Power Consumption

For our power consumption metrics, we use a Prime 95 blend on fixed threads to generate a strong load, and then poll the internal power registers that determine power state calculations to get the power consumption. Each processor is different in how it reports its power, which depends on the level of control the processor has: some of the more advanced CPUs, such as Ryzen, will provide per-core power numbers, while the latest Intel CPUs only give a figure for the CPUs as a whole but also include DRAM controller and uncore power consumption.

An interesting element to the power consumption on the Ryzen APUs, due to the unified power delivery subsystem in play feeding the CPU and the integrated graphics, is that the power registers only report half the power consumption when probed (e.g. when 14W, shows 7W). As of yet, we are unsure if this has a knock-on effect on how the processor adjusts its turbo modes in response to power consumption. Nonetheless, a simple scaling factor gives the following results.

Total Package: The Whole Processor

For this data, we take the values of the processor as a whole, which includes all the interconnect, memory controllers, PCIe root complexes, etc. The system is still only loading the CPU cores with minimal effect on the rest of the system, however depending on how the power is managed, some of the sub-systems still remain enabled.

Power: Total Package (1T)Power: Total Package (Full Load)

At full load, the difference between the Ryzen 5 and the other Ryzen CPUs shows that the 2400G is using more of its upper margin, compared to the 1400 which is rated at the same power (note TDP is only determined at the base frequency), but the extra frequency of the 2400G means that there is extra power draw overall. Part of this is due to the Infinity Fabric, which we will see below. But what these tests also underline is that in a quad-core configuration, the Intel CPUs are still very power efficient.

Cores Only: Pure Work

For the processors that split out the data, we can look at the power consumption of the cores on their own, without any of the sub-systems, like uncore, mesh, or infinity fabric. This usually paints a different picture to the package power.

Power: Cores Only (1T Load)Power: Cores Only (Full Load)

For the core only power, the Ryzen 5 2400G uses less power than the Core i3-8350K, despite the situation being reversed when considering the whole package. This means that Infinity Fabric takes a lot of power here, and the ring bus solution that Intel uses benefits from being simpler, and Intel can push more power to its individual cores.

Benchmarking Performance: CPU Legacy Tests Conclusion: Cutting Low-End Discrete Graphics
Comments Locked

177 Comments

View All Comments

  • haplo602 - Tuesday, February 13, 2018 - link

    Finally one review where I can see the driver version ... So this is the same driver used also for the Ryzen mobile APUs. Can you check if you can force/manual install the latest Adrenaline drivers ? That works on some of the Ryzen 2500u chips and actually increases the performance by some 15+% ...
  • haplo602 - Tuesday, February 13, 2018 - link

    I hope there's a memory scaling article in the future with frequency and CL scaling for the APU part ...
  • crotach - Tuesday, February 13, 2018 - link

    What about HTPC use?

    I was considering GT 1030 + Intel route for H265 and HDR10 playback and was really looking forward to Zen APUs, but there doesn't seem to be any motherboards with HDMI 2.0?!

    Also, I wonder if the chips can be undervolted and underclocked to bring them to a near silent noise level for the living room.
  • Lolimaster - Tuesday, February 13, 2018 - link

    You can undervolt and underclock ANY intel or amd cpu.
  • forgerone - Tuesday, February 13, 2018 - link

    What most writers and critics of integrated graphics processors such as AMD's APU or Intel iGP all seem to forget, is not EVERYONE in the world has a disposable or discretionary income equal to that of the United States, Europe, Japan etc. Not everyone can afford bleeding edge gaming PC's or laptops. Food, housing and clothing must come first for 80% of the population of the world.

    An APU can grant anyone who can afford at least a decent basic APU the enjoyment of playing most computer games. The visual quality of these games may not be up to the arrogantly high standards of most western gamers, but then again these same folks who are happy to have an APU also can not barely afford a 750p crt monitor much less a 4k flat screen.

    This simple idea is huge not only for the laptop and pc market but especially game developers who can only expect to see an expansion of their Total Addressable Market. And that is good for everybody as broader markets help reduce the cost of development.

    This in fact was the whole point behind AMD's release of Mantle and Microsoft and The Kronos Group's release of DX12 and Vulkan respectively.

    Today's AMD APU has all of the power of a GPU Add In Board of not more than a several years back.
  • krazyfrog - Tuesday, February 13, 2018 - link

    Why did you leave out the 8400 and the 1500X in these comparisons?
  • Kamgusta - Wednesday, February 14, 2018 - link

    Because these CPUs, while having the same price range, outperform these Raven Ridge chips. That would have been a bad press for AMD and it seems like Anandtech wants to remains extremely loyal to AMD in these days.
  • msroadkill612 - Tuesday, February 13, 2018 - link

    "the data shows both how far integrated graphics has come, and how far it still has to go to qualify for those 'immerse experiences' that Intel, AMD, and NVIDIA all claim are worth reaching for, with higher resolutions and higher fidelity. "

    This assumes a static situation which is rot.

    what it reveals is that in the current paradigm, coders have coded accordingly for satisfactory results. If the paradigm changes and other ways work better, then code evolves.

    This unprecedented integration of new gen, sibling cpu & gpu, offers many performance upsides too for future code.

    picture a mobo with a discrete gpu like an equivalent 1030, then picture a ~matchbox footprint apu - there is a huge difference in the size of the respective circuits - yet they both do the same job & have to send a lot of data to each other.

    it's not hard to figure which is inherently superior in many ways.

    I strongly disagree with your blinkered bias.
  • Pork@III - Tuesday, February 13, 2018 - link

    There is something unfinished, something inconsolable.
  • elites2012 - Tuesday, February 13, 2018 - link

    anything this chip lost to intel at, was most likely outdated. adobe, fcat, dolphin, pov are all outdated benchmarks.

Log in

Don't have an account? Sign up now