Clarify this for me, please

[Particle]

Yes, I know the game. It's an endless loop and it applies to more than just AMD vs. Intel. It is the typical pattern observed on any message board on any debate topic.

OP: <Concept 1> or <concept 2> is more <comparison>?
Person 1: Opinion about a comparison.
Person 2: Opposite opinion about comparison.
Person 1: OPINION ABOUT COMPARISON!
Person 3: On-topic response to OP that nobody will notice.
Person 2: NO! OPPOSITE OPINION ABOUT COMPARISON!
Person 1: OPINION ABOUT COMPARISON AND INSULT TO PERSON 2!
Person 2: YALALALA OPPOSITE OPINION AND BIGGER INSULT TO PERSON 1!
Person 3: Request for clarification on why Person 1 and Person 2 are fighting. Acknowledgement that both have points. [nobody notices]
Person 1: OH YEAH WELL SLIGHT TOPIC DEVIATION AND POORLY CONSTRUCTED ANALOGY! CLAIM OF ABSOLUTE CORRECTNESS!
Person 2: LALALALALALA STATEMENT OF OWN CORRECTNESS AND REFUSAL TO ACKNOWLEDGE PERSON 1 ANYMORE.

And so on, and so on, and so on, and so on until a discussion about two telephones' signal quality ends up with a discussion about brands of peanut butter and why one makes a person supporter of terrorism if they purchase it.

In the end, neither person learns anything, they both still absolutely believe their original opinions are correct, and will defend it rigidly until the day they die despite the ever-changing world of technology or whatever topic their discussion was about.

[keithlm]

Ouch.

You know, I just defended these smoothness statements that were being shot at by intel believers. But to call me an Intel fanatic? Really. I just am sceptical about everything, not just benchmarks that declare Intel the winner. You are right though that I could search for my own, but I figured it could take days before I find that specific review you mentioned earlier.

Anyway, indeed back on topic.

Sorry. You didn't actually call me a liar... the other guy did. But that's enough to make me start swinging at anyone that is playing the game. (You just happened to be there.)

Besides what happened yesterday is more like this: (PARAPHRASED)

ME: "This is a stupid game that is played way too often on this forum. I'm not playing. It's pointless and the result is always the same."

HIM: "Aha... see you won't play the game correctly so I WIN!".

ME: "Uh.. no... I'm not playing anymore so there is no winner or loser."

HIM: "See I WIN because you lie! Liar!"

(You just added some comments in between a few of the lines... )

[Devil's Prophet]

Yes, I know the game. It's an endless loop and it
In the end, neither person learns anything, they both still absolutely believe their original opinions are correct, and will defend it rigidly until the day they die despite the ever-changing world of technology or whatever topic their discussion was about.

What I notice a lot of times is that both opinions are correct, but that they cover a differenct perspective or ideal about how to approach the problem that is being opinionated.

Maybe they just make their statement and refuse to move on before the opponent in the discussion acknowledges the validity of that statement. The problem is that both do that same thing and none will acknowledge before their own statement is acknowledged. And just like I mentioned before, those opinions could seem to be about the same thing, but actually they are just a small scope of the entire problem, with each statement covering a different part of that problem.

Do you feel me?

Ok, that was my opinion about it.

[gallag]

Sorry. You didn't actually call me a liar... the other guy did. But that's enough to make me start swinging at anyone that is playing the game. (You just happened to be there.)

Besides what happened yesterday is more like this: (PARAPHRASED)

ME: "This is a stupid game that is played way too often on this forum. I'm not playing. It's pointless and the result is always the same."

HIM: "Aha... see you won't play the game correctly so I WIN!".

ME: "Uh.. no... I'm not playing anymore so there is no winner or loser."

HIM: "See I WIN because you lie! Liar!"

(You just added some comments in between a few of the lines... )

It was me, because you told a lie.

You said "The majority of the reviews show the 9850 beating the heck out of the Q6600"

This one statement, nothing else, I have a problem with as it is a lie. Simple.

[SkullCracka]

It was me, because you told a lie.

You said "The majority of the reviews show the 9850 beating the heck out of the Q6600"

This one statement, nothing else, I have a problem with as it is a lie. Simple.

heres the deal online benchamrks and reviews are flawed and scewed. period.

they can be twisted to make the outcome how ever the reviewer wants.

Now get over the uber smoothness of the amd systems. Run what you brung punk. No matter the system. It wont make you win any games. You will still be a lo0ser. now run along with your head between your legs back to the intel or what ever thread or forum you came from or did they run you off aswell.


there now we have gone full circle. acording to particles endless loop theory.

[LIKMARK]

Yes, I know the game. It's an endless loop and it applies to more than just AMD vs. Intel. It is the typical pattern observed on any message board on any debate topic.

OP: <Concept 1> or <concept 2> is more <comparison>?
Person 1: Opinion about a comparison.
Person 2: Opposite opinion about comparison.
Person 1: OPINION ABOUT COMPARISON!
Person 3: On-topic response to OP that nobody will notice.
Person 2: NO! OPPOSITE OPINION ABOUT COMPARISON!
Person 1: OPINION ABOUT COMPARISON AND INSULT TO PERSON 2!
Person 2: YALALALA OPPOSITE OPINION AND BIGGER INSULT TO PERSON 1!
Person 3: Request for clarification on why Person 1 and Person 2 are fighting. Acknowledgement that both have points. [nobody notices]
Person 1: OH YEAH WELL SLIGHT TOPIC DEVIATION AND POORLY CONSTRUCTED ANALOGY! CLAIM OF ABSOLUTE CORRECTNESS!
Person 2: LALALALALALA STATEMENT OF OWN CORRECTNESS AND REFUSAL TO ACKNOWLEDGE PERSON 1 ANYMORE.

And so on, and so on, and so on, and so on until a discussion about two telephones' signal quality ends up with a discussion about brands of peanut butter and why one makes a person supporter of terrorism if they purchase it.

In the end, neither person learns anything, they both still absolutely believe their original opinions are correct, and will defend it rigidly until the day they die despite the ever-changing world of technology or whatever topic their discussion was about.

The best summary ever. Sticky material

[keithlm]

heres the deal online benchamrks and reviews are flawed and scewed. period.

they can be twisted to make the outcome how ever the reviewer wants.

Now get over the uber smoothness of the amd systems. Run what you brung punk. No matter the system. It wont make you win any games. You will still be a lo0ser. now run along with your head between your legs back to the intel or what ever thread or forum you came from or did they run you off aswell.


there now we have gone full circle. acording to particles endless loop theory.

BINGO.

To call somebody a liar because their subjective interpretation of questionable results does not match your opinion shows a definite lack of maturity.

(Besides it is obvious that he is jumping onto the one thing he can criticize from my posts... since he can't refute anything else I stated.)

[stocius]

Yes, I know the game. It's an endless loop and it applies to more than just AMD vs. Intel. It is the typical pattern observed on any message board on any debate topic.

OP: <Concept 1> or <concept 2> is more <comparison>?
Person 1: Opinion about a comparison.
Person 2: Opposite opinion about comparison.
Person 1: OPINION ABOUT COMPARISON!
Person 3: On-topic response to OP that nobody will notice.
Person 2: NO! OPPOSITE OPINION ABOUT COMPARISON!
Person 1: OPINION ABOUT COMPARISON AND INSULT TO PERSON 2!
Person 2: YALALALA OPPOSITE OPINION AND BIGGER INSULT TO PERSON 1!
Person 3: Request for clarification on why Person 1 and Person 2 are fighting. Acknowledgement that both have points. [nobody notices]
Person 1: OH YEAH WELL SLIGHT TOPIC DEVIATION AND POORLY CONSTRUCTED ANALOGY! CLAIM OF ABSOLUTE CORRECTNESS!
Person 2: LALALALALALA STATEMENT OF OWN CORRECTNESS AND REFUSAL TO ACKNOWLEDGE PERSON 1 ANYMORE.

And so on, and so on, and so on, and so on until a discussion about two telephones' signal quality ends up with a discussion about brands of peanut butter and why one makes a person supporter of terrorism if they purchase it.

In the end, neither person learns anything, they both still absolutely believe their original opinions are correct, and will defend it rigidly until the day they die despite the ever-changing world of technology or whatever topic their discussion was about.

QFT! I think this ought to be stickied as well. Particle, would you mind if I posted this (giving you credit of course) in my Offensive Fanboyism thread? It illustrates exactly what I was trying to say.

PS. I win! I win! What I won, I don't really know but I won!!!

[Particle]

I'm glad you guys liked my summary of the endless cycle. Yeah, go ahead and post it or do whatever you want with it.

[gallag]

BINGO.

To call somebody a liar because their subjective interpretation of questionable results does not match your opinion shows a definite lack of maturity.

(Besides it is obvious that he is jumping onto the one thing he can criticize from my posts... since he can't refute anything else I stated.)

"The majority of the reviews show the 9850 beating the heck out of the Q6600"
"The majority of the reviews show the 9850 beating the heck out of the
Q6600"
"The majority of the reviews show the 9850 beating the heck out of the
Q6600"
"The majority of the reviews show the 9850 beating the heck out of the Q6600"
"The majority of the reviews show the 9850 beating the heck out of the Q6600"

Just one question. Is this true

No smartness just a yes or no.

In fact to analise your words in more depth, I see this a an admition of guilt

"Besides it is obvious that he is jumping onto the one thing he can criticize from my posts"

What will be have if we let lies go unchecked?

Also you say that I stick to this because i can not refute any thing else you said. This is simply because I cannot refute anything else you said, It is entirely plausible that amd cpus are smother than Intel cpus due to the reasons people have put forward in this thread. This is the art of debate.

[gallag]

heres the deal online benchamrks and reviews are flawed and scewed. period.

they can be twisted to make the outcome how ever the reviewer wants.

Now get over the uber smoothness of the amd systems. Run what you brung punk. No matter the system. It wont make you win any games. You will still be a lo0ser. now run along with your head between your legs back to the intel or what ever thread or forum you came from or did they run you off aswell.


there now we have gone full circle. acording to particles endless loop "theory.

This is without exaggeration the most puerile response I have ever seen on xs.

well done.

If I can just address this point.

"heres the deal online benchamrks and reviews are flawed and scewed. period.

they can be twisted to make the outcome how ever the reviewer wants."

This is a plausible opinion, and all I could do is put forward my opinion and discuss it with you, after all this is what we are here for, No?

But if you said that all benchmarks show amd thrashing Intel, then this is a blatant lie and should be addressed as such. Do you not agree?

[stocius]

"The majority of the reviews show the 9850 beating the heck out of the Q6600"
"The majority of the reviews show the 9850 beating the heck out of the
Q6600"
"The majority of the reviews show the 9850 beating the heck out of the
Q6600"
"The majority of the reviews show the 9850 beating the heck out of the Q6600"
"The majority of the reviews show the 9850 beating the heck out of the Q6600"

Just one question. Is this true

No smartness just a yes or no.

In fact to analise your words in more depth, I see this a an admition of guilt

"Besides it is obvious that he is jumping onto the one thing he can criticize from my posts"

What will be have if we let lies go unchecked?

Also you say that I stick to this because i can not refute any thing else you said. This is simply because I cannot refute anything else you said, It is entirely plausible that amd cpus are smother than Intel cpus due to the reasons people have put forward in this thread. This is the art of debate.

Didn't you hear? I won! I won! I won!

[keithlm]

Just one question. Is this true

No smartness just a yes or no.

Very simple answer: Yes.

I have said many times I do not include reviews that are shown to be biased in my personal summations and analysis. So we have a situation where I am completely "correct" from my viewpoint and did not lie as you seem to enjoy claiming.

I might also add that you are also "correct" if you have decided to include any biased reviews in your viewpoint. I might not respect your choice of reviews... but I wouldn't call you a "liar" due to your choices even if I consider them to be obviously wrong.

You are more than welcome to not respect my choices. It won't change my mind or opinion in any manner. It won't even change my opinion of you.

[gallag]

Very simple answer: Yes.

I have said many times I do not include reviews that are shown to be biased in my personal summations and analysis. So we have a situation where I am completely "correct" from my viewpoint and did not lie as you seem to enjoy claiming.

I might also add that you are also "correct" if you have decided to include any biased reviews in your viewpoint. I might not respect your choice of reviews... but I wouldn't call you a "liar" due to your choices even if I consider them to be obviously wrong.

You are more than welcome to not respect my choices. It won't change my mind or opinion in any manner. It won't even change my opinion of you.

Thanks, I get what you are saying now. and it kind of makes sense, If you had explained it like this earlier it would have saved me some typing. I hope you can see where the confusion came from though, If you only count reviews that support your view point and make statements that most review sights support you clams it gets confusing real fast.

[grog]

Maybe this will help.... maybe not?

When I was torture testing my system I noticed it really could take just about anything and still keep up.

Benchmarking has one flaw. Benchmarking looks at the average over time. If there are 'gaps' in the system due to processes hogging the system the overall benchmark may not show the issue..

I want to know the 'real time response' of my system.

How can we test it.

How about this.

1. Load the system down big time. How about 100% CPU usage using Mersenne Prime 95 torture test on all four cores.

2. Keep loading the system as much as possible.

3. Record HD content in 1080i in 'real time' for an hour.

Stop the testing.... Check if the show we recorded has any issues.
In my case I had zero issues on the recorded program.
I would have loved to try this in Intel based Quad Core. I know my Intel Core Duo would not handle this task.

The 'real time process' testing we are doing in the case fits well with what users are saying about the AMD experience.

Since I don't have an Intel Quad Core system to compare this technique against I guess the Intel system 'might' be able to do the same thing... Maybe someone could try it?



What's running

1. Mersenne Prime 95 torture test version 2.56 with thread support so we can load all core to 100% usage quickly!http://www.mersenneforum.org/showthread.php?t=9779
2. ATI MMC recording a HD OTA broadcast in 1080i.
3. Windows Media Player 11 playing 'Amazon_1080.wmv' a nice 1080P clip. We have media player set to play the film in loop mode.
4. Cyberlink PowerDVD 7 is playing the movie "Wild Hogs".
5. GraphEdit is playing a divx movie.

The following test ran for over an hour. I did restart the MMC recoding but we got a good 1/2 recording in after the restart. The show I was recording ended.

Note: Everything played perfect even during the torture test!
The video I recorded played back fine with no dropped frames or other issues.

You can thow a hell of a lot at this system and it just handles it!

[keithlm]

Thanks, I get what you are saying now. and it kind of makes sense, If you had explained it like this earlier it would have saved me some typing. I hope you can see where the confusion came from though, If you only count reviews that support your view point and make statements that most review sights support you clams it gets confusing real fast.

I don't just count reviews that support my opinion or viewpoint.

I mainly discount the ones that are obviously biased or show some major flaws. Actually... to be specific I generally don't count or discount any review as an entity. I look at each individual benchmark and try to decide how important that benchmark might be.

For example I would consider the 3DMark06 overall score to be more important than the FPS of a game or even several games. Especially when most games are within a few FPS of each other regardless of CPU anyway.

Honestly: To me what would be one of the most important benchmarks would be how long it takes to do a build world in FreeBSD. But that's not a very common benchmark. Or OS.

ANYWAY: Each benchmark needs to be considered as individual entity and it's importance should be considered. I saw one "reputable" site that took the percentage differences from all of the benchmarks and added them up. Creative math to get results you want to see and not anything that actually shows anything important. (Unless you always run the exact same benchmarks; which they don't.)

Maybe this will help.... maybe not?

GOOD show... that's EXACTLY what needs to happen. We need to define a standard set of tests to run and then see how they work.

The only problem I see: People might not post things that make their CPU look less than stellar. I think one popular site tried to do a test with several things running in an attempt to "stress" the system. But neither of the tasks actually stressed the system so it was inconclusive.

For example: I recommend that ALL game benchmarks be run with 1 copy of Prime95 running per core. (So 4xPrime95 running... how are your FPS versus when the game runs alone.) I'm tired of seeing game benchmarks where there are 8 compared systems... and they all score within a few FPS of each other. That's useless. Perhaps with 1x copy of Prime95 per core running... we might actually SEE something different.

(Although I know many will not like this... since systems that rely on larger caches to get their high scores might be at a disadvantage.)

[stocius]

Maybe this will help.... maybe not?

When I was torture testing my system I noticed it really could take just about anything and still keep up.

Benchmarking has one flaw. Benchmarking looks at the average over time. If there are 'gaps' in the system due to processes hogging the system the overall benchmark may not show the issue..

I want to know the 'real time response' of my system.

How can we test it.

How about this.

1. Load the system down big time. How about 100% CPU usage using Mersenne Prime 95 torture test on all four cores.

2. Keep loading the system as much as possible.

3. Record HD content in 1080i in 'real time' for an hour.

Stop the testing.... Check if the show we recorded has any issues.
In my case I had zero issues on the recorded program.
I would have loved to try this in Intel based Quad Core. I know my Intel Core Duo would not handle this task.

The 'real time process' testing we are doing in the case fits well with what users are saying about the AMD experience.

Since I don't have an Intel Quad Core system to compare this technique against I guess the Intel system 'might' be able to do the same thing... Maybe someone could try it?



What's running

1. Mersenne Prime 95 torture test version 2.56 with thread support so we can load all core to 100% usage quickly!http://www.mersenneforum.org/showthread.php?t=9779
2. ATI MMC recording a HD OTA broadcast in 1080i.
3. Windows Media Player 11 playing 'Amazon_1080.wmv' a nice 1080P clip. We have media player set to play the film in loop mode.
4. Cyberlink PowerDVD 7 is playing the movie "Wild Hogs".
5. GraphEdit is playing a divx movie.

The following test ran for over an hour. I did restart the MMC recoding but we got a good 1/2 recording in after the restart. The show I was recording ended.

Note: Everything played perfect even during the torture test!
The video I recorded played back fine with no dropped frames or other issues.

You can thow a hell of a lot at this system and it just handles it!

If I get a moment I will try this test. Do you want the Intel and AMD oc'ed or stock?

[keithlm]

If I get a moment I will try this test. Do you want the Intel and AMD oc'ed or stock?

Stock for starters?

Maybe try some GAMES on both with 1 copy of Prime95 per core. (And without)

ALRIGHT: Let's design some XTREME BENCHMARK routines!

[Devil's Prophet]

BUT we have to make sure that the HD content recording doesn't have some kind of buffer that allows the system to get hogged from time to time without consequence for the eventual recording's flawlessness. Something like BURN-proof for cd/dvd recorders.

I'm guessing the HD source is a TV broadcast? Then I'm also guessing that system hogging is quite desasterous for the quality of the recording, no?

Anyway, if this method is very good at detecting subtle freezes of the system, then let's go!

[Particle]

It isn't possible to form a standardized set of benchmarks. Certain applications and games don't mean anything to Person A but might be all that Person B does with his computer.

As such, when I'm reading through benchmarks, I try to only pay attention to apps I'm likely to use. Optimization, coding, and compiler differences between different (but similar) applications can yield drastic swings in performance between two dissimilar processors. A true "composite" score just isn't practical. I hope that that approach is on its way out. I'd like to see review sites start putting ALL of their review results into a SQL database and allow users to search the results for just the bits they care about. That would be infinitely more useful than this Web 0.5 approach review sites currently take--forcing users to read through hundreds of pages of reviews just to get the bits that matter. I don't think they do it on purpose. It's just the way reviews have been setup for ages, and nobody has thought to update the routine to current web methods (dynamic pages).

[keithlm]

It isn't possible to form a standardized set of benchmarks. Certain applications and games don't mean anything to Person A but might be all that Person B does with his computer.

As such, when I'm reading through benchmarks, I try to only pay attention to apps I'm likely to use. Optimization, coding, and compiler differences between different (but similar) applications can yield drastic swings in performance between two dissimilar processors. A true "composite" score just isn't practical. I hope that that approach is on its way out. I'd like to see review sites start putting ALL of their review results into a SQL database and allow users to search the results for just the bits they care about. That would be infinitely more useful than this Web 0.5 approach review sites currently take--forcing users to read through hundreds of pages of reviews just to get the bits that matter. I don't think they do it on purpose. It's just the way reviews have been setup for ages, and nobody has thought to update the routine to current web methods (dynamic pages).

I like what Futuremake has done with the "ORB" but would like to see even more benchmarks and more data about each system. (And this data should be auto-generated during the submit process... with the ability for the user to add various notes etc.)

A central database where anyone can post results for the benefit of all.

You could also add extra things that are not usually revealed in a review or forum post... a sysinfo or DXDIAG listing so that it is easy to see a persons machine, what they have and what driver versions. (And also user table that stores all the specifics about a machine for search purposes.)


Then you could do advanced benchmark searches with compares:

EXAMPLES:
Show me results from all the QUAD machines clocked at 2.5Ghz that have 8800GTS video cards, run vista, have 2x2Gb of memory.
OR
Compare (highlight) systems that have exactly the same components so that you can come up with a standard deviation for statistics. (Throw out anomalous scores.)
OR
Compare systems that have the same components except for the soundcard. Highlight the different soundcards so I can decide how much difference onboard makes versus discrete for myself instead of relying on others; show only game benchmarks.

=============
Then when someone creates a "unique" benchmark like has been proposed in this thread the results of all of the work don't get wasted. (This subject seems to come up every month or two... and some unique things are tried... and then the results are lost to time.)

It would also completely stop the "show me the links" types of posts: "Compare all the same benchmarks from these two specific CPU and average the results; discard anomalous results that deviate too much from the average." (Kind of what I do right now... but I have to do it "by hand" and look through a ton of review sites; most of which I don't bookmark so I can't easily just tell people where to look.)

I hate spending 6 or 8 hours doing various tests and benchmarks... only to have the results be lost forever. And I don't just mean for overclocking and purposes of "bragging". This means anytime I run a bunch of tests. If I build a new machine and run a bunch of benchmarks and/or stability tests... I want those results stored so that when I overclock or update the bios... I can compare. (Right now I do that... it's on a bunch of pieces of paper that will get lost when I really need them.)

Plus others could benefit from my efforts and they wouldn't be lost; providing their own results and efforts. So ALL BENCHMARK EFFORTS would be saved.

The first thing to do is decide WHAT needs to be saved for each and every benchmark. The rest is pretty easy.

NOTE: As a start... all the results from various reviews that currently exist could be put into a single database. Then it would be easy to see if a "trusted" site has results that completely contradict 3 or 4 other sites. I've seen too much of this and I get tired of people pointing to the most offensive sites as "proof" for their argument.

I might have to do at least this last bit just for my own use. Extending it to add the other stuff would take more work. But I think it would be a worthwhile effort. (I'll think about it.)

[Devil's Prophet]

It isn't possible to form a standardized set of benchmarks. Certain applications and games don't mean anything to Person A but might be all that Person B does with his computer.

As such, when I'm reading through benchmarks, I try to only pay attention to apps I'm likely to use. Optimization, coding, and compiler differences between different (but similar) applications can yield drastic swings in performance between two dissimilar processors. A true "composite" score just isn't practical. I hope that that approach is on its way out. I'd like to see review sites start putting ALL of their review results into a SQL database and allow users to search the results for just the bits they care about. That would be infinitely more useful than this Web 0.5 approach review sites currently take--forcing users to read through hundreds of pages of reviews just to get the bits that matter. I don't think they do it on purpose. It's just the way reviews have been setup for ages, and nobody has thought to update the routine to current web methods (dynamic pages).

In light of this manner of personal way of comparing hardware, reviewsites should review their hardware in every possible aspect they can come up with, and make a system which allows users to select a subset from that broad range of performance benchmarks, adjusted to everyone's taste. Also they could make up some "profiles" with preset subsets to make things easier for the quick reader. Also it would be nice if users had the option to add weights to certain single benchmarks to make them less or more important to the whole number. Is this what you had in mind?

[Particle]

Yeah, something like that. Imagine a database that starts with a few columns for like RAM type, manufacturer, [other ram details], cpu, [other cpu detail columns], video card, [other video card details], ...more comp specs..., Quake 4 @ 1600x1200 w/ 4xAA & 8xAnsio, another game, another game, another game, etc. Keep adding columns for new software tests. Quit filling in new columns for old machines if desired. Quit testing old software as hardware advances. You would end up with an entirely cross-reference(able) database even if two rows don't share results in certain columns. The fact remains that you could in theory do it, and it won't come up as a problem when comparing same-gen hardware.

You could select the CPUs (rows) and software (columns) to get a dynamic graph just comparing what you want to see the difference between for instance.

It wouldn't be too tough to make such a system. Getting lots of data that is comparable would be a challenge though.

[gosh]

Could the smoothenes have something to do with the thread scheduler? AMD support NUMA and windows is NUMA aware. Better performance using I/O operations should also effect thread scheduling.
Intel processors may have (more) difficulties to switch from one thread to another because they are optimized to run one single thread. Invalidating data from the cache getting new data used by another thread or when all traffic for all cores goes through one single channel may have some optimizations done. Intel may have fever switches between threads compared to AMD processors. The reason could be to gain performance I various tests.
Just speculating, don’t know the right answer.

Links:
http://technet.microsoft.com/en-us/m.../cc194386.aspx
http://en.wikipedia.org/wiki/Non-Uniform_Memory_Access
http://developer.amd.com/pages/1162007106.aspx

[Devil's Prophet]

A valid point!

That is indeed an interesting perspective. We'll have to look into to that some more. Is here anybody with an intel and amd quadcore setup and with some knowledge of how to use windows perfmon (to monitor the threadswitches, etc) and is willing and able to do some tests?