my recommended configuration for >= 8-cores

[alpha754293]

I don't have an official statement with regards to HTT only because I do not have a system to test HTT with.

(I wished I did, but the closest thing I've got is Q9550, and there might be some substantial differences in architecture between that and the Core i7).

However, in my simulated results (I have a server with 8 native cores, and it is NOT HTT capable (AMD)), I have found that you take a 4% performance penalty when I ran with simulated HTT.

Well, since HT takes a 50% performance hit straight out of the box due to saturation of the FPU, I don't beleive your "simulated HTT" is meaningful.

By the way, HTT is HyperTransportTM Technology. What kind of credentials do you have to tell us otherwise?

toTOW asked me the same thing. See thread: viewtopic.php?f=44&t=8027#p79367

See references here, here, here, and here.

On HyperTransport (HT):

See references: here, and here.

[codysluder]

Boy you're quick. I guess you've got nothing more important to do but to ask questions and then tell the people who respond that they're wrong.

See the edits to my previous post.

[alpha754293]

Boy you're quick. I guess you've got nothing more important to do but to ask questions and then tell the people who respond that they're wrong.

See the edits to my previous post.

I'm a serial multi-tasker.

It's the only way that I can get stuff done fast enough. I think I've got over 200 tabs open spread across 4 comps (including this one).

I do my research and make sure that my comments, replies, posts, etc. are researched to the best of my capability with the appropriate sources and citations (albeit I've been TOTALLY slacking by not following APA citation rules at all), and make sure that I'm always in a position ready to defend them like it's my dissertation.

If I can't or there's something that I don't know, or aren't sure about or for which it will fail the above clause, I disclaimer it so, and I've also written SafeHarbor statements before for technical and financial projections for which I cannot realistically predict the real world response/behavior.

And I see to it that I learn the best that I can, as much as I can even though I've long realized that it would be impossible for me to actually know it all, but that certainly hasn't stopped me from learning as much as possible about anything and everything.

*edit*
Certainly helps to have addressed the question before as well.

[7im]

No offense intended alpha754293, but common usage dictates how people abbreviate and use acronyms. Even Intel abbreviates Hyper-Threading Technology as HT Technology (see some of your own citations). And in common usage, most people drop the formal "Technology" part of the Intel brand name (do you ask for a coke? A Coke? or a Coca-cola?), and thus HT is Hyper-Threading. Unfortunately, and probably not accidentally (IMO), a direct competitor also coined a similar "technology" with identical initials. Go figure. HT is an acceptable and common use for both, regardless of citation.

In a location such as a forum, even a scientific forum such as this, the writing style still tends to be informal. Even the project FAQs tend to be informal. Common speaking language is easier to understand when explaining complex concepts. (Ya, sure, the hardcore grammarians freak out about it, but it's a FAQ, not a dissertation.) I have myself asked for people to back up their claims with a reference article on occasion. Though typically, it's not because I don't believe them, it's because I want to read the article too.

Let's keep the forum fun and informal, and leave out the citations unless someone asks for one. Otherwise the place just becomes pendantic and fussy, and that's not who we are here (though opinions vary).

[alpha754293]

No offense intended alpha754293, but common usage dictates how people abbreviate and use acronyms. Even Intel abbreviates Hyper-Threading Technology as HT Technology (see some of your own citations). And in common usage, most people drop the formal "Technology" part of the Intel brand name (do you ask for a coke? A Coke? or a Coca-cola?), and thus HT is Hyper-Threading. Unfortunately, and probably not accidentally (IMO), a direct competitor also coined a similar "technology" with identical initials. Go figure. HT is an acceptable and common use for both, regardless of citation.

In a location such as a forum, even a scientific forum such as this, the writing style still tends to be informal. Even the project FAQs tend to be informal. Common speaking language is easier to understand when explaining complex concepts. (Ya, sure, the hardcore grammarians freak out about it, but it's a FAQ, not a dissertation.) I have myself asked for people to back up their claims with a reference article on occasion. Though typically, it's not because I don't believe them, it's because I want to read the article too.

Let's keep the forum fun and informal, and leave out the citations unless someone asks for one. Otherwise the place just becomes pendantic and fussy, and that's not who we are here (though opinions vary).

Convention doesn't make it right. Just look at the Romans.

Sadly, the HyperTransport Consortium is a group of companies.

Quote:
"There has been some marketing confusion between the use of HT referring to HyperTransport and the later use of HT to refer to Intel's Hyper-Threading feature of some Pentium 4 based microprocessors. Hyper-Threading is officially known as Hyper-Threading Technology (HTT) or HT-Technology. Because of this potential for confusion, the HyperTransport Consortium always uses the written out form: "HyperTransport"."

Also if you notice in the documentations that you are referring that does not use the term "HTT" to refer to HyperThreading Technology, you will find that it uses the term "HT Technology" instead, as addressed as a proper noun. Conversely, HyperTransport is strictly JUST HT.

Additionally, HyperTransport (HT) was introduced April 2, 2001 while the earliest introduction of HyperThreading Technology (HTT) that I've been able to find was dated October 3, 2002.

Think of this kind of like wikipedia. If people want to read the references, they're at the bottom (or embedded via hyperlinking). If they don't want it, they don't have to read it.

Suffice it so say that I find it most peculiar how one would be asking me if I am asking for the sake of asking or if I am asking for a different purpose.

Citations and references are used in order to support a well and properly researched response. If they do not wish to have the citations included, then it would be highly suggested that they conduct their own cross examination research; although it's often more likely you'll find that it would be easier for the person responding to just submit the citations and references at the time of the response anyways instead of waiting to be asked for it later. So you might as well include it anyways.

[codysluder]

Because of this potential for confusion, the HyperTransport Consortium always uses the written out form: "HyperTransport".

So by your own reference, HT does NOT refer to HyperTransport Technology.

To some extent it's due to market penetration. If I order a "Coke" and they give me a Pepsi, it's no big deal. If Windows invents a term for something, who cares what it's called in Linux. If Intel creates HyperThreading (even if it's after HyperTransport was marketed) we adopt their slang, no matter which should rightfully be adopted, and the Consortium has to back down to avoid confusion, simply because they're a minority.

If I say my i7 has HT most people will figure out what I mean from context and if it confuses you, I apologize in advance.

[alpha754293]

Because of this potential for confusion, the HyperTransport Consortium always uses the written out form: "HyperTransport".

So by your own reference, HT does NOT refer to HyperTransport Technology.

To some extent it's due to market penetration. If I order a "Coke" and they give me a Pepsi, it's no big deal. If Windows invents a term for something, who cares what it's called in Linux. If Intel creates HyperThreading (even if it's after HyperTransport was marketed) we adopt their slang, no matter which should rightfully be adopted, and the Consortium has to back down to avoid confusion, simply because they're a minority.

If I say my i7 has HT most people will figure out what I mean from context and if it confuses you, I apologize in advance.

Also from my own reference, HyperTransport's official acryonym is HT. HyperThreading Technology's official acronym is HTT.

"HyperTransport (HT)" - from http://www.hypertransport.org/default.c ... Technology
"Hyper-Threading is officially known as Hyper-Threading Technology (HTT)"

It doesn't confuse me, but I like to use PROPER terminology. The references clearly addresses that question.

Note also that HyperTransport is NEVER referred to as HTT. And HyperThreading Technology is the only one that is officially referred to, even by Intel, as HTT.

See their respective technical publications.

BTW, in most restaurants now, if you order Coke and they have Pepsi, I THINK/believe that technically they are required, by law, to ask if you would like Pepsi. Welcome to the litigous society that is America. Windows doesn't invent ANYTHING. If it does, be afraid. Be very afraid. The Consortium didn't HAVE to do anything. They did it out of common courtesy. Besides, I think the average population is so very sadly ignorant.

"...in the absence of genuine leadership, they'll listen to anyone who steps up to the microphone. They want leadership. THey're so thirst for it they'll crawl through the desert toward a mirage, and when they discover there's no water, they'll drink the sand."
"...People don't drink the sand because they're thirsty. They drink the sand because they don't know the difference."
- The American President (1995)

In this case, one asked for and about verification and validation on the meanings of the acronyms. Despite providing the appropriate references that state the obvious and correct usage of the terms, it is not my duty to enforce it. People are going to do what they want to do. Those who don't know learn. Those who fail to learn or refuse to learn are just ignorant. Unforutnately, in technical and research fields, ignorance is a luxury we can ill afford.

The principle of "when in Rome, do as the Romans do" completely fails Kant's Golden Rule and moral imperative. My point is that just because the vast majority of the population does something, doesn't necessarily make it right. And if you don't have the guts to stand up to your convictions against the tide, to do what IS right, then you will never be anybody of significance. And if people are willing to accept complacency, then you lose all meaning for pushing and promoting moral superiority as a nation, and worst of all; you'll never "be a man, my son." (Rudyard Kipling, "If").

[alpha754293]

Getting back on topic.

Here are the test results for 2x"-smp 8"

(luck of the draw, production run level)
1x4*a1+1x8*a2 = 4945.85 PPD (or ~83.2%)

No results yet for 2x8*a2. Seems like that so far, the best configuration is still 2x"-smp 4". (At least from a PPD perspective).

[Flathead74]

...Besides, I think the average population is so very sadly ignorant....

Sadly, this is obvious in the way that you address other folks.

...Those who fail to learn or refuse to learn are just ignorant....

Being ignorant is to be lacking education or knowledge.
To be unable or to be unwilling to learn is not ignorance, stupidity maybe, but not ignorance.

[alpha754293]

...Besides, I think the average population is so very sadly ignorant....

Sadly, this is obvious in the way that you address other folks.

If a proper, well thought out, intelligent response is too great of an expectation here, tell me now so that I can stop looking for one.

[7im]

Convention doesn't make it right. Just look at the Romans.

Doesn't make it wrong either. Convention doesn't give a crap about what is proper, or what some stuffed shirt in the marketing department complains about. And we still use a LOT of ancient Roman or Roman derived conventions and technology today. BTW, what superbowl number is it this year?

The only thing that becomes obsolete faster than technology, is the marketing name for that tech. But please, do continue preaching to the choir about the proper use of Marketing names for technology.

[alpha754293]

Convention doesn't make it right. Just look at the Romans.

Doesn't make it wrong either. Convention doesn't give a crap about what is proper, or what some stuffed shirt in the marketing department complains about. And we still use a LOT of ancient Roman or Roman derived conventions and technology today. BTW, what superbowl number is it this year?

No clue. I don't follow football.

Tell you what. We'll all carpool to visit you at the Federal Penitentiary for following a Roman's convention of The Rape of Lucrece. SPQR! Porro ago Romanorum res publica.

Now, would that be before or after you flagellate some dude name Jesus, and crucify him at Calvary?

Cheers to Roman convention.

Being ignorant is to be lacking education or knowledge.
To be unable or to be unwilling to learn is not ignorance, stupidity maybe, but not ignorance.

If lacking in education or knowledge is a result of unwilling to learn, then one is MOST DEFINITELY ignorant. (By causality).

So even though, those who know what the right answer SHOULD be, but choose to ignore it; then they are ignorant (as a generic condition).

[7im]

'course I'm just an ignorant 'tard, so can you tell me which is worse? A citation-happy self-proclaimed academic, or the use of selective citations to support untenable positions?

Did you really expect to bring in the downside of the Romans without any consideration for their contributions to modern society? You only need to look at our country's capital to see their far reaching influence on architecture, and their influence on our system of government with 3 balanced branches of democratic power. Many of the world's languages are derived from their use and spread of Latin, including English. Running water and sewer systems permitted mankind to gather in larger cities, promoting cultural, artistic, and academic advancements around the world. Even the concrete and mortar to build those cities originated in Roman culture. And despite Rome's attempts to prevent the spread of Christianity and its human rights violation of one particular individual, it's empire of culture, roads, and lines of communication resulted in the faster and wider spread of Christianity more than any other influence in history. But since this board is agnostic, not just for hardware (Intel vs. AMD), we won't go any farther in to such things. But let's continue with central heating, wine making (Romans spread grape planting in to both France and Italy), the Roman alphabet and legal system, the Julian calendar we use, many of our festivals, and even your good old 3 course meal of appetizer, main course, and desert came from Rome. I don't suppose you follow any of these either?

But then we're way off topic again, and a Mod will probably lock this thread too.

[alpha754293]

'course I'm just an ignorant 'tard, so can you tell me which is worse? A citation-happy self-proclaimed accademic, or the use of selective citations to support untenable positions.

Did you really expect to bring in the downside of the Romans without any consideration for their contributions to modern society? You only need to look at our country's capital to see their far reaching influence on architecture, and their influence on our system of government with 3 balanced branches of democratic power. Many of the world's languages are derived from their use and spread of Latin, including English. Running water and sewer systems permitted mankind to gather in larger cities, promoting cultural, artistic, and accademic advancements around the world. Even the concrete and mortar to build those cities originated in Roman culture. And despite Rome's attempts to preven the spread of Christianity and it's human rights violation of one particular individual, it's empire of culture, roads, and lines of communication resulted in the faster and wider spread of Christianity more than any other influence in history. But since this board is agnostic, not just for hardware (Intel vs. AMD), we won't go any farther in to such things. But let's continue with central heating, wine making, the Roman alphabet and legal system, the Julian calendar we use, many of our festivals, and even your good old 3 course meal of appetizer, main course, and desert came from Rome. I don't suppose you follow any of these either?

But then we're way off topic again, and a Mod will probably log this thread too.

From what I saw and what I remember seeing, many of the buildings at (what I would presume to be Washington, D.C.) are hints of neoclassicalism in its architecture (both Greek AND Roman). U.S. Capitol is an example of such. The White House is possibly Georgian, possibly neoclassical. Course, I'm just guessing here since I'm not about to statistically study the architectural types of the buildings of Washington, D.C..

If you're REALLY interested -- here: http://www.loc.gov/rr/print/adecenter/essays/Scott.html

The concept of Separation of Powers is traced back to Aristotle's "Politics" (Greek), a student of Plato, the latter of whom wrote the Socratic dialogue, "The Republic". I've never actually read either of these works although in Politics, Books 3 and 4, deal with classification of constitutions, just distribution of political power, types of monarchies, tasks of political theory, why there are many types of constitutions, types of democracies, types of oligarchies, polity as the optimal constitution, government offices,

(From the little that I've read just now, apparently it says that the separation of power is to be divided through the different constitutional forms in order to best develop the law, execute the law, and to protect the law.)

The identification and association with executive branch, legislative branch, and judiciary branch didn't come about until the Roman Republic and the codification of the Constitution of the Roman Republic; which drew on the idea of Aristotle such that the governance wasn't a strict form of democracy, aristocracy, or monarchy. Interestingly enough, I haven't been able to find when the Constitution was actually enacted to take on that form since there were 5 distinct phases throughout the Constitution of the Roman Republic. (Or so it appears.) I am GUESSING that it had to be some time after 322 BC. I am also GUESSING that it probably came into effect around 287 BC, but I don't know. Anyways, the point of which is that 3 branches in its current form as you know it is from ancient Rome. The concept, however, of Separation of Power is from ancient Greek. Sorry. You're not as Roman as you thought.

I will agree with you that the Roman Empire was definitely MASSIVE. Interestingly enough, there aren't that many people in the world who know Latin anymore, and the two most common form of Latin that remains is Church Latin and New Latin (often used in medicine). Also interestingly enough, there's a popular belief that English is actually etmyologically 70% German. *shrug* I'm not an etmyologist, so I have no idea.

0.o? Julian? I thought that we were using Gregorian. FRICK! No wonder why I kept showing up late!

Well...I COULD write to you in Cyrillic if you really like. I don't know if this board supports UTF-8. I pretty much work throughout entire year, including statutory holidays (mostly because I think that I'm atheist on a technicality). I don't celebrate New Years (either of them), and it wouldn't phase me one single bit if I missed the countdown. I know that I DEFINITELY don't celebrate the Ides of March, or any of the Roman festivals. (http://en.wikipedia.org/wiki/Roman_festivals) I've never seen "Roman Holiday" and don't really have any real plans to do so anytime soon. (Although I've heard that it's supposed to be a pretty good movie with Audrey Hepburn and Gregory Peck.) *shrug*. I only eat once a day (it's all I really need), and when I do have more than one "course", it's only cuz I'd otherwise need about 11 competent chefs in order to be able to serve it all up simultaneously. So, cooking in serial rather than parallel is mostly due to practical reasons, although I'm known to start cooking the desserts first and always. (because they take the most time).

And if I were literate, I'd write to you in Arabic, Farsi, Chinese, Japanese, or Korean. (Sadly, I only know one or two words in each of those languages.) Tried to learn a bit of Russian, German, Italian, and French (because of musical studies). Didn't stick. Growing up, I was more focused on Western European languages while my parents spoke to me in Chinese.

[alpha754293]

Okay. The results are in. It's official.

2x"-smp 8" clients = BAD.

client1 was running 4*a1 core (luck of the draw) and was getting around 1245 PPD.
client2 was running 8*a2 core and was getting an average of 3431 PPD.

Combined peak was somewhere between 4600 and 4900 PPD.

I've switched it back so that it runs 2x"-smp 4" clients.

client1, still runinng 4*a1 is now getting 1851.58 PPD
client2 (unfortunately), is now getting 3177.93 PPD

which represents a performance loss of 33-38% (compared to previous known peak values for the full potential of the cores).

Checking back on the previous results, it looks like that client2 is actually running nearly about as fast as it can for 4*a2, so there isn't much change there (3177.93 vs. avg of 3177.98). So I guess that a2 picks up a bit of speed, at the expense of a1. Hmmm....ok. Data is data.

I was hoping that it would pick up 2 a2 WUs so that I would be able to watch what would happen when you have 16 threads competing for computational resources.

I don't think that that's going to happen now.

Overall computational efficiency drops 17% running with two "-smp 8" clients. Didn't check the effects on WU speed (although I am guessing that we should be able to deduce it based on the PPD values).