Some Bulldozer and Bobcat articles have sprung up

[William Gaatjes]

So wait, I'm probably just looking at this slide wrong. Will this appear as 1 core or 2? If it is 2, than I stand by my claim that highly threaded FP performance might suffer. But if it is 1 then my claim is off the wall and you can kindly ignore me .

Well, i have not kept track for a while with respect to cpu's and gpu's but
do you not think this is also a start of another fusion idea ? Slowly dumping the ancient FPU for the raw calculation power of the general programmable GPU. As long as legacy code requires it the FPU will be present but sooner or later the programmers will be made more and more enthusiastic to program number crunching algorithms for the g-gpu or how it is called. But the real advantage will only come afcourse when calculations can be done in parallel. Then the gpu( in double precision i think) can really shine.

 

[Idontcare]

Sometimes I never learn. If you were unfortunate enough to read my post, you at least understand how I feel. However, that post really added nothing worthwhile to this discussion.

we are all human, the important thing is that we learn from our mistakes and give others the patience and tolerance needed so they can do the same when they commit similar gaffes...good on ya for thinking about twice and taking the higher path :thumbsup:

 

[JFAMD]

@JFAMD,

Great info on Round 2 (better than Round 1, but it's probably a personal thing). The second question there is actually something I've had in mind but never got around to asking. Like how modern OSes are HT-aware now, I would imagine it would be a performance guarantee to make sure modern OSes are somehow also "module-aware" such that 2 threads don't always end up in a single module to completely negate any performance drop caused by resource sharing

Why exclude Apple, though? You only mentioned Windows and Linux operating systems specifically.

I don't have processors in apple systems.

 

[JFAMD]

@JFAMD
I'm sorry, that was a joke, as I indicated in the disclaimer.

I realize now that if some people don't realize that, they might actually make your job harder.

So to anybody who didn't get that I was kidding because they didn't read the disclaimer, I was interpreting relevant PR data and "spun" it to the best possible outcome to get an incredible single-thread boost, and it was all a joke. Please do not quote that as any reasonable figure or bother JFAMD for any comment about it.


EDIT:


Actually, I did take into account for "rush hour traffic", by mentioning the 20% dual-thread penalty and adjusting the single-thread as necessary. I am now confused if you only have problem with the "low" estimate (13%) and not the adjusted high 40% estimate. Or do you mean it should even be much higher than 40%?

Regardless, it was not meant to be taken seriously at all, whether your protestations are about 13% or (more rightfully) the basis of the figures (a server benchmark, no context at all regarding test environment, apps involved, etc - my basis for regarding the 40% figure as not founded in fact).

I hope I didn't cause you too much grief Sorry, man. Would you prefer I edit out that post? I am not into the habit of ever editing out anything I post (it screws up the flow of the thread), but if it will help make your job not harder, I suppose it won't be too much of a bother for me, especially since I was only kidding.

The problem with jokes is that they spread very quickly and I have to debunk it in 20 different forums and on my blog.

 

[jvroig]

The problem with jokes is that they spread very quickly and I have to debunk it in 20 different forums and on my blog.

Yes, sorry. I don't want to give you flashbacks of the April Fools incident. It shall disappear.

 

[jvroig]

I don't have processors in apple systems.

You are right, for a moment there I was in la-la land where I could get an Xserve with Interlagos. Well, at least they are using your graphics now, maybe Bulldozer will win you some Apple contracts in the future.

 

[bryanW1995]

Actually in a multithreade FPU environment, we would have an advantage.

In AVX we will have 8 256-bit units
In non-AVX we will have 16 128-bit units.

Compared to everything I have seen on the server Sandybridge, they will have 8 256-bit AVX units, so we are generally tied on AVX code, but on non-AVX code they will only have 8 128-bit units, or half the FP capability.

Remember that most apps will not be recompiled to take advantage of AVX right away, so we have an advantage.

Also, unless they have changed their scheduler, they have 1 that covers 2 integer threads and the FPU. We have one for each integer thread plus one for the FPU, so in a multithreaded environement I would bet on Bulldozer.

are you talking about the 16 core server version of bulldozer or the 8 core consumer version? are we going to see more than 8 cores at a later date for the consumer version?

edit, nevermind, as mentioned later it was the 16 core interlagos.

 

[Cerb]

This post was completely unnecessary. We're just trying to figure out where Bulldozer will stand compared to current CPUs (both Intel and AMD), based on what we know so far, and what we deem possible and impossible, barring the use of pixie dust and magic unicorns.

Except for barring the use of pixie dust and unicorns, I disagree . You were dismissing budget, as though it were not a worthy concern, when a few minutes browsing GH would clearly prove otherwise, and also show that AMD is recommended quite a bit, and for good reason.

It's not a thing of being in an AMD camp (really, we're not; the underdog is just more interesting). It's truly interesting how they plan to make risky compromises that, without being able to make transistors as small and fast as Intel, and now technically not making them at all, could net them being competitive in the actual market, to such a degree that they might be able to really make some money from CPUs, again. To do this, they also have to be improving performance.

The CPU people actually get isn't usually the one at the top of the performance charts. And, unless AMD takes a change as radical as Niagara to target servers, the performance needs of servers, workstations, and gaming PCs are not all that different, even though the relative performance of newer hardware can be (IE, throw cheap RAM and a graphics card in with a server CPU, and you have a good gaming CPU). Now, we're seeing that come around in software, by way of game engines being multithreaded from the get-go, which will lead to more games capable of using 4+ cores in the near future (which could also help BD, in the possible case that single-thread performance isn't quite there). I contend that if BD can live up to most of its promises, relative to current AMD CPUs, it aught to be competitive at gaming, and competitive at that in large part due to Intel's margin addiction, much like today.

As long as it turns out to be true that per-thread IPC is actually higher, on average, than a Phenom II, by a significant margin (>10%, but preferably around 20%), then I don't see why BD can't also be a compelling gaming CPU, since we can pretty much be guaranteed that Intel will only bring its prices down to a certain level.

 

[Scali]

You were dismissing budget, as though it were not a worthy concern

Not in this case, since we don't know what the CPUs are going to cost in the first place. Heck, we don't even know what they will perform like...
I think you're in the wrong place here, to be honest.
We are discussing the performance potential of an upcoming chip. Ofcourse we're not interested in budget at this point.
You can't really put a price on it until you know how it performs anyway.
We're just interested in whether or not performance-wise it can become competitive.
Because if it does, it can be a completely different situation from today...
If not, then it's the same as today: AMD's CPUs aren't competitive from a performance point-of-view... It's just that because AMD is practically giving them away for free, that they're still interesting to a certain group of computer users.
Obviously that is of no interest at all to people who are performance enthusiasts or interested in the engineering side of things. And quite frankly I'm a bit fed up with the frugal crowd trolling every enthusiast/engineering discussion with "But AMD is the underdog and it's cheaper!"
WE KNOW! WE DON'T CARE! GET A JOB SO YOU CAN AFFORD A HIGH-END SYSTEM AS WELL! OR JUST BE EXCITED ABOUT COOL HARDWARE EVEN THOUGH YOU CANNOT AFFORD IT YOURSELF! DO YOU ALSO GO TO THE FERRARI FORUMS AND TELL EVERYONE THAT A SMART IS THE BETTER CHOICE BECAUSE IT'S CHEAP AND IT'S THE UNDERDOG AND BLAHBLAHBLAH?
I think the large majority of people who love Ferrari's are people who will probably never own one themselves. They just love the engineering, the aesthetics and that sort of thing. They want to know how it works, and why. You see what I mean here? I may not necessarily buy a Nehalem, Sandy Bridge or Bulldozer, but I am interested in how they work and how they perform nonetheless.

 

[Idontcare]

Not in this case, since we don't know what the CPUs are going to cost in the first place. Heck, we don't even know what they will perform like...
I think you're in the wrong place here, to be honest.
We are discussing the performance potential of an upcoming chip. Ofcourse we're not interested in budget at this point.
You can't really put a price on it until you know how it performs anyway.
We're just interested in whether or not performance-wise it can become competitive.
Because if it does, it can be a completely different situation from today...
If not, then it's the same as today: AMD's CPUs aren't competitive from a performance point-of-view... It's just that because AMD is practically giving them away for free, that they're still interesting to a certain group of computer users.
Obviously that is of no interest at all to people who are performance enthusiasts or interested in the engineering side of things. And quite frankly I'm a bit fed up with the frugal crowd trolling every enthusiast/engineering discussion with "But AMD is the underdog and it's cheaper!"
WE KNOW! WE DON'T CARE! GET A JOB SO YOU CAN AFFORD A HIGH-END SYSTEM AS WELL! OR JUST BE EXCITED ABOUT COOL HARDWARE EVEN THOUGH YOU CANNOT AFFORD IT YOURSELF! DO YOU ALSO GO TO THE FERRARI FORUMS AND TELL EVERYONE THAT A SMART IS THE BETTER CHOICE BECAUSE IT'S CHEAP AND IT'S THE UNDERDOG AND BLAHBLAHBLAH?
I think the large majority of people who love Ferrari's are people who will probably never own one themselves. They just love the engineering, the aesthetics and that sort of thing. They want to know how it works, and why. You see what I mean here? I may not necessarily buy a Nehalem, Sandy Bridge or Bulldozer, but I am interested in how they work and how they perform nonetheless.

I'd venture to say, and risk being slapped with a wet-fish for stating the obvious, that the one thing we can say with some certainty about Bulldozer-based SKU's is that they will most likely be priced such that their price/performance is competitive...be that competitive with a Q9650 or a 990X or something even grander still.

Price/performance is kinda the trivial question...we know the answer and the answer is that unless AMD trumps Intel from top-to-bottom then it will be Intel deciding what price at which Bulldozer sells because Intel will set the price/performance tiers for the market and AMD will move to adopt them for BD so they can maximize margins. (obviously there is a bit of wiggle in this, 10% either way, but still...unless they do another GT200 vs RV700 again, expect GF100 vs Evergreen price management)

The things we can't possibly know are absolute performance...we have zero insight into the clockspeed potential for Bulldozer's architecture combined with GloFo's 32nm HKMG...as well as power-consumption for the same reasons.

 

[Scali]

I'd venture to say, and risk being slapped with a wet-fish for stating the obvious, that the one thing we can say with some certainty about Bulldozer-based SKU's is that they will most likely be priced such that their price/performance is competitive...

Yes, but I already said that a few posts above, so I thought we would not need to get into that again.

However, I think that although the absolute performance won't be there, AMD will probably be able to have competitive performance/watt, and they WILL have the better pricing.

 

[bryanW1995]

Not in this case, since we don't know what the CPUs are going to cost in the first place. Heck, we don't even know what they will perform like...
I think you're in the wrong place here, to be honest.
We are discussing the performance potential of an upcoming chip. Ofcourse we're not interested in budget at this point.
You can't really put a price on it until you know how it performs anyway.
We're just interested in whether or not performance-wise it can become competitive.
Because if it does, it can be a completely different situation from today...
If not, then it's the same as today: AMD's CPUs aren't competitive from a performance point-of-view... It's just that because AMD is practically giving them away for free, that they're still interesting to a certain group of computer users.
Obviously that is of no interest at all to people who are performance enthusiasts or interested in the engineering side of things. And quite frankly I'm a bit fed up with the frugal crowd trolling every enthusiast/engineering discussion with "But AMD is the underdog and it's cheaper!"
WE KNOW! WE DON'T CARE! GET A JOB SO YOU CAN AFFORD A HIGH-END SYSTEM AS WELL! OR JUST BE EXCITED ABOUT COOL HARDWARE EVEN THOUGH YOU CANNOT AFFORD IT YOURSELF! DO YOU ALSO GO TO THE FERRARI FORUMS AND TELL EVERYONE THAT A SMART IS THE BETTER CHOICE BECAUSE IT'S CHEAP AND IT'S THE UNDERDOG AND BLAHBLAHBLAH?
I think the large majority of people who love Ferrari's are people who will probably never own one themselves. They just love the engineering, the aesthetics and that sort of thing. They want to know how it works, and why. You see what I mean here? I may not necessarily buy a Nehalem, Sandy Bridge or Bulldozer, but I am interested in how they work and how they perform nonetheless.

pot meet kettle.

 

[Scali]

pot meet kettle.

Excuse me?
In case you didn't know, this thread is about some of the technical background that AMD has published on their Bobcat and Bulldozer architectures.
Hence it is purely a technical/engineering subject.
AMD has not published any prices. If you want to discuss prices, make a new thread.

 

[Idontcare]

Yes, but I already said that a few posts above, so I thought we would not need to get into that again.

Scali, forgive us our idle prattle and past transgressions, we are but humble and meager folk

There are forums where such adjunct discourse would warrant an individual such as myself being taken to task...but please don't hold it against us mere AT forum goers that we tend to wander in our thoughts, topics, and general aim of the discussion.

Not all of us have laser focus on the topic at hand, not all of us come equipped with the educational pedigree necessary to contribute down to the technical detail, but grant us this, pretty much all of here us are here for the sake of enjoying ourselves while being here, in good company and having a good time.

:beer: :rose:

 

[bryanW1995]

Excuse me?
In case you didn't know, this thread is about some of the technical background that AMD has published on their Bobcat and Bulldozer architectures.
Hence it is purely a technical/engineering subject.
AMD has not published any prices. If you want to discuss prices, make a new thread.

that was more of a shot at your incessant trolling in this thread than a comment about the technical background that amd has published on their upcoming architectures. just because you are qualified to discuss a subject doesn't mean that you can do so in a friendly, non-confrontational way. You've been rarely friendly and/or non-confrontational in this thread, so in spite of your obvious technical expertise you're still trolling.

 

[Idontcare]

guys, none of this is helping. pm, pfi, mod discussions, report post...so many alternative options to work this out versus duking it out in an escalating manner. This will go nowhere if we all don't soften our tone and minimize our rhetoric.

Look at this thread, our posts, and ask yourself if you were a new poster, lurking at the moment, perhaps a person who worked in the industry and could contribute some value to the topic...would you bother?

Who wants to cast their pearls before swine? If we behave like this then we are only going to get what we deserve, we'll all wallow here in our own ignorance flaming each other out of arrogance, all the while being ignored by the folks who actually do this stuff for a living (their are bobcat and bulldozer engineers in this forum...we all do know this, yes?).

 

[lopri]

Single thread performance and multi-thread performance (or rather "throughput") are very different as JFAMD analogized to traffic, especially given a specific product. A single-thread workload on a core would not need to contend for bandwidth or L3 with other cores, for example, if other cores are idle.

 

[CTho9305]

K10 barely improved IPC over K8... perhaps just 5% or so.

Um..what? I don't know how to include images, but:

3.2 GHz K8 getting beat by 2.4 GHz Phenom 9750 http://images.anandtech.com/graphs/phenom9850_032708031830/16755.png
3.2 GHz K8 getting demolished by 2.3GHz Phenom, even losing to a 2.2GHz Phenom 9500 http://images.anandtech.com/graphs/phenom9850_032708031830/16757.png
3.2 GHz K8 losing to 2.5 GHz Phenom http://images.anandtech.com/graphs/phenom9850_032708031830/16759.png
3.2 GHz K8 getting demolished by any Phenom http://images.anandtech.com/graphs/phenom9850_032708031830/16764.png
More of the same http://images.anandtech.com/graphs/phenom9850_032708031830/16771.png
http://images.anandtech.com/graphs/phenom9850_032708031830/16767.png

Unfortunately I can't find reviews that compare at the same clock speeds, but if you look through this old Phenom article you'll see that even in the cases where K8's clock speed helps it, it doesn't win by nearly as much as you'd expect from the clock speed scaling alone.

 

[bryanW1995]

ah, yes. back in those days it wasn't accepted that intel was completely and utterly dominant, it was still somewhat close.

of course, intel hasn't exactly abused their dominant position. well, other than $900 for a 970.

 

[Tsavo]

ah, yes. back in those days it wasn't accepted that intel was completely and utterly dominant, it was still somewhat close.

of course, intel hasn't exactly abused their dominant position. well, other than $900 for a 970.

AMD used to charge that much for their top of line CPUs when they could.

 

[bryanW1995]

true, but the 970 isn't "top of the line". it's not even unlocked. how bad are things at amd that their 2nd best cpu is $199 and intel's is $900?? no wonder intel's GM is through the roof these days...

 

[Idontcare]

it is priced at whatever the market is willing to bare. You can't blame Intel for their prices, nor AMD for theirs, the consumer decides where the supply/demand price equilibrium will be. Who buys the 970 @ $900? Apparently enough people do.

 

[bryanW1995]

uh huh, looks like at least 2 people agree with you

ok, try this link instead:

http://www.newegg.com/Product/Produc...&name=Six-Core

apparently one of those reviews was written by a math-challenged individual, but you get the point.


I must admit, I don't understand why they are so busy selling their 32nm dual cores for as little as $89 but won't sell a 970 for less than $899. Maybe they want people to upgrade to skt 1155 next year, then upgrade again for 2011?

 

[Idontcare]

uh huh, looks like at least 2 people agree with you

ok, try this link instead:

http://www.newegg.com/Product/Produc...&name=Six-Core

apparently one of those reviews was written by a math-challenged individual, but you get the point.

 

[Tsavo]

I must admit, I don't understand why they are so busy selling their 32nm dual cores for as little as $89 but won't sell a 970 for less than $899.

That's because there's lots of competition in the $200 and under space and zero competition above it. AMD has nothing on a 970 and a 980, thus the price for each. Why are you confused that Intel doesn't price a reedy little dual core similarly to a monster 6 core?