Announcement

Collapse
No announcement yet.

Intel Core i3 8100: 3.6GHz Quad-Core With UHD Graphics For Less Than $120 USD

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #31
    Originally posted by torsionbar28 View Post

    Not sure what you're saying here. I've never known it to matter what the price of the board is vs. price of the CPU. Sometimes one is more, sometimes not. Personally, I'm more likely to invest in a high-end board with an economy CPU, than the other way round. Economy CPU will work perfectly and reliably all day every day, but the same cannot be said about bottom-dollar motherboards.
    Hiding the platform cost in a performance per dollar metric is misleading at best.

    The main reason the platform needs to be included is that nobody already has one of these motherboards. You may have the RAM, disk, case, PSU, display, peripherals, but you can't get the performance listed in the review without springing for a mobo, and the mobo is a significant portion of the consumer outlay to obtain the reviewed performance. If this CPU was an upgrade on an existing platform then I'd say fair enough, review the CPU performance by itself, but that is absolutely not the case here. This is also the 2nd CPU family release Intel has put out in a row which has required a new mobo, so the chances of you amortising the mobo cost across multiple CPU upgrades seems way less likely than with past CPU reviews.

    It's relevant because if you get 120 of X with $170+120 and that is then rated 1X/$, and you get 175 of X with $170+$175 and that is also rated 1X/$ then the consumer who needs lots of X may buy 3 of the former ($510+360=$870) to get 360, while they could have bought 2 of the latter (340+350=$690) to get 350. And there it becomes really obvious that the second choice is giving much better performance per cost.

    Normally consumers avoid scale-up because of sharply reduced performance per cost. In this case, scale-up (moving from i3 to i5) is a huge win for the consumer, especially considering that future workloads are likely to be more highly threaded and will take even better advantage of the extra cores than they do today... so then suggesting to consumers that the lower core part is performing the same per cost, when it clearly is not, and encouraging consumers to make purchases that are less future-proof, is doubly wrong.
    Last edited by linuxgeex; 10-11-2017, 11:58 PM.

    Comment


    • #32
      Originally posted by linuxgeex View Post

      Hiding the platform cost in a performance per dollar metric is misleading at best.

      The main reason the platform needs to be included is that nobody already has one of these motherboards. You may have the RAM, disk, case, PSU, display, peripherals, but you can't get the performance listed in the review without springing for a mobo, and the mobo is a significant portion of the consumer outlay to obtain the reviewed performance. If this CPU was an upgrade on an existing platform then I'd say fair enough, review the CPU performance by itself, but that is absolutely not the case here. This is also the 2nd CPU family release Intel has put out in a row which has required a new mobo, so the chances of you amortising the mobo cost across multiple CPU upgrades seems way less likely than with past CPU reviews.
      Agree with everything you're saying, the performance per dollar is not that informative when comparing intel vs AMD. It's far more useful when comparing intel vs. intel or AMD vs. AMD while keeping all other platform hardware the same. Personally I never buy just a new mobo and CPU. I buy the mobo, CPU, memory, case, and PSU all at once. Build the barebones rig, get it up and running, issues worked out, and then I migrate my data and peripherals over to it. That's simply too many variables to devise a meaningful performance per dollar metric.

      Comment


      • #33
        Originally posted by torsionbar28 View Post
        Agree with everything you're saying, the performance per dollar is not that informative when comparing intel vs AMD. It's far more useful when comparing intel vs. intel or AMD vs. AMD while keeping all other platform hardware the same. Personally I never buy just a new mobo and CPU. I buy the mobo, CPU, memory, case, and PSU all at once. Build the barebones rig, get it up and running, issues worked out, and then I migrate my data and peripherals over to it. That's simply too many variables to devise a meaningful performance per dollar metric.
        I disagree. PP$ is nowhere near as difficult as you're making it out to be. The reason it's a metric is for people looking for the best value. If you're not on a budget, PP$ doesn't really matter, so at that point your priorities come to things like performance-per-watt, IPC, maximum clock speeds, number of threads, and most importantly, the demands of your software.
        If value/PP$ is a high priority to you, the only things that matter are the CPU and motherboard; everything else can be the same. We're told what the PP$ is of the CPU, so that just leaves the motherboard as the unknown variable. Generally speaking, AMD motherboards vs competing Intel motherboards are priced roughly the same and offer the same features. So, you pretty much just have to spend what suits your needs. On the other hand, AMD sockets tend to live longer, so AMD motherboards tend to always win in terms of value.

        And sure, Ryzen does benefit more from higher RAM frequencies (implying you skew the value when you buy higher-speed modules), but it isn't a slouch at speeds below 3GHz.

        Comment

        Working...
        X