Recently, I have heard that AMD has released the new Vishera series of FX processors which run at 5 GHZ. My question is whether there is any upper bound of clock rate of a processor? That is can we keep increaing the clock rate for ever? What electrical problems will we face at higher clock rates?
Asked
Active
Viewed 7,986 times
2 Answers
12
EDIT: This question led to a long discussions. It is crucial to understand that the fact that CPUs speeds haven't been increasing over the last years is related to commercial aspects, and not directly related to any engineering or physical problem. You can check this link for the topmost frequencies achieved with existing CPUs by overclocking and supercooling.
From the invention of the first PC and until early 2000's the main parameter of each CPU was its frequency (maximal frequency of operation). Manufacturers tried to come up with new technologies which will allow for higher frequencies, and chip designers worked very hard to develop micro-architectures which will allow to the chip to run on a higher frequency.
However, as chips became smaller and faster, the problem of heat dissipation arose – when the whole amount of heat generated by switching transistors couldn't be dissipated, the chips got damaged. Engineers started to attach heat sinks to processors, then fans, but eventually they concluded that the approach of increasing CPU's frequency is no longer practical in terms of added performance per added cost.
In other words: CPU frequencies can be raised, but this makes CPUs (in fact, not the CPUs but the cooling mechanisms) too expensive. Consumers won't buy expensive computers if there is an alternative.
In general, current technological processes allow very high frequency operation (way above ~3GHz which Intel usually uses, and even AMD's 5GHz is not the ceiling). However, the associated cost of cooling devices which are required at these high frequencies is too high.
I'd like to emphasize this: there is no physical effect that prevents development of 8-10GHz processors with current technology. However, you'll have to provide a very expensive cooling mechanism in order to prevent such a processor from burning out.
Moreover, processors usually work in "burst" – they have very long idle periods, followed by short, but very intensive (and therefore high energy consuming) periods. Engineers could build a 10GHz processor that works at the highest frequencies for short periods of time (and no additional cooling is required because the periods are short), but this approach was also declined as worthless (high investments in development as compared to questionable gains). However, following future micro-architectural improvements, this approach may be reconsidered. It is my belief that this 5GHz AMD processor does not work constantly at 5GHz, but raises its internal clock to a maximum during short bursts.
PHYSICAL LIMIT: There is a physical limit to a maximal achievable clock rate for each process technology (which depends on technology's minimal feature size), however I think that the last Intel's processor which had been really pushed to this limit was Pentium 4. This means that today, when the technology advances and the minimal feature size is reduced (meanwhile in accordance with Moore's law), the only benefit from this reduction is that you can fit more logic into the same area (engineers no longer push CPU frequency to the limits of the technology).
BTW, the above limit can't increase forever. Read about Moore's law and the problems associated with its further application.
-
So, you say that according to Moore's law, since the number of components will increase, we will soon run out of space for an efficiently cool and functioning processor? – Torsten Hĕrculĕ Cärlemän Jul 21 '13 at 12:06
-
@AnuragPallaprolu, I don't know what do you mean by "will soon run out of space". We have already crossed the point where the increase in frequency (which is possible) is not commercially beneficial. However, I don't believe that heat dissipation issue will reduce the operating frequencies in the future - while the number of transistors is going up, the active power dissipated by each transistor is going down. We can discuss the implication of sharp rise in leakage power, but it is not the topic of this question. – Vasiliy Jul 21 '13 at 13:20
-
Won't other components be affected by higher frequencies? They too should clock up to synchronize right? – Torsten Hĕrculĕ Cärlemän Jul 21 '13 at 14:09
-
Not necessarily. Modern CPUs use many clocks (internally). Some are faster, some are slower. There are solid solutions for synchronization problems (commonly known as Clock-Domain-Crossing problems). – Vasiliy Jul 21 '13 at 14:19
-
@AnuragPallaprolu, I edited my answer. It is the most complete answer I can give. I even found something new to myself - there are world records monitored for CPUs frequencies. You can see there that even the processors which weren't designed to work at 8GHz can reach these frequencies. – Vasiliy Jul 21 '13 at 14:25
-
Thanks a lot mon cher ami @VasiliyZukanov! I have understood. Marking as answer! – Torsten Hĕrculĕ Cärlemän Jul 21 '13 at 14:26
-
I suspect a major reason CPU speeds aren't being pushed to the limits is that because of caching effects, it's more cost-efficient to use a larger number of slower cores than a smaller number of faster ones. The limitation isn't with CPU speeds per se, but rather with the fact that improving the execution speed of a CPU by a factor of a billion wouldn't make programs run a billion times as fast, but would instead increase the fraction of its time the CPU wasted while waiting data to be moved to/from memory. – supercat Feb 21 '14 at 20:11
-
@supercat, 2x clock speed never-ever gained 2x performance. What you say sounds reasonable, but it was also reasonable 15 years ago - still clock frequencies rose. Furthermore, today's cache schemes are very sophisticated and attain hit rate way above 90% - only small fraction of load/store instructions result in accesses to off-chip memory. – Vasiliy Feb 25 '14 at 06:10
-
@Vasiliy: What do you mean it "never" gained 2x performance? Switch a Commodore 128 from 1.0227MHz to 2.0454MHz and all processor-based operations will be twice as fast. Performance will actually slightly more than double, since the timer-tick overhead will be cut in half. Cache hit rates have improved with time, but cache-miss penalties have increased too. If one has a 99% cache hit rate, but a 100-cycle penalty on a miss, one will spend as much time waiting for memory as doing everything else. Doubling the "everything else" speed will only improve performance by 33%. – supercat Feb 25 '14 at 13:39
-
@supercat, even for this old computer doubling CPU frequency will not lead to doubling in performance. There are many reasons, but just think about the speed of the floppy drive - it is independent of CPU speed, therefore the latency associated with reads from floppy will not scale. Anyway, the question is not whether 2x clock speed leads to 2x performance. What you are saying is correct, but it is not the reason for not increasing CPU speeds. – Vasiliy Feb 25 '14 at 14:14
-
@Vasiliy: On machines which lacked DMA hardware for their floppy drives, doubling CPU speed could likely have increased floppy speed by at least double. My point is that on newer machines, the benefits from doubling CPU speed would be *insufficient to justify the cost*. That is in some measure a result of increasing costs, but also a result of decreasing benefits. – supercat Feb 25 '14 at 14:49
-
@supercat, this is exactly what I wrote in my answer - it is possible, but is too expensive for mass production. I'm glad we agree on this point ;) – Vasiliy Feb 25 '14 at 15:23
1
There are phyical limits.
The processor frequency is limited by:
- the speed of the electric current (for example in copper)
- the switching speed of the transistors
- the size of the processor
Let's say you have a multiplier and a register on the CPU. Some input variables are multiplied and then stored in the register.
The electrical signal needs time to travel through the signal lines and transistors.
If you increase the clock rate too much then the multiplication has not finished when the next cycle sets in. And you might want to use the result of the multiplication in the next instruction!
So if the CPU is smaller then you can put a higher frequency on it.
Also see: Propagation delay Interconnect bottleneck
zomega
- 607
- 1
- 6
- 11