CPU Tests: Microbenchmarks

Core-to-Core Latency

As the core count of modern CPUs is growing, we are reaching a time when the time to access each core from a different core is no longer a constant. Even before the advent of heterogeneous SoC designs, processors built on large rings or meshes can have different latencies to access the nearest core compared to the furthest core. This rings true especially in multi-socket server environments.

But modern CPUs, even desktop and consumer CPUs, can have variable access latency to get to another core. For example, in the first generation Threadripper CPUs, we had four chips on the package, each with 8 threads, and each with a different core-to-core latency depending on if it was on-die or off-die. This gets more complex with products like Lakefield, which has two different communication buses depending on which core is talking to which.

If you are a regular reader of AnandTech’s CPU reviews, you will recognize our Core-to-Core latency test. It’s a great way to show exactly how groups of cores are laid out on the silicon. This is a custom in-house test built by Andrei, and we know there are competing tests out there, but we feel ours is the most accurate to how quick an access between two cores can happen.

When we first reviewed the 10-core Comet Lake processors, we noticed that a core (or two) seemed to take slightly longer to ping/pong than the others. These two parts are both derived from the 10-core silicon but with two cores disabled, and we still see a pattern of some cores having additional latency. The ring on the 8-core parts still acts like a 10-core ring, but it all depends on which cores were disabled.

Frequency Ramping

Both AMD and Intel over the past few years have introduced features to their processors that speed up the time from when a CPU moves from idle into a high powered state. The effect of this means that users can get peak performance quicker, but the biggest knock-on effect for this is with battery life in mobile devices, especially if a system can turbo up quick and turbo down quick, ensuring that it stays in the lowest and most efficient power state for as long as possible.

Intel’s technology is called SpeedShift, although SpeedShift was not enabled until Skylake.

One of the issues though with this technology is that sometimes the adjustments in frequency can be so fast, software cannot detect them. If the frequency is changing on the order of microseconds, but your software is only probing frequency in milliseconds (or seconds), then quick changes will be missed. Not only that, as an observer probing the frequency, you could be affecting the actual turbo performance. When the CPU is changing frequency, it essentially has to pause all compute while it aligns the frequency rate of the whole core.

We wrote an extensive review analysis piece on this, called ‘Reaching for Turbo: Aligning Perception with AMD’s Frequency Metrics’, due to an issue where users were not observing the peak turbo speeds for AMD’s processors.

We got around the issue by making the frequency probing the workload causing the turbo. The software is able to detect frequency adjustments on a microsecond scale, so we can see how well a system can get to those boost frequencies. Our Frequency Ramp tool has already been in use in a number of reviews.

Both processors ramp from idle to full turbo in about six milliseconds, well within a single frame of standard gaming.

Power Consumption CPU Tests: Office and Science
Comments Locked


View All Comments

  • blckgrffn - Thursday, January 21, 2021 - link

    Thanks for the review, as it basically shows what other reviews already show, namely if you set aggressive PL1 and PL2 values across K & non K SKUs then you'll get similar performance.

    I am curious why you said the performance is much lower with a 65W power limit and then didn't include those results.

    I feel like it is common knowledge, especially with 10th Gen Intel CPUs that you need to manually configure PL1 and PL2 in keeping with your cooling solution, but perhaps not.
  • blckgrffn - Thursday, January 21, 2021 - link

    I mean, published PL1 for this CPU is 10700 is 65, PL2 is 224 for 28 seconds.

    Running outside of those value is essentially turbo overclocking (yeah, I know Intel has also redefined that term).

    If your motherboard auto overclocks the CPU via a ridiculous PL1 value then ¯\_(ツ)_/¯
  • Tunnah - Thursday, January 21, 2021 - link

    Intel has no incentive to change their policy and label their products with the actual power draw they'll be using because they'll show how much more they suck up compared to AMD. People are constantly looking for metrics to compare the 2 "teams", and Intel getting to keep the labels of 65w and 125w lets the fans say "see it has the same power usage as AMD!"
  • yeeeeman - Thursday, January 21, 2021 - link

    this looks just fine to me as long as it is clear for the user.
  • magreen - Friday, January 22, 2021 - link

    But it is not.
  • porina - Thursday, January 21, 2021 - link

    From my observations of Zen 2 at stock operation, 65W TDP models tended to sit continuously at the 88W PPT limit under most all-core load conditions. Has this changed with Zen 3? Do they not hit 88W so easily, or is another (current) limiter taking over? Or is the limit a different value now?
  • Smell This - Friday, January 22, 2021 - link

    Presumably, the Zen3 would operate under the same 'constraints' ----- The constraints are as follows:

    ♦ Package Power Tracking (PPT): The power threshold that is allowed to be delivered to the socket. This is 88W for 65W TDP processors, and 142W for 105W TDP processors.

    ♦ Thermal Design Current (TDC): The maximum amount of current delivered by the motherboard’s voltage regulators when under thermally constrained scenarios (high temperatures). This is 60A for 65W TDP processors, and 95A for 105W TDP processors.

    ♦ Electrical Design Current (EDC): This is the maximum amount of current at any instantaneous short period of time that can be delivered by the motherboard’s voltage regulators. This is 90A for 65W TDP processors, and 140A for 105W TDP processors.

    "Looking at the total power consumption of the new 3700X, the chip is very much seemingly hitting and maintaining the 88W PPT limitations of the default settings, and we’re measuring 90W peak consumption across the package."
  • Olaf van der Spek - Thursday, January 21, 2021 - link

    Why would one want to limit turbo budgets? Thermals? If there's no thermal headroom the CPU won't turbo (as far).
  • Calin - Friday, January 22, 2021 - link

    The motherboard has power limits - both in instant maximum current from the voltage regulation phase (remember, the mainboard receives 3.3 Volts, 5V, 12V and maybe -5V and -12V from the PSU and has to convert that to processor voltage), and in cooling capacity for the VRM (Voltage Regulation Module).
    Regardless of the power limits, the processor will slow down if its internal temperature is too great.
    So yes, the "my mainboard's power delivery module cannot deliver more than 80 amps" is a possible reason. Another would be "My case has bad cooling and I want to keep the processor colder". Another would be "As soon as the sustained power goes over 140 watts, the fans in the case start whirring and I hate the sound".
  • DominionSeraph - Thursday, January 21, 2021 - link

    >This does come with a reasonably good default cooler.

    No. Just no. The Ryzen coolers are utter trash and you're doing a disservice to your readers who may not have ever had a quiet cooler to say otherwise. I build PCs and I've had several Ryzens go through and I have never seen one where I would call the acoustics livable. My first sale was a 1700 with the stock cooler since I didn't have any other AM4 compatible ones at the time and I still feel bad about selling it that way. It was just terrible. The 212 EVO seems to be within its thermal envelope for quiet cooling up to a stock 3700X, so I'd highly recommend one of those over the stock cooler. Going above the ~85W of a 3700X you should spring for a Fuma 2.

Log in

Don't have an account? Sign up now