September 22nd ETA for AMD FX processors

[krumme]

If there's problems with getting high integer performance, then why would they ship for revenue Interlagos already? I understand floating point performance matters more in server than desktop, but integer performance is still a huge factor for servers as well.

I dont know why they are releasing it because as i understand it, the major part by far, of the servermarket is dependant on integer performance. The weak selling of MC is a proof of that (you do get a lot of brute FP perf for your investment here if you keep them running - just look Johans reviews here at anands). And thats why we have the module approach sharing frontend and FP. I remember when Paul Demone first say the diagram, and he right away said it was a bandwith, integer moving machine, made for high clocks. So beats me why they release it?? - one can only think it must have some advantages after all

 

[BlueBlazer]

As Charlie at www.semiaccurate.com say, it will not be a barcelona nor a 4870 launch, http://semiaccurate.com/2011/09/07/bulldozer-finally-shipped-last-week/
There is a bug in the integer part, and the clock is not where they should be. The next stepping should be the one looking for. I think his remarks fit very well with the lack of rumours/BM. The performance is not there yet.
We have to remember that this part is for servers, its supposed to be moving those integers big time, and with a high clock. And when there is problems here, well, its not good.
I dont think this cpu was intended for the very small high-end desktop market. If video encoding is not your primary consern, just go and buy the sb. What BD will do is just force the price down a little.

Integer issue? More likely Bulldozer cache issues that I had highlighted earlier in another thread. Here is the clue >> Looks like Linus isn't happy with a fix AMD proposed for a cache aliasing problem that hurts Linux performance on Bulldozer. It also happens that Cray runs a customized Linux operating system. Remember Barcelona? AMD shipped them to Cray knowingly there was a TLB bug (I'm sure AMD have provided a fix). Well, that's my hypothesis... :hmm:

It is true that you, me, and most of the other forums around really hate AMD not letting more details out, but, it wouldn't be unheard of for them to have 0 heads up about the release date. Interlagos was announced on September 7, and nobody had a heads up as far as I can tell. And that date was a Wednesday.

Since the folks @overclock.net will have a BD streaming event at or pretty close to the launch (and they said it will be on A Saturday) it is sounding like a Monday release.

Speaking of OCN, this one is epic :thumbsup: .....

AMD is like a plane scheduled for takeoff at 3pm, then at 2:59 it announces: takeoff in afternoon. Then at 5:59, while still in the seat, it accounces: takeoff in evening.

"excuse me captain, what's causing the delay"

"can't tell you that, takeoff at takeoff"

"can you atleast give a more accurate expected delay"

"that will be announced at takeoff. Again, takeoff at takeoff"

"but I'll be late for my meeting"

"we have a business to run, you're less than 5% of our customer base"

"but SB Airlines always tells us about expected delays"

"well, we NEVER announce delays before takeoff"

That was the response to JF-AMD's continuous analogy about the plane....

 

[statikregimen]

This could be oopening a can of flesh eating killer worms here, but lets consider this-

If AMD was selling BD as a quad core with 'an alternative to weak hyperthreading' and the 3rd party marketing bumpf was discussing it as such, then how would this effect the discussion here?

This, in my opinion, is how it should be marketed. Then they should do a 6- and/or 8-core(module) design next year when, they perfect 22nm process.

EDIT: Probably the reason they are not doing it this way, is 8-cores is all the rave now, thanks to SB...But if they were to have ignored this and called it a 4-core, 8-thread CPU, and all the benchmarks compared it to other quadcore CPUs, then there would be no question that BD would absolutely wreck the competition and greatly outperform the Phenom II quads. Personally, I will be making my decision to purchase by comparing its performance to quadcores, since this seems most logical, based on its design (and pricing). It just needs to "generally" outperform my 1100t by a small factor, but otherwise, I'm very excited about the new architecture and am very hopeful that as software/compilers catch up, we'll really see it shine!

 

[Elixer]

Speaking of OCN, this one is epic :thumbsup: .....That was the response to JF-AMD's continuous analogy about the plane....

LOL.
Nice!

 

[BlueBlazer]

EDIT: Probably the reason they are not doing it this way, is 8-cores is all the rave now, thanks to SB...But if they were to have ignored this and called it a 4-core, 8-thread CPU, and all the benchmarks compared it to other quadcore CPUs, then there would be no question that BD would absolutely wreck the competition and greatly outperform the Phenom II quads.

Bulldozer "cores" are not true cores. This is what Andy Glew, creator of CMT, had to say over here.....

Apparently AMD is calling my clusters their cores, and my core their cluster. It has been suggested that this change of terminology is motivated by marketing, so that they can say they have twice as many cores.

And Chew* (overclocker at AMD FX overclocking event) had to say over here......

BD is physically a 4 core 8 thread part.

As the saying goes, "no such thing as free lunch".

Personally, I will be making my decision to purchase by comparing its performance to quadcores, since this seems most logical, based on its design (and pricing). It just needs to "generally" outperform my 1100t by a small factor, but otherwise, I'm very excited about the new architecture and am very hopeful that as software/compilers catch up, we'll really see it shine!

We shall see (hopefully soon)!

LOL.
Nice!

Yups, sums up what we have been put through..

 

[classy]

Bulldozer "cores" are not true cores. This is what Andy Glew, creator of CMT, had to say over here.....And Chew* (overclocker at AMD FX overclocking event) had to say over here......As the saying goes, "no such thing as free lunch".

We shall see!

Yups, sums up what we have been put through..

Here's what I'll say. I don't know you or half the people who have comments on this. But I can say this, I do know MM from Xtreme. If he says he seen it across many benches and it performs as well as a 980x, I'll take his word. Because truth be told he could care less, no axe to grind whatsoever. And Kyle from Hard, says he thinks with OC included people will be happy with it, tells me despite all the speculation, those two making positive statements is enough to nullify much of the speculation. Could they be wrong? Anything is possible, but highly doubt it.

This is still a blackeye for AMD, maybe its worst handling of bringing a product to market ever. But still I have very little doubt that this will put them back in the game so to speak. I can't see how much more can be said. I think just using some patience for the product to be released is best.

 

[guskline]

From the latest post from a competitor forum (Tom's Hardware) seems the release is at least 4 to 5 weeks from now. If so, is it really shocking? Not for me as it has seemed all along that AMD was struggling to meet production.

 

[Nemesis 1]

If I see this one more time I think I'll scream. Single thread performance is quickly becoming a thing of the past. A person still talking about single thread performance today should still be using a modem. Its a dying horse that is getting closer to death.

NO one but no one is buying your BS on this . Look were not talking about a single core . What part don't you get . IF core a is faster than core b by fifty % , Than a 6core cpu in B model it takes to compete with a 4 core model A.

To say single thread performance is over is just ignorance . Because 4 cores on 1 die will also be faster than a slower core with less IPC than the faster. AS it stands right now . a 4 core 2500K is 50% faster than PHII so it takes 6 AMD cores to do the work of 4 intel sb cores . That the importantance of single threaded performance . Everyone can see this but you . WHY do you think that is? I would rather have 4 cores doing the same amount of work as a 6 core model . AS would most reasonable people .

 

[BlueBlazer]

Here's what I'll say. I don't know you or half the people who have comments on this. But I can say this, I do know MM from Xtreme. If he says he seen it across many benches and it performs as well as a 980x, I'll take his word. Because truth be told he could care less, no axe to grind whatsoever.

I know a lot of "hopefuls" are hinging on Movieman's single vague comment. But did you check one of his later statements especially this part.........

I was there in Austin and saw with my own eyes a lot of what these can do BUT
that is different from setting up your own machine and then KNOWING what it will do.
Same with Intel, they all want to show you their products in the best light possible and that they should be doing.
After all, this is business right guys?

Open to interpretation....

And Kyle from Hard, says he thinks with OC included people will be happy with it, tells me despite all the speculation, those two making positive statements is enough to nullify much of the speculation. Could they be wrong? Anything is possible, but highly doubt it.

IMHO Kyle is spot on. Let's see what Chew* said.....

There are many bios options that can effect the outcome of benches.

HPET is 1 for example, it stops the cpu from throttling back in mulithreaded apps.

Running pi on a cluster, versus a core ( 2 threads ) versus being able to disable a single cluster in a core ( which 99% boards/bios's do not have implemented so resources are not shared ) can all influence the results in single threaded.

Knowing all this tells you one thing for sure, you can make it look worse or make it look better all depending on your knowledge of the chip and or your intentions.

As far as PI it's an antiquated bench and has not been AMD's strong point for quite some time.

Granted some results shown tend to lead to the fact that 1m times are bad but looking at the bigger picture we also know that in many cases you can validate 1000mhz higher in many cases with BD, which would point to the fact that you can run 1m at alot faster speeds than current AMD tech.

Things that make you go hmm like what kind of times will we see at 8 gig or even comparing BD to deneb/thuban when same cooling is used.

Again open to interpretation....

This is still a blackeye for AMD, maybe its worst handling of bringing a product to market ever. But still I have very little doubt that this will put them back in the game so to speak. I can't see how much more can be said. I think just using some patience for the product to be released is best.

Right now Bulldozer is a hot topic, all for the wrong reasons unfortunately. Like this quote from OCN.....

All year we've been told different deadlines and just when you think it's about to be released we are hit with ANOTHER 90 day delay, AFTER the 60-90 day deadline had passed. Delays are the #1 topic surrounding Bulldozer how can you expect for it to not be discussed?

 

[BlueBlazer]

From the latest post from a competitor forum (Tom's Hardware) seems the release is at least 4 to 5 weeks from now. If so, is it really shocking? Not for me as it has seemed all along that AMD was struggling to meet production.

IMHO AMD is scrambling to fix performance issues (with another stepping/revision), which is why AMD for the past few months had delays and hasn't been able to announce any shipping or launch dates. That's my hypothesis anyway.... :hmm:

 

[aigomorla]

its exactly like what my friends have all been saying.

Your sitting behind the wheel.
You rev the engine and go WOW. <--- cuz they overclock like MAD!!!
Infact u even see the nice bling bling 12k engine RPM redline and even more go WOW. <-- AMD got the highest GHZ World Record u got to give them that....

But your car is making all that noise.. and you got all that great engine.. yet a granny in an intel car is passing you like she doesnt even know your there.
^ cuz all that speed translates to a tad bit more then a stock 2600K!

 

[guskline]

And the beat goes on ... Sonny & Cher

 

[aigomorla]

And the beat goes on ... Sonny & Cher

i was actually thinking more the rythem of the night~

http://www.youtube.com/watch?v=cAQSZhazYk8

seeing how were going back into a type of P4 netburst style arch...

 

[BlueBlazer]

But your car is making all that noise.. and you got all that great engine.. yet a granny in an intel car is passing you like she doesnt even know your there.
^ cuz all that speed translates to a tad bit more then a stock 2600K!

That noisemaker isn't a car, its a big Bulldozer!

Just joking!

 

[Arg Clin]

<opens fortune cookie>
You are going to be very disappointed.

I hope not. But really, a fortune cookie is as qualified a guess at BD performance as any I've seen so far :biggrin:

 

[classy]

NO one but no one is buying your BS on this . Look were not talking about a single core . What part don't you get . IF core a is faster than core b by fifty &#37; , Than a 6core cpu in B model it takes to compete with a 4 core model A.

To say single thread performance is over is just ignorance . Because 4 cores on 1 die will also be faster than a slower core with less IPC than the faster. AS it stands right now . a 4 core 2500K is 50% faster than PHII so it takes 6 AMD cores to do the work of 4 intel sb cores . That the importantance of single threaded performance . Everyone can see this but you . WHY do you think that is? I would rather have 4 cores doing the same amount of work as a 6 core model . AS would most reasonable people .

I never said it was over. Let me put it this way, so babbling babies can finally get the burp out. When all this got started, which is fun by the way, we have heard a lot speculation. We have heard all these different things, some true, some not. When it appeared that AMD really had miserable performance even quite worse than Thuban the specualtion was weak IPC, poor design, yadda yadda yadda. Then when it came to light that maybe this cpu really is going to be good, then the Intel team members started with well its still slower in single thread.

I stated that the argument of single thread is a dead one and it is. I stated that if having a cpu core that has a slightly lower single thread performance but allows more better scaling with more threads, I think that is better. What part of that do you not understand? A top notch single thread core doesn't always mean it can be built to work perfectly in a multi-core cpu. The arguments are just plain dumb. You got people who have heard a longer pipeline and have literally, lol, brought a P4. P4's biggest problems was not its long pipeline.

As far as cpu performance is concerned I have said this before, we buy FINAL OUTPUT, period. At the end of the day we don't really care about the type of detergent used as long as the clothes are clean. At the end of the day 99% will buy based on PRICE, except maybe in the server market. If a $200 bulldozer puts up the numbers using other aspects to increase FINAL PERFORMANCE equal to or better than a $250 Intel cpu, every swinging pipe is going to buy the AMD. Comparing architectures have really been truly worthless apples to apples since socket 7 to be honest.

That is really it in a nutshell. I have owned mostly Intel CPUs, just finally broke down my Q6600 to sellout, so I am not partial one way or the other. At the end of the day if an AMD cpu is on par and even cheaper than an Intel cpu, which one would you buy?

 

[Arkaign]

Classy, with all due respect the problem we may face with a fairly low IPC (assumed given the pricing structure) is that we won't have any cut and dry winners per se.

Reason :

Mediocre IPC + lots of pseudo-cores (I don't consider a 2600k an 8-core either, not even close) = great at apps that can use all 8 cores, not so great in others, should be good at heavy multitasking so long as the cache and memory performance is adequate.

vs.

Maximum IPC + quad core (forget HT) = excellent at almost everything, and can even get a boost on heavy multithreading because processes will finish more quickly if performance per core is higher.

So what we're left with is this :

Games may run XX&#37; faster on 2500/2600k.
App A may run XX% faster on 2500/2600k.
App B may run XX% faster on BD
App A+B+C may run X% faster on BD
APP C may run X% faster on 2500/2600k.

See where this is going?

Add to that the fact that we may have 4ghz+ BD parts competing with stock speed 2500/2600k parts in many things, yet the Intel K chips can hit mid-4ghz OCs with no problem. How much headroom on air will we get from BD? If not an equal % from stock, then the BD will fall behind Intel after both are OC.

All of this is a mess. All I'm hoping for, reading between the lines, is that the BD midrange line will hit stock 2500k levels after OC, and be notably more affordable, thereby negating the reason to buy the stupid locked Intel chips like i5-2400 for enthusiasts. Most people don't have a GPU setup stout enough to really need a 4ghz SB anyway, so it'd be a great option. Save $100 on CPU/Mobo, put $100 saved towards SSD, better GPU, beer, whatever. This would be acceptable. Anything less will be disastrous.

The delays tell a dark tale though. AMD's ASPs have been in the gutter for a long time now. If BD was decent and ready to go 6 months ago, they could have made a TON of money that they lost out on during that timeframe. That is just huge. If it was ready and good, they would have released it. The delay tells us that there were significant problems to overcome, and the problems were so large that it was more important to re-spin or chase bugs than to just release it to staunch the wound.

Here's hoping they're about to release something at least halfway decent.

 

[aigomorla]

At the end of the day if an AMD cpu is on par and even cheaper than an Intel cpu, which one would you buy?

i most definitely would.

But at the end of nearly half a decade, has AMD been able to do this yet?

 

[Andrew528]

i was actually thinking more the rythem of the night~

http://www.youtube.com/watch?v=cAQSZhazYk8

seeing how were going back into a type of P4 netburst style arch...

Gets up and dances !! :biggrin:

 

[Andrew528]

i most definitely would.

But at the end of nearly half a decade, has AMD been able to do this yet?

Its funny I can remember when Intel was playing catchup with AMD now AMD have fallen so far behind with processors its not even funny. Sure some of AMD's mid to high range can be competative with Intels low to mid range but (and dont get me wrong I love competition) I still dont think we will see an AMD outright outperform an intel CPU for a long time to come but hey I hope they prove me wrong !!!

 

[Genx87]

This, in my opinion, is how it should be marketed. Then they should do a 6- and/or 8-core(module) design next year when, they perfect 22nm process.

EDIT: Probably the reason they are not doing it this way, is 8-cores is all the rave now, thanks to SB...But if they were to have ignored this and called it a 4-core, 8-thread CPU, and all the benchmarks compared it to other quadcore CPUs, then there would be no question that BD would absolutely wreck the competition and greatly outperform the Phenom II quads. Personally, I will be making my decision to purchase by comparing its performance to quadcores, since this seems most logical, based on its design (and pricing). It just needs to "generally" outperform my 1100t by a small factor, but otherwise, I'm very excited about the new architecture and am very hopeful that as software/compilers catch up, we'll really see it shine!

After looking at the processor diagrams. I agree with this assessment. These are not 8 cores in a traditional sense. The two integer cores share an FPU. So for integer loads it will be able to work on 8 threads but anything floating point 4 threads.

 

[krumme]

Integer issue? More likely Bulldozer cache issues that I had highlighted earlier in another thread. Here is the clue >> Speaking of OCN, this one is epic :thumbsup: .....That was the response to JF-AMD's continuous analogy about the plane....

Only 3% for the aliasing! Then Intel is going shit in their pants, because the server profit is down the drain. Perhaps we will again experience Michael-Dell sending an email to Otellini-Intel like he did in the good old days quote "We have lost the performance leadership and it's seriously impacting our business ...", and Otellini-Intel will be irritated and shout "is he asking for more money again".
Well thats just my hypothesis for the future

 

[kcidmai]

Bulldozer "cores" are not true cores. This is what Andy Glew, creator of CMT, had to say over here.....And Chew* (overclocker at AMD FX overclocking event) had to say over here......As the saying goes, "no such thing as free lunch".

We shall see (hopefully soon)!

Yups, sums up what we have been put through..

Y our comments are pissing me off the point where I actually registered here.

the adjoining of parts of the "core" was to save die space, it is still a dual core. the FPU can either be s 128 bit fpu's (comparable to intel) or one massive 256b fpu. It was a great way to save die space. the pipeline is a bit longer then stars, but it is also WIDER.

the whole design is actually very optimized. It should be a capable performer.

 

[BlueBlazer]

Only 3% for the aliasing!

That's under Linux, but unknown under Windows. It could be similar to the TLB bug, the performance hit could vary depending on application or operating system. At this point, its effects are unknown yet as the lack of SPEC benchmarks yet to be published or leaked. :hmm:

Then Intel is going shit in their pants, because the server profit is down the drain. Perhaps we will again experience Michael-Dell sending an email to Otellini-Intel like he did in the good old days quote "We have lost the performance leadership and it's seriously impacting our business ...", and Otellini-Intel will be irritated and shout "is he asking for more money again".
Well thats just my hypothesis for the future

Some habits die hard!

 

[krumme]

After looking at the processor diagrams. I agree with this assessment. These are not 8 cores in a traditional sense. The two integer cores share an FPU. So for integer loads it will be able to work on 8 threads but anything floating point 4 threads.

Then i ran windows 3.11 without a single core...

(btw: i think its a marketing decision anyway - and a bad one for the reasons already mentioned)