were not able to figure out how. They could be wrong and it might not be possible or made any more practical but I wouldn't go round saying Spectre two outright is not an issue for AMD processors.

If that is the case, wouldn't that mean some bad legal ramifications for AMD because they were lying in their initial statement?? 
And cause some serious payouts - if they survive that...

[79wjd]

1 minute ago, Stefan Payne said:

If that is the case, wouldn't that mean some bad legal ramifications for AMD because they were lying in their initial statement?? 
And cause some serious payouts - if they survive that...

Because companies never lie. Companies are never wrong.

[leadeater]

1 minute ago, Stefan Payne said:

If that is the case, wouldn't that mean some bad legal ramifications for AMD because they were lying in their initial statement?? 
And cause some serious payouts - if they survive that...

Their statement isn't counter to what I or the researchers said, near zero risk is the current situation as it stands. No risk means never possible and that is not something AMD have said.

[leadeater]

On topic please

[JoostinOnline]

Just now, leadeater said:

On topic please

I'll leave the topic (or at least only comment about Spectre/Meltdown), but, respectively, I hope you and the rest of the staff will consider taking a look at his behavior.  So many topics turn into fights because of him.

[Zodiark1593]

Come to think of it, perhaps the most straightforward means of reducing the ill-effects of the Meltdown/Spectre vulnerabilities and maintain a strong single-thread performance lead is to ramp up the clocks further, which Intel seems to have been progressively pursuing as of late, with their latest efforts hitting 4.8 GHz out of the box.

 

While I've doubts that proper mitigations will be implemented in Intel's next generation, I believe there's a high probability that we'll be breaking the 5 GHz mark quite soon.

 

[Hiya!]

Ughh i think i have to sell my rig and switch to amd.

[mr moose]

Holy crap, this thread got derailed twice in one day.

 

meltdown spectre meltdown (just making sure I am on topic :D)

[leadeater]

1 hour ago, mr moose said:

meltdown spectre meltdown (just making sure I am on topic :D)

 

Two on topic comments in a row, we're back on track now 

[leadeater]

3 hours ago, Zodiark1593 said:

Come to think of it, perhaps the most straightforward means of reducing the ill-effects of the Meltdown/Spectre vulnerabilities and maintain a strong single-thread performance lead is to ramp up the clocks further, which Intel seems to have been progressively pursuing as of late, with their latest efforts hitting 4.8 GHz out of the box.

 

While I've doubts that proper mitigations will be implemented in Intel's next generation, I believe there's a high probability that we'll be breaking the 5 GHz mark quite soon.

I'm not too sure the frequencies can be reliably pushed much higher than what we have now, we're already seeing 'golden core' boosting now and I don't think node shrinks are going to help at all.

 

Maybe more reactive boost tables with more core combinations and finer control on the boost amount would help, you wouldn't see it in synthetic benchmarks though only in application work loads.

 

One thing Intel can do rather quickly is take a page out of AMD's book and not use 12bit alias BTB, it won't fix the issue but I'd guess it would be simple and quick to do for existing architecture designs too late in the design process to make any more significant changes.

[LAwLz]

11 hours ago, Stefan Payne said:

Could you please stop that FUD/Propaganda? 

That isn't true!

ARS Technica has an article about that and it shows that Zen is not really suseptible to Spectre.

 

https://arstechnica.com/gadgets/2018/01/heres-how-and-why-the-spectre-and-meltdown-patches-will-hurt-performance/

You might also want to read AMD's own statement:

Quote

Google Project Zero (GPZ) Variant 1 (Bounds Check Bypass or Spectre) is applicable to AMD processors.

 

GPZ Variant 2 (Branch Target Injection or Spectre) is applicable to AMD processors.

 

GPZ Variant 3 (Rogue Data Cache Load or Meltdown) is not applicable to AMD processors.

It is confirmed by AMD themselves that AMD processors are vulnerable to both Spectre 1 and 2.

 

I think it is amazing that you actually posted this same link a bit later, and somehow missed this. Or maybe you ignored it on purpose?

[LAwLz]

9 hours ago, Stefan Payne said:

Its not like AMDs statement on their website and to the press doesn't have any legal implications, now is it?! 

Nothing in AMD's statement contracited what I said though. They used very vague terms but if you actually read it they confirmed that they were vulnerable to both Spectre 1 and 2. AMD made it more clear which is which in their updated statement. People also seem to interpret "near zero risk" as "zero risk" which is incorrect.

"Near zero risk" means that they are vulnerable, but the risk of exploitation is very low (which could be said for all processors).

 

 

9 hours ago, Stefan Payne said:

So to say that AMD is as badly affected as Intel is just FUD because of the AMD statement that was out on the 3rd of January. And nobody could prove AMD wrong, so to say that 'AMD is as affected as Intel', was a lie and FUD because not true.

Nothing in my post was incorrect. It was you who added the "is as affected as Intel" part, not me.

All I said was that both Intel and AMD are affected, which is true and has been confirmed by AMD themselves.

If you can find a sentence in my post which is incorrect then go ahead and quote it. I don't think you will though, without adding in extra words I never wrote. I can however find several factual errors in your replies to and about me.

 

9 hours ago, Stefan Payne said:

Now with the added stuff, it should be clear that AMD is _NOT_ affected in any way close to how Intel is.

 

And the legal stuff:
IF AMD is wrong, its very very bad for them!

So you should trust that statement because if they are wrong, they are in serious trouble!!

Could you please stop moving the goalpost? The only one who has ever brought up the degrees of how badly affected certain processors are is you.

I made the statement that both Intel and AMD processors are vulnerable, which they are. This is correct.

You made the statement that they aren't. This is wrong because they are.

Then you slightly shifted the goalpost to being about "they are not affected in the same way" or "not affected as much". This might be true but I never made a post stating otherwise.

 

I also think it is a bit foolish to assume AMD processors are less susceptible to variant 2. They are right now with the knowledge we got at hand, but it might be discovered in the future that they are even easier to exploit than Intel, or it might be discovered that some Intel processors are harder to execute on.

Google had to put in quite a lot of time reverse-engineering the Haswell branch prediction in order to get it working. Different versions of Intel's branch predictors (each processor has several) may not be vulnerable to the same attack methods either. So in order to successfully carry out an attack you might need to know the processor architecture of the target machine (not too hard to figure out), as well as which branch prediction mechanism will be used in your specific case, Haswell has 3 or more (might be hard to figure out).

The generic predictor Google wrote had a good success rate for some cases, and less than 1% in other. They had to write processor and case specific exploits to get the percentage up. Is this what AMD means when they say there is a "near zero" risk? If so, pouring time into analyzing the Ryzen branch predictor techniques might make it easier to do, just like it did on Haswell.

 

 

9 hours ago, Stefan Payne said:

It is fair when the Discussin started with someone saying that AMD isn't affected by that and LAwLz trying to push the narrative that AMD is affected as well as Intel - wich they are not.

AMD are affected. This is fact. You might not want to believe it for some mysterious reason, but your feelings doesn't change facts. Even AMD themselves say they are affected.

 

9 hours ago, Stefan Payne said:

And luigi was right, LAwLz was wrong and tried to push the narrative that it was as bad as Intel on AMD processors - wich it is not and that was known at the time as well because of AMDs legally binding statement 3 days prior to luigi's post.

Can you please link to when I said something factually incorrect? I can't find it.

 

 

8 hours ago, Stefan Payne said:

Why do you lie and not mention that only Spectre one may (or may not) affect AMD??

Spectre 2 and Meltdown isn't a Problem for Ryzen.

Can you please stop accusing others of lying when the only one lying is you? Here is a quote from AMD:

Quote

GPZ Variant 2 (Branch Target Injection or Spectre) is applicable to AMD processors.

 

 

I would like an apology from you, for calling me a liar and someone who spreads FUD when I did neither thing (same can't be said for you though).

[jagdtigger]

5 hours ago, Zodiark1593 said:

Come to think of it, perhaps the most straightforward means of reducing the ill-effects of the Meltdown/Spectre vulnerabilities and maintain a strong single-thread performance lead is to ramp up the clocks further, which Intel seems to have been progressively pursuing as of late, with their latest efforts hitting 4.8 GHz out of the box.

 

While I've doubts that proper mitigations will be implemented in Intel's next generation, I believe there's a high probability that we'll be breaking the 5 GHz mark quite soon.

 

Well they can do it but then k CPU's would be meaningless, at those clocks the CPU's would be already at their limit. Intel is ran into a dead and. And their IPC gains arent that bright either, a 4670k clocked 300 MHz above the clock of the 7600 and it gives the almost the same performance....

[IgnorantBastrd]

I really don't like the way Intel has handled this whole situation.  If I had the money, I'd switch to an AMD rig.  At least they're getting on their shit, and their product line is nowhere near as ridiculous as Intels has become.

[Drak3]

On ‎1‎/‎21‎/‎2018 at 2:46 AM, leadeater said:

Two on topic comments in a row, we're back on track now

Is that my queue to derail this thread into oblivion?

On ‎1‎/‎20‎/‎2018 at 11:15 PM, Zodiark1593 said:

I believe there's a high probability that we'll be breaking the 5 GHz mark quite soon.

I doubt it, unless Intel or AMD push another Bulldozer, or become second sources for IBM's POWER arch.

On ‎1‎/‎20‎/‎2018 at 11:33 PM, Hiya! said:

Ughh i think i have to sell my rig and switch to amd.

The realistic impact of the meltdown fix is basically zero for most users, and so far, the worse version of Spectre requires physical access to really exploit. There's not really any reason to switch to AMD as a responce to this, especially since something as/more sinister could be lurking in AMD's processors that hasn't been discovered yet.