Could a 12.8ghz cpu equal a quad core 3.2ghz?

[Maximilian]

Seeing some listings on ebay etc where they do this "AMD cpu 3.2ghz x4 = 12.8ghz" which isnt how it works at all but it still got me thinking. Would a 12.8ghz single core cpu be able to match a 3.2ghz quad? Same architecture etc.

Im guessing it would probably heavily depend on what you are doing but i dont know. Anyone have a better idea which would be fastest?

 

[Candymancan21]

I would assume yes. The reason they have many core's is because there are clock speed limits on silicon. At least thats what i think is the reason they are putting multiple core.

But then im probbly wrong

 

[Kenmitch]

Seeing some listings on ebay etc where they do this "AMD cpu 3.2ghz x4 = 12.8ghz" which isnt how it works at all but it still got me thinking. Would a 12.8ghz single core cpu be able to match a 3.2ghz quad? Same architecture etc.

Im guessing it would probably heavily depend on what you are doing but i dont know. Anyone have a better idea which would be fastest?

It would kick the quad to the curb and stomp on it's head in alot of things. But not sure how it would do with heavy multi-tasking situations. I'm thinking that it would most likely hold it's own.

It would be a fun chip to play with forsure!

 

[Cogman]

in single threaded apps, yeah it would dominate. However, because of the overhead of running more then one thread on a single threaded CPU, it wouldn't be better then a multi-cored CPU in heavily threaded applications.

Think of it this way, if you have 2 threads on a single core, then each time you switch from one thread to the next, you have to save the state of the last thread and load the state of the next thread, this takes a few cycles, that along with the fact that the OS has to do all its tasks inbetween thread switching means that the CPU is spending a decent amount of time doing routine stuff. With a multi-cored CPU, however, one CPU could do the OS stuff and the scheduling (and some other processing) while the other cores just follow what it already has set up.

 

[Chiropteran]

With a multi-cored CPU, however, one CPU could do the OS stuff and the scheduling (and some other processing) while the other cores just follow what it already has set up.

"could" being the keyword. I'm pretty sure I have seen running tasks switch between cores constantly, apparently at random, on most windows machines. I know you have manually set affinity to force a program to execute on a certain core or cores, but most people don't do this. I think a 12.8ghz single would dominate a 3.2ghz quad, all else being equal, because even the most multi-threaded applications and heaviest multitasking users typically don't fully utilize every core equally, often one is pushed to it's max while other cores are half-idle. Even the few tasks that will push every core 100%, while this is not based on any facts other than what makes sense in my mind, I have a feeling even these tasks would run faster on a single core because they wouldn't have to deal with the overhead of splitting the threads between CPUs.


Edit: Actually, why are we even theorizing? This can be easily tested in the real world. Simply underclock a 3.2ghz quad to 800mhz, and benchmark, then benchmark the same CPU at it's native 3.2ghz with 3 cores disabled. Or, just look at some past benchmarks comparing quads to dual cores. Which is faster, a 4ghz dual core or 2ghz quad of the same architecture? I believe the dual wins in almost every situation based on memory, but I don't have benchmarks handy.

 

[jtisgeek]

Yeah it still comes down to raw power so speed is king. Now by adding cores you are upping the efficiency being able to handle more tasks at the same time.

Clocks have leveled out now so amd and intel are just working on the efficiency to get more stuff done.

 

[Cogman]

"could" being the keyword. I'm pretty sure I have seen running tasks switch between cores constantly, apparently at random, on most windows machines. I know you have manually set affinity to force a program to execute on a certain core or cores, but most people don't do this. I think a 12.8ghz single would dominate a 3.2ghz quad, all else being equal, because even the most multi-threaded applications and heaviest multitasking users typically don't fully utilize every core equally, often one is pushed to it's max while other cores are half-idle. Even the few tasks that will push every core 100%, while this is not based on any facts other than what makes sense in my mind, I have a feeling even these tasks would run faster on a single core because they wouldn't have to deal with the overhead of splitting the threads between CPUs.


Edit: Actually, why are we even theorizing? This can be easily tested in the real world. Simply underclock a 3.2ghz quad to 800mhz, and benchmark, then benchmark the same CPU at it's native 3.2ghz with 3 cores disabled. Or, just look at some past benchmarks comparing quads to dual cores. Which is faster, a 4ghz dual core or 2ghz quad of the same architecture? I believe the dual wins in almost every situation based on memory, but I don't have benchmarks handy.

Well, to be fair, a dual vs a quad isn't really the same comparison as all the OS stuff and scheduling is still only occurring on 1 core (that is not saying OS services happen on a single core, I'm talking about the really low level OS stuff, such as scheduling).

I would love to test it, unfortunately my Quad is a 2.4 GHz one, and becomes unstable around 3ghz .

 

[Rezist]

I think the 12.8 GHz would beat the quads at 3.2

 

[Rubycon]

If they can build a single core doing 12GHz they can build a twelve core cpu running at 12GHz. Sounds gross but if you can have a cake you better be able to eat it too!

 

[JAG87]

"could" being the keyword. I'm pretty sure I have seen running tasks switch between cores constantly, apparently at random, on most windows machines. I know you have manually set affinity to force a program to execute on a certain core or cores, but most people don't do this. I think a 12.8ghz single would dominate a 3.2ghz quad, all else being equal, because even the most multi-threaded applications and heaviest multitasking users typically don't fully utilize every core equally, often one is pushed to it's max while other cores are half-idle. Even the few tasks that will push every core 100%, while this is not based on any facts other than what makes sense in my mind, I have a feeling even these tasks would run faster on a single core because they wouldn't have to deal with the overhead of splitting the threads between CPUs.


Edit: Actually, why are we even theorizing? This can be easily tested in the real world. Simply underclock a 3.2ghz quad to 800mhz, and benchmark, then benchmark the same CPU at it's native 3.2ghz with 3 cores disabled. Or, just look at some past benchmarks comparing quads to dual cores. Which is faster, a 4ghz dual core or 2ghz quad of the same architecture? I believe the dual wins in almost every situation based on memory, but I don't have benchmarks handy.

You cant make that comparison because no CPU goes doesn to 800M without bringing down the base frequency, which makes the comparison biased.

Second of all, that's not really how it works. Without comparing single threaded programs for obvious reasons, multi threaded programs run stupidly faster on four 3.2G cores than one 12.8G. The reason is that multiple threads make better use of the resources inside a cpu than one thread.

Read about pipelining and OOO and you will understand.

http://en.wikipedia.org/wiki/Instruction_pipeline
http://en.wikipedia.org/wiki/Out-of-order_execution

In layman terms what gets more people to work in the morning? A one lane autobahn, or a 4 lane regular highway? If only you were driving to work, the one lane autobahn is best, but as soon as you have more cars with you, everybody has to drop speed to avoid collisions (which compares to parts of the cpu being inactive waiting on data). Think about it.

 

[veri745]

Second of all, that's not really how it works. Without comparing single threaded programs for obvious reasons, multi threaded programs run stupidly faster on four 3.2G cores than one 12.8G. The reason is that multiple threads make better use of the resources inside a cpu than one thread.

Read about pipelining and OOO and you will understand.

http://en.wikipedia.org/wiki/Instruction_pipeline
http://en.wikipedia.org/wiki/Out-of-order_execution

In layman terms what gets more people to work in the morning? A one lane autobahn, or a 4 lane regular highway? If only you were driving to work, the one lane autobahn is best, but as soon as you have more cars with you, everybody has to drop speed to avoid collisions (which compares to parts of the cpu being inactive waiting on data). Think about it.

I disagree. The 12.8 GHz single core would be far superior to the 3.2GHz quad. The only reason they started making multicores in the first place was because frequency stopped scaling well.

Yes, there would be a mild amount of overhead when switching threads on a 1-core, but there's plenty of cache-coherency overhead on a quad core. It's hard to pin a number down because of varying applications, but if you had the same amount of cache per thread, the 12GHz CPU would kick ass. You're looking at the same amount of throughput with 1/4 the latency, who wouldn't want that?

 

[Cogman]

You cant make that comparison because no CPU goes doesn to 800M without bringing down the base frequency, which makes the comparison biased.

Second of all, that's not really how it works. Without comparing single threaded programs for obvious reasons, multi threaded programs run stupidly faster on four 3.2G cores than one 12.8G. The reason is that multiple threads make better use of the resources inside a cpu than one thread.

Read about pipelining and OOO and you will understand.

http://en.wikipedia.org/wiki/Instruction_pipeline
http://en.wikipedia.org/wiki/Out-of-order_execution

In layman terms what gets more people to work in the morning? A one lane autobahn, or a 4 lane regular highway? If only you were driving to work, the one lane autobahn is best, but as soon as you have more cars with you, everybody has to drop speed to avoid collisions (which compares to parts of the cpu being inactive waiting on data). Think about it.

While I somewhat agree with what you are saying, OOO and pipelining have nothing to do with a CPU being multi-cored or not. Single core CPU's have long been able to take advantage of OOO and pipelining.

HT might be an example where using CPU resources more fully results in faster operations, but again, that is on single cored CPU's.

The difference might be in register usage, with 4 cores you have effectively 4 x more registers available for doing operations with.

If they can build a single core doing 12GHz they can build a twelve core cpu running at 12GHz. Sounds gross but if you can have a cake you better be able to eat it too!

Well... That is unless the power requirements of the single core CPU are too obscenely high to replicate to 4 cores. 120W x 4 would be a beast to cool.

 

[JAG87]

I disagree. The 12.8 GHz single core would be far superior to the 3.2GHz quad. The only reason they started making multicores in the first place was because frequency stopped scaling well.

Yes, there would be a mild amount of overhead when switching threads on a 1-core, but there's plenty of cache-coherency overhead on a quad core. It's hard to pin a number down because of varying applications, but if you had the same amount of cache per thread, the 12GHz CPU would kick ass. You're looking at the same amount of throughput with 1/4 the latency, who wouldn't want that?

And why do you think frequency stopped scaling? It's almost as if you didn't read my post at all. Developers cannot program something to take advantage of all those clock cycles. If all executions were random than maybe you would be right. There is going to be a ton of wasted clock cycles because the code is just waiting on other executions to finish so that part of the processor is freed up.

While I somewhat agree with what you are saying, OOO and pipelining have nothing to do with a CPU being multi-cored or not. Single core CPU's have long been able to take advantage of OOO and pipelining.

HT might be an example where using CPU resources more fully results in faster operations, but again, that is on single cored CPU's.

The difference might be in register usage, with 4 cores you have effectively 4 x more registers available for doing operations with.

Yes and pipelining and OOO fail to provide any advantage once we reached a certain clock speed. Which is why I am bringing them up. HT is another very good example thank you for bringing it up. HT fails to provide improvement in the majority of applications (because they are coded such that freeing up certain processor resources provides no gains until other executions finish). And it sucked especially when it was implemented on single core P4s. All this should help you draw conclusions as to why a 12.8G single core cpu would suck hard.

 

[StrangerGuy]

And why do you think frequency stopped scaling? It's almost as if you didn't read my post at all. Developers cannot program something to take advantage of all those clock cycles. If all executions were random than maybe you would be right. There is going to be a ton of wasted clock cycles because the code is just waiting on other executions to finish so that part of the processor is freed up.

Frequency stopped scaling because CPUs are limited by the underlying silicon and in turn, waste heat output.

 

[pitz]

Back in the old days, I had a dual P3-450, and I had a P3-800 or P3-933 around here. The dual CPU P3-450 was, by far, a more responsive machine to use (on NT 4.0/Win2k/XP).

The multiple CPU's facilitated multiple execution queues, which meant that one process that was CPU-intensive, didn't hog everything.

So yeah..give me the 3.0GHz quad-core CPU, instead of the 12GHz single-core CPU, for everyday, desktop work. Maybe there's a game or scientific app that runs as one thread, and works better on single core, but for most users, IMHO, they're better off with at least dual cores for overall system responsiveness and performance.

In fact, after that experience...I never bought a single-core machine again for myself.

 

[veri745]

And why do you think frequency stopped scaling? It's almost as if you didn't read my post at all. Developers cannot program something to take advantage of all those clock cycles. If all executions were random than maybe you would be right. There is going to be a ton of wasted clock cycles because the code is just waiting on other executions to finish so that part of the processor is freed up.

And it's almost as though you're using points that directly contradict what you're trying to argue.

Complexity for code development is an argument AGAINST multicores

Ability to execute random (independent) code is an argument IN FAVOR of multicores

Code synchronization (wasted clock cycles because of waiting for a previous instruction thread to complete) is an argument AGAINST multicores.

The only reason a 12GHz core would be bottlenecked is because of I/O latency, and that's just as much of a problem on current quad-core machines.

*edit* Oh, and frequency stopped scaling because the silicon couldn't switch any faster, not because it wasn't useful.

 

[CurseTheSky]

This is actually very easy to test. Take a modern quad core processor (like an i7 920, i7 860, i5 750, etc.), disable two cores, and run a benchmark. Then, enable the two cores again and drop the multiplier so it runs at exactly half the speed of the previous test and run the benchmark again. Regardless of how heavily threaded the benchmark is, my bet goes on the "dual core."

You can do the same with a true dual core processor - disable one core for the first test, and drop the multiplier for the second. This time, my bet goes on the single core if the benchmark is poorly threaded, and on the dual core (due to background services, etc.) if it's not.

Honestly, nearly all computers today would do just fine with a dual core. Quad cores are great for encoding any other heavily-threaded, CPU-intensive tasks, but they're complete overkill for 90% of the systems out there. I just bought one because they're relatively cheap.

 

[JAG87]

And it's almost as though you're using points that directly contradict what you're trying to argue.

Complexity for code development is an argument AGAINST multicores

Ability to execute random (independent) code is an argument IN FAVOR of multicores

Code synchronization (wasted clock cycles because of waiting for a previous instruction thread to complete) is an argument AGAINST multicores.

The only reason a 12GHz core would be bottlenecked is because of I/O latency, and that's just as much of a problem on current quad-core machines.

*edit* Oh, and frequency stopped scaling because the silicon couldn't switch any faster, not because it wasn't useful.

Dude, do you even know the difference between simple assembly code instructions and threads? I am talking about dependency of assembly code executing through a pipeline, and you are talking about dependency of multiple threads in a programming language executing through the operating system's scheduler. You are on a completely different page. Take some programming courses for the love of god.

A 12G single core CPU would need an 80 stage pipeline to get any work done, and would still fail miserably.

 

[JAG87]

This is actually very easy to test. Take a modern quad core processor (like an i7 920, i7 860, i5 750, etc.), disable two cores, and run a benchmark. Then, enable the two cores again and drop the multiplier so it runs at exactly half the speed of the previous test and run the benchmark again. Regardless of how heavily threaded the benchmark is, my bet goes on the "dual core."

You can do the same with a true dual core processor - disable one core for the first test, and drop the multiplier for the second. This time, my bet goes on the single core if the benchmark is poorly threaded, and on the dual core (due to background services, etc.) if it's not.

Honestly, nearly all computers today would do just fine with a dual core. Quad cores are great for encoding any other heavily-threaded, CPU-intensive tasks, but they're complete overkill for 90% of the systems out there. I just bought one because they're relatively cheap.

Good call,

Mind you, I didn't know we were talking about everyday computing. I am talking about pure throughput.

Try to run 4x 1.6G cores and encode a video with a multithreaded encoder.
Then run 2x 3.2G cores and encode a video with a multithreaded encoder.
My money is on the quad. And mind you that a dual is already in a much better position than a single core.

 

[nyker96]

oh no, not one of these discussions again. maybe they should just have a sticky for the topic!

 

[Cogman]

Yes and pipelining and OOO fail to provide any advantage once we reached a certain clock speed. Which is why I am bringing them up. HT is another very good example thank you for bringing it up. HT fails to provide improvement in the majority of applications (because they are coded such that freeing up certain processor resources provides no gains until other executions finish). And it sucked especially when it was implemented on single core P4s. All this should help you draw conclusions as to why a 12.8G single core cpu would suck hard.

Umm.. So long as we have multi-clock cycle instructions, Pipelining and OOO will ALWAYS result in faster processing speeds vs non-pipelined/ out of order processors.

(unless you are referring to higher clock speeds achieved through deeper and deeper pipes, in which case it is somewhat true. However, that applies directly to both multi-cored and single-cored technologies. Even then, OOO benefits both architectures regardless of clock speed.)

We are talking about the same architecture which implies that the pipelines are equally deep. While clockspeed does depend on the longest stage in the pipeline, the length of that stage depends on several factors and can take more/less time depending on the tech used to implement it.

 

[JFAMD]

That is like asking can Michael Jordan beat the 1985 Chicago Bears.

 

[Fox5]

The 12.8Ghz single core would easily outperform the 3.2Ghz quad core.

But the quad core could potentially still be more responsive due to poor schedulers. (though linux has a much better scheduler than windows when it comes to interactivity, a run away process generally doesn't get to monopolize the processor, and I/O is often handled asynchronously instead of causing waits like on windows)

Also, the quad core could be faster if the single core is cache starved. If the quad core has 4x as much cache (or more), that's a major advantage. A 12.8Ghz cpu with no L2 cache would be quite slow, and even a limited L2 cache (like the old durons with 64KB) wouldn't fair well.

 

[aigomorla]

dude the 12ghz single core would dominate.

I dont know wtf u guys are argueming about.

First off GHZ does not scale lineally.
Why does it not scale linear?
Because 12ghz would require either a massive multi, or a massive FSB/QPI.

Having your fsb or qpi that ramped up will lead to an insane boost in overall system speed.

If you ask me, id take the single 12ghz core over 4 x 3ghz core ANY day of the week.
My gulfy @ 4.4ghz x 6 = 26.4ghz w/o HT and 26.4+13.2(cuz ht is about 50% of real cores) = 39.6ghz :O

Give me a single 39.6ghz machine and i'll rule the world.

That is like asking can Michael Jordan beat the 1985 Chicago Bears.

Only if he has bugs bunny on his team.
Remember that movie?

 

[Chiropteran]

Also, the quad core could be faster if the single core is cache starved. If the quad core has 4x as much cache (or more), that's a major advantage. A 12.8Ghz cpu with no L2 cache would be quite slow, and even a limited L2 cache (like the old durons with 64KB) wouldn't fair well.

Modern cache runs at the same speed as the CPU though. Remember back in the days of the celeron A? It was the first cpu with on-die l2 cache running at full speed. Other competing CPUs at the time (such as pentium 2), had cache running at 50% core speed. The pentium pro also had full speed cache, but it was very expensive and only available at 200mhz as the fastest speed, so it wasn't in the same market. Even though the celeron A had less cache than other current CPUs, the cache running at full speed helped compensate for the fact there wasn't as much of it.

Anyway, point is that the single core with 1/4 as much cache might still be faster because it's cache is 4X faster than the cache of the hypothetical quad core.