AMD's Bobcat and Bulldozer

[Opteron146]

if you miss a branch you have to flush all n stages of the pipelines. the probablity of an misprediction increases exponentially with pipeline length.

Not if you have a redirect recovery cache:

it's obvious that amd knows this but saying intel did it wrong and amd did it right/better is a foolish way to look at it.

Yes it is foolish, but it could still be the truth

[Sn0wm@n]

omg... cant wait to see a bulldozer at 5ghz then

[informal]

It's interesting terrace doesn't mention his "significant single thread loss" "argument" any more,since we know that it applies to workload components when module is fully loaded.It was funny how he tried to quote it partially and use it to "prove" what he wishes for

[Sn0wm@n]

There would be no point for anyone to choose Zambezi over octal Sandy, except price, possibly.

Also, JF admitted elsewhere that the 50% / 33% claim for Interlagos / MC has been substantially juiced by including, in the aggregate used to measure, serial workloads that benefit from the Turbo that MC lacked. While adding Turbo is a good thing, this means the "fully-parallel" throughput improvement is considerably less than 50%. In order to make sense of the claim, you're going to need to see exactly what they've chosen to average over, and at that point, you'll likely have access to simpler zambezi benchmarks of all sorts.

there he goes again .... how much intel is paying you .... because to type that much bullstuff they must be paying you quite a big sum i presume ....

[Hornet331]

ble.. to many crystal ballers and imaginativ readers in here... no more real information can be gathered in here.. im out.

[terrace215]

there he goes again .... how much intel is paying you .... because to type that much bullstuff they must be paying you quite a big sum i presume ....

I'm sure you trust informal-- why not ask him what he thinks the chance is of zambezi outperforming an octal-core SB on... well, anything, really, but let's say desktop/client applications...

[Manicdan]

like most of us we think the advancement of CPUs for 2011 are going to be very impressive.

competition leads to reduced prices, if you want AMD to suck it hard, then good job on buying your cpu for an extra 100$ cause amd didnt deliver

[terrace215]

It's interesting terrace doesn't mention his "significant single thread loss" "argument" any more,since we know that it applies to workload components when module is fully loaded.It was funny how he tried to quote it partially and use it to "prove" what he wishes for

Really? I mean, you can practically rephrase those slide bullets as "when designing BD, we prioritized throughput on multithreaded workloads over single-threaded performance."

AMD has said over and over again it is designed for the "throughput era" of computing, blah blah. The signs are all there. And the design elements that have just been revealed point that way, too. (And remember how just a few days ago, all the fans were convinced each core would have 4 ALUs, because having 2 would be a step back? Now that AMD says "2", suddenly, oh, 2 is wonderful! More would be a waste! )

Of course one of their marketing guys is going to push back on this to a client-based audience, that's what marketing folks get paid to do.

[Mats]

I'm sure you trust informal-- why not ask him what he thinks the chance is of zambezi outperforming an octal-core SB on... well, anything, really, but let's say desktop/client applications...

That's not even fair, Zambezi have 8 threads, and an 8 core SB have 16 threads. It would be a huge price difference.
Nice try.

[informal]

I'm sure you trust informal-- why not ask him what he thinks the chance is of zambezi outperforming an octal-core SB on... well, anything, really, but let's say desktop/client applications...

Client workloads rarely task all cores,so 16 threads on SB won't help it there much.Only a handful of applications scales well to 8 cores,the rest are sub 6 or heck even sub 4 thread aware. And in this case you will have an all around new AMD core vs somewhat upgraded Westmere core(SandyBridge).AMD needs around 15-20% IPC(the sort of jump they got with 10h from tweaking the *existing* Hammer core ) and a new and improved Turbo boost,which they will have (power gated Turbo on 32nm HK/mg;4Ghz Turbo is totally plausible)-all this come close enough to SB. So they are not bringing a pocket knife to client desktop fight .

[crazydiamond]

you say that so much and i think it's easier to just either visit another forum, not get so upset, ignore the posts, or even just set certain users to ignore. notice it's not a big deal to anyone else. the shintai era is over dude.

some people dont like amd and you'll just have to deal with it.

guess im not the only one sick of it

Why? Does this boil down to your special insight with regards to GF's 32nm process?

Maybe I should join a Britney Spears forum and see what being negative in a forum is all about, because I honestly don't know.

Eh? The way I see it, AMD added tons of stuff to enhance single threaded performance. It's not just one or two things. Are you sure you're reading the same slides I am reading?

This is simply not true. No matter how many times you say it.

You always doubt AMD's methodology, sales, technical details, release dates and... etc

terrace love only Sandy bridge

fixed.

indeed ... tired of people making fail prediction in advance with no hard benchmark numbers to put their claims to the test

Quoted for truth. We need LESS BAD information.

LOLL terrace you make me lol all the time .. keep it up

XS should have troll of the year awards or something!

Your understanding skills are ridiculous.

Good god..

I'm simply godsmacked at JF-AMD's patience to keep posting here with so much trolling around.

terrace215: get a life and if you love Intel so much, the Intel forum is a few steps below...

ble.. to many crystal ballers and imaginativ readers in here... no more real information can be gathered in here.. im out.

[MrMojoZ]

Of course one of their marketing guys is going to push back on this to a client-based audience, that's what marketing folks get paid to do.

And you get paid to do what, spread mis-information? Troll? Steer AMD threads away from anything but fanboy stupidity? You seem to have an issue with marketing people while trying to push your own marketing campaign. Sweet, sweet hypocrisy.

[AKM]

WOW, Fear uncertainty and doubt galore in this thread..

[generics_user]

guess im not the only one sick of it

woooow and you only got 20% of all posts written in such a manner

[blindbox]

Um, measure the performance. Note the power used (for the per W) part. Note the core size used to run it. Compare to previous offering.

The "per mm^2" is from AMD's claim, btw, check the slide you posted.

Heh, looks like you're looking way too deeply into things. That's simply implying it's either perf/W or perf/mm^2. Obviously an increase in perf/W with respect to single-threaded performance means faster clock-for-clock. Unless you're saying AMD found a way to get 6 GHz in the same thermal envelope. That would be the biggest speculation of the year.

guess im not the only one sick of it

+1, good quote compilation.

And this year's XS's biggest troll award goes to....

[Chumbucket843]

i dont read or reply to his posts and it works well for me. i just find it ironic that the people who are most interested in this stuff are the ones who fall into the trap and ruin potentially good threads. the people not bothering with him outnumber the ones who reply. choose your battles.

Not if you have a redirect recovery cache:




Yes it is foolish, but it could still be the truth

not if you miss the recovery cache. that's pretty interesting though.

well, from reading just this forum i would assume that AMD did dual core right first, they did na(t)ive quadcore first, they did mcm better and now pipelining? what else cant intel do better?

i dont mind people liking/preferring amd, intel or nvidia but when they start insulting competitors designs it is annoying. they all do some very cool stuff and saying it is wrong or inferior is foolish, especially something from 5 years ago.

[Mats]

i dont read or reply to his posts and it works well for me. i just find it ironic that the people who are most interested in this stuff are the ones who fall into the trap and ruin potentially good threads. the people not bothering with him outnumber the ones who reply. choose your battles.

Agreed. I'll try not to make any replies, and I hope others will too.

[richierich]

everyone just needs to put him on the ignore list.

[IDANPE]

50% improvement from MC with 33% more cores ?

so 16 core MC@ 2.6GHZ = interlagos..

bad ? no.
good ? maybe, dependents what is the TDP & price.
revolutionary? not imo.

[qcmadness]

Now another question.

Why AMD did not go after 6-module Bulldozer CPUs?

AFAIK, the size of a Bulldozer module @ 32nm should be similar or smaller than a K10 core @ 45nm.
It would not be a difficult task to integrate 6 modules into a CPU die with ~300mm^2 size.

[terrace215]

That's not even fair, Zambezi have 8 threads, and an 8 core SB have 16 threads. It would be a huge price difference.
Nice try.

It's not fair to compare the 2 products Intel and AMD will be offering to the desktop enthusiast market segment because the Intel part supports twice the number of threads?

I'm sure the Intel parts *will* cost more, but they'll perform better, and in that, help justify the price differential.

[terrace215]

guess im not the only one sick of it

That's a nice collection of posts that should be moderated, for the most part.

[HelixPC]

I love new hardware, no matter who is releasing it, exciting and good times ahead, especially since ATi has been on a roll since 2008!. The people who try to talk down new hardware without knowing anything about it should re-evaluate the situation, i like both nvidia, intel, amd, and ati, why would anyone want any of the companies to do bad? that would be horrible for the customer, i hope all companies continue to make money and make good products!

[terrace215]

Now another question.

Why AMD did not go after 6-module Bulldozer CPUs?

AFAIK, the size of a Bulldozer module @ 32nm should be similar or smaller than a K10 core @ 45nm.
It would not be a difficult task to integrate 6 modules into a CPU die with ~300mm^2 size.

Might be yields, with core logic being more susceptible to defects (and that would be 12 cores), might be the L3 cache interconnect work required, or an analysis of the stress on the L3 that 12 cores would exert. AFAIK, AMD has not gone the ring-bus route with their L3 that Intel has done with N-EX and SB, but we'll have to wait and see to be sure.

[Hans de Vries]

Agreed. I'll try not to make any replies, and I hope others will too.

Not replying to "Terrace215" will help him with his "addiction". He has well
over 100,000 posts on the Intel-vs-AMD subject under dozens of different
nicknames all over the Internet.

The last few years he manages to avoid getting personal which is a good
thing but he's clearly getting carried away again now.

The funny thing is that he is still the "all time AMDroid world record holder"
as well, relentlessly trolling Intel fans.... as this older thread shows:
http://www.xtremesystems.org/forums/...ad.php?t=55672


Regards, Hans