Announcement

Collapse
No announcement yet.

Intel Announces Core 14th Gen 35/65 Watt Desktop CPUs, 14th Gen HX Mobile CPUs

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

  • #11
    Originally posted by vladpetric View Post
    It's called hate, mate

    If one starts with a strong hate position, one can easily spin and cherry-pick the available information to match their pre-existing hate.
    The facts are what they are. If you think I'm cherry-picking, then please show us what information I've omitted that would change the picture.

    I don't hate Intel, but I don't like how far they & AMD have gone in using boosting to extract a few extra % of performance. It also makes the advertised power consumption a fiction, especially on their "unlocked" CPUs.

    Comment


    • #12
      Originally posted by coder View Post
      LOL at the "65 W" i9-14900 that has a max turbo of 219 W. The "35 W" i9-14900T is a little more reasonable, with a max turbo of only 106 W, but then again those are really intended for small form factor PCs with tiny fans and heatsinks, and would probably sound like a hairdryer when dissipating 106 W.
      The kind of "mobile workstation" laptops which usually feature HX-tier CPUs indeed sometimes do have 4 DIMM slots. And yes, you can buy DDR5 SODIMMs with 48 GB, today!

      Shop 48GB per Module Laptop Memory on Newegg.com. Watch for amazing deals and get great pricing.

      i do not think your argument has any kind of rationality the max turbo could be 1000watt or more if only the cpu only use that for spikes less than a second to bypass bottlenecks only.

      in the end the average power consumtion counts even on battery its the average power consumtion what sucks out the battery.

      the cpu could have 10 000 watt max turbo if only they boost it for much less than a second.

      i would never buy such intel products but the argument agaist the max turbo TDP is really nonsense.

      i am pretty sure that upcoming cpus for 2024 and 2025 will have a max turbo TDP of 1000watt and it will only be used in case of need for less than a second. it will not increase the average power consumtion in any way. but it will be able to fix some nasty bottlenecks.
      Phantom circuit Sequence Reducer Dyslexia

      Comment


      • #13
        Originally posted by coder View Post
        65 W is easily enough for a single thread. Most responsiveness is dependent on single-thread performance.
        Moreover, the Tau limit for the i9-13900 is 28 seconds (don't yet know about 14th gen). That makes it seem to me less about interactivity and more about scoring higher on benchmarks.If you don't mind screaming fans or the cost & bulk of an overkill cooling solution for a "65 W" CPU, then it would help for moderately long tasks like incremental software builds.
        what exactly is your electrical/physics education? if you use 220TDP as a boost for 0,5seconds you clearly do not need a "overkill cooling solution" ...

        you could have a 10 0000watt TDP boost if you only boost it for a nanosecond you will not need any "overkill cooling solution"

        i give you a hind cooling solutions need multiple minutes to reach its temperature with the maximum cooling capacity

        thats why if you thest the cooling system with furmark and prime95 you need to run it for more than 15 minutes to get the end temperature.

        if you boost for 220watt for 0,5 seconds the cooling system plain and simple does not need to be big
        Phantom circuit Sequence Reducer Dyslexia

        Comment


        • #14
          Originally posted by coder View Post
          A few reasons. In the K-series Intel CPUs, it requires a lot of cooling capacity to keep them from thermal-throttling. You basically need either big water cooling radiators or a huge air cooler. A lot of the reviewers use the former, producing benchmark results that you can't replicate with a more modest setup.
          Second, it makes your fans spin up at max speed, which can be obnoxious. If you limit your fan speeds or power limits, then you're reducing your performance even further.
          If you do use a big cooler, it comes at the cost of a bigger and more expensive PC. Plus, there's the heat output, which I personally don't welcome in the summertime.
          On most mainstream motherboards, the K-series CPUs will boost for unlimited duration, by default! This makes a complete lie of the TDP figure, replacing it with whatever amount of the PL2 you can cool.
          I believe non-K CPUs can't run with unlimited Tau, but 28 seconds is still long enough to create a lot of drama.
          Turbo isn't inherently bad. However, Intel has gone way overboard with it. Do you really think it's okay to have a "65 W" CPU burning 3.4x that much power for up to 28 seconds?
          If you want to talk about really short term spikes, the i9-13900K has a PL4 of 420 W!
          you have a physical inertia of the mass ​of your cooling system and this is not 28 seconds its more like 15 minutes

          even 20 years ago if you tested with furmark or prime95 you would need to run your system over 15minutes to reach the max temperature of your system.

          a boost for 28 seconds is nothing compared to these 15 minutes. they could build 1000watt TDP cpus if they would only boost it short periot of time it would not overload the cooling systems.

          Phantom circuit Sequence Reducer Dyslexia

          Comment


          • #15
            Originally posted by coder View Post
            The facts are what they are. If you think I'm cherry-picking, then please show us what information I've omitted that would change the picture.

            I don't hate Intel, but I don't like how far they & AMD have gone in using boosting to extract a few extra % of performance. It also makes the advertised power consumption a fiction, especially on their "unlocked" CPUs.
            its not fiction that you can optimise nasty bottle necks by boosting as high as possible even 1000watt would be a good idea.

            and no of you boost for 0.5 seconds you would not need any fancy cooling system.

            any cooling system has some indolence​ what makes it tolerate​ high watt boosts for short periot of time.
            Phantom circuit Sequence Reducer Dyslexia

            Comment


            • #16
              Originally posted by coder View Post
              A few reasons. In the K-series Intel CPUs, it requires a lot of cooling capacity to keep them from thermal-throttling. You basically need either big water cooling radiators or a huge air cooler. A lot of the reviewers use the former, producing benchmark results that you can't replicate with a more modest setup.

              Second, it makes your fans spin up at max speed, which can be obnoxious. If you limit your fan speeds or power limits, then you're reducing your performance even further.

              If you do use a big cooler, it comes at the cost of a bigger and more expensive PC. Plus, there's the heat output, which I personally don't welcome in the summertime.


              On most mainstream motherboards, the K-series CPUs will boost for unlimited duration, by default! This makes a complete lie of the TDP figure, replacing it with whatever amount of the PL2 you can cool.

              I believe non-K CPUs can't run with unlimited Tau, but 28 seconds is still long enough to create a lot of drama.


              Turbo isn't inherently bad. However, Intel has gone way overboard with it. Do you really think it's okay to have a "65 W" CPU burning 3.4x that much power for up to 28 seconds?

              If you want to talk about really short term spikes, the i9-13900K has a PL4 of 420 W!
              So... because of a misunderstanding?
              The K-series (which is not what's being discussed in this thread, but we'll let that slide), require nothing. They will run just fine with stock cooling, they'll just not boost as high or for as long periods of time. What Intel did is actually really clever: add enough sensors to the CPU can detect how effective the cooling is, and stretch its legs as appropriate. It's not that different from what overclockers used to do, only now it's automatic.

              And yes, as I have already stated, I don't mind burning through more that the TDP rating, as long as it averages out over longer periods. I don't like the unlimited PL1/PL2, but that's just a default setting, I've long changed that in my BIOS.
              The TDP still has its meaning, it's the minimum a compatible cooler is supposed to handle.

              Comment


              • #17
                Originally posted by bug77 View Post
                So... because of a misunderstanding?
                The K-series (which is not what's being discussed in this thread, but we'll let that slide), require nothing. They will run just fine with stock cooling, they'll just not boost as high or for as long periods of time.
                1. They will deliver less performance than people see in benchmarks posted online, because most reviewers are testing with water cooling and 420 mm radiators. So, you're not getting all the performance you expect. Knowing how much performance people would actually get, they might've made a different purchasing decision.
                2. It's burning more power than expected. It's sold as a 125 W CPU, but will happily boost like all damn day at up to 253 W, in most motherboards (depending on how much of that you can dissipate).

                Originally posted by bug77 View Post
                What Intel did is actually really clever:
                Oh yeah, lying to consumers and misrepresenting your products is like totally genius.

                Originally posted by bug77 View Post
                It's not that different from what overclockers used to do, only now it's automatic.
                Overclockers go in with eyes open. They expect to need lots of cooling and burn lots of power, for tiny marginal gains.

                Originally posted by bug77 View Post
                as I have already stated, I don't mind burning through more that the TDP rating, as long as it averages out over longer periods.
                First, it's not an average. These are limits. It's not going to compensate for boosting by later running at a depressed speed, in order to maintain some long-term average.

                Second, they're time-bound, except that motherboards typically default to unbounded boost, for K-series models.

                Originally posted by bug77 View Post
                The TDP still has its meaning, it's the minimum a compatible cooler is supposed to handle.
                To maintain the advertised base frequency, but the problem is that consumers don't actually know how that would perform, because they're never tested that way.

                Again, I don't have a fundamental problem with the idea of turbo boost, but the situation has just gotten way out of hand.

                Comment


                • #18
                  Originally posted by coder View Post
                  1. They will deliver less performance than people see in benchmarks posted online, because most reviewers are testing with water cooling and 420 mm radiators. So, you're not getting all the performance you expect. Knowing how much performance people would actually get, they might've made a different purchasing decision.
                  2. It's burning more power than expected. It's sold as a 125 W CPU, but will happily boost like all damn day at up to 253 W, in most motherboards (depending on how much of that you can dissipate).


                  Oh yeah, lying to consumers and misrepresenting your products is like totally genius.


                  Overclockers go in with eyes open. They expect to need lots of cooling and burn lots of power, for tiny marginal gains.


                  First, it's not an average. These are limits. It's not going to compensate for boosting by later running at a depressed speed, in order to maintain some long-term average.

                  Second, they're time-bound, except that motherboards typically default to unbounded boost, for K-series models.


                  To maintain the advertised base frequency, but the problem is that consumers don't actually know how that would perform, because they're never tested that way.

                  Again, I don't have a fundamental problem with the idea of turbo boost, but the situation has just gotten way out of hand.
                  You lose less than 9% if you stick to 125W: https://www.techpowerup.com/review/i...to-35-w/2.html
                  If you run it unconstrained, if will burn through more than 125W, but only if the system can supply that much and the cooling can handle it. It's a minor inconvenience, nothing more.

                  Comment


                  • #19
                    Originally posted by bug77 View Post
                    You lose less than 9% if you stick to 125W: https://www.techpowerup.com/review/i...to-35-w/2.html
                    First of all, that's according to their broad, multi-application average. If you look at a heavily-threaded workload (e.g. Blender), the i9-14900K takes 24.9% longer, at 125 W. (also, 80.2% longer at 65 W and 227.7% longer at 35 W).

                    Second, to the extent we believe their multi-application profile fits typical usage, 9% is quite a small gain for over 2x the power. Heck, even 25% is a rather small performance gain for double the power! This highlights just how ridiculous and desperate Intel's decision really was.

                    Originally posted by bug77 View Post
                    If you run it unconstrained, if will burn through more than 125W, but only if the system can supply that much
                    When you say "if the system can supply that much", that only applies to the motherboard. If you overtax the PSU, it's lights out.

                    I stand by my claims, even if you consider it just "a minor inconvenience". That's a value judgement, which you cannot make for other people.

                    Some people might've considered Volkswagen's emissions cheating "a minor inconvenience", but it undermined consumer choice, the company got fined billions of dollars, and individual executives faced criminal prosecution.
                    Last edited by coder; 11 January 2024, 05:32 AM.

                    Comment


                    • #20
                      Originally posted by coder View Post
                      First of all, that's according to their broad, multi-application average. If you look at a heavily-threaded workload (e.g. Blender), the i9-14900K takes 24.9% longer, at 125 W. (also, 80.2% longer at 65 W and 227.7% longer at 35 W).
                      Second, to the extent we believe their multi-application profile fits typical usage, 9% is quite a small gain for over 2x the power. Heck, even 25% is a rather small performance gain for double the power! This highlights just how ridiculous and desperate Intel's decision really was.
                      When you say "if the system can supply that much", that only applies to the motherboard. If you overtax the PSU, it's lights out.
                      I stand by my claims, even if you consider it just "a minor inconvenience". That's a value judgement, which you cannot make for other people.
                      Some people might've considered Volkswagen's emissions cheating "a minor inconvenience", but it undermined consumer choice, the company got fined billions of dollars, and individual executives faced criminal prosecution.
                      well your comparison is really nonsense the Volkswagen emission cheating benefits no one and also harms people.
                      a hypothetical 1000watt TDP boost clock cpu does not harm anyone but benefits these people who invest in a good lets say overpowered cooling solution.
                      intel in generall has a problem of power efficiency anyone who knows this buy AMD or ARM cpus instead.
                      intel intels fabrication node this 10nm node is garbage but to not do boost clock with high TDP would make their situation even worst.
                      les say instead of 10nm intel now switch all cpus to a 3nm TSMC node then they would produce higly power efficient cpus
                      but with 3nm they would lose again if they do not do 1000watt TDP boost clock ---

                      whatever intel does if they follow your advice they lose... if they don't give a shit about your advice here and make a 1000watt TDP cpu they stay relevant and make sales. and people are happy for super responsible system because many apps only need this performance for less than a second.

                      your open the app you need the boost after this 1 second the system idle again.
                      Phantom circuit Sequence Reducer Dyslexia

                      Comment

                      Working...
                      X