How Come AMD CPU's Have much Higher GHz but Worse Performance Than Intel?

[PcMe]

Question says it all... Intel beats AMD CPUs performance wise in benchmarks although most of the time AMD has higher GHz than Intel. How is that? AMD has over 5 GHz CPUs now WITHOUT EVEN OVERCLOCKING and CPU hasn't even hit 4 yet.

 

[HazBot240599]

Ghz is not necessarily a good way to judge a cpu's speed. A hert is a wave of energy and a Giga-hert is 1,000,000,000 waves of energy. In intel cpus more energy and data it transfered per wave than amd therefore getting more performance to the Giga-hert. This does not necessarily mean amd cpus it just means they are clocked differently.
Hope this helps

 

[kirilmatthew]

The answer is GHz isn't everything. The design of the cores is the most important thing. Intel's design is better, so their cores can process more information per cycle(Hz). This is to better performance per Hz. Therefore AMD's CPUs have to have more Hz then Intel's to process the same amount of information. Hope this helps

 

[Jaxem]

This all comes down to instruction piplines. In an Intel processor it takes less instructions to accomplish something than the current crop of AMD processors because it has a much shorter pipeline. during each cycle of the processor Intel processors are much more efficient, to the point of being able to do more at a lower speed. Think of it like doing a dot-to-dot against someone, you can be doing yours 15% faster, but if you have 20 dots and they have 15, they'll finish first.

 

[kirilmatthew]

This all comes down to instruction piplines. In an Intel processor it takes less instructions to accomplish something than the current crop of AMD processors because it has a much shorter pipeline. during each cycle of the processor Intel processors are much more efficient, to the point of being able to do more at a lower speed. Think of it like doing a dot-to-dot against someone, you can be doing yours 15% faster, but if you have 20 dots and they have 15, they'll finish first.

Kinda the opposite way around. They both have to accomplish the same work, but the Intel ones do it faster. I don't think your explanation really works too well 😛

 

[miggtt699]

Architecture is an important role on performance. Plus, they usually say that the FX line sucks compared to the intel one, yet they forget that the FX-8350 isn't being compared apples to apples. In my view, the FX DEMOLISHES the core i5 in heavy multitasking due to having twice the number of cores, yes weaker, but more cores can do more stuff. It's not for some reason that Sony and Microsoft are going with an Octa Core design for their new systems, to be able to handle the UI and play games and also download games and update the system all in the background without much issues. In all fairness, the FX is a processor that isn't supposed to be better than intel when talking about Performace Per Clock or Performance Per Core. The FX is a multitasking processor, and most of the benchmarks where you see the intel stepping on the FX, they aren't multitasking scenarios, but performance oriented benchmarks.

This is my opinion on it.

 

[8350rocks]

This all comes down to instruction piplines. In an Intel processor it takes less instructions to accomplish something than the current crop of AMD processors because it has a much shorter pipeline. during each cycle of the processor Intel processors are much more efficient, to the point of being able to do more at a lower speed. Think of it like doing a dot-to-dot against someone, you can be doing yours 15% faster, but if you have 20 dots and they have 15, they'll finish first.

Kinda the opposite way around. They both have to accomplish the same work, but the Intel ones do it faster. I don't think your explanation really works too well 😛

Actually this is the best breakdown I can give you...

Think of an AMD CPU processing instructions like Wal-Mart checkout lines, except that 1 checkout line has a line a mile long and 2-3 other lines have a few people each in them, and the rest are empty. Not terribly efficient, right?

Think of an Intel CPU processing instructions like BestBuy, you have 1 line a mile long, and as each person comes to the front they go to the open register. Far more efficient in this scenario right?

What it boils down to is this...current software is not optimized to use the AMD architecture because it is a new and revolutionary design. Intel is using a slight tweak of a 15 year old design, so it is better suited to current software.

However, as software engineers begin to figure out how to solve problems working things out with multiple cores working on the same issues to get them done faster, the AMD architecture will get rapid performance increases, where the Intel architecture will see some increases, it won't be as dramatic.

The issue is not so much in the hardware, as much as it is in the software. Software now is lagging about 2-3 years behind hardware because they have to cater to the lowest common denominator. The world is still predominantly running on 2-3 year old architecture in their PCs. This means you cannot build software for something cutting edge if you intend to appeal to a big audience.

Good software can make a bad CPU look good, and bad software can make a great CPU look terrible. It's all about optimization.

Hope that helps.

 

[Jaxem]

This all comes down to instruction piplines. In an Intel processor it takes less instructions to accomplish something than the current crop of AMD processors because it has a much shorter pipeline. during each cycle of the processor Intel processors are much more efficient, to the point of being able to do more at a lower speed. Think of it like doing a dot-to-dot against someone, you can be doing yours 15% faster, but if you have 20 dots and they have 15, they'll finish first.

Kinda the opposite way around. They both have to accomplish the same work, but the Intel ones do it faster. I don't think your explanation really works too well 😛

It was the best i could come up with on the spot, you're just seeing it from a different perspective that i thought of it from:

Work to be done= 1 dot to dot.
Speed hand is moving=clock speed
number of dots=number of steps to complete the work done

The way i thought of it, it is the same amount of work, The intel does it faster, but not because of speed, because of efficiency.

 

[HazBot240599]

This all comes down to instruction piplines. In an Intel processor it takes less instructions to accomplish something than the current crop of AMD processors because it has a much shorter pipeline. during each cycle of the processor Intel processors are much more efficient, to the point of being able to do more at a lower speed. Think of it like doing a dot-to-dot against someone, you can be doing yours 15% faster, but if you have 20 dots and they have 15, they'll finish first.

Kinda the opposite way around. They both have to accomplish the same work, but the Intel ones do it faster. I don't think your explanation really works too well 😛

It was the best i could come up with on the spot, you're just seeing it from a different perspective that i thought of it from:

Work to be done= 1 dot to dot.
Speed hand is moving=clock speed
number of dots=number of steps to complete the work done

The way i thought of it, it is the same amount of work, The intel does it faster, but not because of speed, because of efficiency.

Think of an amd cpu as a diesel engine and an intel cpu a petrol engine. There is not much difference in the hardware but the petrol engine runs more efficient (and cooler).

 

[kirilmatthew]

This all comes down to instruction piplines. In an Intel processor it takes less instructions to accomplish something than the current crop of AMD processors because it has a much shorter pipeline. during each cycle of the processor Intel processors are much more efficient, to the point of being able to do more at a lower speed. Think of it like doing a dot-to-dot against someone, you can be doing yours 15% faster, but if you have 20 dots and they have 15, they'll finish first.

Kinda the opposite way around. They both have to accomplish the same work, but the Intel ones do it faster. I don't think your explanation really works too well 😛

It was the best i could come up with on the spot, you're just seeing it from a different perspective that i thought of it from:

Work to be done= 1 dot to dot.
Speed hand is moving=clock speed
number of dots=number of steps to complete the work done

The way i thought of it, it is the same amount of work, The intel does it faster, but not because of speed, because of efficiency.

The thing is, Intel CPUs don't move less "dots", they move dots a shorter distance. Think of an Intel CPU as a direct route and an AMD CPU as a indirect route. AMD CPUs will move faster on the road(ex 3.8mph) and Intel will move slower(ex. 3.4mph) at a certain speed the AMD CPU "car" will move fast enough to offset the difference with the Intel CPU "car". But that's basically how it works. The Intel route is shorter, that's why its faster. But think if FX-8350 as an 8 lane highway and Intel as a 4 lane. If there are 8 cars(instructions) at the same time the FX will deliver more cars in the same time. If there are only 4 cars the Intel will. Make sense?

 

[Jaxem]

This all comes down to instruction piplines. In an Intel processor it takes less instructions to accomplish something than the current crop of AMD processors because it has a much shorter pipeline. during each cycle of the processor Intel processors are much more efficient, to the point of being able to do more at a lower speed. Think of it like doing a dot-to-dot against someone, you can be doing yours 15% faster, but if you have 20 dots and they have 15, they'll finish first.

Kinda the opposite way around. They both have to accomplish the same work, but the Intel ones do it faster. I don't think your explanation really works too well 😛

It was the best i could come up with on the spot, you're just seeing it from a different perspective that i thought of it from:

Work to be done= 1 dot to dot.
Speed hand is moving=clock speed
number of dots=number of steps to complete the work done

The way i thought of it, it is the same amount of work, The intel does it faster, but not because of speed, because of efficiency.

The thing is, Intel CPUs don't move less "dots", they move dots a shorter distance. Think of an Intel CPU as a direct route and an AMD CPU as a indirect route. AMD CPUs will move faster on the road(ex 3.8mph) and Intel will move slower(ex. 3.4mph) at a certain speed the AMD CPU "car" will move fast enough to offset the difference with the Intel CPU "car". But that's basically how it works. The Intel route is shorter, that's why its faster. But think if FX-8350 as an 8 lane highway and Intel as a 4 lane. If there are 8 cars(instructions) at the same time the FX will deliver more cars in the same time. If there are only 4 cars the Intel will. Make sense?

I'm not talking about moving dots. The dots analogy is there to highlight steps/efficiency, which is efficient because it's a shorter distance/more direct route with less dots (steps). And from the OP question, i don't think we're addressing multiple threads right now which is something totally different (and i know how it works). I think we're both explaining the same thing, but from our own understanding which lead to slightly different analogies and points of emphasis

 

[kirilmatthew]

This all comes down to instruction piplines. In an Intel processor it takes less instructions to accomplish something than the current crop of AMD processors because it has a much shorter pipeline. during each cycle of the processor Intel processors are much more efficient, to the point of being able to do more at a lower speed. Think of it like doing a dot-to-dot against someone, you can be doing yours 15% faster, but if you have 20 dots and they have 15, they'll finish first.

Kinda the opposite way around. They both have to accomplish the same work, but the Intel ones do it faster. I don't think your explanation really works too well 😛

It was the best i could come up with on the spot, you're just seeing it from a different perspective that i thought of it from:

Work to be done= 1 dot to dot.
Speed hand is moving=clock speed
number of dots=number of steps to complete the work done

The way i thought of it, it is the same amount of work, The intel does it faster, but not because of speed, because of efficiency.

The thing is, Intel CPUs don't move less "dots", they move dots a shorter distance. Think of an Intel CPU as a direct route and an AMD CPU as a indirect route. AMD CPUs will move faster on the road(ex 3.8mph) and Intel will move slower(ex. 3.4mph) at a certain speed the AMD CPU "car" will move fast enough to offset the difference with the Intel CPU "car". But that's basically how it works. The Intel route is shorter, that's why its faster. But think if FX-8350 as an 8 lane highway and Intel as a 4 lane. If there are 8 cars(instructions) at the same time the FX will deliver more cars in the same time. If there are only 4 cars the Intel will. Make sense?

I'm not talking about moving dots. The dots analogy is there to highlight steps/efficiency, which is efficient because it's a shorter distance/more direct route with less dots (steps). And from the OP question, i don't think we're addressing multiple threads right now which is something totally different (and i know how it works). I think we're both explaining the same thing, but from our own understanding which lead to slightly different analogies and points of emphasis

I think the car analogy addresses multiple threads to an extent.

 

[Jaxem]

This all comes down to instruction piplines. In an Intel processor it takes less instructions to accomplish something than the current crop of AMD processors because it has a much shorter pipeline. during each cycle of the processor Intel processors are much more efficient, to the point of being able to do more at a lower speed. Think of it like doing a dot-to-dot against someone, you can be doing yours 15% faster, but if you have 20 dots and they have 15, they'll finish first.

Kinda the opposite way around. They both have to accomplish the same work, but the Intel ones do it faster. I don't think your explanation really works too well 😛

It was the best i could come up with on the spot, you're just seeing it from a different perspective that i thought of it from:

Work to be done= 1 dot to dot.
Speed hand is moving=clock speed
number of dots=number of steps to complete the work done

The way i thought of it, it is the same amount of work, The intel does it faster, but not because of speed, because of efficiency.

The thing is, Intel CPUs don't move less "dots", they move dots a shorter distance. Think of an Intel CPU as a direct route and an AMD CPU as a indirect route. AMD CPUs will move faster on the road(ex 3.8mph) and Intel will move slower(ex. 3.4mph) at a certain speed the AMD CPU "car" will move fast enough to offset the difference with the Intel CPU "car". But that's basically how it works. The Intel route is shorter, that's why its faster. But think if FX-8350 as an 8 lane highway and Intel as a 4 lane. If there are 8 cars(instructions) at the same time the FX will deliver more cars in the same time. If there are only 4 cars the Intel will. Make sense?

I'm not talking about moving dots. The dots analogy is there to highlight steps/efficiency, which is efficient because it's a shorter distance/more direct route with less dots (steps). And from the OP question, i don't think we're addressing multiple threads right now which is something totally different (and i know how it works). I think we're both explaining the same thing, but from our own understanding which lead to slightly different analogies and points of emphasis

I think the car analogy addresses multiple threads to an extent.

I agree. It does address that at the same time and addresses it well. I was trying to keep my explanation to a single thread for simplicity's sake, but it could be expanded to encompass threads...like teams doing dot-to-dots. a team of 4 people doing 6 dot-to-dots at 115% speed will finish after a team of 8 people doing 6 dot-to-dots at 100% speed...though i have no idea why you'd have teams of people doing dot-to-dots...

 

[kirilmatthew]

In other words we understand, but its very hard to fully explain 😛

 

[Jaxem]

Exactly. I've taken my stab at it and hope PcMe has gleaned something on the subject. Let us know if you need more/better explanation.

 

[PcMe]

Exactly. I've taken my stab at it and hope PcMe has gleaned something on the subject. Let us know if you need more/better explanation.

Thank you for your answers. It is much appreciated. I'm glad I learned something new.

 

[kirilmatthew]

Exactly. I've taken my stab at it and hope PcMe has gleaned something on the subject. Let us know if you need more/better explanation.

Thank you for your answers. It is much appreciated. I'm glad I learned something new.

Glad we could all help somewhat!