Etc.

Howdy, folks. I’m back from Nvidia GTC and am pretty well exhausted. As you might expect, I have plans to write a bit about the GK110 chip, the “big Kepler,” that Nvidia talked about quite a bit at GTC. However, I may not have that finished today.

In the meantime, we have a few of other bits of news.

First, in case you missed it, the folks at HCW have become the first outside website to adopt our latency-focused game benchmarking methods. They did a quick blog post explaining their decision early in the week, and now they’ve followed up with a review of sub-$200 video cards that considers individual frame times as well as FPS averages. Nice to see some other folks jumping on the bandwagon.

Since I know from the web stats that almost nobody read the pages I’m about to link, I’d also like to point out that we deployed our “inside the second” methods to test CPU performance in several recent games for our Ivy Bridge review. The results start on this page. Our Skyrim findings were quite enlightening, in my view; they suggest that AMD’s desktop FX-8150 processor has a real weakness in gaming performance. My sense is that the FX may have an Amdahl’s Law problem, that its relatively low performance in individual threads may impede the consistent delivery of low frame latencies.  In response to your requests, we also concocted a nifty multitasking test that’s worth a look.

Next, TR’s MacHole blogger, Jason Fox, has just revealed today that he is the culprit behind a very successful fake Twitter account pretending to be someone else’s facial hair: Lee Clow’s Beard. Yes, I just wrote that sentence. If it sounds positively loony to you, then you have a good sense of the matter. I knew back in high school that Jason would go on to big things, but I didn’t suspect how truly strange those things might be. Regardless, Jason has been profiled at a fancy website today, and his tweets have been turned into an upcoming book, available for pre-order at Amazon. There’s even a Lee Clow’s Beard iOS app, a fact I’m having a hard time fully accepting. At any rate, congrats to Jason on the book and all the rest. May the beard remain well-trimmed, or something like that.

Comments closed
    • cygnus1
    • 7 years ago

    [quote<]Since I know from the web stats that almost nobody read these pages[/quote<] Damage, please don't feel like it's unappreciated work. I know I'm guilty of skipping over a lot of the benchmark results pages, but I read your methodology pretty in depth and usually skip those benchmark pages because I know your excellent methodology is going to come up with an excellent and accurate conclusion. Usually if I'm surprised by your conclusion I go back and check out the actual benchmark details to see what happened, but knowing the nitty gritty just doesn't appeal to me. There's a flattering way to interpret it. Just look at it this way, I bet most people that skip those pages are like me and trust you and the work you do. BTW, whenever you guys want to setup a subscription/donation option, I'm in.

    • marraco
    • 7 years ago

    HCW just don’t get it. They present confusing charts. His comparison is useful because that review is a clear case when 2 cards have similar average framerates, but one card is clearly better than the other… but they don’t present the worst framerates (like TechReport does on percentiles).

    • spuppy
    • 7 years ago

    Scott thanks for the plug, and thanks again for the inspiration. Hopefully this leads to being able to review high end GPUs once again. We’ll see if I can make that happen at Computex. Hope to see you there!

    • Bensam123
    • 7 years ago

    Did you guys patent using sub second measurements in benchmarking? XD

    “Since I know from the web stats that almost nobody read the pages I’m about to link…”

    lol…

    • UberGerbil
    • 7 years ago

    I wonder if part of the reason almost nobody reads those pages is because the received wisdom for the past few years has been that desktops are GPU-bottlenecked when gaming at any of popular higher resolutions, so gaming benchmarks don’t tell you anything interesting once you’re past some minimum CPU threshold. Clearly that’s no longer the case (to the extent it ever was) but the perception may persist for a while.

    • OneArmedScissor
    • 7 years ago

    [quote<]My sense is that the FX may have an Amdahl's Law problem, that its relatively low performance in individual threads may impede the consistent delivery of low frame latencies.[/quote<] While it's worse off than the Phenom, it's not by much. There's a similarity there: They both have L3 caches clocked more than 1 GHz, and in some cases, almost 2 GHz, below any SB or IB i7. I'd be interested to see you run the frame latency tests with the FX's northbridge turned up, and also against a Nehalem based i7 and Core 2 for those of us not running the latest and greatest.

      • BobbinThreadbare
      • 7 years ago

      If you call 36 vs 4 frames above 36 ms not much.

        • khands
        • 7 years ago

        It depends on how much I assume, an issue in stability yes, but possibly one that’s not far from correction.

    • codedivine
    • 7 years ago

    Your “inside the second” accounts are great. Would also like to see more experimentation with the 240fps camera. My concern is that the frame-time measurements using software techniques may themselves not be 100% accurate, but a camera would solve that. Its early days still, but its great to see TR pioneer and take a lead role in the techniques. Highly apreciated.

      • odizzido
      • 7 years ago

      The camera will tell you when the frame is drawn, but it might not take into account the changing animation speed if there is any sort of frame smoothing like nvidia apparently does with SLI. Maybe you could notice it but both methods may be useful.

    • Stickem
    • 7 years ago

    “Since I know from the web stats that almost nobody read these pages”

    Which pages? Your Etc. or HCW?
    I read everything that you issue. Your writing style just “works for me.”

      • Damage
      • 7 years ago

      The pages of the Ivy review I then reference.

      There, clarified.

      • torquer
      • 7 years ago

      This sounds like the beginning of a beautiful bromance.

        • sweatshopking
        • 7 years ago

        WE SHOULD ALL HUG

          • Duck
          • 7 years ago

          Use your indoor voice.

      • fredsnotdead
      • 7 years ago

      I read almost everything published on TR. The writing style is known as standard, North American English, unfortunately not yet mastered by most of the competition. Wouldn’t matter though if the content wasn’t good.

Pin It on Pinterest

Share This