Analyzing Creator Mode and Game Mode

Way back on page 3, this review explained that AMD was promoting two modes: Creator Mode with all cores enabled and a uniform memory access (UMA) architecture, and Game Mode that disabled one of the dies and adjusted to a non-uniform memory architecture (NUMA). The idea was that in Creator Mode you had all the threads and bandwidth, while Game Mode focused on compatibility with games that freaked out if you had too many cores, but also memory and core-to-core latency by pinning data as close to the core as possible, and keeping related threads all within the same Zeppelin die. Both methods have their positives and negatives, and although they can be enabled through a button press in Ryzen Master and a reboot, most users who care enough about these settings are likely to set it and forget it. (And then notice that if the BIOS resets, so does the settings…)

*This page has been edited on 8/17, due to a misinterpretation in the implementation of Game Mode. This original review has been updated to reflect this. We have written a secondary mini-article with fresh testing on the effects of Game Mode.

Power Consumption and Distribution 2017: The Core Wars (Conclusions)
Comments Locked

347 Comments

View All Comments

  • imaheadcase - Thursday, August 10, 2017 - link

    So you lost respect for a website based on how they word titles of articles? I think you don't understand advertising at all. lol

    If you want to know a website that lost respect, look at HardOCP and you know why people don't like them for obvious reasons.
  • Alexey291 - Thursday, August 10, 2017 - link

    No offence but HardOCP is far more respectable than what we have in ATech these days.

    But that's not hard. AT website is pretty much a shell for the forum which is where most of the traffic is. I'm sure they only so the reviews because 'it was something we have always done'
  • Johan Steyn - Thursday, August 10, 2017 - link

    You may not understand how wording is used to convey sentiments in a different way. That is what politicians thrive on. You could for instance say "I am sorry that you misunderstood me." It gives the impression that you are sorry, but you are not. People also ask for forgiveness like this: "If I have hurt you, please forgive me." It sounds sincer, but it is a hidden lie, not acknowledging that you have actually hurt anybody, actually saying that you do not think that you did.

    Well, this is a science and I cannot explain it all here. If you miss it, then it does not mean it is not there.
  • mikato - Monday, August 14, 2017 - link

    I thought I'd just comment to say I understand what you're saying and agree. Even if a sentence gives facts, it can sound more positive one way or another way based on how it is stated. The author has to do some reflection sometimes to catch this. I believe him whenever he says he doesn't have much time, and maybe that plays into it. But articles at different sites may not have this bias effect and it can be an important component of a review article.

    "Intel recently announced that its new 18-core chip scores 3200 on Cinebench R15. That would be an extra 6.7% performance over the Threadripper 1950X for 2x the cost."

    These 2 sentences give facts, but sound favorable to Intel until just the very end. It's a subtle perception thing but it's real. The facts in the sentences, however, are massively favorable to AMD. Threadripper does only 6.7% less performance than an announced (not yet released) Intel CPU for half the cost!

    Here is another version-

    "Intel recently announced that its new 18-core chip scores 3200 on Cinebench R15. So Threadripper, for half the cost of Intel's as-yet unreleased chip, performs only 6.7% slower in Cinebench."

    There, that one leads with Threadripper and "half the cost" in the second sentence, and sounds much different.
  • Johan Steyn - Thursday, August 10, 2017 - link

    HardOCP and PCPer is more respected in my opinion. Wccftech is unpredictable, sometimes they shine and sometimes they are really odd.
  • mapesdhs - Thursday, August 10, 2017 - link

    I've kinda taken to GamersNexus recently, but I still always read AT and toms to compare.

    Ian.
  • fanofanand - Tuesday, August 15, 2017 - link

    WCCFtech is a joke, it's nothing but rumors and trolling. If you are seriously going to put WCCFtech above Anandtech then everyone here can immediately disregard all of your comments.
  • Drumsticks - Thursday, August 10, 2017 - link

    Fantastic review In. I was curious exactly how AMD would handle the NUMA problem with Threadripper. It seems that anybody buying Threadripper for real work is going to have to continue being very aware of exactly what configuration gets them the best performance.

    One minor correction, at the bottom of the CPU Rendering tests page:

    "Intel recently announced that its new 18-core chip scores 3200 on Cinebench R15. That would be an extra 6.7% performance over the Threadripper 1950X for 2x the cost." - this score is for the 16 core i9-7960X, not the 7980XE.
  • Drumsticks - Thursday, August 10, 2017 - link

    Ian*. Can't wait for the edit button one day!
  • launchcodemexico - Thursday, August 10, 2017 - link

    Why did you end all the gaming review sections with something like "Switching it to Game mode would have made better numbers..."? Why didn't you run the benchmarks in Gaming mode in the first place?

Log in

Don't have an account? Sign up now