Noob question about processors

[JanuS-1995]

I am fairly new to the world of CPU's, Motherboard and Memory and I have a question. This is probably a noob question but if someone could answer it I would be very happy.

 

When two CPU both have 4 cores and run at the same speed. Why isn't it obvious that they deliver the same FPS's in games or get similar scores in benchmarks?

 

Again, this is probably a noob question but I still want to know the answer

[Godlygamer23]

DIfferent graphics card or different CPU architecture.

[Vitalius]

I am fairly new to the world of CPU's, Motherboard and Memory and I have a question. This is probably a noob question but if someone could answer it I would be very happy.

 

When two CPU both have 4 cores and run at the same speed. Why isn't it obvious that they deliver the same FPS's in games or get similar scores in benchmarks?

 

Again, this is probably a noob question but I still want to know the answer

Different architecture's, an architecture is how the CPU is designed and laid out, what resources it has, etc etc.

Some are more efficient then others.

Then there is die size. Some are 32nm, some are 40nm, and so on (there are smaller ones too). That's how much area the die of the actual CPU is IIRC, but I could be wrong.

@Glenwing.

[Lord Pantaloons]

Because core count and efficiency of the CPU's architecture matter just as much if not more so than actual clock speed.

[Godlygamer23]

Then there is die size. Some are 32nm, some are 40nm, and so on (there are smaller ones too). That's how much area the die of the actual CPU is IIRC, but I could be wrong.

Actually, what you're referring to is the transistor lithography. That's the individual size of the transistors. So when Intel says their lithography is 14nm, they're saying each transistor is 14nm.

[TheSLSAMG]

CPUs are different when it comes to efficiency. For example, an Intel Core i5 4670K crushes an AMD FX 4300 or an AMD Athlon X4 760K because the Intel's cores are more efficient and powerful.

[JanuS-1995]

Wow, thanks for the fast replies. Didn't thought of the architecture.

 

Actually, what you're referring to is the transistor lithography. That's the individual size of the transistors. So when Intel says their lithography is 14nm, they're saying each transistor is 14nm.

But does that mean that the die size is not important if you compare two CPU's?

[Guest]

I am fairly new to the world of CPU's, Motherboard and Memory and I have a question. This is probably a noob question but if someone could answer it I would be very happy.

 

When two CPU both have 4 cores and run at the same speed. Why isn't it obvious that they deliver the same FPS's in games or get similar scores in benchmarks?

 

Again, this is probably a noob question but I still want to know the answer

Everything is different

[SMURG]

The simple analogy for this is that 4 oxen will pull a plough better than 4 chickens. Not all CPU cores are equal - in fact they can have vastly different features and work in really quite different ways. AMD's latest CPUs for example use cores that are divided into modules with 2 cores in each, whereas Intel's cores operate independently and can even have hyperthreading which provides double the logical threads - essentially allowing the CPU to queue a second task before the first is completed, thereby minimising idle time and improving performance.

Frequency simply denotes how fast the transistors are able to cycle, it doesn't give any insight into how tasks are dealt with or how computation is done.

[TheArium]

Yes, I can confirm what the others are saying. Not all cores are created equal. This means that you cannot just compare CPU's and see which is faster. You need to see actual benchmarks to determine which is faster in a given scenario. You can only compare two (or more) CPU's if they use the same architecture, but to be sure, find a review of the product.

 

I hope this makes sense, otherwise feel free to ask. 

[Godlygamer23]

Wow, thanks for the fast replies. Didn't thought of the architecture.

But does that mean that the die size is not important if you compare two CPU's?

Let's say you take two CPUs. They are both the same size in reference to the die, but CPU B uses smaller transistors. This means that more transistors can be crammed into the same amount of space.

[Lays]

Let's say you take two CPUs. They are both the same size in reference to the die, but CPU B uses smaller transistors. This means that more transistors can be crammed into the same amount of space.

 

Or like potatoes. The more potatoes you can fit into the bag

 

:]

[Glenwing]

Basically a "core" is a generic term for a cluster of transistors that works together on a calculation.  There is not a defined amount of performance that a "core" gives.  Some cores are more powerful than others, it is about transistor count, layout, design, and many other things.  But some applications just have a ton of tiny calculations, so benefit from a larger number of weaker cores (see: GPU.  Approaching 3000 cores.) which are not powerful, but can do many independent calculations at the same time.  So performance depends on the type of application too; Intel CPUs usually win in games with a strong quad-core against an AMD weaker 8-core, but the AMD CPU will beat the Intel handily in video rendering and CAD.

 

The fabrication process (45nm, 32nm, 28nm, 22nm, etc.) refers to the distance between the gates on each transistor.  It isn't a direct measurement of the size of the whole transistor but smaller gates usually mean smaller transistors as a whole, the larger measurement just isn't published.  When transistors switch, which is how they perform operations, voltage is applied across the gate (or removed from it), causing current to flow (or stop flowing) between the source and drain terminals (bear with me).  That current flowing or not flowing constitutes a binary 1 or 0 signal, and lots of transistors signalling lots of 1's and 0's creates digital communication.  A smaller gap between the gates means the voltage switch can occur more quickly, and the current flow or cutoff can occur more quickly, and so digital signalling can occur more quickly.  It also means less voltage is needed to get current to flow, reducing power consumption.  On a larger level, it means the transistors are smaller.  Which means you can either fit more transistors in the same amount of space, or fit the same amount of transistors in a smaller space, which means every silicon wafer will yield a larger number of CPU dies in the same area, which means you can manufacture more CPUs with the same resources, decreasing costs.

 

If that was all gibberish to you I can rephrase  :rolleyes: I get carried away sometimes...