Why isn't there any improvement in CPU speeds recently

[ali8]

Hello,

I want to know why didn't the CPU clock speed increase recently?

The farthest it reach, I think, is around 4GHz.

What I know is that current manufacturing technologies have reached 45nm-size CPUs.

So decreasing the transistor size will eventually hit a wall.

Also, in the atomic world, we can no longer ignore the effects of Quantum mechanics.

So are these the only reasons (if really they are reasons) ?

-Ali

 

[yadavvlsi]

Moore's law was on the number on devices that can be fabricated on a chip. While the speed with which a device can operate dependens on some other fators like its material, its dimentions (i.e distance between source,drain & metal layer). Its the device capacitance that plays major role in its speed performance.

As we know that : capacitance(C) = E.A/D less d means more C.

As by moore law if the device density is increasing it means its D is decreasing hance its leading to increasing C. More C means more leackgae , more power consumption and more noise prone device. Hance it becomes a challenge for designers to oprate a device on same clock frequency after scaling down it to by a factor of half. That's why more efforts are needed to make a faster device with the dimentions less than its previous version.
If you want to make a processor that can run on high clock frequencies then you need to make fast logic-devices/flip-flops etc.
Which in turn needs high speed transistors to implement these blocks. If you try to run slow devices on high clock speeds then data voilation will be there and the processor will stop working.

Now the device speed performance depends on various factors.

1. It should have less device capacitance (As i explained in previous reply)
2. It should be made up of material capable of transporting the cariers with high speed.
3. Device dimentions (Channel length etc).
4. Operating voltage.
..
.. Many More

All these fators needs more R&D in device physics as some of these factors are acting complementry to each other for eg.
if you will decrease the channel length and device dimentions to increase the speed of device. doing this will inturn increase the
device capacitance and cause the device to be slow again. Also the leackage through device will increase and make the device unstable. There are many more factors which are dominating nowdays when we are trying to make faster devices will more device
density(Trying to obey Moore's Law).

You can even make a processor of 10 GHZ or more with today’s available ---Si--technology at 45 nm node but why to do so .......???? when it is of no use...... we are shrinking the size not to increase the speed of processor now a days .... it is only for increasing the storage capacity (i.e Memories) for their storage capabilities and operation frequency......

Tell me what is the need of increasing the speed beyond 4 GHZ as you are not even able to utilize half of it? the limiting factor is the outer peripherals and interconnects which fetch and receive the data to Processor i.e Cash and RAM memories, and HDD. see at what clock frequencies they works on...... and you found that..... their is a drastic change in speed of these devices and Processor that you are using today. therefore these are not even competent enough to tackle the data at this speed.....

To understand it more clearly you may take an analogy to real world,
Imagine that you are going to drive a car from your home town to some city on national highway. You speed up every time you are out of city road and slow down while a city came in between on the road. This is because of the conjution in the town the traffic moves slowly but out of the town the traffic is less and you can move faster. But in case of processor it is other way round. Now consider the Processor as the cities on the road and interconnecting peripheral track on PCB and RAM memory and HDD etc. as roads outside the cities. Here the data travels fast in packed processor and move slows in the roads of interconnecting peripherals.
Thus the limitation is the operating frequencies of the interconnecting peripherals to the processor not the processor. It is true as if for today and for at least in future 20 years or so. This limits the useable frequency of the processor although somehow, this has been tackled with the piplining architecture approach while designing the processor but still not enough to completely utilize its capabilities of processor. You can achieve it by reducing the interconnect delays by changing them from electrical to optical fibers. This is true for interconnects at PCB as well as in ICs and Processor interconnects too………

 

[ali8]

Thank you so much, your reply is really helpful.

If possible, I want to ask why is the operating frequency of "the
other prephirals" still slow?

I mean, If we right now have the technogoloy to develop a table
selicon device that operates on, say, 2 GHz, then why did cpu manufacturers benefit from this while memory manufacturers
didn't benefit from this, as shown by the current memory speed
which is somewhat below 1 GHz ?

 

[yadavvlsi]

Cache (pronounced cash) memory is extremely fast memory that is built into a computer’s central processing unit (CPU), or located next to it on a separate chip. The CPU uses cache memory to store instructions that are repeatedly required to run programs, improving overall system speed. The advantage of cache memory is that the CPU does not have to use the motherboard’s system bus for data transfer. Whenever data must be passed through the system bus, the data transfer speed slows to the motherboard’s capability. The CPU can process data much faster by avoiding the bottleneck created by the system bus.
As it happens, once most programs are open and running, they use very few resources. When these resources are kept in cache, programs can operate more quickly and efficiently. All else being equal, cache is so effective in system performance that a computer running a fast CPU with little cache can have lower benchmarksthan a system running a somewhat slower CPU with more cache. Cache built into the CPU itself is referred to as Level 1 (L1) cache. Cache that resides on a separate chip next to the CPU is called Level 2 (L2) cache. Some CPUs have both L1 and L2 cache built-in and designate the separate cache chip as Level 3 (L3) cache.

Memory access time (speed) is limited by large capacitance and inductance due to long wires running on motherboard. Processor running at high speed requires large memory to process data faster. Large memory integration on chip is not feasible.

 

[ali8]

That makes sense!

Thank you