AMD to start Bulldozer AM3+ production by March 2011, launch in April 2011

[DarthShader]

The main advantage of bulldozer is therefore its eight (pseudo-)cores.

Pseudo? What do you mean by "pseudo"?

[mutantmagnet]

The main advantage of bulldozer is therefore its eight (pseudo-)cores.

God this terminology inconsistency is frustrating. Bulldozer should have 16 cores in 8 modules. If it came out with 8 cores it wouldn't be competitive with Sandybridge.

[Lightman]

God this terminology inconsistency is frustrating. Bulldozer should have 16 cores in 8 modules. If it came out with 8 cores it wouldn't be competitive with Sandybridge.

4 modules - 8 cores on desktop
8 modules - 16 cores on server only for now

[Sn0wm@n]

so amd has pseudo cores now ??? LOLL

[zalbard]

so amd has pseudo cores now ??? LOLL

So does Intel with HT on. Welcome to 2002.

[Sn0wm@n]

So does Intel with HT on. Welcome to 2002.

since when does amd has pseudo cores???? intel has ht and yes i know about that ...


but please help me figure out why amd has pseudo cores

[zalbard]

Also, Google is your friend... Same for the search function on these forums.

[mAJORD]

not this again

[Sn0wm@n]

Also, Google is your friend... Same for the search function on these forums.

and those slides proves your right??? LOLL



i thought that it showed the full physical hardware of 2 real cores ...

[informal]

AMD won't use module terminology when it ships Zambezi. It will be cores only(since it *really* has 8 real cores).

[Nintendork]

zalbard
Modules consistes of two cores, not the virtual ones found on HT.

So by given an equal performance (asuming here and there)
1 core --> 100%
1 core + HT --> <100%, 120% (as long as the core are not fully loaded such as Linx bench)
2 core --> 180%~
1 module (two core) --> 160%?~


This was funny:



8 cores Zambesi under Vision Black FX label
4-6 cores Zambesi under Ultimate Vision FX label

http://www.donanimhaber.com/islemci/...en-doguyor.htm

[Tomasis]

Dr Who pisses in his pants

[FlanK3r]

:-d

[amdsempron_xs]

Are there benchmark numbers leaked?

It's strange that we're 4 months before the launch and we don't have any number

At least these numbers keep me away from buying 2500K/2600K

[Hell Hound]

Zambesi am3+ only?

[Sn0wm@n]

Zambesi am3+ only?

yes ... its been know for months

[JF-AMD]

AMD keep silence. Four months before launch - no tests. The most likely scenario is that zambezi has single-thread performance level compared to nehalem (clock-for-clock), in other words zambezi is faster than K10.5 by the same per cent K10.5 is faster than K8 (10 per cent)

The main advantage of bulldozer is therefore its eight (pseudo-)cores.

A. No benchmarks before launch, that is standard AMD policy.
B. They are real cores. Period.

No tests, no previews = Sandy bridge road cleaning ...

See above.

zalbard
Modules consistes of two cores, not the virtual ones found on HT.

So by given an equal performance (asuming here and there)
1 core --> 100%
1 core + HT --> <100%, 120% (as long as the core are not fully loaded such as Linx bench)
2 core --> 180%~
1 module (two core) --> 160%?~

[/url]

Not quite. 1 module with 2 cores would be ~180%. 2 threads running on 2 different modules would be ~190-195%.

Are there benchmark numbers leaked?

It's strange that we're 4 months before the launch and we don't have any number

At least these numbers keep me away from buying 2500K/2600K

No, those are not benchmarks, benchmarks come out at launch.

Zambesi am3+ only?

yes

[Motiv]

A. No benchmarks before launch, that is standard AMD policy.
B. They are real cores. Period.

See above.

Not quite. 1 module with 2 cores would be ~180%. 2 threads running on 2 different modules would be ~190-195%.

No, those are not benchmarks, benchmarks come out at launch.

yes

Your biggest problem will be the fud slung from the 'opposite' side. There are some so called experts on this forum, who struggle with the core/module type thing. I fully expect that to be exploited by the other camp.

It's a nice elegant solution to having extra cores with less die space, however intel on a smaller node will be able to compete core for core at some point, without the performance tradeoff.

The only question is can you keep the 'core' lead.

[MAS]

A. No benchmarks before launch, that is standard AMD policy.
B. They are real cores. Period.

wrong policy. I'm quite sure Intel already has bulldozer chips and is aware of its performance level.

So nondisclosure bulldozer info doesn't help AMD

Your company needs brief zambezi testing, which can show strong sides of new cpu

[flippin_waffles]

Your biggest problem will be the fud slung from the 'opposite' side. There are some so called experts on this forum, who struggle with the core/module type thing. I fully expect that to be exploited by the other camp.

Exactly. However I think it's pretty clear that the other camp is already trying to exploit it. A handful of well placed shills can have a pretty big affect.

It's a nice elegant solution to having extra cores with less die space, however intel on a smaller node will be able to compete core for core at some point, without the performance tradeoff.

The only question is can you keep the 'core' lead.

Which tradeoff? I think it's the other way around, intel is trying to pseudo mock a dual core with dual threads. I see the tradeoff being slightly increasing performance while reducing die space.

I'm not sure what the tradeoff of replacing 2 full cores with 2 full but faster cores would be. There is no evidence whatsoever to point to and say a SB die size is a good tradeoff over an Interlagos, Valencia or Zambezi die size.

wrong policy. I'm quite sure Intel already has bulldozer chips and is aware of its performance level.

So nondisclosure bulldozer info doesn't help AMD

Your company needs brief zambezi testing, which can show strong sides of new cpu

There is the Osbourne Affect, however I wonder how big that really is. CPU's are dirt cheap anyway.

[JF-AMD]

wrong policy. I'm quite sure Intel already has bulldozer chips and is aware of its performance level.

So nondisclosure bulldozer info doesn't help AMD

Your company needs brief zambezi testing, which can show strong sides of new cpu

So that people can immediately stop buying current product and wait? This isn't about intel.

[freeloader]

So that people can immediately stop buying current product and wait? This isn't about intel.

Great. You get to piss off customers twice. The ones who will wait will be pissed if they find out that BD sucks and the ones who don't wait and buy SB will be pissed off if they find out that BD is great.

AMDs' policy is totally ed up. Release a few numbers or loose customers.

Let me get this straight.....I'm average Joe who wants to purchase a computer one week from now. I have a look at Intel and discover this fantastic chip called Sandy Bridge, great performance for the value. I say to myself, "I wonder what AMD has to compete?" Looks around, nothing to compete agains SB. I think I'll buy SB. So how are you losing sales when the person has no intention of buying AMD chips anyhow? Don't you ever think to yourself that by releasing a few benches, you can stem the tide of people (even a half of 1%) from purchasing an Intel setup? Think outside the box man.

And I thought Hector ran the company badly!

[Roadhog]

Let me get this straight.....I'm average Joe who wants to purchase a computer one week from now. I have a look at Intel and discover this fantastic chip called Sandy Bridge, great performance for the value. I say to myself, "I wonder what AMD has to compete?" Looks around, nothing to compete agains SB. I think I'll buy SB. So how are you losing sales when the person has no intention of buying AMD chips anyhow? Don't you ever think to yourself that by releasing a few benches, you can stem the tide of people (even a half of 1%) from purchasing an Intel setup? Think outside the box man.

And I thought Hector ran the company badly!

Your average joe wouldn't even know about BD, or be reading forums like these to look for some supposedly leaked benchmarks that might be true or false. They also wouldn't know what SB is. Your average joe would just walk into best buy and purchase a computer there. AMD can either release their CPU's at the same time as SB, or just keep doing what they are doing since it seems to work fine.

[Sn0wm@n]

wrong policy. I'm quite sure Intel already has bulldozer chips and is aware of its performance level.

So nondisclosure bulldozer info doesn't help AMD

Your company needs brief zambezi testing, which can show strong sides of new cpu

if intel has acces to an amd cpu that is still withing nda ... dont you think it could be seen as corportate espionage ... a thing that is punishable with a prison sentense if its proven to be true .... come on ....


and the nda thing is the same with all big corporations .... saying too much hurts big corporations ...

Great. You get to piss off customers twice. The ones who will wait will be pissed if they find out that BD sucks and the ones who don't wait and buy SB will be pissed off if they find out that BD is great.

AMDs' policy is totally ed up. Release a few numbers or loose customers.

Let me get this straight.....I'm average Joe who wants to purchase a computer one week from now. I have a look at Intel and discover this fantastic chip called Sandy Bridge, great performance for the value. I say to myself, "I wonder what AMD has to compete?" Looks around, nothing to compete agains SB. I think I'll buy SB. So how are you losing sales when the person has no intention of buying AMD chips anyhow? Don't you ever think to yourself that by releasing a few benches, you can stem the tide of people (even a half of 1%) from purchasing an Intel setup? Think outside the box man.

And I thought Hector ran the company badly!

so an average joe that cant wait surely knows how to control a multi billion dollars corporation .... sure ...

[Manicdan]

this thread went lol