AQUARIAN

Of what is a processor made up?

Basically one can say that it is composed of a plate of Silicon.
The Silicon, noted If in table periodicals of the elements, at summer discovered scientifically in 1823 by Jons Berzelius Jacob. After oxygen it is the chemical element most abundant on the Earth.
It is today an essential element in electronics because of its characteristics, it is indeed, one of best the semiconductor.

It is contained in quartzose sand (dioxide of silicon, already known since antiquity, silica), and one extracts some in a pure form between 96% and 99.99999% by heating it at very high temperatures (about 1700°c approximately).
However in the use that makes electronics of it, and in particular our use, the silicon extracted from this treatment should not contain more than one impurity for thousand billion (10^12) atom of If
It thus undergoes several treatments, and one obitent of Silicochloroforme.On amalgamates then this Silicochloroform to obtain Silicon bars.

It is of these Silicon bars that one manufactures Wafers of our processors. The manufacture of Wafer is composed of several succésives, hard and expensive stages, from their numbers and their complexities.
All these operations entrainent a margin of error and differences. It is what explains the differences between the series of the processors, and which some of them would be more or less favourable with the overclocking. The differences between Wafers make it possible to work out more or less powerful processors (while taking for base that a faster processor will be more powerful, which in reality will not be true forcing) according to the purity of this one.


- What increases the temperature of a processor?

The first factor which influences the temperature, it is the frequency. As opposed to what one can believe, the fact of increasing the frequency of the processor increases the operating temperature of this one.
To include/understand why, it should be known that the TDP (Thermal Power Design) is calculable thanks to a formula, this formula contains a specific constant to each processor, the coefficient of adjustment.

If one connait the frequency, the tension and the TDP of a processor, one can calculate it.

That is to say:

P: Power to be dissipated
V: The tension
F: The frequency
K: The coefficient of adjustment

The formula to find the coefficient of adjustment.



One can thus deduce from it that:

AQUARIAN

Let us take a concrete example.

A given rhythm processor has 2400 MHz, having a TDP of 60 Watts and a tension of origin of 1.4v.
If one compares it with another processor, whose characteristics will be the same ones except the frequency, 2200 MHz for example, one calculates a coefficient of different adjustment for these two processors.

- Processor 1:

k1 = p (V ² *f)

k1 = 60 (1.4 ² *2400)

k1 = 0,012755102040816326530612244897959

- Processor 2:

k2 = p (V ² *f)

k2 = 60 (1.4 ² *2200)

k2 = 0,013914656771799628942486085343228

It is observed that coeffcient it adjustment of processor 2 is higher than that of processor 1.
Let us see now what occurs in several common cases from overclocking.


Overclocking case N°1. Increase in the frequency of 500 MHz.

By calculation,

P = (f+500) *V ² *k

processor 1 develops ~ 72.49 Watts
processor 2 develops ~ 73.63 Watts

Obervations: One can notice that the tension was not modified, and that however the processors dissipate more Watts. The temperature thus will increase.



Overclocking case N°2. Increase in the frequency of 500 MHz and the tension of 0.2v.

By calculation,

P = (f+500) * (V+0.2) ² *k

processor 1 develops ~ 94.69 Watts
processor 2 develops ~ 96.17 Watts

Obervations: One can notice that the processor with the lowest coefficient of adjustment dissipates less. And this, although its frequency of operation is higher. Note retain the importance of K here.


Overclocking case N°3. Increase in the two processors at a frequency of 3000 MHz with a tension of 0.3v.

By calculation,

P1 = (f+600) * (V+0.3) ² *k
P2 = (f+800) * (V+0.3) ² *k

processor 1 develops ~ 110.58 Watts
processor 2 develops ~ 120.17 Watts

Obervations: One notes a variation of 10 Watts between these two processors has equal frequencies/tensions. One can deduce from it, that more the difference between the coefficients of adjustment (and thus instantaneous frequency deviation, but not tension and TDP of origin) will be large, more the equal variation with overclocking will be tall.
A strongly overclocké processor will consume always more than one little overclock processor, has frequency/tension equal, it is what will not interest us the continuation.


- Why the temperature, and why necessity increases to provide a system with powerful cooling?

Each system of cooling in our use will have a thermal resistance.
Thermal resistance notes °C/W, the design and the composition of our system of cooling will determine its thermal resistance. More it is low and best cooling will be. It is its capacity to dissipate Watts produces by the processor, and more precisely, to bring its temperature closer to the ambient temperature. That is to say x°C/W increase in x°C per dissipated Watt.

It is noted that the elementary materials have a more or less raised thermal conductivity.
That is to say 401 W (m*K) (Watt by meter-Kelvin) for Copper, 429 W (m*K) for the Money, and 237 W (m*K) for Aluminium, these values are given has 20°c. The more this value is raised, the more the material is conductive thermically.

Also let us note that the values however above vary according to the temperature and from moisture, and that thermal conductivity is directly in connection with electric conductivity, for example, the conductivity of a diamond is close to the null one.

With similar design a Copper ventirad will be more éfficace that a ventirad in Aluminiu.

Two points:
it does not use money, or diamond (between 895-2300 W (m*K), because they are too much expensive, and that their propritétés, although strongly intéréssantes, would not allow much better results than those obtained with copper. Let us recall that the limiting factor is the temperature of the ambient air.
it ventilator which blows on the radiator, of watercooling, or the ventirad, is also an important factor. Its capacity has renouveller the air on the surface of the exchanger is a factor determining in the capacity of this one to evacuate heat. It as should be known as when the air velocity increases, it cooled.

In a given system, thermal resistance is defined by the ambient temperature and the temperature of the processor, and its load to be dissipated.

For a system whose thermal resistance is of 0.15°C/W, with an ambient temperature of 20°c and one 60 Watts has to dissipate, one will have a temperature of 9°c to the top of the ambient temperature,

either 29°c, or (0.15*60) +20 = 29.

With a load of 100 W, one obtains 35°c, but while trying to dissipate this same load with a less powerful system half, let us say whose thermal resistance is of 0.30°C/W, one obtains 50°C. That packs quickly in summer, with an ambient temperature of 32°c for example, that would give us 62°C.

This data is non-néglieable in a passive system for example, knowing that thermal resistance will be relatively high.

By increasing the load to be dissipated, one thus increases the temperature, by decreasing thermal resistance, one decreases the temperature (in certain proportions).
Only the systems with phase shift propose negative thermal resistances, however to speak about negative resistance is scientifically false.

AQUARIAN

- Why shouldn't a certain tension be exceeded when one overclock?


As we saw, the fact of increasing the tension causes to make increase the load to be dissipated. Out the silicum, which composes Wafers of our processors, as any semiconductor is more “éfficace” at low temperature which has élévée temperature.

When the temperature increases on the surface of the core (the heart of the processor), of the electromagnetic disturbances are more important. They pertubent the path of the current within this one, and of is done, the processor will generate errors beyond a certain margin, which one calls max Operating Temperature. When the frequency, or the tension exceed the tolerance level of the processor, it to create errors, it is instability. The operating system is put in protection, it is the blue screen (for Windows), or the freeze.

At a certain moment, one arrives at a wall such as the frequency ratio/temperature/cooling is too high so that the increase in the tension changes there though it is. Worse, it makes heat the processor, and makes it still more unstable. The only manner of finding a stability, is to improve cooling.

If, even with a processor of which all the adjustments are defined by defect, the temperature exceeds a certain threshold, this one will generate érreurs.

The second point which makes that the temperature influences the stablity of the processor is related to the resistivity. The resistivity of silicon, like known as higher, returns it less éfficace. It less lets pass the current, and once again, that creates érreurs. We will see low why the allied temperature perhaps invaluable in the rise in temperature.

- What is resistivity?

Resistivity of a material with for unit ohm.meter, i.e. the electric resistance which it develops per meter.
The resistance of a material is calculated graçe with the formula:




That is to say:

R: Resistance
p: Resistivity
L: The length in meter
S: The section in meter-square

The resistivity of materials depends on their temperature:

That of metals varies according to heat, one will be able to take for example the tungsten filament of a bulb, whose cold resistivity is lower than that hot. It is besides for that that it reddens (like any resistance of heat) and thus which it produces of the light. If it let as much pass the hot current that cold, it would not occur a light.
For the semiconductors, the resistivity diminiue strongly with the temperature. One speaks about supraconductivity (quasi null resistivity) when that the temperature of a semiconductor approaches the absolute zero (0 K/-273.15°c).

What interests us here, it is precisely the relation temperature/resistance/resistivity:
Within the framework of the overclocking, one with tendency to increase tension and frequency, and thus thermal load.
With traditional systems of cooling, one will never go down in lower part from the ambient temperature. The only systems able to go down under the ambient temperature are the systems with phase shifts.


Briefly, to include/understand their operations:
One with a closed loop which contains a diphasic element (freon), three key points in this system, compression, condensation, and relaxation. It is at the time of this relaxation, that one “creates cold”, éxactement, slackens it (liquid state with gas = Evaporation) of the freon, which intervenes, it intervenes at the time of a change of pressure, forces it to cool, and so it absorbs a certain quantity of heat in its environment.

AQUARIAN

This environement in question is an evaporator, which can be of two types:
- Posed directly on the processor, as a waterblock in Watercooling in the case of a DOD (Direct One Die) who applies directly to the processor
- Of helicoid form, in Waterchill, or it will come to cool a liquid (often alcohol, because it freezes has less than -100°c) which will be the coolant of a system with Watercooling.


Each gas with these proper properties of condensation and relaxation, most used in the “Phase exchange” is R290, R22, R134a, R404, R507, other gases with a temperature of lower evaporation (thus which “create more cold” can-to be used, but under certain conditions because it is necessary to be able to condense them, and that is at low temperature made, whereas the gases quoted before condense at temperatures close to that of the ambient temperature (count on 20°C), if one considers a pressure of 6/8 bars approximately. That is done with several stages, therefore several systems which use gases whose beach of use is increasingly low.

The systems on simple floors, Single Training course, are less complicated and less “powerful” that the systems on multiple floors, Cascade, (as a cascade which contains several stages) which allow temperatures close to -100°c in certain cases.
Another “simple” system more, consists in evaporating LN2, liquid nitrogen, in a tank, I do not speak about machine of war, but of a copper container generally which one poses on the CPU or GPU to cool it. Or that becomes intéréssant, it is that the LN2 evaporates with -196°c, that is to say only 77.15 K.

One observes that with sytèmes such as those, gone up in frequency is more important, that is always related to the resistivity of silicon. It diminiue enormously with such a temperature, moreover one will not exceed the temperatures where the processor will create errors.

Thus with same vcore but at lower temperature, a processor will gain in frequency compared to a traditional cooling, because one will improve the passage of the current in this processor.
However, any processor with its limit which is the éffet of skin (Skin Effect), which increases with the frequency.


- How is the frequency of one processor to the sale defined?

To include/understand, it should initially be known how one defines under which label one will sell a processor.

It should be known that one does not create a processor according to his needs, one creates a processor according to the results which it gives to a given tension.

, One creates a processor which will hold such or such frequency, and if one bases oneself on only one and even tension (it is generally the case) it will often be necessary to test hundreds of processor before finding of them one which will hold a high frequency with the same tension as the other processors of its range. That costs money, and thus these processors are expensive and “rare”.

Only, there is always a more or less important margin that the manufacturer asserts itself to sell his processors on the market. It is for that that, even if a certified processor with 2.4 ghz and a tension of 2.0v can turn to 3.2 ghz with this same tesnsion, a processor which is certified to him with 3.2 ghz and 2.0v will go up theoretically more in final frequency.

The processors are then noted according to a code which contains the week of production, and of another figures related to the production.
It is the stepping, the code that the overclockers seek because the processors which it door will have a propention with the more or less marked overclocking. However there will be always a difference between two processors carrying a similar stepping because the quality of engraving will be never the same one.

The goods steppings will be defined by:

stroke it of certification, plus it is important, more one is likely to go up in frequency without handling the vcore.

- Need for the market, one certifies processors at a frequency lower than that presented in the tests to be able to face the market, it results from it from the processors “masked” under a appelation lower than that for which they were intended.

- Support, one attaches a processor to limit the blows of production, one limits his functions and it is sold less expensive. One can take for example the duron 1.6 ghz, which was in fact Athlon to which one had cut down part of the L2 mask.

This file is not finished, and I will add there questions/answers, I would ask you to announce me the errors whatever they are.

It is necessary also that I replace the formulas by more readable images.

AQUARIAN

Source:Techarena.in