AMD cuts to the core with 'Bulldozer' Opterons

[kl0012]

Of course you can't accurately predict the outcome based on the diagram,but you can get a idea where intel is going(and it's going after evolution of Core design,again).

I doubt you can get any idea about the performance of SB. I remember similar predictions about the performance of Conroe since it was "based" on Pentium3 core which was not much faster then Athlon (version 1).

Remember it was you who started to throw out the bs percantage numbers

You're taking my words out of the context again. Why? Still you didn't answered my question.

Other poster is implying that intel is reworking current SB core to make it better for BD time...

Pure speculations based on what...?

[informal]

I doubt you can get any idea about the performance of SB. I remember similar predictions about the performance of Conroe since it was "based" on Pentium3 core which was not much faster then Athlon (version 1).

This is not Pentium 3 we are talking about,but a highly optimized Core design.There is always a room for improvement,but all that intel emphasized and talked about was IGP integration in the SB design and new AVX(minus FMA) instr. set support + a couple of tweaks. This can very well bring a notable improvement in performance,but it is to be seen. Westmere for example is considered a tweak ,but per clock and per core is barely faster than Nehalem,bar the AES supported apps,even though it has more L3 and other things intel considers improvements. This all proevs that it is getting ever so harder to get more IPC out of highly optimized high performance cores such as Nehalem.

You're taking my words out of the context again. Why? Still you didn't answered my question.

No i am not. I simply don't want to speculate in percentage figures when we have no samples of either designs. BD is definitely more radical change,whether it will be radical perf. improvement or not is to be seen. SB is not radical in any way except new instr. set support(which will be supported by BD cores as well..).

Pure speculations based on what...?

You have to ask him that,he claims he has some sources.I don't know anything about that.

[Helmore]

+1

Thats what i mean...

Bollocks I say. Just because I like that word though.

They already have SB samples running and nothing so far has indicated any major departures from what we know about it thus far.

On the subject of complete architecture overhauls, isn't Haswell supposed to be Intel's major new microarchitecture?

[informal]

Bollocks I say. Just because I like that word though.

They already have SB samples running and nothing so far has indicated any major departures from what we know about it thus far.

On the subject of complete architecture overhauls, isn't Haswell supposed to be Intel's major new microarchitecture?

Exactly.I highly doubt this SB rework in process thing... This is not a trivial matter at all and if intel wants to add,say FMA support,it will take them A LOT of time and validation to do so.

[FischOderAal]

I doubt you can get any idea about the performance of SB. I remember similar predictions about the performance of Conroe since it was "based" on Pentium3 core which was not much faster then Athlon (version 1).

We had an idea of Conroe thanks to Dothan (Pentium M) and Yonah (Core Duo). Even before the ES were widely available it was expectable that Conroe would be a hit.

[demonkevy666]

informal that looks like it has like more 1 execution over core 2 duo but still the same as core 2 duo ? ?

[Helmore]

informal that looks like it has like more 1 execution over core 2 duo but still the same as core 2 duo ? ?

It is, it's pretty much the same configuration as on Nehalem: http://www.hardwaresecrets.com/article/535/4

[ajaidev]

I doubt you can get any idea about the performance of SB. I remember similar predictions about the performance of Conroe since it was "based" on Pentium3 core which was not much faster then Athlon (version 1).


You're taking my words out of the context again. Why? Still you didn't answered my question.


Pure speculations based on what...?

Bollocks I say. Just because I like that word though.

They already have SB samples running and nothing so far has indicated any major departures from what we know about it thus far.

On the subject of complete architecture overhauls, isn't Haswell supposed to be Intel's major new microarchitecture?

It not FMA its something else, i did not say it was FMA there is a different thing which comprises of slight rework of the core which is not that difficult....

Haswell will be introduced sooner than you all think if SB does not work out and one can quote me on this.

[generics_user]

I would rather expect Magny Cour

Thuban is based on Sao Paulo and Magny Cours is composed of 2 Sao Paulo cores on one package so it is quite possible that they carry an E stepping as Shanghai + Phenom are C2

PS: istanbul is a D0 stepping

CPUID
Fn0000_0001_EAX
(Mnemonic)
Six-Core
AMD Opteron™
Processor
00100F80h (HY-D0) X

[Drwho?]

Let's wait and actually see what these systems do before we bash them ok?
AMD has been doing pretty damned well lately from what I've seen.
All I'm saying is give them a chance to show what they have before you close your mind. They might just surprise you..

I totally agree. You guys (not Movieman) usually under estimate the complexity of those new generation of processors. You always forget how formidable the Penryn/Nehalem/Westmere cores are. While we were doing Prescott and Cedarmill we were working like animals on the cores of Merom, it toke a long long time to get to Merom, then, Penryn was the optimized version of Merom. Later on, with nehalem and westmere,we putted incredibly fast mem subsystem andcutted the latencies. We added a lot of prefetcher, a lot of things that by themself are not so easy to design.

Give AMD a chance and time, this is not an easy task ... try to put 100 transistor together to do a regular NAND circuits, and then, tell t yourself that what you just did is millions of time simpler than what Intel and AMD do.
I need AMD to be around with cool toy to be able to ask for more toys to my Bosses

be patient!
Francois

[JF-AMD]

So you mean AMD is going to scarify single-thread performance in favour for multi-thread performance in coming architectures?

And what is the estimation of transistor count for each module? About 160 mil?

No, I am not saying that. Until the products are out it is impossible for anyone to say what it will be relative to the competition.

My point is that you will have multiple architectures with multiple core counts.

What if you have a V4 and a V6 engine? Do you worry about the "per cylinder" performance or the whole engine performance.

If you are going to head down the path of single core performance, then you have to be fair and also address the per core price and power. Otherwise you are cherry picking your data points and you are not getting a fair comparison.

As for transistor count, we don't release that until launch.

[JF-AMD]

Thuban is based on Sao Paulo and Magny Cours is composed of 2 Sao Paulo cores on one package so it is quite possible that they carry an E stepping as Shanghai + Phenom are C2

PS: istanbul is a D0 stepping

Thuban is a desktop processor, not a server processor. It will have desktop features and some things that are not in server parts.

[AliG]

What if you have a V4 and a V6 engine? Do you worry about the "per cylinder" performance or the whole engine performance.

I think I might just have to find that out for my family's cars

[salad]

DrWHo and JF-AMD in the same thread, posting consecutively?

Ooooh, feel the heat rising. Or maybe they're sitting in Starbucks together commiserating over the iPad not using either company's CPUs

[railer]

We had an idea of Conroe thanks to Dothan (Pentium M) and Yonah (Core Duo). Even before the ES were widely available it was expectable that Conroe would be a hit.

QFT
I still have one of these Little dothans. Use it as a server now. But in the days it was a fun little sys to play with. If i remember correctly my dothan at 2.9 was like fx-51 at 3.4 in Aquamark. Those were the days when Athlon was king in SPI and then Dothan edged it out. In fact someone on this forum broke the SPI WR using Dothan.
It was not official yet as you had to run this cpu with an adapter for MB but AMD back then already got the hint of whats coming with core2.

[JF-AMD]

DrWHo and JF-AMD in the same thread, posting consecutively?

Ooooh, feel the heat rising. Or maybe they're sitting in Starbucks together commiserating over the iPad not using either company's CPUs

No, actually I just got back from a trip to europe. Monday night I was sitting in a pub in london with my counterpart from intel having a pint after a joint customer event thrown by one of our mutual partners.

There is nothing wrong with being sociable, they aren't bad people, we just have different philosophies. There is always plenty to talk about without getting into business.

[ajaidev]

No, actually I just got back from a trip to europe. Monday night I was sitting in a pub in london with my counterpart from intel having a pint after a joint customer event thrown by one of our mutual partners.

There is nothing wrong with being sociable, they aren't bad people, we just have different philosophies. There is always plenty to talk about without getting into business.

Yep CMT philosophie to SMT philosophie

[LightSpeed]

There is nothing wrong with being sociable, they aren't bad people, we just have different philosophies. There is always plenty to talk about without getting into business.

Nicely said. Its how things should be, but sadly...

[Helmore]

Nicely said. Its how things should be, but sadly...

In my (very limited) experience, those working at both AMD or Intel (or NVIDIA) are usually more civil towards each other than some fanboys on the internet.

[Chumbucket843]

Give AMD a chance and time, this is not an easy task ... try to put 100 transistor together to do a regular NAND circuits, and then, tell t yourself that what you just did is millions of time simpler than what Intel and AMD do.
I need AMD to be around with cool toy to be able to ask for more toys to my Bosses

be patient!
Francois

its not hard for one person to design some alu's with NAND and NOR gates. after that it gets hard.i have designed my own cpu, it is very primitive and when i see intel and amd chips i can only drop my jaw at the level of complexity and speed. with that said how many chips has AMD made by just themselves? K5,K8, and K11. all the rest were from buying other companies.

[qcmadness]

its not hard for one person to design some alu's with NAND and NOR gates. after that it gets hard.i have designed my own cpu, it is very primitive and when i see intel and amd chips i can only drop my jaw at the level of complexity and speed. with that said how many chips has AMD made by just themselves? K5,K8, and K11. all the rest were from buying other companies.

cpus after k7 are all made from themselves

[Brother Esau]

Let's wait and actually see what these systems do before we bash them ok?
AMD has been doing pretty damned well lately from what I've seen.
All I'm saying is give them a chance to show what they have before you close your mind. They might just surprise you..

its great to see that some things never change always a pleasure to see happy positive people Dave......as usual

[madcho]

its not hard for one person to design some alu's with NAND and NOR gates. after that it gets hard.i have designed my own cpu, it is very primitive and when i see intel and amd chips i can only drop my jaw at the level of complexity and speed. with that said how many chips has AMD made by just themselves? K5,K8, and K11. all the rest were from buying other companies.

Nice troll

[Drwho?]

DrWHo and JF-AMD in the same thread, posting consecutively?

Or maybe they're sitting in Starbucks together commiserating over the iPad not using either company's CPUs

http://www.youtube.com/watch?v=lsjU0K8QPhs
or
http://www.youtube.com/watch?v=_n7ZyS_g0DQ



PS: this reflect my person humour, my employer is not responsible.

[Aerwidh]

^ nice videos!