System Performance: Multi-Tasking

One of the key drivers of advancements in computing systems is multi-tasking. On mobile devices, this is quite lightweight - cases such as background email checks while the user is playing a mobile game are quite common. Towards optimizing user experience in those types of scenarios, mobile SoC manufacturers started integrating heterogeneous CPU cores - some with high performance for demanding workloads, while others were frugal in terms of both power consumption / die area and performance. This trend is now slowly making its way into the desktop PC space.

Multi-tasking in typical PC usage is much more demanding compared to phones and tablets. Desktop OSes allow users to launch and utilize a large number of demanding programs simultaneously. Responsiveness is dictated largely by the OS scheduler allowing different tasks to move to the background. Intel's Alder Lake processors work closely with the Windows 11 thread scheduler to optimize performance in these cases. Keeping these aspects in mind, the evaluation of multi-tasking performance is an interesting subject to tackle.

We have augmented our systems benchmarking suite to quantitatively analyze the multi-tasking performance of various platforms. The evaluation involves triggering a ffmpeg transcoding task to transform 1716 3840x1714 frames encoded as a 24fps AVC video (Blender Project's 'Tears of Steel' 4K version) into a 1080p HEVC version in a loop. The transcoding rate is monitored continuously. One complete transcoding pass is allowed to complete before starting the first multi-tasking workload - the PCMark 10 Extended bench suite. A comparative view of the PCMark 10 scores for various scenarios is presented in the graphs below. Also available for concurrent viewing are scores in the normal case where the benchmark was processed without any concurrent load, and a graph presenting the loss in performance

UL PCMark 10 Load Testing - Digital Content Creation Scores

UL PCMark 10 Load Testing - Productivity Scores

UL PCMark 10 Load Testing - Essentials Scores

UL PCMark 10 Load Testing - Gaming Scores

UL PCMark 10 Load Testing - Overall Scores

All PCMark 10 workload components see the relative ordering being maintained even after the addition of the concurrent loading.

Following the completion of the PCMark 10 benchmark, a short delay is introduced prior to the processing of Principled Technologies WebXPRT4 on MS Edge. Similar to the PCMark 10 results presentation, the graph below show the scores recorded with the transcoding load active. Available for comparison are the dedicated CPU power scores and a measure of the performance loss.

Principled Technologies WebXPRT4 Load Testing Scores (MS Edge)

Despite a 50% performance loss, the NUC13ANKi7 retains a signficant lead over the other systems in webXPRT4 in the presence of the transcoding workload.

The final workload tested as part of the multitasking evaluation routine is CINEBENCH R23.

3D Rendering - CINEBENCH R23 Load Testing - Single Thread Score

3D Rendering - CINEBENCH R23 Load Testing - Multiple Thread Score

We see a 60%+ drop in the single thread case, and a 40%+ drop for the multi-threaded case, but the raw numbers are still well in favor of the NUC13ANKi7.

The latest version of Windows 11 (22H2) apparently has some additional thread scheduling improvements that can help Alder Lake and Raptor Lake deliver better / more balanced performance for multi-tasking scenarios. We may get a chance to evaluate that a few quarters down the line.

GPU Performance: Synthetic Benchmarks HTPC and Digital Signage Credentials
Comments Locked

46 Comments

View All Comments

  • PeachNCream - Tuesday, March 28, 2023 - link

    I congratulate you. Your bait caught a lot of them this time. And, like a true troll, you didn't waste keystrokes doing it. Though, admittedly, computer enthusiasts are low hanging fruit that get riled up with minimal work.
  • MrCommunistGen - Monday, March 27, 2023 - link

    I'm pretty pleased that there are P-series chips and not just U-series - as many (but not all) previous SFF NUCs used 15W TDP chips rather than 28W and higher TDP mobile CPUs. As long as the cooler can handle the heat load and isn't too terribly loud (or can have its performance/fan curves adjusted) then this could be a really cool SFF PC.

    I currently have a NUC8i7BEH i7-8559U (28W 4c/8t + eDRAM) which I'm not YET looking to upgrade, but this new 13th Gen NUC is the first model that's gotten me excited about a new NUC in this form factor.
  • Samus - Tuesday, March 28, 2023 - link

    I agree. The P-series was a solid move on Intel's part to stay competitive against Ryzen mobile parts. Even though most notebooks don't seem to utilize the TDP Up (if its a thin plastic laptop kiss PL2 goodbye entirely) the single-threaded boost performance is above U-series parts, and general workflow at low-loads benefit from the stronger E-core headroom. Intel competitively prices the P chips against the U chips as well, which is why you see them in cheaper laptops.
  • abufrejoval - Tuesday, March 28, 2023 - link

    The latest U-series to me represent an attempt to have at least a toenail in a fully passive market and that's not an economy niche, even if Intel can put more of them on a wafer: sometimes Intel is strangely insensitive to wasting die area!

    And since NUCs have always used active cooling even for their Atom models, U-series SoCs simply have no business there.

    I am also a fan of the great NUC fan curves, because they allow you to achieve unobtrusiveness, which is only beaten by "never need cleaning out fans and grills", but not at the price of crippled performance or an Akasa case.

    I have the same NUC and it's younger cousins, a hexacore NUC10 and a Tiger Lake NUC11. And that last "custom Apple" SoC with the Iris 655 iGPU in the i7-8559U spends more than half its die area on an iGPU that Intel never charged for, but mostly because it didn't actually pay of in performance.

    The Tiger Lake trounces NUC8 and 10 by 25% on scalar peformance while the 6 cores of Gen10 and 4 cores of Gen11 draw even on multicore.

    The Tiger Lake's 96EU Xe iGPU puts the Iris 48EU iGPU to shame, which only gets 50% performance increase over an 24EU iGPU from the NUC10 for twice the EUs and without needing the eDRAM, which doesn't help nearly as much as it must have cost in production. It marks the turning point for Apple, which wanted much more GPU power at mobile power budgets.

    Yes, you can expect a tempting performance uplift from one of these, should you make the switch. But a gaming engine it won't be, because 4x the speed of a lame duck is still pretty lame.
  • ekon - Monday, March 27, 2023 - link

    I like the multitasking benchmarks; don't see that attempted very often as it's not an off-the-shelf test and involves the time and risk of coming up with your own methodology, but it's teased out some interesting results.
  • cruiseliu - Monday, March 27, 2023 - link

    Kinda surprised it's still DDR4.
  • Samus - Tuesday, March 28, 2023 - link

    I was surprised to see that too. Even 6 months ago I wouldn't be since DDR5 SODIMMs were still elusive unicorns but nowadays they are plentiful and price competitive with DDR4. I've seen 32GB 16x2 kits of Crucial DDR5 5200(?) SODIMMs from B&H for $100 shipped yet nothing mobile seems to take DDR5, and if it does, it's non-upgradable (looking at you XPS 13)
  • Fenturi - Tuesday, March 28, 2023 - link

    U690 6800hx based with swappable ddr5
  • MrCommunistGen - Tuesday, March 28, 2023 - link

    Me too. I was going to write a follow-up comment about that, but then got a bit derailed writing my anti-rant comment.

    I suspect that the re-use of DDR4 was a cost saving measure so that they could (mostly) re-use the memory traces, etc from previous NUC boards instead of having to rework that for DDR5.

    DDR5 doesn't help all workloads, particularly at JEDEC speeds, but there's definitely some where it helps. Having that little bit of extra performance in this form factor would be appreciated.
  • Einy0 - Tuesday, March 28, 2023 - link

    In response to all the Mac mini comments. I've wanted to dip my toe into the OSX pond for a while now. I do some moonlighting for a business that uses a lot of Macs, and I feel like more hands on time with a Mac would be useful. The ancient Macbook own gives me some exposure but a lot was changed since OSX 10.6.8 . I just can't bring myself to get a mini with only 8GB or RAM. My Macbook has 4GB and a decent SSD and it is SLOW! I imagine newer versions of OSX are more memory hungry, like everything else. I also WILL NOT pay $200 more for $25 worth of RAM. It is plain insulting! Shame on Apple...

Log in

Don't have an account? Sign up now