In this article, we discuss the reasons for excess power consumption of Intel processors, which is one of the most talked-about issues in the computer world.
The problem of high power consumption continues in the 10th and 11th generations since the 9th Gen Intel processors, and it is a very talked about subject in the computer world. Intel’s latest processors still draw an average of more than 100W TDP, and we often see some processors exceeding incredible values like 180W . In this article, we will explain why this is the case, why Intel processors still need such high power consumption.
What is TDP (Thermal Design Power)?
Just before we start to explain the reasons for the high power draw of Intel processors, it will surely make the subject more understandable if we mention what TDP is , which we often hear .
Intel guarantees a certain frequency rate at a certain power for each processor by taking into account warming. This is an application not only made by Intel, but by all companies interested in semiconductor production. In order to decide that all processors from a factory are suitable, a model-based standard must be determined.
As you know, because the processors use silicon structures and electrical switching logic too much, they get very hot. If the heat collected on the processors is not dissipated, the processor will not be able to function properly, so it must be cooled. Many computer enthusiasts naturally consider the TDP value to be the same as the maximum power draw , since they realize that the ratio of heat to be dissipated and the power drawn by the processor are almost the same . But normally the TDP rate represents a certain power.
To put TDP in its simplest form, we can say a value related to the ability of a cooler to dissipate heat. In other words, it refers to the CPU cooler capacity required to achieve the performance guaranteed by the manufacturer. That is, if you install a cooler with 85W TDP support to a processor with 95W TDP value, you will not get the promised performance because the heat on the processor cannot be removed sufficiently.
Considering that a little more heat is absorbed with the help of the socket and motherboard, we can technically say that the actual heat value may be lower than the TDP value. However, in many places TDP and power consumption are mentioned together; TDP = How much power a central processor draws under load.
In fact, it is a value that can be adjusted and manipulated via TDP firmware. If a processor uses TDP as the limit value expressing the maximum power, if we compare the performance values achieved by the processors at various power values, we will encounter a table as follows.
This visual summarizes Intel’s definition of TDP for the last 3-4 years . In the case of any processor, Intel guarantees the base frequency for a given power and a specific TDP value. For example, a 65W processor such as the i7-8700 , which can reach 3.2 GHz base frequency and 4.7 GHz turbo frequency, only runs at 3.2 GHz, which means 65W and below are guaranteed. Any performance level above 3.2 GHz is not guaranteed by Intel.
In addition to these core values, Intel also takes Turbo into account. As you know, a processor like the Core i7-8700 can reach a turbo frequency of 4.7 GHz by drawing more power instead of running at 3.2 GHz. But for a processor like the i7-8700, the turbo frequency guaranteed by Intel and valid for all cores is 4.3 GHz . What is very interesting and confusing is that none of the turbo modes fall to the fundamental frequency.
In other words, this means that your processor can always work above the TDP value, the 65W cooler you purchased, even the processor itself, may be insufficient and cause a bottleneck. In this case, you should remove the coolers that meet the same factory or pure TDP value and throw them aside in order to get higher performance and take full advantage of the processor.
Yes, most of the manufacturers do not tell you this fact. Nor will they say . When there is insufficient coolant for your processor in the turbo, your processor will reach its temperature limit by limiting its power and only offering you a certain speed instead of real performance. In other words, your processors cannot fully exhibit their real performance because the cooler is not at a level to meet this.
So Is TDP Just a Meaningless Value?
While the usage of the term “TDP value ” has not changed much in the last 10 years , the way the processors use power has changed a lot. Since the introduction of six- and eight-core, and sometimes even ten-core, end-user-targeted processors that can exceed 4.0 GHz, we see that our processors trying to cope with heavy workload have exceeded the said TDP values.
We know that most of the quad-core processors in the past years had a TDP value of 95W, but even with the turbo turned on, it only draws around 50W at full load. As the cores are added, something needs to be given without changing the TDP value.
Hidden Numbers Do Not Write on the Box
Intel defines several power levels for each processor to match its capabilities and processing capacity. This sounds pretty good, but these power and capability specifications mean that OEM PC manufacturers can tailor the processors to how they should perform. This creates a huge uncertainty as to what the exact power consumption will be when a processor is finally plugged into the system.
To simplify this point, you should pay attention to three main numbers. Intel names these numbers as PL1 (power level 1), PL2 (power level 2) and T (also known as Tau).
PL1 value is the long-term constant power consumption value expected from a processor. PL1 is generally defined as the TDP value of the processor, according to all modes of operation . So if the TDP value is 90W, the PL1 value is 90W.
PL2 refers to the maximum power a processor draws in the short term. This number is much higher than PL1 anyway, and when faced with high-performance workloads, the processor switches to this mode, allowing the turbo to draw power only up to PL2 value. Turbo never works in PL1 mode.
Tau can be called a timing variable. Tau determines how long a processor stays in PL2 mode before switching to PL1 mode. Note that Tau is a value that depends on power consumption and processor temperature. If the processor reaches its thermal limit, an automatic adjustment to a different low voltage is expected to occur.
Intel’s official processor power specifications are as follows.
Based on this, we can say the following.
The processor first starts up in PL2 mode. If single thread is used, it should reach the highest turbo value on this list. In normal times, the power consumption of a single core cannot be close to the PL2 value of the entire processor package . Our processor cores currently work load to cope with the Intel core determined by the head of the turbo tries to respond appropriately to the value of reducing the turbo frequency. When the CPU package’s power consumption reaches PL2, the frequency is set again. Thus, the PL2 value is never exceeded.
If the system is heavily loaded for too long , the firmware will switch to PL1 to limit power thanks to the “Tau” . Turbo values are no longer used because the turbo only works in PL2 . If the workload will cause power consumption above PL1, the frequency and voltage values are adjusted so that the processor’s total power consumption value is the same as PL1. Thus, power consumption during the workload is pulled from PL2 to PL1. At the same time, the temperature of the processor has been reduced and it is provided to have a longer life.
The PL1 value remains the same until the workload is removed from the processor and one CPU core reaches an idle state (typically under 5 seconds). If another workload is to be applied, the system may necessarily switch back to PL2 mode.
Here we can start from the values that Intel lists on its website. Take an i7-8700K processor for example. We know the following values for this processor.
- PL1 = TDP = 95 W
- PL2 = TDP * 1.25 = 118.75 W
- Tau = 8 seconds
Based on these values, we can say that the system consumes approximately 119W of power for 8 seconds before switching back to 95W . Intel has applied this technique in many new generation processors. Therefore, the power consumption for the entire processor, even at full load, was often underestimated, as it was well below the PL1.
This is exactly where it gets to be extremely silly. Motherboard manufacturers are also involved in power tuning. Because PL1, PL2 and Tau values can be adjusted on the firmware. For example, as you can see in the graphic above, you can set PL2 to an unlimited value and PL1 to 165W or 95W.
Random Numbers Confusion
To be honest; PL1, PL2 and Tau values are carefully checked and calculated in mobile devices such as laptop computers where there is a very limited thermal situation . It was also spoken by many tech publications at the time that a number of top laptop manufacturers have set the PL2 value to TDP. In this way, as long as the core loads of the processors do not exceed TDP, it is possible to work in more turbo.
However, in all processor reviews, since the introduction of six-core processors, we have generally seen much higher power consumption than PL1 and PL2 values . And these values usually remain so until the processor reaches very high temperatures. So why is this so?
Most of the modern BIOS found in all today’s motherboards now have specific power limits that motherboard manufacturers offer them. Long-term power limits and short-term power limits are examples of this. These are mostly adjustable. Many users do not know exactly what these are when browsing the BIOS, so they choose the “recommended auto” settings and continue.
Automatic settings are actually not fully automatic. It has been tested by the motherboard manufacturers in accordance with the relevant processor model and then some values written into memory are used in “automatic” settings. In this way, the PL2 value and the Tau value can be adjusted to any infinite value depending on the motherboard manufacturer. This means that the processor will run in turbo mode for all days and weeks as long as it does not reach thermal limits.
But why do motherboard manufacturers do this? Every manufacturer has an appropriate answer or excuse to give this.
First of all, it may be because users think that they always want to use turbo, and that they will want to benefit from turbo for all processor cores every day and always. This comparison can often be due to the fact that the test scores of the processors are seen to have peaked or the higher numbers appear better when comparing numbers among users.
The second reason is that the product is designed accordingly. As Intel generally releases every new processor, it places some default features on motherboards that are expected to function properly in a given lifecycle. Manufacturers can thus make some adjustments to regulate power efficiency without exaggerating. For example, if the processor on a motherboard can always turbo all cores, why not do it?
The third is at least for expensive models aimed at many computer enthusiasts. Computer enthusiasts who will buy high-end motherboards and processors already know that they will use a high-level cooling with them. So, for example, if the processor goes above 160W but the user can provide enough cooling, why not experience a high all-core turbo experience? The standards set by Intel itself have been defined at length under the coolers recommended by Intel.
So What The Right Way? Who Can Be Trusted?
Intel sets a certain standard for every processor it produces. PL1, PL2, Tau, and motherboard circuitry and firmware settings all use Intel recommended default values. Some of these are public, as published by Intel, and some are carefully kept secret . These are still recommended values. Regardless, at the end of the day, motherboard manufacturers do what they want.
This makes it very difficult for tech publishers to test the hardware. It is important to note that the stock tests in almost every processor review are run out of the box at values obtained or “recommended by Intel” .
|Unlimited||4096W (Unlimited)||999 seconds (Unlimited)||4096W (Unlimited)||%one hundred|
|Intel, 165W||207W||8 seconds||165W||98%|
|Constant 165W||165W||1 second||165W||94%|
|Intel, 95W||118W||8 seconds||95W||84%|
|Constant 95W||95W||1 second||95W||71%|
Recently, some motherboard manufacturers are said to have changed their strategy regarding PL1 / PL2 / Tau values and set the Tau to a value of 30 seconds. When users make a comparison while using these motherboards, the values they see may appear lower than they are used to. Although this is true and in line with the values announced by Intel.
The main problem is that many motherboards that say they are running at “auto” settings often don’t tell you what values are set in the background. This makes it very difficult to report power and temperature values.
As we have stated at length in the article, the only reason why Intel processors have a high TDP value and often work on these values is also due to motherboard manufacturers who completely manipulate the values inappropriately .
Does Intel have no share in this? Of course there is. As long as the processor manufacturers do not adjust their processors to be used only at the appropriate value, users will continue to be victimized and oil will be put on the hands of motherboard manufacturers who run improper hardware.
We think Intel should set a peak value for PL2 and some continuous values for PL1 to prevent this. Thus, many problems arising from heating and power consumption will be solved by themselves, as the power to force the processor to the fullest cannot be given.