• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Bulldozer Finally!

Martini1991 said:
Just a thought to throw out there.
Due to the modules design, when the two parts of the module are being used, there's a performance drop as opposed if there was two "halfs" from two modules being used.

It's very possible than each half of a module has close IPC to Lynfield (best case scenario, maybe 5-10% in it), the 8150 in 4 threaded app's would perform close to an i5 750 clock for clock, but when a full module is used, its IPC is closer to two deneb cores.

So, while the FX4100 would pwn is 2 threaded app's, in 4 threaded app's it may not be much faster than a Phenom II x4, just higher clocked.

Obviously that's a lot of conjecture, but it's fact when the two parts of the module as used, it takes a hit.
Click to expand...

AMD said ages ago that a bd module will be 80% as fast as two classic cores.

bulldozer-module.jpg


http://techreport.com/articles.x/19514
 
Mario said:
Their record was set with 2 cores / 2 threads. So far, nacho_arroyo has gotten 7.1GHz with 6 cores / 12 threads.
That the Intel 7.1GHz eats the 8.4GHz amd for breakfast.
Not to mention that the 980X debuted well over a year ago, and is still going strong
Smoke screen from amd - phenom's also OC'ed better than C2D,how phenom's finished we know LOL :D

Not sure it should really count considering it is multiple processors on a single chip created to work together.
Divide it by the number of cores they use, then you get the actual speed of a single processor and that would be no where near what they are saying.
Click to expand...

Its fastest core record, not fastest cpu package. Would it only be valid if it was a single module, 2 core production chip (achieved by fusing off modules).

Im more than sure you will see benchers performing similar feats on 4,6 and 8 core categories with bulldozers also.
 
Last edited:
Martini1991 said:
That's what I'm getting at.
Define classic core though?

A module may only be the same as two Phenom II cores but a little faster, anyone expecting better is kidding themselves.
Click to expand...

I posted that to backup what you were saying, not argue against it.

By classic core I meant a hypothetical complete core thats the same speed as a single bulldozer core. As for how fast this hypothetical core would be compared to the competition, I have no idea.
 
Last edited:
ng93 said:
I posted that to backup what you were saying, not argue against it.

By classic core I meant a hypothetical complete core thats the same speed as a single bulldozer core. As for how fast this hypothetical core would be compared to the competition, I have no idea.
Click to expand...

Exactly. It has been stated by AMD this is not in comparison to a Phenom II core. Anyway,even if a 8 core Bulldozer is the same as 8 Phenom II cores then it is not going to be slow. However,we still have no idea of what the IPC will be with half the number of threads running. One thread will mean full access to all the resources of the module.
 
Last edited:
ng93 said:
I posted that to backup what you were saying, not argue against it.

By classic core I meant a hypothetical complete core thats the same speed as a single bulldozer core. As for how fast this hypothetical core would be compared to the competition, I have no idea.
Click to expand...

Oh I know :D

CAT-THE-FIFTH said:
Exactly. It has been stated by AMD this is not in comparison to a Phenom II core. Anyway,even if a 8 core Bulldozer is the same as 8 Phenom II cores then it is not going to be slow. However,we still have no idea of what the IPC will be with half the number of threads running. One thread will mean full access to all the resources of the module.
Click to expand...

I'm hoping that it's lynfield IPC on a single "core". That'd make it possibly a worth while contender to a 2500k in 4 v 4 app's.
Though, does it work like HT? The proper cores before the logical ones?
 
Last edited:
mmj_uk said:
I'm confused so is the highend BD 8 conventional cores like Phenom II or just four of these fancy 'split' cores?
Click to expand...

It's 4 modules. 2 "mini cores" if you would is what a module consists of.
Which when fully loaded may only be as good as an 8 core Phenom II.
Though when the modules are only "half" loaded, could be similar IPC to lynfield.
I doubt it'd be as fast as lynfield like.
 
Last edited:
Incase anyone hadn't pointed it out, 5Ghz was basically the lowest the chips went on air, the best chips hit 5.5Ghz on air still as they tried a bunch of chips.

I think its fair to say that given a batch of 2600k's, the highest won't do 5.5Ghz on air, nor would they all come close to 5Ghz, so Bulldozer certainly seems to have a 500-1Ghz advantage on Sandy at the moment. Will Ivy hit the same clocks, will be interesting to see as clock speed advantages from process alone seem almost dead. Bulldozer is clocking higher than Phenom 2 as it was always going to regardless of process due to its design.
 
drunkenmaster said:
Incase anyone hadn't pointed it out, 5Ghz was basically the lowest the chips went on air, the best chips hit 5.5Ghz on air still as they tried a bunch of chips.

I think its fair to say that given a batch of 2600k's, the highest won't do 5.5Ghz on air, nor would they all come close to 5Ghz, so Bulldozer certainly seems to have a 500-1Ghz advantage on Sandy at the moment. Will Ivy hit the same clocks, will be interesting to see as clock speed advantages from process alone seem almost dead. Bulldozer is clocking higher than Phenom 2 as it was always going to regardless of process due to its design.
Click to expand...

Doesn't mean much until we actually know if they were stable, by our definitions that is.

Guess we'll see.
BD's clocking might win me over, but we'll see.
 
Martini1991 said:
It's 4 modules. 2 "mini cores" if you would is what a module consists of.
Which when fully loaded may only be as good as an 8 core Phenom II.
Though when the modules are only "half" loaded, could be similar IPC to lynfield.
I doubt it'd be as fast as lynfield like.
Click to expand...

Ok understood thanks for the clarification.

drunkenmaster said:
Incase anyone hadn't pointed it out, 5Ghz was basically the lowest the chips went on air, the best chips hit 5.5Ghz on air still as they tried a bunch of chips.
Click to expand...

Was that with all of the modules enabled or just a single one like that 8ghz+ attempt that is being lauded?
 
drunkenmaster said:
Incase anyone hadn't pointed it out, 5Ghz was basically the lowest the chips went on air, the best chips hit 5.5Ghz on air still as they tried a bunch of chips.

I think its fair to say that given a batch of 2600k's, the highest won't do 5.5Ghz on air, nor would they all come close to 5Ghz, so Bulldozer certainly seems to have a 500-1Ghz advantage on Sandy at the moment. Will Ivy hit the same clocks, will be interesting to see as clock speed advantages from process alone seem almost dead. Bulldozer is clocking higher than Phenom 2 as it was always going to regardless of process due to its design.
Click to expand...

Not sure I really agree with that Drunkenmaster, the pre-release hype (same as we have here for AMD) was that Sandy would happily do "5ghz on air".
The only person I recall saying it wouldn't was the Asus rep on the Hard OCP forum who said most chips were capable of 4.4-4.6 and abou 10% (I think) would do more...

I think Sandybridge > BD > Sandybridge E in terms of value. And I doubt BD will be that much better at overclocking than Sandy was/is.
Those samples that were benched would've been cherry picked. No doubt at all.
We are just hearing the same hype just from the other camp...
I hope BD is good though... That way I can upgrade...
 
lollyhayes said:
Not sure I really agree with that Drunkenmaster, the pre-release hype (same as we have here for AMD) was that Sandy would happily do "5ghz on air".
The only person I recall saying it wouldn't was the Asus rep on the Hard OCP forum who said most chips were capable of 4.4-4.6 and abou 10% (I think) would do more...

I think Sandybridge > BD > Sandybridge E in terms of value. And I doubt BD will be that much better at overclocking than Sandy was/is.
Those samples that were benched would've been cherry picked. No doubt at all.
We are just hearing the same hype just from the other camp...
I hope BD is good though... That way I can upgrade...
Click to expand...

AFAIK, this was a bunch of unused chips handed out, stuck in systems and see how far they'd go. Its not hype when you can see it happen and see the results. Intel send out one sample and have someone tell you how well it did. AMD gave a bunch of overclockers a bunch of unused chips, and said had a go, and the overclockers gave us the results.

The other part being, again that the "hype" as you said around Sandy was that some would do 5Ghz on air, some do, most don't. We've seen Bulldozers do 5.5Ghz, so again a 500Mhz clock advantage on what we've seen from Sandy.

Hence if the best few Sandy's do 5Ghz on air, by the same assumption the best few Bulldozers will do the same, 4.5Ghz is "easy" on Sandy's and it seems 5Ghz will be easy on a Bulldozer aswell.

The 5-5.5Ghz overclocking on air is all 8 cores working, almost all the overclocking was everything enabled, world records are done with as much turned off as possible, thats just the way it is, I don't much care for unsustainable overclocking, and even phasechange isn't particularly interesting due to the pain to set up and fairly high noise levels, vibrating compressors, need for decent amount of airflow and space. But its pretty likely that while one module did 8.4Ghz, all 4 modules would probably have been fine at not much lower than that. I'd be surprised if they couldn't do 7.5Ghz+ with all cores active.

Quite clearly with all cores active these are going significantly higher than Phenom 2's with water and phase, and Phenom 2 hexcores have done 6.9Ghz with everything enabled.

Theres not much to say they couldn't have done 8.4Ghz with all cores enabled, but end of the day it was a press event and went for a record, the 7Ghz Phenom/Intel hexcore benches weren't done in the same way, basically with limited time and tweaking.

Of course its also possible that around2v with all 4 modules drawing current that the mobo simply wouldn't be able to supply enough juice.

I was sure I read that they hit 7.5-7.7Ghz with all cores enabled but I can't remember where I read that. Basically its about a Ghz ahead of Phenom at every stage, and so I'd expect all cores max overclock to be at similar distance ahead of Phenom, unless it becomes mobo current limited.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom