Understanding Quad- vs. Dual-core

[idea_hamster]

So I'm standing by to instantly buy the new MBP now that I have gotten C-suite approval and financing for this IT upgrade (read: my wife said OK ) and as an inveterate Mac rumors follower, I have a pretty good handle on what we expect.

We are all but certain that there will be a top-end dual-core option (2620M). But there is a chance that there will be a quad-core option (2720QM). The differences are really only in the core count and clock speed. So here's my question:

Does the turbo capacity of the 2720QM allow it to perform equally well compared to the 2620M despite the difference in base clock speeds (2.2GHz vs. 2.7GHz respectively) across all tasks? I ask because there was a relatively well-developed discussion regarding the better performance of the 3.33GHz 6-core Mac Pro compared to lower-clock-speed 8- and 12-core models on tasks that were not easily distributed across cores.

I understand that for some processor-intensive tasks that can be broken up across cores (video encoding, which I do, albeit rarely) the quad-core will always outperform. Rather, I'm wondering to what extent the turbo boost feature will allow the quad-core processors to scale performance up to match a high-clocked dual core on tasks in which (in the past) the higher-clocked dual core would have outperformed.

Thx!

(I guess the thread title should be "Understanding turbo boost." Honestly, I'm a bit embarrassed to even post this thread -- I always kind of thought I had a handle on the hardware side of things! Oh, well....)

 

[mulo]

depending on how long your are going to keep it, the quad may be the better choice as more software gets support for more cores.

in regards to turboboost in the MBP there is a big but.
turboboost adjusts depending on the temperature of the core, and as we all know things get very heated inside the MBPs, so the amount of turboboost we are going to see is most likely not very high.

 

[Rodimus Prime]

Like the above posted stated it really depends. If the code is multicore aware and can use it then the quad 720 is going to beat out the 640. If is not then the 640 is going to be faster and even then it depends.
All and all the 2 chips are pretty much equal. Just the dual core one is a little easier on the battery.

People are starting to turn cores into like the Mhz myth. More cores is not always better and I will tell you right now in terms of programing multi threading and multi core coding is a hell to do and generally not done until it really brings in some big gains because it is a lot more difficult to track down errors bugs and it is REALLY easy to screw things up in a multithread multicore design big time if the different threads need to shared access to the same data.

 

[mark28]

Like the above posted stated it really depends. If the code is multicore aware and can use it then the quad 720 is going to beat out the 640. If is not then the 640 is going to be faster and even then it depends.
All and all the 2 chips are pretty much equal. Just the dual core one is a little easier on the battery.

People are starting to turn cores into like the Mhz myth. More cores is not always better and I will tell you right now in terms of programing multi threading and multi core coding is a hell to do and generally not done until it really brings in some big gains because it is a lot more difficult to track down errors bugs and it is REALLY easy to screw things up in a multithread multicore design big time if the different threads need to shared access to the same data.

What Mhz myth? A 3.33 ghz 6-core Westmere i7 is faster than a 2.4 ghz 6-core i7. More Ghz = faster. The whole point of overclocking is making the CPU faster.

 

[axu539]

What Mhz myth? A 3.33 ghz 6-core Westmere i7 is faster than a 2.4 ghz 6-core i7. More Ghz = faster. The whole point of overclocking is making the CPU faster.

The MHz myth is that more MHz = more performance, and that is the only factor. While that is true for two processors of different clock speeds, but have the same architecture. Of course, if the two processors have different numbers of cores or have different architectures (i.e. Arrandale v Sandy Bridge), then performance is dependent on all the factors.

 

[Rodimus Prime]

What Mhz myth? A 3.33 ghz 6-core Westmere i7 is faster than a 2.4 ghz 6-core i7. More Ghz = faster. The whole point of overclocking is making the CPU faster.

I suggest you Google it. Intel pushed the Mhz myth big time and it was AMD that broke it but it been broken now for at least 6-7 years so chances are people are already forgetting about it.

Give you an idea I built my desktop in summer of 2004. It it is an Athlon 3000+ clocked at 2 Ghz. That CPU is faster than a P4 at the time of around 3.5-4Ghz.

Intel pushed the more Ghz the better the chip but did not tell the fact that their is a lot more to speed than raw clock rate. Intel CPI (clocks per instruction) was a lot higher than Athlon.

Mhz/CPI gives you a much better view of how fast a CPU is.

Now is the CPI is the same then yes MHz matters. I might like to point out that sometimes you can do a 10-15% over clock and only gain like 5% because of other bottle necks and might increase you CPI a little.

Sorry for nerding out on you. Just do not buy clock rate. I am hoping intel does not go back to the clock rate is everything BS again because that is when their chips truly started sucking.

 

[iPhone1]

Quad core processing primer.

 

[Tconroy135]

What Mhz myth? A 3.33 ghz 6-core Westmere i7 is faster than a 2.4 ghz 6-core i7. More Ghz = faster. The whole point of overclocking is making the CPU faster.

This is also Dependant on the architecture. I have a Desktop x58 980x and a p67 2600k and clock for clock core for core (980(6 cores) 2600k(4 cores)) the Sandy Bridge processor is faster.

 

[idea_hamster]

depending on how long your are going to keep it, the quad may be the better choice as more software gets support for more cores.

in regards to turboboost in the MBP there is a big but.
turboboost adjusts depending on the temperature of the core, and as we all know things get very heated inside the MBPs, so the amount of turboboost we are going to see is most likely not very high.

Hmm -- so in theory, a 2.2GHz 4-core chip that could turbo to 3.4GHz on a single-core task could match a 2.7GHz 2-core chip that can also turbo to 3.4GHz. However, since the turbo may be throttled back due to heat, that 2.7GHz chip could still outperform on clock-intensive single-thread tasks.

Thanks! It makes things "clearer" and ultimately I think that the long-term picture would suggest the quad. Really, I am loathe to use a computer less than four years. My current 2006 2GHz white MacBook is 4.5 years old and before that I had an original TiBook for nearly 6.

Ultimately, the point may be moot if our top processor choice is the 2620M, but here's to hoping! (Oh, and how about a monster video card for OpenCL, Steve -- since I know you follow my posts!)

 

[kfscoll]

This is also Dependant on the architecture. I have a Desktop x58 980x and a p67 2600k and clock for clock core for core (980(6 cores) 2600k(4 cores)) the Sandy Bridge processor is faster.

"Clock for clock, core for core" means you ran both processors at the same speed and disabled two of the six cores on the 980x to do your comparison. Is that what you did? If so, it goes without saying that the newer processor would be faster. If that's not how you did your comparison, then I call into question your results. For example, the 980x would almost certainly be quite a bit faster on multithreaded-optimized tasks like video encoding, but the 2600k is probably faster on single-threaded tasks like gaming. What kind of apps did you run to carry out your comparison? How much memory/how fast did each system have? What about GPUs?

In other words, your statement that the 2600k is faster omits a lot of information that enables that statement to mean very much.

 

[idea_hamster]

OK, so I probably should have called the thread

2620M 2.7 vs. 2720QM 2.2 -- Does Turbo Make the Quad an Across-the-Board Winner?

I'll know for next time!

 

[mac jones]

Great this is exactly the discusssion I want.

The grey area for me is:

1) exactly how many cores is the task going use (I guess i'll have to try and find this out)

2) the time frame that turbo boost is active.

Because if the the turbo boost remains active for the quad, then the discussion is settled: the quad wins, as it's boost is the same as the dual 2.7.

I need fast burst speed, so if the boost remains active for the duration, than the quad will suffice.

I guess?

 

[Hellhammer]

Great this is exactly the discusssion I want.

The grey area for me is:

1) exactly how many cores is the task going use (I guess i'll have to try and find this out)

2) the time frame that turbo boost is active.

Because if the the turbo boost remains active for the quad, then the discussion is settled: the quad wins, as it's boost is the same as the dual 2.7.

I need fast burst speed, so if the boost remains active for the duration, than the quad will suffice.

I guess?

The reported Turbo speeds are theoretical. It is unlikely that you will ever reach those. Your best bet is to wait for more reviews and benchmarks, I know that at least Anand is working on one and he usually has great real world benchmarks.

 

[mac jones]

The reported Turbo speeds are theoretical. It is unlikely that you will ever reach those. Your best bet is to wait for more reviews and benchmarks, I know that at least Anand is working on one and he usually has great real world benchmarks.

Yes thanks.

I just learned that the boost is TDP gauged. This is great, as that would obviously be the best way to do it. ( I didn't know how it worked).

The question of course then is how long can the quad use the boost in these new Macbooks, before it shuts it down.

Really is ambiguous , as after all, these are notebooks and have little bitty fans in them. Maybe the boost only lasts for nano seconds