What reason could you possibly have had for modding me down? If I squint really hard I can maybe make out a reason for modding me offtopic... but troll? Isn't this a legitimate question?
IBM drives have always treated me really well in the past (and when I did have a problem I was impressed by their replacement policy and speed) so I unhesitatingly bought two 76.8 gig 75gxps a few months ago... now we have all this nonsense coming out... Do I just keep good backups and wait for the failure or what? I'm way past the 30 days on the drives...
Hah... on flac's comparison page he tested his program on a cannibal corpse song! I'd never use his program anyway, but for that he gets a thumbs up from me for sure
SPEC benchmarks are all legitimate, real world applications (or do you consider gcc to be a toy benchmark?). Where are you getting your information from?
Immediately after I hit submit I realized what you meant when you were talking about different motherboards giving different numbers. To be honest, though, I don't think there would be a way around that with any benchmarking suite. Processors are starting to incorporate data prefetching so we need to test their performance with benchmarks that have a significant data cache miss rate to fairly compare them. That in mind, there will be differences based on the motherboard used.
The question I would pose is, what benchmarks is AMD using? I bet they are programs with, for example, bad branch prediction rates, so as to put the P4 at a disadvantage.
Now that I think about it, I'd still feel better with AMD branding their chips with the SPEC rating because at least that is a set of benchmarks not specifically chosen by the vendor doing the speed comparison. Perhaps to make things fairer they should compare SPEC point per $ spent (I'm not sure how since prices change).
Ok, maybe I have a solution. What should happen is systems you buy are sold rated by their SPEC (or some other benchmark rating), but individual processors you buy give the MHz rating. That way, the 95% who don't know much can buy on a fair performance comparison (based on the actual SPEC numbers of the system they're buying) and the rest who know MHz isn't everything can still get all the information they want.
It may be highly dependent on the motherboard, but I hardly see that as a reason to invalidate SPEC as a benchmark suite. One of Intel's advantages to this point has been the performance of the motherboards they create. Who really cares if the AMD processor is faster 'on paper' but in any real machine it will be slower?
If AMD doesn't like the numbers, they're going to have to bite the bullet and publish some numbers using a nonAMD motherboard. The fact of the matter is that SPEC is still the most widely industry recognized and accepted benchmark suite.
The reason I thing SPEC is a good measure (even allowing for the issue you raised) is because the vendor has complete control over the numbers he publishes--he gets to use his choice of hardware, his compiler etc. We don't have any of these complaints that Oh quake 3 wasn't fairly optimized etc.
Just my opinion I suppose, and I do acknowledge the large number of data streaming floating point benchmarks is going to give a processor using RDRAM and advantage, but that's just life.
The "proper" way to compare processors would be to use the SPEC benchmarks (http://spec.org) (since they are supposed to "define" processor performance). However, AMD won't use these benchmarks use them. If we look at the highest published scores from AMD we see
The athlon was Advanced Micro Devic Gigabyte GA-7DX Motherboard, 1.4GHz 1 cpu
The P4 was Intel Corporation Intel D850GB motherboard (2.0 GHz, Pentium 4 processor) 1 cpu
Obviously, this isn't a totally fair match (this is the most recent numbers I could find from both on this page http://www.spec.org/osg/cpu2000/results/cpu2000.ht ml) (there is no CPU2001 benchmarks... we'll probably stick with CPU2000 until 2005 or so... we stuck with the SPEC95 benchmarks for 5 years). The newer althonxps will have improved numbers, but the 1.4ghz part was beaten pretty badly (and those numbers came from AMD themselves, so their setup was optimal in their eyes).
I can only speculate that the proof in this case has not been made public because it would endanger future intelligence gathering efforts. however
1) Every other country has been 100% convinced by the evidence we showed them
2) Evidence has already been presented tying him to other terrorist acts (USS Cole and Embassy bombings)
Yes, I would *hope* the US would hand over someone with a lot less evidence than we have on bin Laden
Their market share is a lot worse off than it has been in past years... their highest end processor (Itanium) shipped years late and with rather poor performance...
In fact, the only thing that is going well for them is that their marketing is being successful in getting people to waste money on p4s (including someone I know at my school).
Yeah but 40 hours battery life is pretty nice from MD (don't own one myself so I can't verify his numbers). I don't know how long the mp3 cd players go but it can't be too long since they have both moving parts and a mp3 decoder to power...
Quit yer bitchin.... You want to get out in four years? Then give up the wild parties and loose women like I did... I finished in four years taking MORE classes than I needed to graduate (several grad level courses) and graduated summa cum laude...
Another thing I would add is that it takes a hell of a lot of work to get the same IPC (instructions per cycle) on a processor with a 13 stage pipeline than it does on a 5 stage pipeline (not to mention, again, that the ppro was a 3 wide, out of order machine, unlike the 486 which was 1 wide, in order, so the PPro actually had far greater IPC).
Honestly, why do people feel they can flippantly comment on something that took for more than 100 man years to develop when they have absolutely no background in the area...
Re:A bit of a naive statement on your part.
on
More On Tragedy
·
· Score: 1
My good friend dwain... got bored fooling mods with fake computer architecture language so now you're claiming to be from south africa?
Well, that quote has already become incredibly overused, but even with what he said... are you ignoring the term 'temporary' in there? What leads you to believe that giving up some of the specific liberties relating to this issue would provide only a temporary benefit?
Whenever people start talking about marketing leading engineering at intel it makes me ask if that person actually knows anything about computer architecture...
Slashdot Mirror
What reason could you possibly have had for modding me down? If I squint really hard I can maybe make out a reason for modding me offtopic... but troll? Isn't this a legitimate question?
IBM drives have always treated me really well in the past (and when I did have a problem I was impressed by their replacement policy and speed) so I unhesitatingly bought two 76.8 gig 75gxps a few months ago... now we have all this nonsense coming out... Do I just keep good backups and wait for the failure or what? I'm way past the 30 days on the drives...
losers
Hah... on flac's comparison page he tested his program on a cannibal corpse song! I'd never use his program anyway, but for that he gets a thumbs up from me for sure
SPEC benchmarks are all legitimate, real world applications (or do you consider gcc to be a toy benchmark?). Where are you getting your information from?
Immediately after I hit submit I realized what you meant when you were talking about different motherboards giving different numbers. To be honest, though, I don't think there would be a way around that with any benchmarking suite. Processors are starting to incorporate data prefetching so we need to test their performance with benchmarks that have a significant data cache miss rate to fairly compare them. That in mind, there will be differences based on the motherboard used.
The question I would pose is, what benchmarks is AMD using? I bet they are programs with, for example, bad branch prediction rates, so as to put the P4 at a disadvantage.
Now that I think about it, I'd still feel better with AMD branding their chips with the SPEC rating because at least that is a set of benchmarks not specifically chosen by the vendor doing the speed comparison. Perhaps to make things fairer they should compare SPEC point per $ spent (I'm not sure how since prices change).
Ok, maybe I have a solution. What should happen is systems you buy are sold rated by their SPEC (or some other benchmark rating), but individual processors you buy give the MHz rating. That way, the 95% who don't know much can buy on a fair performance comparison (based on the actual SPEC numbers of the system they're buying) and the rest who know MHz isn't everything can still get all the information they want.
It may be highly dependent on the motherboard, but I hardly see that as a reason to invalidate SPEC as a benchmark suite. One of Intel's advantages to this point has been the performance of the motherboards they create. Who really cares if the AMD processor is faster 'on paper' but in any real machine it will be slower?
If AMD doesn't like the numbers, they're going to have to bite the bullet and publish some numbers using a nonAMD motherboard. The fact of the matter is that SPEC is still the most widely industry recognized and accepted benchmark suite.
The reason I thing SPEC is a good measure (even allowing for the issue you raised) is because the vendor has complete control over the numbers he publishes--he gets to use his choice of hardware, his compiler etc. We don't have any of these complaints that Oh quake 3 wasn't fairly optimized etc.
Just my opinion I suppose, and I do acknowledge the large number of data streaming floating point benchmarks is going to give a processor using RDRAM and advantage, but that's just life.
The "proper" way to compare processors would be to use the SPEC benchmarks (http://spec.org) (since they are supposed to "define" processor performance). However, AMD won't use these benchmarks use them. If we look at the highest published scores from AMD we see
t ml) (there is no CPU2001 benchmarks... we'll probably stick with CPU2000 until 2005 or so... we stuck with the SPEC95 benchmarks for 5 years). The newer althonxps will have improved numbers, but the 1.4ghz part was beaten pretty badly (and those numbers came from AMD themselves, so their setup was optimal in their eyes).
Integer: 495 base, 554 peak
FP: 426 base, 458 peak
For P4
Integer: 640 base, 656 peak
FP: 704 base, 714 peak
The athlon was Advanced Micro Devic Gigabyte GA-7DX Motherboard, 1.4GHz 1 cpu
The P4 was Intel Corporation Intel D850GB motherboard (2.0 GHz, Pentium 4 processor) 1 cpu
Obviously, this isn't a totally fair match (this is the most recent numbers I could find from both on this page http://www.spec.org/osg/cpu2000/results/cpu2000.h
Anyway, there you go
But he has already *admitted* to being involved in other terror attacks. That is more than enough reason to extradite him...
Maybe you should keep up on the news. We showed our evidence to NATO members, Pakistan and others and all were convinced of his involvement...
I can only speculate that the proof in this case has not been made public because it would endanger future intelligence gathering efforts. however
1) Every other country has been 100% convinced by the evidence we showed them
2) Evidence has already been presented tying him to other terrorist acts (USS Cole and Embassy bombings)
Yes, I would *hope* the US would hand over someone with a lot less evidence than we have on bin Laden
Their market share is a lot worse off than it has been in past years... their highest end processor (Itanium) shipped years late and with rather poor performance...
In fact, the only thing that is going well for them is that their marketing is being successful in getting people to waste money on p4s (including someone I know at my school).
SO WHERE IS THE MARKETING FUCKUP?
Where is the marketing mistake? Intel made money, AMD didn't... fill me in on this if you don't mind...
It's called capitalism... deal with it...
or get a job there like I will once I finish my phd...
Clearly a questions that couldn't be answered in any way but an ask slashdot
Yeah but 40 hours battery life is pretty nice from MD (don't own one myself so I can't verify his numbers). I don't know how long the mp3 cd players go but it can't be too long since they have both moving parts and a mp3 decoder to power...
Quit yer bitchin.... You want to get out in four years? Then give up the wild parties and loose women like I did... I finished in four years taking MORE classes than I needed to graduate (several grad level courses) and graduated summa cum laude...
BLAH
No, it prompts the question. Begging the question is something else...
*sigh*
Another thing I would add is that it takes a hell of a lot of work to get the same IPC (instructions per cycle) on a processor with a 13 stage pipeline than it does on a 5 stage pipeline (not to mention, again, that the ppro was a 3 wide, out of order machine, unlike the 486 which was 1 wide, in order, so the PPro actually had far greater IPC).
Honestly, why do people feel they can flippantly comment on something that took for more than 100 man years to develop when they have absolutely no background in the area...
My good friend dwain... got bored fooling mods with fake computer architecture language so now you're claiming to be from south africa?
happy to hear you signed up with the army
Well, that quote has already become incredibly overused, but even with what he said... are you ignoring the term 'temporary' in there? What leads you to believe that giving up some of the specific liberties relating to this issue would provide only a temporary benefit?
Actually all processors retire instructions in order... just some of the execute them out of order...
And theres no way someone has said exactly what you just did every day for the last 30 years either
Whenever people start talking about marketing leading engineering at intel it makes me ask if that person actually knows anything about computer architecture...
well... do you?