Thuban gets turbo...kinda

[jennyh]

http://www.xbitlabs.com/news/cpu/display/20100205073010_AMD_Phenom_II_X6_Thuban_Processors_to_Get_Dynamic_Speed_Boost_Technology.html

Switching off unused cores could be even better, or well should be even better in terms of gaining clock speed.

We're probably looking at 2.8ghz max on the top Thuban, overclocking to 3.6ghz+ in single threaded stuff. Should be interesting times ahead.

 

[jennyh]

I would assume it will be renamed 'Overdrive', keeping in line with the software.

 

[someguy7]

This is good news. Would be even nicer if AMD could release a new stepping of the current Phenom II's with the feature as well.


It is going to be interesting to see what all of the AMD loons who call Turbo a gimmick and all that nonsense say.

 

[clement4413]

If I have understood, The thuban won't be able to be o/c due to TDP (6 core in the same time) so if one software is using 1 thread, the thuban will o/c the first core used at 3.6 ghz (par example) instead of 2.8 ghz

 

[jennyh]

Nobody said turbo was a gimmick, just that it was unfair when benchmarking and 'clock for clock' comparisons.

http://www.freepatentsonline.com/20090235108.pdf

According to that, it could have overclocking potential based on how good your cooling is, which is a radical departure from intels turbo boost tech.

 

[Atranox]

I expect that all of the "turbo boost is a useless ploy" posts over at AMDZone will be disappearing very shortly.

Lets all hope that Thuban makes a positive splash on the market - AMD gaining market share, even if only in regards to enthusiasts, spells good news for everyone (including Intel fans).

 

[jennyh]

If I have understood, The thuban won't be able to be o/c due to TDP (6 core in the same time) so if one software is using 1 thread, the thuban will o/c the first core used at 3.6 ghz (par example) instead of 2.8 ghz

You will still be able to overclock it as normal, probably even past 4ghz on all 6 cores.

The reason it has this automatic overclocking is so that it can keep within the 140w tdp limit for motherboards. AMD could be on to a winner here, especially if my assumption on how far it overclocks itself is true.

Think how 4.2ghz on a single threaded benchmark is going to look compared to 3.2ghz on an i5, or 2.8ghz on an i7.

 

[keithlm]

I expect that all of the "turbo boost is a useless ploy" posts over at AMDZone will be disappearing very shortly.

That is not likely to happen since the arguments against "Turbo" and "Overdrive" or dynamic overclocking in general have nothing to do with brand name.

Unless any benefit received from this type of "feature" applies in all situations, conditions and real time environments then it is a useless feature only good for benchmarketing. (I.e., fooling people that don't have enough knowledge to know any better.)

 

[roofus]

Actually, turbo has been called a marketing gimmick over and over in these very forums. I don't think it is a gimmick and glad to see AMD adopt optimization for lower thread count operations but it could very well be the most useless feature to put in a chip designed for a server (assuming that a 6 core is designed for that market as targeting desktop market that doesnt take advantage of 4 cores would be asinine ). I am guessing this is first implementation and this product feature will find its way to the quad cores and lower.
.

 

[jennyh]

I agree that anything like this is basically just a benchmarking tool with limit uses. Turbo and SMT were devised as methods for intel to break up their market into nice little sections.

However, I'm excited because AMD are doing exactly what they need to be doing, which is taking on intel on the benchmarks.

Depending on how aggressive this 'overdrive' is set, we might even see Thuban beating Gulftown in single or double threaded apps. I would sure be able to appreciate the irony in AMD beating intel in something that intel devised, and AMD perfected.

 

[someguy7]

Jenny. You did not say it was a gimmick. But a good portion of the AMD loons did. And you just used language like "which is a radical departure from intels turbo boost tech.". Which is the AMD spin in action already.

Turbo was not used in a clock for clock comparison. That defeats the purpose of clock for clock itself. It was not unfair during benchmarking. It and what ever AMD ends up calling this should be benchmarked at stock settings(turbo and AMD version running). After that you can disable them both and and do all your clock for clock test you want.

All that is going to happen is the usual. The AMD lovers that hated on turbo will just now say AMD's version is better/radically different and so on.

I love this technology and am thrilled that AMD is adding it to its CPU myself.

 

[roofus]

I do not believe AMD is using this for benchmarks any more than Intel. It is a very useful function that makes sense. It can be disabled at will so I see no reason to cry foul over it. it comes out of the box on and I am certain AMD will do the same. Turbo wasn't something with enthusiasts in mind. It was the folks who have no knowledge or desire to tweak their systems to make use of power that they otherwise would never take advantage of. Calling it something different doesn't make it something different.

 

[jennyh]

Well we don't know for sure yet, just looking at the patents (check out fig 4) - http://www.freepatentsonline.com/20090235108.pdf it looks like AMD's solution will allow a constant increase depending on how good the cooling is.

Intels turbo increases by a bin or two, that's it. It isn't really 'dynamic' overclocking in the way the AMD one is. Why I am excited about this is, imagine future chips that can determine which cores are best for overclocking and how far they can go etc.

That being said, this kind of 'dynamic' overclocking must carry certain risks with it.

 

[roofus]

Sorry but for all intents and purposes, it is generally the same. This isn't me knocking it. It is a great idea and hope they go all out with it across all product lines they possibly can.

 

[cjl]

That is not likely to happen since the arguments against "Turbo" and "Overdrive" or dynamic overclocking in general have nothing to do with brand name.

Unless any benefit received from this type of "feature" applies in all situations, conditions and real time environments then it is a useless feature only good for benchmarketing. (I.e., fooling people that don't have enough knowledge to know any better.)

It applies to ALL single threaded CPU-limited applications. Which is still quite a lot to be perfectly honest. You can keep claiming it does nothing (and I'll freely admit that its implementation on Bloomfield is quite weak), but in anything single threaded, a laptop with an i7-720QM will do MUCH better with turbo enabled (>2.5Ghz IIRC) than with turbo off (1.6GHz, again IIRC).

 

[jennyh]

Well it is the same in that it speeds up cores, but that really is the end of the similarities (assuming AMD are doing what the patents suggest).

This feature should be able to determine exactly what the sweet spot is for the cpu at all times. Compared to turbo boost it is far more advanced. All turbo mode does is determine when 1, 2 or 4 cores are being used, then increases the core by 133mhz or so. The AMD version should be constantly checking the cpu temperature and power and which cores are being used, then deciding whether or not to increase/decrease the clock speed.

If this all works to plan, there could be some very interesting benchmarks results appearing soon. 2010 just got a lot more interesting at least.

 

[logainofhades]

Nobody said turbo was a gimmick, just that it was unfair when benchmarking and 'clock for clock' comparisons.

http://www.freepatentsonline.com/20090235108.pdf

According to that, it could have overclocking potential based on how good your cooling is, which is a radical departure from intels turbo boost tech.

Actually iirc, keithlm has called it a gimmick numerous times.

 

[sirrobin4ever]

I agree that anything like this is basically just a benchmarking tool with limit uses.

Obviously you know very little about programming. Multithreading code is often a very difficult process, as some, if not most, processes work much better in a single threaded environment due to concurrency issues. Obviously a 2.4ghz hexacore processor has a lot more processing horsepower than for instance, a 4ghz single core of the same architecture. Unfortunately, harnessing that power for some applications is just not feasible, and certain applications could very well run faster on the single core than the hexacore simply due to the clock speed. For this reason, the clock speed boosting that both Intel and now AMD are implementing make a lot of sense. I see nothing gimmicky or limiting about them.

 

[jennyh]

Yes the thing about turbo is, you can get the same effect simply by doing a minor overclock on the cpu. Take the i5 for instance, there is absolutely no reason at all why the cpu couldn't just be clocked to 3.2ghz all the time.

In the end, all this is doing is overclocking the cpu and I'm not about to change my mind on that just because AMD has a better version incoming. It's a cool feature, and if AMD nail it they will see great results in benchmarks...but it's still just an overclocked 6 core cpu.

 

[theholylancer]

Jenny. You did not say it was a gimmick. But a good portion of the AMD loons did. And you just used language like "which is a radical departure from intels turbo boost tech.". Which is the AMD spin in action already.

Turbo was not used in a clock for clock comparison. That defeats the purpose of clock for clock itself. It was not unfair during benchmarking. It and what ever AMD ends up calling this should be benchmarked at stock settings(turbo and AMD version running). After that you can disable them both and and do all your clock for clock test you want.

All that is going to happen is the usual. The AMD lovers that hated on turbo will just now say AMD's version is better/radically different and so on.

I love this technology and am thrilled that AMD is adding it to its CPU myself.

it would be a radical departure from intel's turbo if it was not just binned by speed, but binned by tdp on the running chip, and thus, if you had LN2 on the damned thing it can reach 6 GHz for you while intel is locked to it's topest bin


granted that is if they actually made it so that it knows how to up the voltages intelligently, if they did a speed thing like intel did then this would be a respin by jenny/amd but if they did it by tdp and unlimited speeds up, then kudos to them and intel is simply using this as a market cutoff and amd is actually trying to help the consumers that don't oc whom have good airflow cases or aftermarket hsf because they are purtty

 

[Chad Boga]

Nobody said turbo was a gimmick

Dopey Keith did.

According to that, it could have overclocking potential based on how good your cooling is, which is a radical departure from intels turbo boost tech.

What BULLSHlT.

And it is good to see that AMD is copying yet another Intel innovation. :lol:

You will still be able to overclock it as normal, probably even past 4ghz on all 6 cores.

:lol: :lol: :lol: What for a suicide screenshot or under LNP.

Why would AMD's Hex core on the same process be any easier to overclock than their Quads?

AMD could be on to a winner here, especially if my assumption on how far it overclocks itself is true.

I think we all know you will be wrong, again.

Think how 4.2ghz on a single threaded benchmark is going to look compared to 3.2ghz on an i5, or 2.8ghz on an i7.

Leaving aside that Thurban clearly won't be able to get 4.2Ghz under air cooling in a non-overclocked system when it gets released in a few months, the i7-860 on a single thread should be at 3.33Ghz.

 

[keithlm]

Actually iirc, keithlm has called it a gimmick numerous times.

That's because... regardless of all the Intel fanboy argument... at the end of the day the most beneficial use of "Turbo-Boot" is benchmarking. Which makes this a benchmarketing gimmick.

BTW: Did you know that Tom's posted a 2.8Ghz clock-per-clock comparison? This comparison can easily be used to show that Turbo-Boot is generally not even as good as overclocking the i7 920 or i5 750 to a constant 2.8Ghz. All you have to do is compare the results in that article with the standard "clock" speed results. OOPS... this reveals that Turbo-Boot is basically worthless. Worthless == Gimmick.

But some people are stupid enough to buy into the marketing hype. (And it makes them really angry when somebody points out that it is merely marketing hype.)

Dopey Keith did..

Hey LOOK! The flame-baiter that likes to insult people because he doesn't agree with them. How mature.

 

[cjl]

Yes the thing about turbo is, you can get the same effect simply by doing a minor overclock on the cpu. Take the i5 for instance, there is absolutely no reason at all why the cpu couldn't just be clocked to 3.2ghz all the time.

In the end, all this is doing is overclocking the cpu and I'm not about to change my mind on that just because AMD has a better version incoming. It's a cool feature, and if AMD nail it they will see great results in benchmarks...but it's still just an overclocked 6 core cpu.

Going back to my example, you actually can't just overclock it. Even if the laptop's cooling was up to cooling an i7-720qm at 2.5+GHz all the time (which it almost definitely is not), the battery life would be atrocious. The best solution for a laptop is the solution Intel has - turbo mode. I agree that it is perhaps less useful in a desktop (though even there it has a purpose), but in a laptop, turbo mode is a great idea.

 

[keithlm]

I agree that it is perhaps less useful in a desktop (though even there it has a purpose), but in a laptop, turbo mode is a great idea.

Actually every time I think of this feature in a laptop I cringe.

The last thing I want in a laptop is something that will decrease battery life in ANY way. I guess if you use the thing plugged into the mains all the time then that would be one thing... but when using a battery... just... NO.

 

[Mousemonkey]

When I saw this I couldn't help but recall that in other threads turbo boost seemed to be a bad thing, cheating or something.

Jenny, keep in mind AMD may be implementing a very similar technology, which you will undoubtedly hail as innovative and cutting-edge.

Well forecast TC!