Intel is still sneakily sabotaging AMD performance

dgz

Supreme [H]ardness
Joined
Feb 15, 2010
Messages
5,838
Pretty sure I read about this at least a couple of weeks ago and I believe it. Also, everyone should use Intel's compiler. Fuck this gcc open source crap. Surely chipzilla would never try to sabotage the competition. Ever.
 

Revdarian

2[H]4U
Joined
Aug 16, 2010
Messages
2,616
The behavior of Intel's MKL checking for Genuine Intel moniker has been widely known for ages. Technically they don't have a reason to change the libraries.

This isn't a secret, this isn't a conspiracy, Intel doesn't go out to say that amd won't get any special feature activated even if compatible they just say that it's optimized for Intel.

https://software.intel.com/en-us/forums/intel-distribution-for-python/topic/777540

You can check how the information stops the second that they got asked directly about if it accepts the supported features of amd.


Edit : there are ways to activate debugging modes with the libraries that can be used to test processors that due to being engineering samples or whatever don't have the moniker, it's the developer that should have set the flag to activate it tho as it was seen on math lab.
 

Dark12

[H]ard|Gawd
Joined
Dec 13, 2003
Messages
1,856
No kidding. I love their instant calls about Nvidia and Intel fanboys when this entire place is AMD circle jerk central.
It's almost as if a company's recent performance effects overall consumer sentiment.

_57c8a1a431a592af806925e57258202f.jpg
 

Arcygenical

Fully [H]
Joined
Jun 10, 2005
Messages
25,068
Yeah, and matlab fucking sucks.

I dare you to find the proper runtime exec for a matlab program written in 2012.

POS software for anything other than very current projects, up to date builds etc. Then... It's pretty dope.

Bah.

/Rant

I have a whole bunch of matlab sims from the... 2008 era and I can't find the redists for them anywhere. They're all discontinued on Matlab's site too.
 

N4CR

Supreme [H]ardness
Joined
Oct 17, 2011
Messages
4,944
LOL at people calling this a conspiracy. It's how the compiler is coded from Intel. It'd be like AMD coding one of their GPU apps to only work on AMD CPUs or severely slow them down if not.
You can do it when you are majority of the market. But after a while, people are going to get sick of that shit when your marketshare drops.. which is coming in the next few years.

Yeah, and matlab fucking sucks.

I dare you to find the proper runtime exec for a matlab program written in 2012.

POS software for anything other than very current projects, up to date builds etc. Then... It's pretty dope.

Bah.

/Rant

I have a whole bunch of matlab sims from the... 2008 era and I can't find the redists for them anywhere. They're all discontinued on Matlab's site too.

Fuck I went through that hell recently, I have the runtimes on an XP machine but not the setup for them. You can't get 09 runtimes anywhere on the net that I found >_<
I don't even run matlab. Just someone wrote a program that uses the runtimes. So I have to use a pretty important and extremely niche piece of software for my biz on a damn XP machine from time to time.
mclmcrrt711.dll may you burn in hell.
 

ThatITGuy

Gawd
Joined
May 5, 2017
Messages
577
No kidding. I love their instant calls about Nvidia and Intel fanboys when this entire place is AMD circle jerk central.
I would agree if you said PC fanboys and anti-Apple/Mac. I dont quite get the Intel / AMD favoritism here. There is definitely some rivalry (as with AMD vs Nvidia) but it feels like there are plenty on both sides.
 
Last edited:

SvenBent

2[H]4U
Joined
Sep 13, 2008
Messages
3,310
This is old news

The ICC did the same things. if its not Intel cpu disabled SSE SSE and SSE3 whatever if it supported or not
faking teh CPUID string suddenly gave magical performance increases on the same CPU

There are dedicted patch solutions that disabled intels CPU check so it has to go by feature checking instead

Again known for at least a decade. multiple lawsuits in the news

-- edit---
at least 10 years in the known
https://www.agner.org/optimize/blog/read.php?i=49#49

-- edit2--
Heck intel has to advertise their pratice after one of the lawsuits
Everyone that calls this conspiracy theory on something that Intel has public announced themselves really need to adjust how they approach new information.
You are building yourself an info bubble/fanboy bias

Reference from Intel
https://software.intel.com/en-us/articles/understanding-cpu-optimized-code-used-in-intel-ipp
"Intel's compilers may or may not optimize to the same degree for non-Intel microprocessors for optimizations that are not unique to Intel microprocessors. These optimizations include SSE2, SSE3, and SSSE3 instruction sets and other optimizations. Intel does not guarantee the availability, functionality, or effectiveness of any optimization on microprocessors not manufactured by Intel. Microprocessor-dependent optimizations in this product are intended for use with Intel microprocessors. Certain optimizations not specific to Intel microarchitecture are reserved for Intel microprocessors. Please refer to the applicable product User and Reference Guides for more information regarding the specific instruction sets covered by this notice."

TLDR: Intel admits it. its no secret. and has been known of over 10 years.
 
Last edited:

Krazy925

Supreme [H]ardness
Joined
Sep 29, 2012
Messages
6,504
I would agree if you said PC fanboys and anti-Apple/Mac. I dont quite get the Intel / AMD favoritism here. There is definitely some rivalry (as with AMD vs Nvidia) but it feels like there are plenty on both sides.
You can say the sky is blue here and find a line of people to argue with.
 
Joined
Oct 11, 2018
Messages
543
Intel vs AMD is older than the 586 arch.

Not sure how it's a surprise to anyone even a tiny bit interested in computers. Far far older than AMD vs Nvidia.

There's plenty of frivolous things people argue about here and everywhere... But Intel vs amd has many legitimate reasons for argument and fans.
 

Aireoth

Supreme [H]ardness
Joined
Oct 12, 2005
Messages
5,934
Same people always touting the same side (most people in here) = Fanboys.

Not that the AMD crowd is wrong at the moment, but even a broken clock is right twice a day.
 
Joined
Oct 11, 2018
Messages
543
Same people always touting the same side (most people in here) = Fanboys.

Not that the AMD crowd is wrong at the moment, but even a broken clock is right twice a day.

in nerd-land:
Fanboys is a derogatory term used to describe fans who ignore all facts and usually parrot lies / marketing nonsense to back their association even when it's been shown to be incorrect or irrelevant.

That doesn't encompass the term fans -- who will back something when it's doing well or not.

you can not agree with being a fan at all to any particular hardware company, but it would be hard to not cultivate a preference if you've had any prolonged contact with computer hardware...both building computers or the low level software interaction with them. It's not like they're the same on any level outside of adhering to certain industry standards.
 

Lakados

Supreme [H]ardness
Joined
Feb 3, 2014
Messages
5,779
So an Intel built DLL developed long before AMD had the feature set looks for an Intel chip that it knows as the function and disables it if it sees a non Intel chip which it doesn't know if it does have it or is compatible with it. I see nothing wrong with this that is a good programming practice from a compatibility and support stand point. There is nothing stopping AMD from developing their own DLL to supply the people at Matlab with, that could do the same for AMD chips so that they could be on equal footing, hell AMD could even develop and suppor a platform agnostic version of it.
 

Master_shake_

Fully [H]
Joined
Apr 9, 2012
Messages
17,795
So an Intel built DLL developed long before AMD had the feature set looks for an Intel chip that it knows as the function and disables it if it sees a non Intel chip which it doesn't know if it does have it or is compatible with it. I see nothing wrong with this that is a good programming practice from a compatibility and support stand point. There is nothing stopping AMD from developing their own DLL to supply the people at Matlab with, that could do the same for AMD chips so that they could be on equal footing, hell AMD could even develop and suppor a platform agnostic version of it.

did you see this part?

I dug into an old FTC investigation from 2009-2010 that determined, not only did the FTC order intel to STOP doing exactly what they're doing today, but they call out Intel's Math Kernel Library by name (which can be found in the last page of the conclusion):

IT IS FURTHER ORDERED that Respondent shall not make any engineering or design change to a Relevant Product if that change (1) degrades the performance of a Relevant Product sold by a competitor of Respondent and (2) does not provide an actual benefit to the Relevant Product sold by Respondent, including without limitation any improvement in performance, operation, cost, manufacturability, reliability, compatibility, or ability to operate or enhance the operation of another product; provided, however, that any degradation of the performance of a competing product shall not itself be deemed to be a benefit to the Relevant Product sold by Respondent. Respondent shall have the burden of demonstrating that any engineering or design change at issue complies with Section V. of this Order.

they were told to stop.

be funny to see AMD drag intel back to court and kick them while they are down.... literally.
 

Lakados

Supreme [H]ardness
Joined
Feb 3, 2014
Messages
5,779
did you see this part?



they were told to stop.

be funny to see AMD drag intel back to court and kick them while they are down.... literally.
Yeah but Matlab is using a DLL version from 2007.... what AMD processors supported SSE3-SSE4, AVX1/2 at that point in time. And what requirement did Intel have to support them. Yeah they were ordered to stop and supposedly did, but they don’t have to go back and back port changes to old code.
Would it be better if the code scanned for the feature set then ran it accordingly sure but then you run the risk of somebody else’s implementation not being compatible with yours. Is it dirty, a smidge, was it a solid implementation back in 2007 probably.

Edit:
AMD didn’t start shipping SSE3 until 2011, 4 years after the program was written. Same goes for the AVX support.

so yes that DLL is a perfect example of IF Intel Yes, ELSE NO, so of course it is going to get called out In court. But that doesn’t mean it was wrong at the time.
 
Last edited:

Aireoth

Supreme [H]ardness
Joined
Oct 12, 2005
Messages
5,934
in nerd-land:
Fanboys is a derogatory term used to describe fans who ignore all facts and usually parrot lies / marketing nonsense to back their association even when it's been shown to be incorrect or irrelevant.

That doesn't encompass the term fans -- who will back something when it's doing well or not.

you can not agree with being a fan at all to any particular hardware company, but it would be hard to not cultivate a preference if you've had any prolonged contact with computer hardware...both building computers or the low level software interaction with them. It's not like they're the same on any level outside of adhering to certain industry standards.

If all you do regardless of evidence is tout one side (and without pointing fingers you can easily see them in threads over the last decade), then you are a fanboy, blinded by a like/dislike (love/hate) relationship with an set of uncaring companies.

Preference to a brand or ecosystem doesn't mean you must be blindly loyal to said company.

I will agree that Intel fanboys are just as bad.
 
Last edited:

DeathFromBelow

Supreme [H]ardness
Joined
Jul 15, 2005
Messages
7,316
Some oddly defensive responses in here.

How well is Intel paying their social media manipulators these days?
 

MMitch

Gawd
Joined
Nov 29, 2016
Messages
807
To be fair, yes they use an old DLL but the fact it's so easy (when known) to actually "fix" this and that mathlab dev turn a blind eye is odd at best.
Don't they want their SW to work as best as possible on all HW ? Why don't they implement a fix for this behavior instead of calling out their outdated DLL.
 

Lakados

Supreme [H]ardness
Joined
Feb 3, 2014
Messages
5,779
To be fair, yes they use an old DLL but the fact it's so easy (when known) to actually "fix" this and that mathlab dev turn a blind eye is odd at best.
Don't they want their SW to work as best as possible on all HW ? Why don't they implement a fix for this behavior instead of calling out their outdated DLL.
The fix would be to write their own math library to replace the Intel one. Not a small undertaking, AMD should have put out their own libraries to provide developers with to counter the old Intel behaviour. Because until somebody builds the more than not you are going to see either sub standard implementations or people falling back on the old Intel ones.
 

MMitch

Gawd
Joined
Nov 29, 2016
Messages
807
The fix would be to write their own math library to replace the Intel one. Not a small undertaking, AMD should have put out their own libraries to provide developers with to counter the old Intel behaviour. Because until somebody builds the more than not you are going to see either sub standard implementations or people falling back on the old Intel ones.

? couldn't mathlab implement the workaround in the preference menu mentioned around the internet lol ? Using a simple command they're able to tap into the blocked power of AMD CPU.
I think the worst commotion is it's not 100% block, it can be bypassed but it's hidden from typical user. Mathlab could implement that workaround or document it properly...
 

Lakados

Supreme [H]ardness
Joined
Feb 3, 2014
Messages
5,779
? couldn't mathlab implement the workaround in the preference menu mentioned around the internet lol ? Using a simple command they're able to tap into the blocked power of AMD CPU.
I think the worst commotion is it's not 100% block, it can be bypassed but it's hidden from typical user. Mathlab could implement that workaround or document it properly...
Yeah that is Matlabs failure, but given the DLL is a black box to them can they would have to put some disclamers up on the work around which would possibly scare some places off regardless.
 
Top