AMD tapes out its Bulldozer CPU architecture

[largon]

Again, flawed equation. If each core was 12.5% faster, there'd be more than 50% increase. 12.5%, times 16 cores= what? 200%? and 12 cores is 150%?

You're making no sense.
What can I say...

Please. FIX. THAT. math.

...ditto.

 

[erocker]

I just hope the AMD fanboys wont get all upsad if they are not what they expect . Nut I sure hope AMD finally gets it right . My hope is they are going to at the very least mach the core i7's . that is a tall order seeing as AMD has been trying for years now .

I hope the term "fanboy" dies, is never heard again, and AMD releases a good/competetive CPU at a good price.

 

[cadaveca]

AMD didn't use number of cores do denote how much faster they are, just 33% more cores make a 50% increase. The 12 and 16 cores I'm assuming were just pulled out since they make that 33% more a very convenient whole number.

I mean no offense by this, but really, your reading comprehension sucks big time.

The number of cores is explicity stated, even in what I quoted, that started the discussion down this avenue.

They said, and I quote,:

From a performance standpoint, if you compare our 16-core Interlagos to our current 12-core AMD Opteron™ 6100 Series processors (code named “Magny Cours”) we estimate that customers will see up to 50% more performance from 33% more cores.

So, again, if each core is 12.5% better, over 16 cores, this would be a 200% boost, not 150%.

You're making no sense.
What can I say...
...ditto.

I understand how that equation was formed, however, clearly, it isn't correct. The math may be properly calculated, however, the example is NOT accurate.


If they get 12.5% per core, AMD has a killer chip here. However, thier own numbers says differently.

 

[trickson]

I agree on that, there past two lineups have yet to truly past an i7. I'm hoping we will see a good resurgence with this architecture. As a person who doesn't cum my pants over a particular company(aka a fanboy), i truly won't be heartbroken if these chips don't succeed.

Same here .

 

[Beertintedgoggles]

Can I at least get an agreement that 50% more performance with 33% more cores means exactly this:

Old Proc:
12 cores / 1000 X's done per unit
16 cores / 1333.3 X's done per unit

So then New Proc:
12 cores / 1125 X's done per unit
16 cores / 1500 threads

Pretty sure we can agree on that. So then that means each core of the old 12 core processor did 83.333 X's per unit. The new 12 core processor then does 93.75 X's per unit. So you get 12.5% increase per core again. The same works if you want to use the math for a 6 to 8 core comparison.

Lets do it the opposite way now, making each core 12.5% faster (93.75 X's per unit rather than 83.33333 since 93.75/83.333333 = 1.125)

6 cores doing 83.33333 X's per unit per core = 500 X's done per unit of time
make them all 12.5% faster: 6 cores * 93.75 X's etc. = 562.5 X's done per unit of time
That's 562.5/500=1.125 (12.5% faster) not 6*12.5% faster (75%)

So then an 8 core would do 8 * 93.75 = 750 X's per unit of time or in other words 50% more performance than the 500 X's of the 6 core with 33% more resources. Comprehension and application, not all can do it well.

 

[CDdude55]

Can I at least get an agreement that 50% more performance with 33% more cores means exactly this:

Old Proc:
12 cores / 1000 X's done per unit
16 cores / 1333.3 X's done per unit

So then New Proc:
12 cores / 1125 X's done per unit
16 cores / 1500 threads

Pretty sure we can agree on that. So then that means each core of the old 12 core processor did 83.333 X's per unit. The new 12 core processor then does 93.75 X's per unit. So you get 12.5% increase per core again. The same works if you want to use the math for a 6 to 8 core comparison.

Lets do it the opposite way now, making each core 12.5% faster (93.75 X's per unit rather than 83.33333 since 93.75/83.333333 = 1.125)

6 cores doing 83.33333 X's per unit per core = 500 X's done per unit of time
make them all 12.5% faster: 6 cores * 93.75 X's etc. = 562.5 X's done per unit of time
That's 562.5/500=1.125 (12.5% faster) not 6*12.5% faster (75%)

So then an 8 core would do 8 * 93.75 = 750 X's per unit of time or in other words 50% more performance than the 500 X's of the 6 core with 33% more resources. Comprehension and application, not all can do it well.

If i was good at math i would probably agree.

 

[Beertintedgoggles]

Sorry for hijacking the thread, just percentages don't always go hand in hand with reasoning sometimes. One last math post and I'll shut the hell up.

Say 1 core does 100 X's per second
6 cores then do 600 X's per second (in a perfectly threaded world)

12.5% faster cores then do 112.5 X's per second
6 of those will do 675 X's per second

That is still only a 12.5% improvement (675/600 = 1.125).

Saying a 6 core then would have 12.5% improvement on each core * 6 total cores gives you 75% more performance seems like it makes sense but it's the wrong way to think about it. (that 75% improvement would mean 600*1.75 = 1050 X's per second)

 

[trickson]

All the math in the world will not make any difference any way , Till we get the chips all this is a moot point any way .

 

[cadaveca]

All the math in the world will not make any difference any way , Till we get the chips all this is a moot point any way .

Ah, thank you.

Sorry for hijacking the thread, just percentages don't always go hand in hand with reasoning sometimes. One last math post and I'll shut the hell up.

Say 1 core does 100 X's per second
6 cores then do 600 X's per second (in a perfectly threaded world)

12.5% faster cores then do 112.5 X's per second
6 of those will do 675 X's per second

That is still only a 12.5% improvement (675/600 = 1.125).

Saying a 6 core then would have 12.5% improvement on each core * 6 total cores gives you 75% more performance seems like it makes sense but it's the wrong way to think about it. (that 75% improvement would mean 600*1.75 = 1050 X's per second)

yes, yes, I said the math was correct, no? I also said the equation was wrong...because we don't know if that "promise" will ever materialize...I call to evidence, a certain blond, hyping the 3GHZ Phenom 1 chips...


The rest was me being silly, as I saw "my way" of looking at it as accurate as your numbers...both are speculation at this point, so both have equal weight on reality. We've got to wait until next year, which is at least 4 months away.

I did tell you to take no offense at my comment about comprehension...obviously you did, so I apologize.

Saying a 6 core then would have 12.5% improvement on each core * 6 total cores gives you 75% more performance seems like it makes sense but it's the wrong way to think about it. (that 75% improvement would mean 600*1.75 = 1050 X's per second)

Actually, I very purposely translated that to 200%, rather than the equivalent of having a 17 core cpu rather than a 16 core. It DOES give you 75% more performance...using your example...of an individual core, not the entire cpu.

 

[pantherx12]

This discussion is confusing as hell, how did people end up adding together percentages?

Just to clarify if you have two processors with the same amount of cores, but one is 12.5% more Efficient per clock cycle per core, you get a net increase of performance of 12.5%

 

[TheMailMan78]

I am bringing back Mailman math.

The new AMD processors will have approximately 150 more Mailman's then the previous generation. Thats a 3000% increase.

 

[Techtu]

I am bringing back Mailman math.

The new AMD processors will have approximately 150 more Mailman's then the previous generation. Thats a 3000% increase.

about time I heard some proper calculations about this

 

[pantherx12]

I am bringing back Mailman math.

The new AMD processors will have approximately 150 more Mailman's then the previous generation. Thats a 3000% increase.

150 more!

OUTSTANDING!

 

[trickson]

What is really more to the point is with out the chips in your hands how can one even try to do the math ? And we all know how well AMD did with the math on the Phenoms so why are you trying to make the point of how good it is with math ? I just do not get it . I mean really you could say they are 1 million times faster but that would not add up to shit unless the chip was in your hands . So lets stop the math talk and get on with the bulldozer !

 

[JATownes]

I am bringing back Mailman math.

The new AMD processors will have approximately 150 more Mailman's then the previous generation. Thats a 3000% increase.

The way I understand it the new Intel will have 152 more Mailmans, so I will get one of those.

 

[pantherx12]

What is really more to the point is with out the chips in your hands how can one even try to do the math ? And we all know how well AMD did with the math on the Phenoms so why are you trying to make the point of how good it is with math ? I just do not get it . I mean really you could say they are 1 million times faster but that would not add up to shit unless the chip was in your hands . So lets stop the math talk and get on with the bulldozer !

It was the maths they were arguing over more than anything

 

[erocker]

I am bringing back Mailman math.

The new AMD processors will have approximately 150 more Mailman's then the previous generation. Thats a 3000% increase.

What is really more to the point is with out the chips in your hands how can one even try to do the math ? And we all know how well AMD did with the math on the Phenoms so why are you trying to make the point of how good it is with math ? I just do not get it . I mean really you could say they are 1 million times faster but that would not add up to shit unless the chip was in your hands . So lets stop the math talk and get on with the bulldozer !

That's 50,000 more Trickson units to you and me.

 

[cadaveca]

It was the maths they were arguing over more than anything

UH, actually...no...I said from the start that the math was correct. We were NOT arguing about it...I even said THE EQUATION was wrong....many just wrongly interpreted what I was on about, and It's not liek I really did anything to change those perceptions...I was being sarcastic, and that never translates well.



However, Mailman got it.


Many are hoping for new details tomrorow, I don't think very much new will be told...AMD has so much info out already, I am really starting to question why...

It really does remind me of how they were, pre-Phenom1. Except this time, they seem to exude far more confidence...with the 3GHZ Phenom1 stuff, it was obvious they were lying.

 

[pantherx12]

English isn't your first language then I'm guessing.

Because from what you wrote on the previous pages you were saying the mathematics were wrong hence the guy getting so uppity dude

 

[trickson]

One thing is for sure , Once we get them there will be no need for all the speculation the math or any BS we will know for sure . I hope they are far better than there predecessors . I do not put must stock in all the math and Hype I want to see real world chips nothing less will do for me .

 

[cadaveca]

English isn't your first language then I'm guessing.

Because from what you wrote on the previous pages you were saying the mathematics were wrong hence the guy getting so uppity dude

Actually, he got uppity becuase I said his reading comprehension sucks. I said that because he said:

The 12 and 16 cores I'm assuming were just pulled out since they make that 33% more a very convenient whole number

So I played a bit. He didn't even read the posts, clearly..he was jumping into the argument for the sake of the argument.

It's no big deal. But now i want to question YOUR reading comprehension...


And I say this because he posted in response:

Comprehension and application, not all can do it well.

And no, I'm not serious. This is a speculation thread. Everything in it needs to be taken with a WHOLE LOT of salt. Including this post.

my first language is Martian, BTW. :shadedshu

 

[Beertintedgoggles]

That's all I've been trying to explain, the math behind it. I never claimed to believe the numbers AMD posted, more than likely they were the best results on a very biased set of tests.... happens all the time. I hope this round of processors finally proves to be the next leap for AMD architecture, I'm still rockin an ancient opteron on skt 939.

Edit: I did get uppidy after you told Steevo he was wrong when he wasn't.

 

[trickson]

You what would really be funny ? If they come out and only 10% faster than my Q9650 I would till I pissed myself .

 

[pantherx12]

Sometimes the internet makes communication harder rather then easier eh?

Cheers for posting the AMD blog last page by the by, not read that one myself yet.

 

[cadaveca]

Sometimes the internet makes communication harder rather then easier eh?

Cheers for posting the AMD blog last page by the by, not read that one myself yet.

LoL. you bet. that's a big part why I don't really take very much of it too seriously.

By the by, there's quite a few blog posts about Bulldozer and Bobcat. If you want the info that's out there, hit up those blogs...best to get the info from the source, as always.