September 22nd ETA for AMD FX processors

[aigomorla]

But calling BD a P4 is completely ignorant.

no they are comparing it because its basically what AMD did.

They increased clock speed while losing the translation on ghz. P3 vs P4

Instead of scaling down, they scaled up. (again P3 vs P4)

And they took a concept like hyperthreading... something intel had to fail once to learn, and adopted, while stamping the label, its better then the non hyperthreading PH2 cpu's.

In reality... hyperthreading owns in things like encoding.
But who do you honestly expect to have it done better?
Intel who once again failed horribly at it.. had to relearn it all over again though C2D and C2Q, got it partially correct in Bloomfield (hyperthreading actually gave you worse scores in some apps), and perfected on Sandy Bridge (i dont recall if i have that problem on SB like on Bloomfield which is why i think intel gave u an obnoxious turbo on multi vs bloomfield's).

Or AMD? whose taken the Ghz scale back where netburst was at?

 

[kcidmai]

no they are comparing it because its basically what AMD did.

They increased clock speed while losing the translation on ghz. P3 vs P4

Instead of scaling down, they scaled up. (again P3 vs P4)

And they took a concept like hyperthreading... something intel had to fail once to learn, and adopted, while stamping the label, its better then the non hyperthreading PH2 cpu's.

In reality... hyperthreading owns in things like encoding.
But who do you honestly expect to have it done better?
Intel who once again failed horribly at it.. had to relearn it all over again though C2D and C2Q, got it partially correct in Bloomfield (hyperthreading actually gave you worse scores in some apps), and perfected on Sandy Bridge (i dont recall if i have that problem on SB like on Bloomfield which is why i think intel gave u an obnoxious turbo on multi vs bloomfield's).

Or AMD? whose taken the Ghz scale back where netburst was at?

This assertion your making isn't provable. come to us with real evidence, you know, actual production hardware and what not. You are repeating the same untrue things as everyone else. The reasons for the failure of P4 and frequency scaling have nothing to do with the design of bulldozer. They also had plenty of time to analyze those potential issues. the failure of your competitor is your opportunity to learn.

The reason for the delays etc, is likely due to 32nm capacity.

 

[aigomorla]

This assertion your making isn't provable. come to us with real evidence, you know, actual production hardware and what not. You are repeating the same untrue things as everyone else. The reasons for the failure of P4 and frequency scaling have nothing to do with the design of bulldozer. They also had plenty of time to analyze those potential issues. the failure of your competitor is your opportunity to learn.

The reason for the delays etc, is likely due to 32nm capacity.

real evidence?

I am placing analogies.

You really think a simple bios fix is going to magically improve all benchmarks by 20%?


Lets drop the fanism here, i am not supporting anyone.
I have bashed intel back and forth.. i have bashed nvidia back and forth.

But i keep coming back to them... do you know why?
Because they have something called crown.


Lastly i have told people who asked me.
If bulldozer was that great... when they broke the WR, why didnt they break records across all benchmarks like the 980X did when it was out?

Why? because it has problems beating a stock 2600K on air overclocked, you think it can take down a 990X (which ARE BETTER then most 980X unless u have a cherry 980X) under LN2?

 

[BlueBlazer]

Again, The cores came from the hardware and software engineers -> then the marketing gets it then they advertise as such


Get with the times....

Integer Clusters have always been considered the cores from a hardware and software standpoint

So is HyperThreaded "logical core" (hardware thread). That one also looks like a "core" from software point of view. Hardware engineers designs the architecture and the chips, not software engineers.

I'm going point this one out....

There is one integer clusters in Llano and and two integer clusters in Bulldozer
Llano is a single core and Bulldozer is a dual core

They are not real cores. Even the original patents shows that. The real core is the "module" (or "compute unit"). See below..

Yes I have, Integer has always been the core, in fact the old days 386 and the like cpu's had external co-processor fpu units.

anything else you care to drivel on about now ?

Guess you didn't see this part?

My original motivation for MCMT was to work around some of the
limitations of Hyperthreading on Willamette. E.g. Willamette had a very
small L0 data cache, 4K in some of the internal proposals, although it
shipped at 8K. Two threads sharing such a tiny L0 data cache thrash.
Indeed, this is one of the reasons why hyperthreading is disabled on
many systems, including many current Nhm based machines with much larger
closest-in caches.

At the time, the small L0s were a given. You couldn't build a
Willamette style "fireball" high frequency machine, and have a much
bigger cache, and still preserve the same small cache latency.

To avoid threads thrashing each other, I wanted to give each thread
their own L0. But, you can't do so, and still keep sharing the
execution units and scheduler - you can't just build a 2X larger array,
or put two arrays side by side, and expect to have the same latency.
Wires. Therefore, I had to replicate the execution units, and enough of
the scheduler so that the "critical loop" of Scheduler->Execution->Data
Cache was all isolated from the other thread/cluster. Hence, the form
of multi-cluster multi-threading you see in Bulldozer.

 

[NostaSeronx]

You really think a simple bios fix is going to magically improve all benchmarks by 20%?

It might not be 20% but there are certain toggles in the BIOS that can improve performance :whiste:

Then if an application is patched/recompiled/fixed(LWP will help a lot here)
http://www.anandtech.com/show/2302/1

You might see a 50% boost or more and even mitigate the whole sharing problem if there was such a problem to begin with

So is HyperThreaded "logical core" (hardware thread). That one also looks like a "core" from software point of view. Hardware engineers designs the architecture and the chips, not software engineers.

If you compared Intel's GPRs you would notice there is only 3 GPR ports...There is only 1 integer cluster in intel that has SMT enabled on it...if 2 GPRs ports are being used by thread 0 and thread 1 also needs 2 GPR ports, thread 1 stalls and on AMD Bulldozer in a compute unit there are two integer clusters each with 2 ALUs(GPR)

One makes sure the GPRs are used and can stall
The other has two integer clusters each with dedicated gprs and won't stall

They are not real cores. Even the original patents shows that. The real core is the "module" (or "compute unit"). See below..

Microarchitecture design doesn't equal core design

The "real" core is what processes the thread and schedules the additives(FPU and GPU)

And there is two such physical cores on the Bulldozer die in each compute unit
while there is only one physical core on the Sandy Bridge die per core that is virtually duplicated

 

[Gundark]

Sandy Bridge and Bulldozer process 8 threads at the same time beyond that you can hit problems

For example Sandy Bridge is very ALU dependent if thread A is going to use ALU B and thread B needs to also use ALU B then you have a conflict of resources(ALU B is Floating Point Multiply and Integer Multiply if I remember correctly(You can't have a Integer Multiply and Floating Point Multiply simultaneously execute))

As i understand, in HT one thread must stall for other to execute. So it's like one hand trying to do two jobs. In BD it shouldn't be the case. It should be two threads in parallel ( or 2 hands doing 2 jobs ). In this light, BD is much closer to 2 full cores than HT, but i have no problems accepting the other side.

 

[kcidmai]

real evidence?

I am placing analogies.

You really think a simple bios fix is going to magically improve all benchmarks by 20%?


Lets drop the fanism here, i am not supporting anyone.
I have bashed intel back and forth.. i have bashed nvidia back and forth.

But i keep coming back to them... do you know why?
Because they have something called crown.


Lastly i have told people who asked me.
If bulldozer was that great... when they broke the WR, why didnt they break records across all benchmarks like the 980X did when it was out?

Why? because it has problems beating a stock 2600K on air overclocked, you think it can take down a 990X (which ARE BETTER then most 980X unless u have a cherry 980X) under LN2?

Have you seen benchmarks of production hardware ? The answer to that question is pretty easy to come by. that answer is a definitive NO.

You have not seen such a benchmark.

so the best thing to do is to shut your mouth until such benchmarks arrive.

 

[aigomorla]

It might not be 20% but there are certain toggles in the BIOS that can improve performance :whiste:

Then if an application is patched/recompiled/fixed(LWP will help a lot here)
http://www.anandtech.com/show/2302/1

You might see a 50% boost or more and even mitigate the whole sharing while achieving the same performance

50% boost? :O

if we get this magical 50% boost i would even move over.

However lets be more realistic.

if they get a 5% boost they will pop champaign and toast.

20% boost AMD engineers would get a FAT bonus because they would have an intel stomping cpu which is affordable.

But once again... the last option has yet to be completed in half a decade and you expect them to do this when money is scarce and without their own FABs even?

 

[Arkaign]

Have you seen benchmarks of production hardware ? The answer to that question is pretty easy to come by. that answer is a definitive NO.

You have not seen such a benchmark.

so the best thing to do is to shut your mouth until such benchmarks arrive.

Hahah noob. Way to go, come in with 12 posts into a forum of largely serious types in the CPU forum, and you tell a legendary member who also happens to be a mod to 'shut your mouth'. All I can say is that you're about as dumb as a box of rocks. All of this 'have you seen final xxx' is just a distraction from the fact that what we DO know for certain about BD doesn't look good at this point. And this discussion forum exists for the purposes of sharing ideas and information. Telling people to shut their mouth because you don't agree with them is derptastic.

 

[aigomorla]

Have you seen benchmarks of production hardware ? The answer to that question is pretty easy to come by. that answer is a definitive NO.

You have not seen such a benchmark.

so the best thing to do is to shut your mouth until such benchmarks arrive.

LuLz...

I have movieman on my cell phone memory... and he has my cell phone number on memory as well.

Ive met chew a few times in las vegas at the XS parties.
Ive even filled his LN2 thermos while he was benching in the side so he didnt have to lose focus on his overclocking.

Ive been sponsored by intel partners to help them do thermal tests on cpu's in extreme conditions and voltages to see what breakpoints they would have under those conditions.

But to answer your questions...

Yes ive seen all the benchmarks, and was forced by my friends to stay shut until they were public.

So when i tell people, i am waiting for the X6 to drop in price... and migrating my friend on a X6, people tend to listen to me more then they would to you.

 

[Arzachel]

Hahah noob. Way to go, come in with 12 posts into a forum of largely serious types in the CPU forum, and you tell a legendary member who also happens to be a mod to 'shut your mouth'. All I can say is that you're about as dumb as a box of rocks. All of this 'have you seen final xxx' is just a distraction from the fact that what we DO know for certain about BD doesn't look good at this point. And this discussion forum exists for the purposes of sharing ideas and information. Telling people to shut their mouth because you don't agree with them is derptastic.

Doesn't look good at this point? What have we actually seen that would indicate such except delays that have just as much to do with GF as with AMD?

Unless you believe benches, where BD has lower IPC than Bobcat, not to mention Stars. If that's the case, I have some snake oil to sell you.

 

[Zstream]

LuLz...

I have movieman on my cell phone memory... and he has my cell phone number on memory as well.

Ive met chew a few times in las vegas at the XS parties.
Ive even filled his LN2 thermos while he was benching in the side so he didnt have to lose focus on his overclocking.

Ive been sponsored by intel partners to help them do thermal tests on cpu's in extreme conditions and voltages to see what breakpoints they would have under those conditions.

But to answer your questions...

Yes ive seen all the benchmarks, and was forced by my friends to stay shut until they were public.

So when i tell people, i am waiting for the X6 to drop in price... and migrating my friend on a X6, people tend to listen to me more then they would to you.

Wait... you seen Intel folks benchmark bulldozer and they received retail chips? Somehow I doubt it was a non es chip.

 

[krumme]

Guess you didn't see this part?

Well that looks like the invention of an extra core

OK, I dont care about the definition, what i have to wonder is, when the Integer unit is 3.8mm2 in BD in a total of 31mm2, what is really the incentive to implement HT in the future for AMD?

I mean for Bobcat i can see it can have a purpose (1-2 core designs) but for BD targeting the server environment, where is the potential?

 

[aigomorla]

Wait... you seen Intel folks benchmark bulldozer and they received retail chips? Somehow I doubt it was a non es chip.

how is Movieman an intel person?

He used to be an intel person, but he's over at the AMD camp as far as i can recall as of reciently.

And Chew always been an AMD person.
He even brings down AMD rigs to setup and display and OC @ the XS parties.
Really cool guy... :biggrin:

And yes i heard about the invite to AMD @ TX b4 u guys even heard about it.
Dave even told me he saw an article writer from AT there, so u can expect AT will have an article on it too.
But once again.... they only broke 1 record... a ghz record.

That is all, why? Because from what i heard from the guys there, and from more then 1 source... all that GHZ translated to erm... disappointment on a real speed scale.

 

[krumme]

Well then how can you know the x6 will fall in price, when all the GHz translated to disappointment?

And how can we know this when you will stay shut until the benchmark are public?

 

[BlueBlazer]

Well that looks like the invention of an extra core

OK, I dont care about the definition, what i have to wonder is, when the Integer unit is 3.8mm2 in BD in a total of 31mm2, what is really the incentive to implement HT in the future for AMD?

I mean for Bobcat i can see it can have a purpose (1-2 core designs) but for BD targeting the server environment, where is the potential?

If they had put more Thuban cores into the CPU, then the power requirements and TDP will increase dramatically (6-core Thuban is already 95W to 125W, imagine 8-core or 10-core Thuban on a single package). AMD wants to reduce core count while increasing thread count. This can reduce power as well as reducing die space. One way to do this is to share some resources (like FPU and cache). Thus Bulldozer is designed as a multi-threading CPU, much like Sun's Niagara design (which also uses CMT). Of course not every design (whether its HT or CMT) is perfect and has its caveats/trade-offs (depends on how well its designed and implemented).

Looks like 19th September has been uneventful (this rumor is busted).

 

[kcidmai]

Hahah noob. Way to go, come in with 12 posts into a forum of largely serious types in the CPU forum, and you tell a legendary member who also happens to be a mod to 'shut your mouth'. All I can say is that you're about as dumb as a box of rocks. All of this 'have you seen final xxx' is just a distraction from the fact that what we DO know for certain about BD doesn't look good at this point. And this discussion forum exists for the purposes of sharing ideas and information. Telling people to shut their mouth because you don't agree with them is derptastic.

did you see me make any assertions on what the IPC etc of bulldozer would be ? No, I don't think so.


Oh gezz, a moderator on a internet website. Wheew, gonna have to call the red phone.

 

[kcidmai]

LuLz...

I have movieman on my cell phone memory... and he has my cell phone number on memory as well.

Ive met chew a few times in las vegas at the XS parties.
Ive even filled his LN2 thermos while he was benching in the side so he didnt have to lose focus on his overclocking.

Ive been sponsored by intel partners to help them do thermal tests on cpu's in extreme conditions and voltages to see what breakpoints they would have under those conditions.

But to answer your questions...

Yes ive seen all the benchmarks, and was forced by my friends to stay shut until they were public.

So when i tell people, i am waiting for the X6 to drop in price... and migrating my friend on a X6, people tend to listen to me more then they would to you.

Let me put you in your place, I drove preproduction cars, cars that are still on the drawing board 2 years ago. A I ain;t going to tell you about it, B I won;t tell my friends about it.

big whoopdee do, you have someones cells phone number. I bet chew thinks of you like the way I think about alot of people.

Shut up and go away.

 

[blackened23]

I hope to god another one of these threads about BD don't pop up again. There's a new one every day. Give it a rest already, sick of hearing about it....

So BD is slower than the 386. Lets wait for official benchmarks, NDA lifts on 22 September doesn't it ? In the meantime give it a rest.

 

[Arkaign]

Let me put you in your place, I drove preproduction cars, cars that are still on the drawing board 2 years ago. A I ain;t going to tell you about it, B I won;t tell my friends about it.

big whoopdee do, you have someones cells phone number. I bet chew thinks of you like the way I think about alot of people.

Shut up and go away.

You and this small band of mentally deficient people that popped out of the woodwork to come here in start fights with mods and long-standing members instead of having a logical discussion on the issues at hand need to stfu and gtfo. If you want to discuss things like a grown up, by all means stay, but this kind of overtly hostile posting style won't fly here for very long.

Fwiw, 'Morla has been an incredibly informative and valued member here for many years now, who has earned a lot of respect, deservedly so. He has proven over the years to indeed have eerily prescient information and a balanced perspective on the issues he discusses. You're a nobody, so telling him to shut up and go away is just laughable.

 

[cantholdanymore]

I hope to god another one of these threads about BD don't pop up again. There's a new one every day. Give it a rest already, sick of hearing about it....

So BD is slower than the 386. Lets wait for official benchmarks, NDA lifts on 22 September doesn't it ? In the meantime give it a rest.

AMD is guilty to make us go through this ordeal. Shame on them for forcing good intentioned consumers to endure month of gibberish, name calling and nonsense

 

[BlueBlazer]

AMD is guilty to make us go through this ordeal. Shame on them for forcing good intentioned consumers to endure month of gibberish, name calling and nonsense

Its not only here, other top forums as well (mudslinging and some signs of "revolution" popping up), like the ones at OCN. :\

 

[RussianSensation]

AMD is guilty to make us go through this ordeal. Shame on them for forcing good intentioned consumers to endure month of gibberish, name calling and nonsense

AMD's constant delays of BD should not be used as an excuse to promote disrespect among forum members, or hardware enthusiasts in general. Often times, it is the person who gets attacked based on the opinion, but instead a proper discussion should revolve around the opinions / premises.

 

[podspi]

Looks like 19th September has been uneventful (this rumor is busted).

Guess we should all start gearing up for the Sept 22nd launch now


Just a note: TDP of the current X6's on 45nm != the TDP of a hypothetical X6 on 32nm.

I still believe BD has got to perform higher than the X6 (and a hypothetical X8) on 32nm. Otherwise, why bother? They've already ported Stars to 32nm. Frankly, I'm kind of surprised (I'm guessing capacity constraints here) they aren't releasing 32nm Phenoms. Some of us stuck with 800-series boards (that were compatible with BD last April, not so much anymore) wouldn't mind a cheap X8...

 

[Zstream]

Its not only here, other top forums as well (mudslinging and some signs of "revolution" popping up), like the ones at OCN. :\

Because you're so well intentioned and all. Only posting the past few days and all anti-bulldozer and on an alt account to I bet.