An Opteron 2xx or 8xx has more functioning coherent HT links, but the FX and Opteron 1xx each have one. I don't think there's any such thing as an FX motherboard, they just run in single Opteron motherboards with or w/o AGP.
People should buy what they can afford that fits their needs. If an FX-51 with ECC memory satisfies their requirements, then there's no point paying more for a 'real' Opteron.
It's my understanding that while AMD64 chips give a nice boost in integer calculations [largely due to extra GPRs], their SSE2 floating point is only about on par with the fastest P4s. Later versions should have improved SSE2.
The socket will be changing to 939pin for next generation dual memory channel 64-bit Athlons but not for Opterons, which the FX-51 essentially is [~Opteron 148 IIRC]. If what you want is a system upgradeable to future Opterons, starting with an FX-51 may be a relatively inexpensive approach.
Good point about RPN subexpressions being readily available as typed. Complicated algebraic expressions with lots of parentheses would probably benefit from being line buffered so you can edit the whole thing before evaluation.
And yeah, RPN easily supports multiple parameters passed to subroutines. For that I'd rather just have formal typed parameters, but I guess that defeats the purpose of a quick and immediate access calculator.
BTW, has anybody noticed the irony that stack based floating point arithmetic is generally deprecated on the latest Intel and AMD processors? The 3DNow! / SSE / SSE2 SIMD instruction sets are often much faster and preferred except in cases where you need transcendental functions or full 80-bit precision. So you may be crunching your RPN calculator's FP stack while your computer isn't.
Yes, you can think through the process of RPN in your head, but my argument is that it's much more natural for people to consider what some abstract idea represents, rather than the nuts and bolts of how to crunch it. Meaning versus process. I'd look at your example and see it as the product of two operands, which can be further detailed as needed.
I don't doubt that you can implement an RPN evaluator in 10 lines of GWBASIC, but we have MIPS and MFLOPS and RAM to spare these days: why not have that same program parse the whole infix expression for us? And while a simple expression occupying only three stack levels can be easily done mentally, it just gets harder when transcribing more complicated expressions. Besides that, once you've written such an RPN expression it becomes inscrutable if you want to revise the expression later on.
Translated as "I think RPN is a silly and archaic UI for humans."
I used to have fun programming my Dad's HP 41C and loved the feel of its keys, but there's no reason humans should have to convert natural infix expressions to simplify life for a calculator. Who honestly thinks spreadsheets would be improved if they contained cell expressions in RPN?
A possibly naive question: Would it make any sense to include some substantial amount of main [non-cache] memory on the CPU die or in the same package? If it were say 128MB to 256MB or so, that would likely be enough to cover the working set of a typical consumer desktop machine. A NUMA-aware OS could be smart enough to migrate data structures to this faster 'on-chip memory' from external DIMMs.
In other words, if it's too difficult or expensive to improve latency of all memory, is it cost-effective to do so for the system's most frequently or LRU data?
You obviously haven't looked very carefully at the AMD64 ISA. In the eyes of many observers, AMD has made quite a number of inspired decisions in the design of their 64-bit architecture. While technical purity may be desirable from an esthetic viewpoint, they've done an excellent job of providing the machine resources most needed by executing programs while eliminating those which were expensive to provide and rarely needed.
One of the [accidental?] virtues of x86 code, especially as compared to RISC or VLIW/EPIC, is its high density. While decoding it is a pain, its compactness takes very good advantage of available cache space and memory bandwidth.
Besides which, releasing a totally incompatible 64-bit ISA wouldn't make sense. Nobody would buy such a consumer chip if it didn't run their existing Win32 software, which will take time to be recompiled for any new architecture. Athlon64 and Opteron run IA32 very competitively without needing any type of emulation, and the decade-long investment in IA32 code generation technologies can be used as a jumping off point for efficient AMD64 code generation, as contrasted with Intel's continuing difficulty to provide efficient compilation for IA64.
You can keep talking about hacks and patches, but in the real world AMD64 is one helluva powerful design.
Thanks for the clarifications. I did just now look it up [getting a headache in the process], and I think your points are valid although it's not clear that "fair use" includes the ability to give digital copies to someone who doesn't already own a legal copy of the material. I guess the law means whatever the courts interpret it to mean.
Okay, sorry for replying to my own post. Just remembered that you can negotiate to include in your employment contract such things as intellectual property ownership, non-compete clauses, and the like.
"The truth is, if you are not self employed, you don't own your own work. Your boss / company does." Of course! That's what you agree to when you accept their employment.
"This is why I try to avoid writing any truly reusable software at work - I don't want to limit myself in the future." There's no reason to write awful code on purpose. In one case where my new company wanted to use software I'd written at my previous company, they just drew up a contract and purchased the rights for a nominal fee. No biggie.
You've already found the solution that works best for you, pursuing open source projects. That's great, but don't assume that everyone should be forced to do the same.
Property can be given or sold, but not taken. That doesn't change when we dislike the owner or when we're envious of their wealth.
I'm glad you liked Fair Warning and followed up by patronizing the band, but it's still not up to us to decide that it's okay to give someone else's property away for free.
I could send copies of a copyrighted material to 10 or 10,000 of my dearest friends as long as they're all first generation copies? And rather than 'sell' them I could have a PayPal account for 'donations'?
Admittedly here's a case which vexes me: How do I legally make a CD compilation for a friend/relative of music I own without having to purchase a dozen full duplicate albums? iTunes and such may provide a way, but I'd want these songs in normal uncompressed high-fidelity form.
A ringer or apologist? RIAA's not my favorite organization, I'm disgusted at the still-high prices of CDs after all these years and am very happy about the proliferation of iTunes and other such services.
Just because I agree with RIAA's legal justification doesn't mean I have to approve of their tactics. You're right, they could've found a better approach. My guess is once they find how the first 260 suits go they'll modify it for future ones.
We don't have the right to declare that it's just as well for a particular band to make $$$ by touring. Perhaps they're not willing or able to for some reason, perhaps their style of music doesn't lend itself to live performance.
And the last thing we need is for concert prices to skyrocket. Not everyone has a lot of disposable income in this economy.
The only legitimate way society has to redefine what's legal is by changing the law. Until that happens, we need to stick with current regulations, except in very rare cases where things like civil disobedience may be called for. If you care about this, then do what you can to elect candidates who see things the way you do.
I don't understand the overwhelming sympathy for those who steal, 12yr old or otherwise. If you were mugged by 50,000,000 mostly teenagers, would you say "aw, stop picking on the kids" ?
Aren't many of you/.ers also software developers like me? If I choose to release an open source project for anybody to use, that's fine. But if I make my living writing code -- or performing music -- then why is it acceptable for people to take my product w/o compensating me?
Does theft suddenly become acceptable when it's done on a massive scale? Would you support a company protecting its rights if there were only dozens of thieves instead of scores of millions?
The other argument of "RIAA is evil, therefore committing crimes against them is okay" doesn't wash either. These aren't victimless offenses, the money has to be coming out of somebody's pocket. If CDs are too expensive, then don't buy them. I don't claim the right to steal an SVT Cobra just because the $35k list price is too expensive.
IMHO, RIAA and other music associations should enforce their rights and are obligated to do so for their stockholders. Dunno of any practical way for them to ensure they pursue only unsympathetic defendants. Seems to make sense for them to ease into it, have a little consideration when negotiating with naive downloaders, gather as much publicity as possible so that music listeners in the future will understand that stealing music is wrong.
BTW, I have an adult friend who insists it's perfectly legal to make copies of her CDs as long as she only gives them to friends. The only way such attitudes will change is when people are prosecuted for it.
Slashdot Mirror
Intel's forthcoming Prescott will have improved HT, will be interesting to see how much it helps on current and new applications.
Most definitely, see here for an overclocking starting point. They're also an excellent bang/buck, currently $84usd at Newegg.
I'm not sure we know how fast G5 actually is, given the lack of a 64-bit OS under which to benchmark it.
An Opteron 2xx or 8xx has more functioning coherent HT links, but the FX and Opteron 1xx each have one. I don't think there's any such thing as an FX motherboard, they just run in single Opteron motherboards with or w/o AGP.
People should buy what they can afford that fits their needs. If an FX-51 with ECC memory satisfies their requirements, then there's no point paying more for a 'real' Opteron.
It's my understanding that while AMD64 chips give a nice boost in integer calculations [largely due to extra GPRs], their SSE2 floating point is only about on par with the fastest P4s. Later versions should have improved SSE2.
The socket will be changing to 939pin for next generation dual memory channel 64-bit Athlons but not for Opterons, which the FX-51 essentially is [~Opteron 148 IIRC]. If what you want is a system upgradeable to future Opterons, starting with an FX-51 may be a relatively inexpensive approach.
Good point about RPN subexpressions being readily available as typed. Complicated algebraic expressions with lots of parentheses would probably benefit from being line buffered so you can edit the whole thing before evaluation.
And yeah, RPN easily supports multiple parameters passed to subroutines. For that I'd rather just have formal typed parameters, but I guess that defeats the purpose of a quick and immediate access calculator.
BTW, has anybody noticed the irony that stack based floating point arithmetic is generally deprecated on the latest Intel and AMD processors? The 3DNow! / SSE / SSE2 SIMD instruction sets are often much faster and preferred except in cases where you need transcendental functions or full 80-bit precision. So you may be crunching your RPN calculator's FP stack while your computer isn't.
Very interesting revelation, I didn't know that. It never hurts to be reminded that one's own frame of reference doesn't apply to everyone.
Domo arigato!
Yes, you can think through the process of RPN in your head, but my argument is that it's much more natural for people to consider what some abstract idea represents, rather than the nuts and bolts of how to crunch it. Meaning versus process. I'd look at your example and see it as the product of two operands, which can be further detailed as needed.
I don't doubt that you can implement an RPN evaluator in 10 lines of GWBASIC, but we have MIPS and MFLOPS and RAM to spare these days: why not have that same program parse the whole infix expression for us? And while a simple expression occupying only three stack levels can be easily done mentally, it just gets harder when transcribing more complicated expressions. Besides that, once you've written such an RPN expression it becomes inscrutable if you want to revise the expression later on.
Translated as "I think RPN is a silly and archaic UI for humans."
I used to have fun programming my Dad's HP 41C and loved the feel of its keys, but there's no reason humans should have to convert natural infix expressions to simplify life for a calculator. Who honestly thinks spreadsheets would be improved if they contained cell expressions in RPN?
People aren't stack machines, we think top down!
A possibly naive question: Would it make any sense to include some substantial amount of main [non-cache] memory on the CPU die or in the same package? If it were say 128MB to 256MB or so, that would likely be enough to cover the working set of a typical consumer desktop machine. A NUMA-aware OS could be smart enough to migrate data structures to this faster 'on-chip memory' from external DIMMs.
In other words, if it's too difficult or expensive to improve latency of all memory, is it cost-effective to do so for the system's most frequently or LRU data?
You obviously haven't looked very carefully at the AMD64 ISA. In the eyes of many observers, AMD has made quite a number of inspired decisions in the design of their 64-bit architecture. While technical purity may be desirable from an esthetic viewpoint, they've done an excellent job of providing the machine resources most needed by executing programs while eliminating those which were expensive to provide and rarely needed.
One of the [accidental?] virtues of x86 code, especially as compared to RISC or VLIW/EPIC, is its high density. While decoding it is a pain, its compactness takes very good advantage of available cache space and memory bandwidth.
Besides which, releasing a totally incompatible 64-bit ISA wouldn't make sense. Nobody would buy such a consumer chip if it didn't run their existing Win32 software, which will take time to be recompiled for any new architecture. Athlon64 and Opteron run IA32 very competitively without needing any type of emulation, and the decade-long investment in IA32 code generation technologies can be used as a jumping off point for efficient AMD64 code generation, as contrasted with Intel's continuing difficulty to provide efficient compilation for IA64.
You can keep talking about hacks and patches, but in the real world AMD64 is one helluva powerful design.
Thanks for the clarifications. I did just now look it up [getting a headache in the process], and I think your points are valid although it's not clear that "fair use" includes the ability to give digital copies to someone who doesn't already own a legal copy of the material. I guess the law means whatever the courts interpret it to mean.
Okay, sorry for replying to my own post. Just remembered that you can negotiate to include in your employment contract such things as intellectual property ownership, non-compete clauses, and the like.
"The truth is, if you are not self employed, you don't own your own work. Your boss / company does." Of course! That's what you agree to when you accept their employment.
"This is why I try to avoid writing any truly reusable software at work - I don't want to limit myself in the future." There's no reason to write awful code on purpose. In one case where my new company wanted to use software I'd written at my previous company, they just drew up a contract and purchased the rights for a nominal fee. No biggie.
You've already found the solution that works best for you, pursuing open source projects. That's great, but don't assume that everyone should be forced to do the same.
Property can be given or sold, but not taken. That doesn't change when we dislike the owner or when we're envious of their wealth.
I'm glad you liked Fair Warning and followed up by patronizing the band, but it's still not up to us to decide that it's okay to give someone else's property away for free.
Nah, I don't own jack. Usually my hiring company would own anything I write.
But what does RIAA being leeches have to do with an entitlement to steal from them? It's a non-starter of an argument.
IANAL, but this doesn't sound plausible.
I could send copies of a copyrighted material to 10 or 10,000 of my dearest friends as long as they're all first generation copies? And rather than 'sell' them I could have a PayPal account for 'donations'?
Admittedly here's a case which vexes me: How do I legally make a CD compilation for a friend/relative of music I own without having to purchase a dozen full duplicate albums? iTunes and such may provide a way, but I'd want these songs in normal uncompressed high-fidelity form.
A ringer or apologist? RIAA's not my favorite organization, I'm disgusted at the still-high prices of CDs after all these years and am very happy about the proliferation of iTunes and other such services.
Just because I agree with RIAA's legal justification doesn't mean I have to approve of their tactics. You're right, they could've found a better approach. My guess is once they find how the first 260 suits go they'll modify it for future ones.
We don't have the right to declare that it's just as well for a particular band to make $$$ by touring. Perhaps they're not willing or able to for some reason, perhaps their style of music doesn't lend itself to live performance.
And the last thing we need is for concert prices to skyrocket. Not everyone has a lot of disposable income in this economy.
The only legitimate way society has to redefine what's legal is by changing the law. Until that happens, we need to stick with current regulations, except in very rare cases where things like civil disobedience may be called for. If you care about this, then do what you can to elect candidates who see things the way you do.
I don't understand the overwhelming sympathy for those who steal, 12yr old or otherwise. If you were mugged by 50,000,000 mostly teenagers, would you say "aw, stop picking on the kids" ?
Aren't many of you /.ers also software developers like me? If I choose to release an open source project for anybody to use, that's fine. But if I make my living writing code -- or performing music -- then why is it acceptable for people to take my product w/o compensating me?
Does theft suddenly become acceptable when it's done on a massive scale? Would you support a company protecting its rights if there were only dozens of thieves instead of scores of millions?
The other argument of "RIAA is evil, therefore committing crimes against them is okay" doesn't wash either. These aren't victimless offenses, the money has to be coming out of somebody's pocket. If CDs are too expensive, then don't buy them. I don't claim the right to steal an SVT Cobra just because the $35k list price is too expensive.
IMHO, RIAA and other music associations should enforce their rights and are obligated to do so for their stockholders. Dunno of any practical way for them to ensure they pursue only unsympathetic defendants. Seems to make sense for them to ease into it, have a little consideration when negotiating with naive downloaders, gather as much publicity as possible so that music listeners in the future will understand that stealing music is wrong.
BTW, I have an adult friend who insists it's perfectly legal to make copies of her CDs as long as she only gives them to friends. The only way such attitudes will change is when people are prosecuted for it.