AMD "Zembezi

[illfindu]

Hey im looking towards the future and I'm eyeing the AMD 8 core bulldozer CPU'S coming down the line. I'v seen some source say there going to use a AM3+ Socket and im wondering if that means youll be able to toss one in a current AM3+ compatible board?
Will my current http://www.newegg.com/Product/Product.aspx?Item=N82E16813130297&nm_mc=OTC-Froogle&cm_mmc=OTC-Froogle-_-Motherboards+-+AMD-_-MSI-_-13130297 msi 870A Fuzion work i noticed just now that its a AM3 not a AM3+ I'm guessing that means ill need a new mother board cause the sockets are comparable?

 

[rolli59]

You are correct on the assumption that you will need a new board!

 

[Kaori Cpu]

Odd are you are going to need a new board.

 

[jimmysmitty]

If you get a AM3+ mobo then you can us a Phenom II for now and get Zambezi but you cannot put Zambezi into a AM3 mobo.

 

[illfindu]

Okay any idea when we will start seeing AM3+ boards im really chomping at the bit to try out a this new archt and 8 core set up.

 

[iceclock]

yes well people forgot one of the first octocore cpu was a xeon octocore, a while back, but was very expensive, so for the mass public, will have to wait for the mainstream 0ctoc0res its unfortunate but heck thats life,

 

[tomate2]

if amd were loyal to their announcements am3+ cpus and mobos should already be on shelves...
there is no way of knowing for sure when those are going to come out...
but i dont think am3+ mobos are going to be available before any am3+ cpus are ready for consumers...
if you happen to find a am3+ mobo be ready for the bulldozer as it will probably come shortly after

 

[jimmysmitty]

Okay any idea when we will start seeing AM3+ boards im really chomping at the bit to try out a this new archt and 8 core set up.

No real 100% solid info has been laid out.

Plus don't forget its not 8 cores. Its 8 modules. Not the same.

Now Intel will have a 8 core CPU out later this year based on Sandy bridge.

 

[EXT64]

Yes, but I can't imagine that an 8 core SB would be affordable considering how well the 4 core is doing.

 

[jimmysmitty]

It all depends really. It will probably come out high to start but then lower end models will hit that will be friendlier in price.

Of course thats the way it always is.

I just wonder what AMD will do if the modular design of BD fails them. Will they go back to the core based or keep trying.

 

[enzo matrix]

No real 100% solid info has been laid out.

Plus don't forget its not 8 cores. Its 8 modules. Not the same.

Wrong. It is the same. Obviously you haven't been keeping up with the news.

All they did is have the pairs of cores (module) share a few more resources. They are still cores. However, each core takes up less space than before.

 

[EXT64]

The 6 core i7s never came down in price (though it would be awesome if 8 core SBs were semi-affordable).

 

[iceclock]

well if they put more cores, and reduce power consumption even more that would be swell

 

[jimmysmitty]

Wrong. It is the same. Obviously you haven't been keeping up with the news.

All they did is have the pairs of cores (module) share a few more resources. They are still cores. However, each core takes up less space than before.

If you compare a 8 core SB chip with a 8 "core" BD chip its not the same. Each SB core has its own resources while each core in BD shares resources. Thats the reason why AMD is calling the modules instead of ccores because they are not each a full core.

It is AMDs version of SMT (called CMT) on a higher level. It means that their version should produce better results than SMT which normally can push 20% better threaded performance, this should be much higher.

But calling the each real cores is a lie because in fact they are not each true cores.

It would be like calling each core in Atom a real CPU core when in fact they are just parts of the CPU core.

 

[iceclock]

could u explain me the difference, because i was shur amds new cores, where actually physical cores, im interested thanks

 

[Hellboy]

If you compare a 8 core SB chip with a 8 "core" BD chip its not the same. Each SB core has its own resources while each core in BD shares resources. Thats the reason why AMD is calling the modules instead of ccores because they are not each a full core.

It is AMDs version of SMT (called CMT) on a higher level. It means that their version should produce better results than SMT which normally can push 20% better threaded performance, this should be much higher.

But calling the each real cores is a lie because in fact they are not each true cores.

It would be like calling each core in Atom a real CPU core when in fact they are just parts of the CPU core.

Hey Jimmy

any news on what sort of speed we expect...

sounds like it wont be too impressive... i mean the 6 core was a bit of a flop... i dont know anyone whos got one.

 

[iceclock]

6 cores is useful, for example u use 4 coreds for 1 task, encoding hd video, and use the 2 other cores, to game, and windows , so it works out quite well, it just most people dont do more than 1 thing at a time honestly that require alot of cpu power

 

[jimmysmitty]

Its hard to fully explain but thi picture does a decent job:

Each core has its own pipeline, L1 DCache and Int scheduler but they share the Fetch, Decode, FP Scheduler, 2 128bit FMACs and the L2 cache. The L3 cache is shared among all the cores.

This may look like a dual core but its not. A Sandy Bridge core has all of that per core, minus the FMAC units, Intel ahsn't implemented FMAC yet and will not till Haswell but are still using a single 256bit unit per core.


Here is a Phenom II (Deneb core) for comparison:

As you can see the Core is just a core, has its own everything including L1/L2 cache and then shares L3 and IMC.

The only thing shared by Sandy Bridge CPUs is the L3 cache and IMC but the IMC in BD is also shared as its part of the CPU as a whole.

Without knowing performance, I cannot guess as to what it would take in order to get 8 real cores worth of performance. If each module gives 1.5 cores worth of performance then it would take more than 10 modules to give the same performance as a real 8 core.

Of course its all speculation which is why I will wait till BD hits to see its performance but it will be hard to say if the 8 "module" BD CPU should go up against a 8 core Sandy Bridge CPU.

Hey Jimmy

any news on what sort of speed we expect...

sounds like it wont be too impressive... i mean the 6 core was a bit of a flop... i dont know anyone whos got one.

For BD, AMD has nothing in expectations of clock speed but people have been "theorizing" that because it is AMDs first step into 32nm and HK/MG that it might not be as impressive as SB is.

I can't say but I would at least expect it to be near stock clocks of Phenom II when it hit, but as for overclocking, I don't see it passing SB. Hell some people patched the BIOS for P67 and were hitting 5.5GHz on air on i7 2600Ks. Of course they are using very nice air coolers but still, 61% stable OC on air is insane.

http://www.xbitlabs.com/news/cpu/display/20101123230924_AMD_s_Bulldozer_Clock_Speeds_May_Be_Higher_than_3_50GHz.html

Thats a rumor of 3.5GHz but I wont hold my breath. First gen process normally doesn't blow away previous generation mature process in stock clock speed or overclocking. Hell Gulftown wasn't better than nehalem but SB is blosing everything previous out of the water.

 

[Hellboy]

Its hard to fully explain but thi picture does a decent job:

http://images.anandtech.com/reviews/cpu/amd/hotchips2010/bulldozermodule.jpg

Each core has its own pipeline, L1 DCache and Int scheduler but they share the Fetch, Decode, FP Scheduler, 2 128bit FMACs and the L2 cache. The L3 cache is shared among all the cores.

This may look like a dual core but its not. A Sandy Bridge core has all of that per core, minus the FMAC units, Intel ahsn't implemented FMAC yet and will not till Haswell but are still using a single 256bit unit per core.


Here is a Phenom II (Deneb core) for comparison:

http://media.bestofmicro.com/AMD-Phenom-II-X4,L-0-174420-13.jpg

As you can see the Core is just a core, has its own everything including L1/L2 cache and then shares L3 and IMC.

The only thing shared by Sandy Bridge CPUs is the L3 cache and IMC but the IMC in BD is also shared as its part of the CPU as a whole.

Without knowing performance, I cannot guess as to what it would take in order to get 8 real cores worth of performance. If each module gives 1.5 cores worth of performance then it would take more than 10 modules to give the same performance as a real 8 core.

Of course its all speculation which is why I will wait till BD hits to see its performance but it will be hard to say if the 8 "module" BD CPU should go up against a 8 core Sandy Bridge CPU.



For BD, AMD has nothing in expectations of clock speed but people have been "theorizing" that because it is AMDs first step into 32nm and HK/MG that it might not be as impressive as SB is.

I can't say but I would at least expect it to be near stock clocks of Phenom II when it hit, but as for overclocking, I don't see it passing SB. Hell some people patched the BIOS for P67 and were hitting 5.5GHz on air on i7 2600Ks. Of course they are using very nice air coolers but still, 61% stable OC on air is insane.

Just as i thought... thats a no then 🙁

Would be nice if AMD got something to have a chance with - oh well, looking towards Windows 8 when it comes out i guess..

Why dont Apple just release their os on pcs too.. Dual boot pc and apple os legally as its been done illegally for ages. I used a Mac mini the other day with Windows on it natively and you know what its quite quick running xp.. for a size of a large burger box.

 

[cbrunnem]

i thought JFAMD? stated that if the first core was being used at 100 percent that the second core could run at 90 percent or both at 95. if that is the case then i think they might have a winner.

 

[jimmysmitty]

Just as i thought... thats a no then 🙁

Would be nice if AMD got something to have a chance with - oh well, looking towards Windows 8 when it comes out i guess..

Why dont Apple just release their os on pcs too.. Dual boot pc and apple os legally as its been done illegally for ages. I used a Mac mini the other day with Windows on it natively and you know what its quite quick running xp.. for a size of a large burger box.

I am not saying BD is going to suck. Just that its hard to say with no real basic info beyond the arch itself.

It may be amazing or it could suck.

i thought JFAMD? stated that if the first core was being used at 100 percent that the second core could run at 90 percent or both at 95. if that is the case then i think they might have a winner.

I think it depends on the actual job thats being done. In servers this may be true. But on consumer desktops it may be different. AMD tends to focus on server design more than consumer design due to the larger profits to be made in that market.

I say wait and see. I think the idea is interesting but I don't think it will achieve 90-95% or a real core since it has to share a lot of the CPUs parts.

 

[iceclock]

i find it not intelligent, or productive to have cpu cores or modules that arent running at 100% capacity, maybe when in idle, but besides that, i dont see the point,

 

[EXT64]

Well, while AMD may not have the resources of Intel, they still have intelligent people (directly in the company and in the other companies working with them). They chose this design for a reason: a balance of performance, power, and cost (die size). While I doubt it blow SB out of the water (in desktops at least) I'm hoping it will provide a solid alternative.

Remember, while it will not be able to 100% utilize all of the 8 x86 (integer) cores at once, it is doing it in less space than 8 full cores. So while total performance won't equal 8 true cores, hopefully the die size reductions will result in better performance per mm^2.

 

[iceclock]

good point, but i still like to see it before i beleive it, honestly,

 

[EXT64]

Definitely. I didn't mean to say it will be awesome (note the liberal use of 'hopefully'), all I meant to say was don't count it out yet. Processors, caches, pipelines, etc. are all such complex things that it is really impossible to look at pics or specs alone and say 'this one is faster'. Even the engineers who designed every part of these chips can't truly say until the first sample comes off. Remember how the ATI engineers were surprised of the performance of the 4xxx when they got the first real chips.