AMD Bulldozer and Llano going to be delayed? GF 32nm troubles?

[DrMrLordX]

Barcelona is K10, not K8. Duh. Furthermore, your statement makes it sound like it was a K8 MCM, which it was not. If you were trying to say that K10 has an insignificant number of improvements over K8, then that's one thing, but Barcelona sure as hell wasn't "two K8 cores glued together".

Also, outside of the TLB issue, the L3 really wasn't bad at all. Hell, the original K10s were very similar to current K10.5 chips, minus DDR3 support and some bug fixes/performance tweaks (and, on the server side, the almighty HT Assist, among other things). Even though I have been (and still am) a big fan of the L3-less K10.5 chips, I have yet to find a situation where the L3 is an actual liability either in "real world" apps or in benchmarks. Sure seems to work okay to me.

 

[Scali]

Barcelona is K10, not K8. Duh. Furthermore, your statement makes it sound like it was a K8 MCM, which it was not. If you were trying to say that K10 has an insignificant number of improvements over K8, then that's one thing, but Barcelona sure as hell wasn't "two K8 cores glued together".

It's four K8 *cores* glued together actually (yes, with insignificant improvements, that was the reason I referred to the K8, to emphasize that fact. Bonus point for reading comprehension)...
The dualcore K8 derivatives were already of the glued kind, as they made no attempt of exploiting the fact that they were on a single die, unlike the Core Duo and Core2 Duo, which exploited that with a shared L2 cache.
K10 tries to finally do something with the L3 cache, but it is so poorly designed that the latency is barely better than what you'd get from a two-socket dualcore K8 system using the hypertransport bus (which cache2cache proves, Anandtech tested it in one of their reviews).
Aside from that the L3 cache was also very small, so the hit ratio was too low to really have much benefit.
I was lenient enough to ignore the TLB issue actually.

'Glued' does not necessarily imply MCM. The original Pentium D Smithfield was also a single-die. Still it was pretty much copy-paste of Pentium 4 cores, not that different from what AMD did with the K8 and K10.

Anyone else who wants to challenge what I said? I think it's pretty useless to try.

 

[evolucion8]

It's four K8 *cores* glued together actually (yes, with insignificant improvements, that was the reason I referred to the K8, to emphasize that fact. Bonus point for reading comprehension)...
The dualcore K8 derivatives were already of the glued kind, as they made no attempt of exploiting the fact that they were on a single die, unlike the Core Duo and Core2 Duo, which exploited that with a shared L2 cache.
K10 tries to finally do something with the L3 cache, but it is so poorly designed that the latency is barely better than what you'd get from a two-socket dualcore K8 system using the hypertransport bus (which cache2cache proves, Anandtech tested it in one of their reviews).
Aside from that the L3 cache was also very small, so the hit ratio was too low to really have much benefit.
I was lenient enough to ignore the TLB issue actually.

'Glued' does not necessarily imply MCM. The original Pentium D Smithfield was also a single-die. Still it was pretty much copy-paste of Pentium 4 cores, not that different from what AMD did with the K8 and K10.

Anyone else who wants to challenge what I said? I think it's pretty useless to try.

Is pretty useless to try because you are stubborn as a mule. Stating that Phenom is just two K8 glued together is like stating that Nehalem is simply Penryn without MCM and Hyper Threading, by your extremeley harmful biased point of view, after all, the execution engines didn't change at all between Conroe and Nehalem.

 

[Scali]

Is pretty useless to try because you are stubborn as a mule. Stating that Phenom is just two K8 glued together is like stating that Nehalem is simply Penryn without MCM and Hyper Threading, by your extremeley harmful biased point of view, after all, the execution engines didn't change at all between Conroe and Nehalem.

Well, go ahead and say that Nehalem is simply Penryn without MCM and Hyper Threading, if that makes you happy.
I don't see how that has any bearing on what I said though.

And stubborn? I just know what I'm talking about, and it's useless to argue when you don't (and read more closely, I already corrected the other guy on four K8 cores, and you do it again). It's like bringing a knife to a gun fight.

The problem here is that people just feel the uncontrollable urge (literally, apparently) to attack me, because they don't like what I'm saying. Even though it is the truth, however painful it may be to some... Grow up.

 

[ModestGamer]

It's four K8 *cores* glued together actually (yes, with insignificant improvements, that was the reason I referred to the K8, to emphasize that fact. Bonus point for reading comprehension)...
The dualcore K8 derivatives were already of the glued kind, as they made no attempt of exploiting the fact that they were on a single die, unlike the Core Duo and Core2 Duo, which exploited that with a shared L2 cache.
K10 tries to finally do something with the L3 cache, but it is so poorly designed that the latency is barely better than what you'd get from a two-socket dualcore K8 system using the hypertransport bus (which cache2cache proves, Anandtech tested it in one of their reviews).
Aside from that the L3 cache was also very small, so the hit ratio was too low to really have much benefit.
I was lenient enough to ignore the TLB issue actually.

'Glued' does not necessarily imply MCM. The original Pentium D Smithfield was also a single-die. Still it was pretty much copy-paste of Pentium 4 cores, not that different from what AMD did with the K8 and K10.

Anyone else who wants to challenge what I said? I think it's pretty useless to try.

Nice use of buzzwords. -1 to you. Obviously you aren't grasping what is so different about bulldozer and why intel should be absolutely terrified if it works as expected. Instructions per clock cycle will be huge if they get what they tried for. while clock speed is nice crunching instructions is where it is at. bulldozer is a total refresh on how to get stuff done. It is a very innovative architecture and I highly doubt AMD would put the resources into a chip that wasn't going to be a substantial overhaul.

do you work for Intel or something ?

 

[ModestGamer]

Well, go ahead and say that Nehalem is simply Penryn without MCM and Hyper Threading, if that makes you happy.
I don't see how that has any bearing on what I said though.

And stubborn? I just know what I'm talking about, and it's useless to argue when you don't (and read more closely, I already corrected the other guy on four K8 cores, and you do it again). It's like bringing a knife to a gun fight.

The problem here is that people just feel the uncontrollable urge (literally, apparently) to attack me, because they don't like what I'm saying. Even though it is the truth, however painful it may be to some... Grow up.

Well nobody cares becuase if you can read you can abslutely debase your arguments. Barcelona is not k8 glued together. In fact they don't even use the same fab process.

Fail

 

[evolucion8]

Well, go ahead and say that Nehalem is simply Penryn without MCM and Hyper Threading, if that makes you happy.
I don't see how that has any bearing on what I said though.

And stubborn? I just know what I'm talking about, and it's useless to argue when you don't (and read more closely, I already corrected the other guy on four K8 cores, and you do it again). It's like bringing a knife to a gun fight.

The problem here is that people just feel the uncontrollable urge (literally, apparently) to attack me, because they don't like what I'm saying. Even though it is the truth, however painful it may be to some... Grow up.

Pleeeease, Nehalem isn't a brand new architecture built from the ground up, the same goes for the K10.5, we all know that..... If you love Intel so much, instead of being impartial like a mature person is (Something that lacks around your posts), just change your name to ScalinTeL and live happy. I'm currently happy with my 4 Pentium M's glued with MCM and improved FPU.

 

[Scali]

Nice use of buzzwords. -1 to you. Obviously you aren't grasping what is so different about bulldozer and why intel should be absolutely terrified if it works as expected. Instructions per clock cycle will be huge if they get what they tried for. while clock speed is nice crunching instructions is where it is at. bulldozer is a total refresh on how to get stuff done. It is a very innovative architecture and I highly doubt AMD would put the resources into a chip that wasn't going to be a substantial overhaul.

do you work for Intel or something ?

Lol, just the other day I worked for nVidia, in the "CPU hobbled PhysX" thread...

And I suppose you can praise the (elusive) features of bulldozer without working for AMD? How does that work exactly?

I'll believe Bulldozer is a success when I see it. AMD doesn't exactly have a good track record with Barcelona. Their predictions of performance and the virtues of native die design were nothing like the final product.
You see, the thing is, I'm an assembly programmer. I actually work with CPUs on a low level, and need to be intimately familiar with how the microachitecture works, in order to optimize code for it. As a result I probably have a much better insight in where AMD and Intel can gain, and how much, than someone like you, with all due respect.

 

[Scali]

Pleeeease, Nehalem isn't a brand new architecture built from the ground up, the same goes for the K10.5, we all know that..... If you love Intel so much, instead of being impartial like a mature person is (Something that lacks around your posts), just change your name to ScalinTeL and live happy. I'm currently happy with my 4 Pentium M's glued with MCM and improved FPU.

I don't even like Intel, and I certainly don't like x86. I think x86 is an abomination of an instructionset, and have mentioned that many times on these forums over the years.
Don't insult me.
It's just a fact that when it comes to x86, AMD is often even worse than Intel.
When AMD is not, I buy AMD, simple as that.

Also, I don't see why you keep dragging Nehalem into the discussion, and try to explain how it's not a brand new architecture. I never mentioned Nehalem, let alone that I claimed it was a brand new architecture. Stop the fallacies.

 

[ModestGamer]

Lol, just the other day I worked for nVidia, in the "CPU hobbled PhysX" thread...

And I suppose you can praise the (elusive) features of bulldozer without working for AMD? How does that work exactly?

I'll believe Bulldozer is a success when I see it. AMD doesn't exactly have a good track record with Barcelona. Their predictions of performance and the virtues of native die design were nothing like the final product.
You see, the thing is, I'm an assembly programmer. I actually work with CPUs on a low level, and need to be intimately familiar with how the microachitecture works, in order to optimize code for it. As a result I probably have a much better insight in where AMD and Intel can gain, and how much, than someone like you, with all due respect.

I program MCU's dialy.

Be glad motorola doesn't make CPU's. becuase niether Intel nor AMD have shit on motorola.

 

[Scali]

Be glad motorola doesn't make CPU's. becuase niether Intel nor AMD have shit on motorola.

I was there when they did... 68k and PPC FTW!
Motorola is one of the reasons why I could never like Intel.

 

[ModestGamer]

I don't even like Intel, and I certainly don't like x86. I think x86 is an abomination of an instructionset, and have mentioned that many times on these forums over the years.
Don't insult me.
It's just a fact that when it comes to x86, AMD is often even worse than Intel.
When AMD is not, I buy AMD, simple as that.

Its not going away and it really isn't the problem your making it out to be.

 

[ModestGamer]

I was there when they did... 68k and PPC FTW!
Motorola is one of the reasons why I could never like Intel.

copperhead and now viper are just dumb. Mips are incrediable.

 

[Scali]

Its not going away and it really isn't the problem your making it out to be.

It is when you have to program in it on a daily basis... which you don't, apparently...

 

[ModestGamer]

It is when you have to program in it on a daily basis... which you don't, apparently...

Its not going away and while a pain in the ass to work with. It isn't cuasing a substantial performance hit unless you want to hold it up against motorla instruction sets but we know that will never happen.

 

[Scali]

Its not going away and while a pain in the ass to work with. It isn't cuasing a substantial performance hit unless you want to hold it up against motorla instruction sets but we know that will never happen.

So what is it now?
First I'm not allowed to have criticism on AMD, and now I'm not allowed to have criticism on Intel either.
Seems like you just can't help yourself, you MUST argue whatever I say!

 

[taltamir]

With software you are kinda locked in.

Intel vs. Microsoft

-If Microsoft prices their software exorbitantly, then people can simply download them online without paying. Microsoft has a limit on how much they can stop such downloads. With microprocessors you can't download them.
-If Windows pricing becomes suddenly unattractive, people will find other ways to try to use Windows, ie, like illegal downloads. When majority of the world's software run on the Windows operating system, you are locked to using that particular OS. With Intel, you can simply use AMD chips if they price it that high. There's no downsides like compatibility issues.

AFAIK, the majority of windows installed are from pirated copies...

 

[ModestGamer]

So what is it now?
First I'm not allowed to have criticism on AMD, and now I'm not allowed to have criticism on Intel either.
Seems like you just can't help yourself, you MUST argue whatever I say!

didn't say that did I. Nice try putting words in my mouth. Basically you need to stop complaining. We are hear to debate the merits of the new AMD processors which to this point. you have not.

Intel fanboi you might be. AMD looks to have a winner if they get it right.

 

[ModestGamer]

AFAIK, the majority of windows installed are from pirated copies...

Largest customer of microsoft. Goverments. We subsidize the piracy via tax's.

 

[bryanW1995]

Nice use of buzzwords. -1 to you. Obviously you aren't grasping what is so different about bulldozer and why intel should be absolutely terrified if it works as expected. Instructions per clock cycle will be huge if they get what they tried for. while clock speed is nice crunching instructions is where it is at. bulldozer is a total refresh on how to get stuff done. It is a very innovative architecture and I highly doubt AMD would put the resources into a chip that wasn't going to be a substantial overhaul.

do you work for Intel or something ?

kind of like barcelona... oh, wait

 

[ModestGamer]

kind of like barcelona... oh, wait

total departures from a design stand point. I really think AMD might have something. been wrong before, likely to be wrong again. Having owned both AMD and intel machines I just go with what works best per $. AMD wins and intel wins. depends on what period of time we are talking about. Barcelona paved the was for the new cpu's and was most likely a quick stop gap to hold the fort. I think they have put consider time and energy into BD and llano. I think alot of the product they have offered the last few years was essentially stop gap make money type stuff.

Lets all try to reserve judgement until the parts hit the test cycles and we see what they can do.

 

[Scali]

Intel fanboi you might be.

I'm not, I'm just realistic... Intel is doing a lot better than AMD, and has been for years now.
I'm a fan of good technology, and Intel has been the main provider for that in the PC market since Core2.
Before that I had an AMD Athlon XP CPU, and a regular Athlon before that, so be careful with your accusations.

AMD looks to have a winner if they get it right.

If they do, then my next upgrade is AMD. It's that simple (I currently have an AMD GPU aswell, gee, amazing!).
Is it okay with you to be sceptic about AMD's upcoming architecture? AMD has a poor track record anyway, their execution is rarely as good as what was originally on the roadmap... and what's on the roadmap doesn't look like it's going to beat Sandy Bridge to me.
Geez.

I don't get people like you, having to cry fanboy all the time. I find it deeply insulting however.

 

[ModestGamer]

I'm not, I'm just realistic... Intel is doing a lot better than AMD, and has been for years now.
I'm a fan of good technology, and Intel has been the main provider for that in the PC market since Core2.
Before that I had an AMD Athlon XP CPU, and a regular Athlon before that, so be careful with your accusations.



If they do, then my next upgrade is AMD. It's that simple (I currently have an AMD GPU aswell, gee, amazing!).
Is it okay with you to be sceptic about AMD's upcoming architecture? AMD has a poor track record anyway, their execution is rarely as good as what was originally on the roadmap... and what's on the roadmap doesn't look like it's going to beat Sandy Bridge to me.
Geez.

I don't get people like you, having to cry fanboy all the time. I find it deeply insulting however.

if you wonder why the last few cpu's weren't as good as they could be. look at the investment in a whole new design. I think resources went elsewhere.

 

[Scali]

if you wonder why the last few cpu's weren't as good as they could be. look at the investment in a whole new design. I think resources went elsewhere.

Yea, like in their new 32 nm production process... oh wait.
Or maybe they didn't have the resources in the first place? Yea, looking at AMD's lack of profit in the past few years, that may just be it!

 

[ModestGamer]

Yea, like in their new 32 nm production process... oh wait.
Or maybe they didn't have the resources in the first place? Yea, looking at AMD's lack of profit in the past few years, that may just be it!

building a processor that work at all is a massive undertaking. I think they put the money elsewhere. Smartly so.