Intel’s TDP Shenanigans Hurts Everyone

Every time we come round to talking about laptop power consumption on a new generation of Intel processors, we find that the goalposts have moved. The physics of power consumption and cooling are always the same (in this universe), however we find that the marketing focal point of Intel’s power consumption has shifted yet again, causing more confusion than I believe is warranted. Tiger Lake, in this case, is no exception.

For Tiger Lake, Intel is offering two sets of processors, as with previous generations. These used to be called the U-series, based at 15 watts, and the Y-series, at 4.5/7/9 watts, however for Tiger Lake the U and Y designations will no longer be used. This time users will have to fully understand Intel’s product SKU list in order to make sense of it. Good luck, Grandma.

What used to be called the U-series is now, from the perspective of engineering rather than marketing, known as the ‘UP3’ product packaging. These UP3 processors are identifiable in two ways: first, the processor name has a ‘5’ before the G, such as Core i7-1185G7; second, by the TDP range of 12-28 watts.

This time around Intel is not giving the historic U-series a fixed TDP value, but instead giving it a range of values, stating that ‘TDP is a poor metric’ for the type of systems these chips go into. Despite this statement, in the next breath Intel goes ahead and quotes the base frequency of the UP3 processors as the peak 28 W value.

For anyone untrained in reading Intel product names, one might easily be mistaken when comparing the previous generation Ice Lake to the new Tiger Lake parts. Suddenly the typical ‘U-Series’ 15 W processor, such as the Core i7-1065G7, was at 1.3 GHz, but is now replaced by the Core i7-1185G7, with a base frequency of 3.0 GHz. The names are so similar, and there’s only a one generation gap, so wow, that’s an amazing jump in one generation! Right? Right??

To make matters worse, the values for the 28 W base frequency are hard-coded into the processor string in the products. For anyone looking at the CPUID for one of these processors, that 28 W value is what they are going to see, even if the product is a fanless notebook running at 12 W. To give some credit, Intel in its online database also provides the base frequency at the lowest TDP value as well. But this is not hard-coded into the processor string like the 28 W value is.

One might forgive Intel if they also provide the 15 W value for these new processors in the online database. This would allow enthusiasts and key influencers to understand the product portfolio in an apples-to-apples comparison over the previous generation in this U-series bracket. However, this is not provided. It was not until I had obtained my review sample, found the value myself, and then returned the sample that Intel gave me this value for the single processor I had tested. This is how far the company seemingly wants to go to ‘remove’ this concept of TDP and power consumption.

So what do we get if we look at Intel’s own processor database?

  • Base Frequency at the lowest PL1 value
  • Base Frequency at the highest PL1 value
  • Maximum Turbo Frequency at a not-given PL2 value

The metrics that Intel leaves out include:

  • The base frequency at the U-series 15 W value (for UP3)
  • The PL2 value for which the maximum turbo frequency is defined
  • The number of cores that the maximum turbo frequency is valid
  • The turbo frequency when all cores are loaded

Intel does sometimes, when products are launched, provide that final value of an all-core turbo. However users will have to be looking at coverage in places like AnandTech at the time, as it is not on the product page at Intel.

The other three values listed as not given are not provided to us, even when requested, with the common response being ‘these values are proprietary’. This is despite the fact that most of these values can be easily discovered by having the hardware at hand and probing the relevant output. It is unlikely that one outlet will have one of each processor to discover this data and provide a table, and even then that only matters if a user can find it, as it is not on Intel’s website. From my point of view, and I’ve told Intel this countless times, it would offer a more complete picture (and a better marketing message that Intel cares for its enthusiast audience that cares about this) if the data was provided in full.

If that mish-mash of data didn’t confuse you, how is this for completeness. Intel also announced the Y-series equivalent processors, which have UP4 style of packaging. Intel listed these products as having a TDP range of 7 watts to 15 watts, and the online database has the base frequency of these values. But in the product launch specification tables, to add insult to injury, Intel provided the base frequency for these processors at 9 watts, allowing for an apples-to-apples comparison. Go figure. It can be understandable when there’s some inconsistency in detail between products from different business units under the same company, however these two client processor groups couldn’t be closer together, given they’re cited in the same table on the same slide of the same presentation.

(Sometimes things like this are caught during press pre-briefs and fixed for the public announcement, however Intel decided not to pre-brief any press this time around.)

A Recap on TDP, PL1, PL2

For any new readers, that previous section can sound very complex. Power consumption, at a high level, should be a very simple thing to discuss. Power goes in, then power goes out - it used to be as simple as this. However to extract the most performance out of the hardware, as well as focusing performance in areas where it is more vital (such as turbo), has made it complex. Discussing elements like power draw and thermal design points has been an iterative process over the last decade. For those who haven’t delved into the topic before, here’s a short summary, mostly with respect to Intel.

 

Intel’s definition of TDP, or the thermal design point, does not mean the peak power draw of the processor as it does with some of the competition.

The thermal design point of an Intel processor is a measure of the expected cooling required for that processor for it to perform within normal operation at its base frequency with a high load. The TDP has the units of watts, for the amount of joules of energy per second that needs to be removed from the system to maintain normal operation.

Because TDP is in watts, the TDP is often equated as the power draw of that processor. This simplification is, for our purposes, a valid identity to make, as we’re dealing with the cooling a processor needs.

With that in mind, Intel defines the TDP as the power draw of a processor at a guaranteed minimum frequency with a sustained highly demanding workload within a suitable environment. What that means is that if you have a compute heavy workload (like rendering) and as long as your system isn’t somewhere abnormal (such as in an oven or at the north/south pole), the processor will not go above the TDP value for power consumption as well as offer the minimum frequency stated with that power - if it does not, Intel will honor its warranty and replace it.

Normally we refer to Intel’s TDP rating as the ‘sustained power draw’ for the given base frequency. It is sometimes referred to as the ‘power level 1’ or PL1 of that processor. So, for example, the Tiger Lake processor we are testing today supports a range of TDP values, such as 12 W, with a given minimum frequency, in this case 1200 MHz at 12 W. Because Tiger Lake offers a range of TDP values, it can get complicated - we will cover this in a bit.

Alongside the PL1 value, these processors also have a PL2 value. This is commonly referred to as the peak power consumption while the processor is running in a turbo performance mode, and the value of PL2 is higher than PL1. Intel rarely quotes an official value for PL2 when announcing a product, but it always announces the frequency associated with PL2, which is a maximum turbo frequency. Because the turbo mode is a limited time mode, these frequencies are reached during user-experience events, such as touching a screen, loading software, or even on key presses. In between these user-experience events, the system often moves out of the turbo mode to save power, and it can do so within 1/60th of a second.

The frequency associated with this PL2 is thus a limited time frequency, rather than a sustained frequency, and can also be limited by how many cores it can apply to at any one time. This is also the frequency that is listed on the box or laptop, alongside the PL1 value, which some users feel is not a genuine reflection of sustained performance.

PL2 is also used in the initial phase of a high-performance workload. Aside from thermal considerations, Intel also has a metric called Tau, in seconds, which is a measure of how long the PL2 turbo mode can be sustained for a ‘typical high performance’ workload. Thus if a PL2 is rated at 50 W, with a Tau of 10 seconds, then the ‘turbo budget’ is 500 joules of energy. If the workload is more strenuous than Intel’s metric, then the actual time for turbo may be shorter as that 500 joules budget is used. The budget is replenished if less than PL1 is needed (the actual calculation is more complex, based on an exponential weighted average time window, but has the same principle).

To summarize:

  • PL1, or TDP, is the sustained power draw for a given minimum frequency
  • PL2, or turbo, is the peak power draw for a maximum frequency under a turbo mode
  • A turbo mode can either be a user-experience event, or the initial phase of a high-performance workload
  • Tau is a time measure for how long PL2 can be sustained in a high-performance workload
  • Tau is based on a typical high-performance workload defined by Intel

As noted before, the PL1 of a processor can be a range of values. When the processor goes into a product however, it is set to a fixed value in that range. This value is often determined by the type of product (thin and light notebook, vs gaming notebook, for example).

Beyond this, Intel notes that its suggested values for PL2 and Tau are only recommendations. Laptop manufacturers are free to choose whatever values for PL2 and Tau they believe are sufficient for the system they are building. As we have seen in some fanless designs, PL2 and Tau can be incredibly small, whereas in a desktop, Tau could be effectively infinite.

Tiger Lake Metrics

How does this play into Tiger Lake? For this review, we have the Core i7-1185G7. Intel gives us the following values on its online database:

  • At 12 W TDP, the Base Frequency guaranteed is 1200 MHz
  • At 28 W TDP, the Base Frequency guaranteed is 3000 MHz
  • In a turbo mode, the peak frequency will be 4800 MHz.

A few more details were given at the time the processor was launched:

  • The peak frequency of 4800 MHz is valid only when one core is loaded
  • The peak frequency is 4300 MHz when more than one core is loaded
  • The maximum power draw in the turbo mode is ~50 W*

*It should be noted that the last figure was not given directly, but through analysis of power consumption data provided in the slide deck accompanying in the launch. The value of 50 W is valid regardless of what TDP mode is in play.

From testing the Core i7-1185G7 we have, we can also provide the following data point:

  • At 15 W TDP the Base Frequency guaranteed is 1800 MHz

Unfortunately we have no insight into the other UP3 Tiger Lake processors. 

Cache Architecture: The Effect of Increasing L2 and L3 Power Consumption: Comparing 15 W TGL to 28 W TGL
POST A COMMENT

252 Comments

View All Comments

  • tipoo - Thursday, September 17, 2020 - link

    “Baskin for the exotic”
    I see what you did there...
    Reply
  • ingwe - Thursday, September 17, 2020 - link

    I didn't get it until I read your comment. Reply
  • Luminar - Thursday, September 17, 2020 - link

    RIP AMD Reply
  • AMDSuperFan - Thursday, September 17, 2020 - link

    "Against the x86 competition, Tiger Lake leaves AMD’s Zen2-based Renoir in the dust when it comes to single-threaded performance." - But I am hoping Big Navi can compete well against this Intel chip. Reply
  • tipoo - Thursday, September 17, 2020 - link

    What does Big Navi have to do with a laptop CPU? Reply
  • AMDSuperFan - Thursday, September 17, 2020 - link

    You care about games don't you? This Intel Tiger won't have an answer for Big Navi. We can look forward to that showing who is the boss. Reply
  • blppt - Thursday, September 17, 2020 - link

    Based on preliminary data, they'll both be about 2 years behind Nvidia, what with Big Navi only matching a 2080ti, and not available for another month at the earliest. Reply
  • hecksagon - Friday, September 18, 2020 - link

    Crazy how you can make that prediction, the only preliminary data that is out is a photograph of the card. Are you a wizard? Reply
  • blppt - Friday, September 18, 2020 - link

    Incorrect.

    https://wccftech.com/amd-radeon-navi-gpu-specs-per...
    Reply
  • HarryVoyager - Friday, September 18, 2020 - link

    I'm not really seeing where you are getting that from. We know that RDNA2 can hit 2.23Ghz from the PS5 implementation, and we have solid rumors that it the top end one will be an 80CU chip, rather than a 40 CU chip. That implies on the order of a 230% improvement over the 5700XT, if their are no other performance improvements. That alone puts it in the 30-40% improvement range over the 2080 Ti. Given we've already seen at least a few AMD benchmarks of unidentified cards showing a 30-40% improvement over 2080 To performance, that sort of lift does seem likely.

    If I had to guess, that RDNA2 that recently showed up with a near 2080 TI performance is probably a 6700 competitor to the 3070, not the top end card. Those do have to be developed and tested too, after all.
    Reply

Log in

Don't have an account? Sign up now