Ryzen 9000

>16c/32t 170w 95°c
>No core upgrade
>No power efficiency upgrade
>No thermal upgrade

AMD seems to be at the 4c/8t stage intel was stuck in.

Seems like the biggest nothingburger™ release from AMD and that's coming from a long term AMD fan.

Nothing Ever Happens Shirt $21.68

DMT Has Friends For Me Shirt $21.68

Nothing Ever Happens Shirt $21.68

  1. 2 weeks ago
    Anonymous

    B-but muh npu anon now i can use copilot+pc more efficiently then before and my life and productivity will increase by x10000

    • 2 weeks ago
      Anonymous

      I'm convinced that all the AI parroting you see in the media is paid PR by big money holding bags. They desperately want AI to take off but the reality is that it has almost no real use cases. Consumers don't want AI. Companies want AI to replace workers but it is nowhere near ready for that level of automation and most of all - accuracy.

      • 2 weeks ago
        Anonymous

        it's gotta be salty venture capitalists desperately trying to force their irresponsible investments to pay off

    • 2 weeks ago
      Anonymous

      Not even applicable here, moron

    • 2 weeks ago
      Anonymous

      There is no NPU in ayymdead desktop

  2. 2 weeks ago
    Anonymous

    The biggest goof up is the fact that AMD expects regular home users to sit next to a 95°c furnace.

    Electricity is expensive. Keeping a procesor and gpu outputting 90+ °c at the same time is not practical with legacy cpu coolers.Imagine what it would be like to use these parts in tropical weather without air conditioning.

    • 2 weeks ago
      Anonymous

      stupid question - why can't we use the heat components output to generate a bit of power that feeds back into the pc?

      • 2 weeks ago
        Anonymous

        I don't have a degree in thermodynamics but I imagine the conversation from heat generated to usable electricity is so low it's basically useless to try and do it
        also maybe that the entire system is relatively small so it would be difficult to set up something efficient

      • 2 weeks ago
        Anonymous

        Not OP but thats because the conversion from direct thermal energy into electricity is extremely inefficient. If it weren't, don't you think cars would have dozens of these things in the engine bay to recapture energy from the exhaust?

        • 2 weeks ago
          Anonymous

          5-10% is better than 0. But even if you used one, you need to dump that electricity somewhere, and it's not easy to determine where the average user would dump it.

    • 2 weeks ago
      Anonymous

      I'm not sure if you're actually fricking moronic or merely pretending. It could go either way on this board.

    • 2 weeks ago
      Anonymous

      >electricity is expensive
      Maybe don't be a poorgay third worlder?
      Literally non issue if you're white.

      • 2 weeks ago
        Anonymous

        it's ok, he's europoor

      • 2 weeks ago
        Anonymous

        Electricity is far cheaper in the third world. Less regulations, don't care about "muh fossil fuels".

        • 2 weeks ago
          Anonymous

          >france among cheapest of the eu
          nuclear bros win again

          • 2 weeks ago
            Anonymous

            Checking people's stance on nuclear energy is a flawless idiocy test.
            The sad part is that nuclear energy takes decades to build; we're going to have to live through this bullshit for decades even if politicians pass the idiocy test.

          • 2 weeks ago
            Anonymous

            If people hadn't been idiots over half a century ago and let nuclear panic overtake them I would've been born in a world powered by nuclear energy
            sigh

        • 2 weeks ago
          Anonymous

          WTF is Deutschland doing? How are they so cucked? Don't they have cheap gas from russia?

          • 2 weeks ago
            Anonymous

            In the process of retiring tons of nuclear plants with no plans to replace them

          • 2 weeks ago
            Anonymous

            Biden blew up our pipelines.

        • 2 weeks ago
          Anonymous

          because they also make a shit load of less money moron

        • 2 weeks ago
          Anonymous

          Well also a lot of third-world countries heavily subsidize residential electricity consumption. If they had to pay the actual cost, most of the third-world would not be electrified

    • 2 weeks ago
      Anonymous

      Normal people don't buy 9950x moron. 9600 and 9700 are 65w chips.

    • 2 weeks ago
      Anonymous

      Maybe stop touching your CPU/GPU die and you won't feel the 95C heat you cretin

      • 2 weeks ago
        Anonymous

        What do shoe soles taste like bootlicker?

        • 2 weeks ago
          Anonymous

          That's besides the point
          You don't understand thermodynamics at all
          95C t-die doesn't translate into 95C atmospheric air or you'd long be dead when using your laptop CPU throttling at 100C

    • 2 weeks ago
      Anonymous

      >still thinks having a bad cooler makes your room hotter
      underage

    • 2 weeks ago
      Anonymous

      total bullshit, you simply dont run the CPU at 100% all the time... even hardcore gaming it does not even exceed 60°C

    • 2 weeks ago
      Anonymous

      95c is tjunctuin. your cou should never be at that temperature to begin with. get proper 170w cooler to get your temps down.

    • 2 weeks ago
      Anonymous

      The processor's thermal limit is 95 degrees Celsius, and it will boost up to this temperature depending on the cooling system. Enabling the eco mode in the BIOS and setting the limit to 65W can help manage the temperature. This approach is similar to what Intel and Nvidia do, allowing the CPU to boost within its thermal limits. For further assistance, please contact our technical support team.

      • 2 weeks ago
        Anonymous

        You know you can adjust the throttle ceiling if you feel that 95C is too toasty? The throttle celling for Intel chips been hovering north of 85C since Kaby Lake by default.

        • 2 weeks ago
          Anonymous

          You know you can throttle my wiener?

    • 2 weeks ago
      Anonymous

      >The biggest goof up is the fact that AMD expects regular home users to sit next to a 95°c furnace.

      Let me guess - American educated?
      100W is 100W, regardless it being off a 100C* surface or 50C* surface. Processors have been idling at 5C and pushing 150W for years now - regardless of make.

      The laughingstock of P4 - the Prescott - ate _LESS_ power than current top tier cpus.

    • 2 weeks ago
      Anonymous

      nobody tell this dorkass how warm the chips in his tv gets lest we get "dumbass tries to heat home via looping video of fireplace" appearing in a list of news articles somewhere

    • 2 weeks ago
      Anonymous

      The temperature of the chip doesn't mean it's outputting a lot of heat. The lower temperature literally doesn't matter for anything but a slight decrease in voltage it needs to sustain the same clock. After 20 years you MIGHT see some silicon degradation in chips with constant extreme temperature variation due to substrate degradation but that's a hypothetical based on some assumption about physics at play.

      • 2 weeks ago
        Anonymous

        that sounds like a good one seeing the amount of cores. would be nice to have it.

        >max temp for 20 years
        i have heard they start to degrade much earlier than that so i keep mine running at 60-65 degrees max.

        since one cant know how long will he use his current machine.

        • 2 weeks ago
          Anonymous

          >i have heard they start to degrade much earlier than that so i keep mine running at 60-65 degrees max.
          I have my doubts that it really matters. Usually when you hear people blaming temps for instability it's always the same overclocking morons who have likely damaged their systems by being ham fisted idiots, configuring settings they don't understand and manhandling the parts until a once working system becomes marginal.
          Hell, apple for the 10 years before the M-series was putting intel chips through the torture racks on a global scale. Inadequate heatsinks shared with the GPU, fans set to idle until the CPU hits 90°C, constant throttling due to 99°C maximum temperatures for the whole time the CPU usage is >80%... And yet, you never heard of CPU degredation. The drives would fail, the LCDs would have issues, keyboards definitely failed a lot, but the CPUs just didn't seem to care.

    • 2 weeks ago
      Anonymous

      >american education

  3. 2 weeks ago
    Anonymous

    OP is a homosexual who doesn't post sources

    • 2 weeks ago
      Anonymous

      Well, technically these still count as rumors, but getting details just a couple days before launch, can't be that far off.

      https://videocardz.com/newz/alleged-amd-ryzen-9000-desktop-zen5-specs-leaked-ryzen-9-9950x-with-16-cores-up-to-5-7-ghz

      • 2 weeks ago
        Anonymous

        So same clocks, same cache, same number of cores, same number of iGPU cores, same TDP... This doesn't make any sense. Between 5xxx and 7xxx at least clocks have gone up.

        • 2 weeks ago
          Anonymous

          It's just on a smaller 3nm node. That's it. Literal "Gib monies" move by AMD.

          • 2 weeks ago
            Anonymous

            Nope, it sounds like uses the same 5nm node. 3nm only because viable production when Zen5 tape out. Zen6 is getting 3nm.

      • 2 weeks ago
        Anonymous

        >Granite ridge
        No way is this real, what is that intel ass name?

        • 2 weeks ago
          Anonymous

          Summit Ridge, Raven Ridge, Pinnacle Ridge, etc
          It has been this way since Zen1 with only a couple outliers. Where do you morons crawl out from where you've been totally oblivious to reality? Imagine going on IQfy and posting
          >I can't believe Toyota really named a car Accord!

          • 2 weeks ago
            Anonymous

            Ever heard of Crystal Ridge?

          • 2 weeks ago
            Anonymous

            The ridge of this sack on your face

          • 2 weeks ago
            Anonymous

            That's no name to be calling your mom's pussy, son

          • 2 weeks ago
            Anonymous

            The ridge of this sack on your face

            Ever heard of Crystal Ridge?

  4. 2 weeks ago
    Anonymous

    >uses less power than the 7950X
    Why is this a loss?

  5. 2 weeks ago
    Anonymous

    Zen 5 is primarily going to be about increasing IPC. We've known this for a long time.
    Before Zen 4 came out, you'd always hear about Zen 5 being the one AMD's engineers were actually excited about, due to IPC. With Zen 4's main purpose being to get Ryzen to AM5, DDR5, PCIe 5.0, and such.
    >No power efficiency upgrade
    >No thermal upgrade
    Hard to say what it'll mean in practice, but except for the 16-core, the TDP of the upcoming Ryzen 9000X CPUs seems to have been lowered, compared to their 7000X counterparts. Pic related.

    • 2 weeks ago
      Anonymous

      You can't trust AMD on TDP anymore
      >https://forum.level1techs.com/t/passive-ryzen-7900-test-and-testimonial/211009
      >The figure of 65W is clearly underestimated

      • 2 weeks ago
        Anonymous

        Ryzen has been PL = 1.35x TDP for a long time
        why are polacks so fricking stupid

  6. 2 weeks ago
    Anonymous

    Why would you need more cores? At this point all apps that ca use more than 16 cores ca also use GPU acceleration that works even better.

  7. 2 weeks ago
    Anonymous

    so glad I switched to Apple M.
    2 years of headache free computing and silence and no heat.

    • 2 weeks ago
      Anonymous

      I'm not doing that because
      my linux distro > windows > mac os

    • 2 weeks ago
      Anonymous

      Apple chips are fine for basic daily driver - browser, videos, music, office. They choke at high performance tasks and M3 & M4 have overheating issues under load.

    • 2 weeks ago
      Anonymous

      if only apple made low/mid tier machine with normal pci-e slots that could fit normal gpus

    • 2 weeks ago
      Anonymous

      I have been thinking about doing this and turning my PC basically into a console, I use macos all day for work and its a treat at my Microsoft oriented MSP. I come home to my desktop running Windows. It's been mostly been fine for 15 months but has all the papercuts one has using a windows device. Random and unexplainable lag, stutter, animation drops, random unexplainable errors the briefly flash on screen. At work, I have dealt with hundreds of obscure and moronic bugs and shit design simply because MS do not care. I cannot stress enough how many windows issues are caused by Microsoft vs MacOS issues always being caused by lazy 3rd party devs, the OS is almost never at fault. Just my pov after supporting both for a couple years. If Chrome becomes a b***h to run adblocker in, I will most certainly do this so I can move to Safari+AdBlock on desktop and just us a kvm to swap display/peripherals to the pc when I want to game.

      I'll wait for the intel 15th gen vs ryzen 9000 comparison before buying either one

      Even if Arrow Lake matches it, I think AL will be better, as a 7950X chad I see where the wind is blowing. TSMC N3B, an expensive node with better efficiency while AMD are once again pushing the 4nm node when the node advantage barely got em over the finish line last time. Thunderbolt 5, 3x faster than USB4. 24-Cores, a better NPU. Just like 13th gen vs Zen, the tables have turned. intel will have the better platform and more efficient chips. It just has all the things you'd want thrown in there on top of the new CPU perf that you'd get out of either of them. Still don't know if I want to build another pc though, kinda burnt out on this crap. Rather wauit for another 60-100% perf jump again. That's why 5090 may be worth it.

      Apple chips are fine for basic daily driver - browser, videos, music, office. They choke at high performance tasks and M3 & M4 have overheating issues under load.

      They don't "Overheat" they're FANLESS and thermal throttle like literally every other piece of silicon out there that hits its thermal envelope and then must start down clocking as it has saturated the cooling element.

      • 2 weeks ago
        Anonymous

        If ARL is more efficient than Zen 5 then Intel have pulled a rabbit out of a hat.
        Meteor Lake was supposed to be a chance for Intel to impress with Intel 4, but the only impressive part was the GPU fabbed by TSMC.

        • 2 weeks ago
          Anonymous

          I'm not clear if the N3B process will be for GPU or CPU, but if it is for CPU, Intel would have to be shitting all over TSMC's advances to make a 3nm node less efficient than a 4nm node. but yeah if N3B is just for the gou and they are still using intel 4 for cpu then it will be ok. Meteor Lake laptops do have decent efficiency, hopefully only improving with ARL, but if the 285K is still gonna hit 250W they are fricking out of their minds.

        • 2 weeks ago
          Anonymous

          Intel is going all TSMC for Arrow Lake. No Intel fabbed chiplets at all, lmao.

          I'm not clear if the N3B process will be for GPU or CPU, but if it is for CPU, Intel would have to be shitting all over TSMC's advances to make a 3nm node less efficient than a 4nm node. but yeah if N3B is just for the gou and they are still using intel 4 for cpu then it will be ok. Meteor Lake laptops do have decent efficiency, hopefully only improving with ARL, but if the 285K is still gonna hit 250W they are fricking out of their minds.

          N3B is the CPU tile, N4 is iGPU and N6something is the rest. It's going to be really interesting to see Intel's designers (undisputably the best in the world) using TSMC's process (undisputable the best in the world) to implement build Intel's consumer chips (what have undisputably the worst CPU architects of any of the majors).

          I've long said that Intel's problems aren't with process (which most people think), and not with their designers (which are great), but rather their architects which do moronic things like desktop big.LITTLE, hetergenous compute, fickle instruction sets, etc. If my theory hold, then Arrow Lake will have great clocks and synthetic performance, and might even win in efficiency, but lose in overall app performance and games.

          • 2 weeks ago
            Anonymous

            >Intel is going all TSMC for Arrow Lake
            Lmao I completely missed that news. Sounds like 20A will only be hitting laptops.

          • 2 weeks ago
            Anonymous

            >I've long said that Intel's problems aren't with process (which most people think), and not with their designers (which are great),
            Whoever's working on their E cores have done an excellent job scaling up those shitty Atoms into something worthwhile, but they've clearly engineered themselves into a corner with P cores. 10nm is the big failure for Intel's processes but everything newer has failed to impress too.
            >but rather their architects which do moronic things like desktop big.LITTLE, hetergenous compute, fickle instruction sets, etc.
            Intel are only doing these things now because they can't slap any more bloated P cores on their outdated dies. AMD don't have to go heterogeneous because a Zen core is like half the transistors for 90% the per-core performance.

          • 2 weeks ago
            Anonymous

            AFAIK the E cores are not Atoms, except in some non-technical press material. Atom couldn't even do OOP which was a pretty big deal breaker for any sort of low latency local compute. The IPC leaks are real but no real-world clocks are shown. Intel's *Lake architecture is already very power hungry; clocking 16 E cores up to the point where their improved IPC makes them a serious P core contender is not an option, even with improved process. We're going to see the regular 8P cores sucking down enourmous power and 16E cores at a much lower clock, but with much better IPC. It's a big improvement over Alder-Raptor Lake, and the 8P+16E CPU is going to be real competition for AMD's 16 core CPUs in thread-heavy workloads for the first time since Zen3.

            Intel's P cores are really fricking good tho. I'm super pissed off they cancelled the 8P+0E CPU. That was going to be a gamer/OCer dream chip. I'd even eat the extra core-2-core latency of a 10P chip if they spun a die without any E cores.

          • 2 weeks ago
            Anonymous

            >That was going to be a gamer/OCer dream chip.
            Not at all. Intel set up the scheduler to prefer P cores and then E cores, only using HT logical threads as a last resort and for a very good reason. I've tested this in multiple games and I always got better performance in 8P + 16E over 8P + 0E. In fact, the best setting I've found is 8P + 16E + no HT. By disabling HT I get a small decrease in nT but I can clock the P-cores 200MHz and avoid all the HT frickery like stalling threads and an increase in cache misses. Meanwhile I can clock P cores as high as I want and nothing I do with E cores interferes with that. I am p95 stable with 5.7GHz P-cores with HT off and 16 E cores on and 10C below throttling, it gets to around 90C at ~300W on my loop. With HT it immediately throttles at lower frequencies.

          • 2 weeks ago
            Anonymous

            No wonder Arrow Lake will be dumping HT. E cores are just plain better.

  8. 2 weeks ago
    Anonymous

    We don't need more cores at this point. They should focus on improving single core/ipc and the infinity fabric garbage latency

  9. 2 weeks ago
    Anonymous

    When you have to push 95C and the bestselling CPU is a gaming 8-core with glued on cache why should they try harder?

  10. 2 weeks ago
    Anonymous

    I'm a dumbass so can you all explain it to me. What is bad about the new AMD CPU? Compared to say the old meta 7800x3d. I was waiting for the generation refresh but if its bad I want to know why.

    • 2 weeks ago
      Anonymous

      We can't say if it is bad or not until we have the product in our hands, but from the spec leaks it is obvious that is not really an upgrade, its more of the same - same core count, tdp, almost same frequency.

      They are stagnating and choosing to take advantage of their market leader position. This is the same reason why consumers were upset with Intel when they pushed the same old 4core /8thread chips with 5% improvement.

      • 2 weeks ago
        Anonymous

        Meaning its possible that it might not be significantly better than the 7800x3d? Would it at least be cheaper or more power efficient?

        • 2 weeks ago
          Anonymous

          It's going to be so funny when 9600 ends up being faster than 7800x3d at 250$.

    • 2 weeks ago
      Anonymous

      Nothing it's Intel spammer going crazy. They are going to be 20% faster while being more power efficient.

  11. 2 weeks ago
    Anonymous

    Show me the performance numbers and I'll make my own conclusions

  12. 2 weeks ago
    Anonymous

    I'll wait for the intel 15th gen vs ryzen 9000 comparison before buying either one

    • 2 weeks ago
      Anonymous

      I've been waiting to pull the trigger on computer parts for 5 years now

      so glad I switched to Apple M.
      2 years of headache free computing and silence and no heat.

      M chips are tempting but also really not at the same time because apple

  13. 2 weeks ago
    Anonymous

    The only difference is it went from Zen 4 to Zen 5, it's a rebadge of the same fricking chip.

  14. 2 weeks ago
    Anonymous

    >No core upgrade
    >No power efficiency upgrade
    >No thermal upgrade
    AMD seems to be at the 4c/8t stage intel was stuck in.
    Seems like the biggest nothingburger™ release from AMD and that's coming from a long term AMD fan.

    • 2 weeks ago
      Anonymous

      stop posting the same thing homosexual

    • 2 weeks ago
      Anonymous

      It's afraid

    • 2 weeks ago
      Anonymous

      Intel, please fix your shillbot

  15. 2 weeks ago
    Anonymous

    that sounds like fx-8000 all over again

  16. 2 weeks ago
    Anonymous

    CPU and GPUs are way more than fast enough these days.

    Now its all about efficency and getting the power draw down while maintaining the same level of performance.

    Imagine running a cpu at 100w but the performance of a 7950x/14900k which is a big deal as it means less electricity, cooling (big coolers) and heat

    • 2 weeks ago
      Anonymous

      You can already limit a 7950x to 105W for like 95% of stock performance.
      Inefficiency is a choice.

      • 2 weeks ago
        Anonymous

        How? Is that a PBO setting or something?

        • 2 weeks ago
          Anonymous

          All Ryzens have ECO mode in BIOS. That's how most people run them.

          • 2 weeks ago
            Anonymous

            Most people don't know what the frick you're talking about. There's no "ECO mode" in X570 I AORUS PRO WIFI's BIOS.

          • 2 weeks ago
            Anonymous

            Most people buy laptops and prebuilds and don't even know what CPU they use. Anyone actually picking parts knows shit like BIOS settings.

          • 2 weeks ago
            Anonymous

            Read your mobo's manual. Every single board has an option if not a fully page dedicated to changing cTDP and power behavior

          • 2 weeks ago
            Anonymous

            That's why I asked if it's related to PBO options because that's the only place I know of where I can adjust PPT and other power-related settings

          • 2 weeks ago
            Anonymous

            MSI owner here, the manual does not document anything other than how to connect the wires on the motherboard. And still, it doesn't say which LGBT header (because there are two - 5V and 12V) is for the LEDs on the CPU cooler

          • 2 weeks ago
            Anonymous

            There's no eco mode in X670E.
            The TDP guardrails do nothing.
            Even when you disable core boost and performance boost, something is still trying to overclock your CPU for shits and giggles.
            t. 7900 owner

      • 2 weeks ago
        Anonymous

        7950x in 105w mode is amazing, the efficiency is something i can't even grasp

      • 2 weeks ago
        Anonymous

        Damn, for a moment there I thought bigger number better and got really confused

    • 2 weeks ago
      Anonymous

      >CPU and GPUs are way more than fast enough these days.
      lol, nothing is ever fast enough
      >Now its all about efficency and getting the power draw down while maintaining the same level of performance.
      Says who, for generations the top of the line GPUs have only had higher and higher TDPs and beefier cooling.
      >Imagine running a cpu at 100w but the performance of a 7950x/14900k which is a big deal as it means less electricity, cooling (big coolers) and heat
      You can, buy 7950X/14900K and undervolt/underclock/power limit to 100W and you'll still have the majority of the performance in most tasks.

    • 2 weeks ago
      Anonymous

      >Imagine running a cpu at 100w but the performance of a 7950x
      I mean, a 7950X3D runs at 120W or some shit at stock with no tweaks or adjustments and it's very close in MT workloads while being ahead in anything that needs the cache.

    • 2 weeks ago
      Anonymous

      I just want cpus that lose minimal performance from power limiting. Amd's monolithic ryzen chips perform that role really well

  17. 2 weeks ago
    Anonymous

    Was hoping to see some funny opinions by people with zero professional experience, and IQfy always delivers

    What happened man this board used to have some genuinely competent people back in the days

  18. 2 weeks ago
    Anonymous

    I'm waiting for AM6

  19. 2 weeks ago
    Anonymous

    NPU?
    also remove AVX-512, give me 32 cores and I'll be happy

    • 2 weeks ago
      Anonymous

      Nah, I'd rather have AVX-512.

    • 2 weeks ago
      Anonymous

      The Xilinx AIE is only featured in the Strix Point APU family.
      The Zen5 desktop parts use the same IO die as Zen4 parts and have the same 2CU RDNA2 IGP which was the big giveaway

      I thought we couldn't go below 7nm because of the physical quantum tunneling limits?

      Stop listening to reddit.
      Quantum tunneling is a theoretical issue. The process engineers who continue to advance CMOS scaling use appropriate materials to keep charges exactly where they're supposed to be. Back in 2013-2014 Applied Materials showed an actual working model including all material selections in all metal layers to produce 2nm nanowire devices.
      inb4 some mouth breather whines about process naming conventions

      • 2 weeks ago
        Anonymous

        So 2nm nanowire devices means we're...two years away from that, and why isn't that the same issue in your head, just stretched out to now instead of five years ago?

        • 2 weeks ago
          Anonymous

          >So 2nm nanowire devices means we're...two years away from that,
          Samsung has taped out a 3nm MBCFET chip already, its right now. This process was one of the last things worked on by the Common Platform Alliance and they've had actual fully functional test chips for years. IBM initially demoed it at 2nm, only a slightly denser pitch than Samsung's now mass market ready 3nm class node.

          People(morons not even in the industry) have been crying about quantum tunneling since the 32/28nm era. When intel's 14nm Trigate process launched hundreds of totally uneducated spergs on the internet began opining about chips dying within 3 years because of aberrant tunneling effects. And yet here we are.

          • 2 weeks ago
            Anonymous

            Those terms, if I'm interpreting things correctly, still set the limit in your mind around 2nm. What do you have in your mind comforting you that this can continue?

          • 2 weeks ago
            Anonymous

            1nm nanowires have been demonstrated in the mid 2010s. CMOS scaling has quite a long way to go yet. Even if FEOL cannot physically scale any more the MOL and BEOL can still become significantly denser. Back end signal and power routing takes up a lot of space compared to the actual switching logic on the front end.

          • 2 weeks ago
            Anonymous

            Back end is neat

            ?feature=shared

          • 2 weeks ago
            Anonymous

            >Those terms, if I'm interpreting things correctly, still set the limit in your mind around 2nm. What do you have in your mind comforting you that this can continue?
            Current 2nm processes still have 20 something nm gate pitch, so we have a long way to go before the gates themselves approach 2nm

          • 2 weeks ago
            Anonymous

            Photonics homie

          • 2 weeks ago
            Anonymous

            2nm, 3nm, 14nm etc are all just marketing terms and has nothing to do with their actual size. If you actually measured the physical gate width, we've been stuck at around 20-30nm for about a decade.
            These fabs just found other ways of increasing density and power efficiency.

          • 2 weeks ago
            Anonymous

            I'm surprised it took this long for one of you midwit mouth breathing idiots to show up.
            You aren't smart for pointing out the obvious that everyone knows. Process naming nomenclature has been consistent since the late 1990s. Shit eating morons going UMMM ACHSHEWALLY is fairly recent, probably because you just found out about this in some youtube video.

          • 2 weeks ago
            Anonymous

            this gay

            2nm, 3nm, 14nm etc are all just marketing terms and has nothing to do with their actual size. If you actually measured the physical gate width, we've been stuck at around 20-30nm for about a decade.
            These fabs just found other ways of increasing density and power efficiency.

            is mostly right. nowadays process names are about tolerances dimensions, while some time ago was the transistor minimum feature sizes.
            anyone that works on manufacturing know that is very important to make such difference.

    • 2 weeks ago
      Anonymous

      Wow super moronic onions itp
      NPUs are garbage vs. literally any GPU released in the last 6 years. Qualcomm's X NPU is so fricking weak that co-pilot will not even work on it without a GPU. Sad, gay feature used to lure idiot investors.
      32 cores are available - on HEDT. If you really want them, they're there.
      And what kind of professional workload are you running that needs NPUs, 32 cores but not big vector math? Nothing.

    • 2 weeks ago
      Anonymous

      AMD does avx512 right, namely without downclocking or impacting performance of non-avx code.

      • 2 weeks ago
        Anonymous

        The downclocking was a lithography problem, not an implementation problem.
        Zen 5 will do AVX512 "the intel way" (aka full 512 bit FPU) but I suspect it won't have downclocking problems because AMD currently have way more headroom in their power budget.

        • 2 weeks ago
          Anonymous

          Its an implementation problem
          Intel's implementation had 1x avx512 + 2x avx256 fused units to process avx512 instructions and had downclocking issues.
          Then they cut it down to only 2x avx256 on consumer alderlake and the downclocking went away ((if you had an earlier 12th gen with avx512 in-tact)). Amd does the same 2x avx256 fused implementation as alderlake which uses less power than full bore avx2 and doesn't downclock as well.

  20. 2 weeks ago
    Anonymous

    Disappointing
    And the 9950X3D probably won't come out for another year as well.

    • 2 weeks ago
      Anonymous

      7950X3D released 5 months after the 7950X
      I hope they bring that down to under 3 months

      • 2 weeks ago
        Anonymous

        Zen4 X3D chips launched under heavier competition. Its expected that Zen5 X3D chips will be a mid gen upgrade since intel has nothing that can compete with them.

        • 2 weeks ago
          Anonymous

          >intel has nothing that can compete with them
          Arrow Lake might actually be a gaming beast because Intel outsourced all parts to TSMC. Yes, Intel didn't use their own fabs, not even for the basic b***h parts. lol

      • 2 weeks ago
        Anonymous

        One would hope so
        7950X3D was the first X3D release so the lag time makes sense.
        This time around though, it doesn't really make sense, hopefully in future they can have release parity.
        I am surprised they did not at least "announce" 9950X3D.

        • 2 weeks ago
          Anonymous

          X3D started with the Zen3 family

          • 2 weeks ago
            Anonymous

            Shit I need to stop smoking weed.

          • 2 weeks ago
            Anonymous

            You should grow up and start putting it in your ass instead

  21. 2 weeks ago
    Anonymous

    I thought we couldn't go below 7nm because of the physical quantum tunneling limits?

  22. 2 weeks ago
    Anonymous

    no thanks. i will stick with my 5950X for another 10 years

  23. 2 weeks ago
    Anonymous

    I wish I got a 7950x instead of this power hungry 13900k. I can power limit it but it loses quite a lot of performance when I tried. Also I lost confidence in even touching the bios setting after I had to RMA my first 13900k since it degraded over serval months at stock settings.

  24. 2 weeks ago
    Anonymous

    god fricking damnit, i just got a 7600X like 2 months ago

    • 2 weeks ago
      Anonymous

      Single generation upgrades are rarely worth it anyway. Fingers crossed AM5 gets Zen 6 X3D on 3nm or better.

  25. 2 weeks ago
    Anonymous

    >still am4
    Sounds good to me.

  26. 2 weeks ago
    Anonymous

    wake me up when 0.1nm are here

  27. 2 weeks ago
    Anonymous

    Threadripper already exists what more do you want

  28. 2 weeks ago
    Anonymous

    >still 16c
    I suppose I can keep the 5950X around for another generation... Some of my workloads do benefit from AVX-512 though, so I'm getting restless. My laptop (Zen 4) currently handles those things significantly faster. No AMD, I will NOT buy a threadripper you israelites.

    • 2 weeks ago
      Anonymous

      >AVX-512
      bro we still have people not utilizing threads or multiple cores you ain't gonna use that shit until 2050 minimum.

  29. 2 weeks ago
    Anonymous

    If AMD wants me to even consider the 9950X3D, put the V-Cache on both CCDs. Otherwise I won't even look at benchmarks and buy Arrow Lake. 7950X3D was such a shitshow compared to 7800X3D it's embarrassing AMD decided to release it. 7800X3D on the other hand was just 8 cores so if you wanted both gaymen perf AND nT, you had no choice but Intel.

    • 2 weeks ago
      Anonymous

      Literally why? V cache only helps games and they shit themselves if they schedilw threads across multiple CCDs.

      • 2 weeks ago
        Anonymous

        Because the scheduler doesn't determine what threads to put on V-Cache CCD, Windows detects if a game runs by checking if Xbox Game Bar engaged. Don't ask me why, ask AMD.
        Theoreticals are very fricking nice because nobody on IQfy has actual hardware but if you go to overclock.net or wccf comments people with 7950X3D and into overclocking will tell you how annoying it is. One guy I know just runs the 7950X3D with one CCD because it clocks higher than 7800X3D. On paper it sounds nice but it doesn't work and I'm tired of beta testing after the TPM and USB stutter on X570. I want the 9950X3D to be foolproof like Intel 13th gen has been for me. I don't care if the 9950X3D PBOs too high and crashes or fricking whatever because I can fix it myself. This happened on Intel 14th gen, sucks for normies but anyone tech literate just adjusts the all core clockspeed and solves it in 5 minutes. However I cannot fix the scheduler or shitty microcode. Stuff like Z490 launch fiasco and X570 (B550 however was perfect) suck because the user can't fix them.

  30. 2 weeks ago
    Anonymous

    they're finally fixing the inter-core latency on this one

    • 2 weeks ago
      Anonymous

      Zen 3 and Zen 4 have excellent intercore latency though?

      • 2 weeks ago
        Anonymous

        Did you forget about the AMDip?

        • 2 weeks ago
          Anonymous

          AMDip is gone from Zen4 and Zen4 3D. Jufes has AMDip on his 7800X3D because his FCLK is unstable. If your Intel SA is unstable you crash, unstable FCLK on AMD causes performance regression first and crashes only in egregious cases. He knows this and knows he fricked up and is now too afraid to admit it which is a shame because his content has otherwise been amazing.
          I'm the last person to defend shitMD and their cultists but Zen4 genuinely doesn't have the dip anymore, it's exclusive to earlier CPUs and all Radeon GPUs.

    • 2 weeks ago
      Anonymous

      what for? it still mogs Intel's sphincter bus

    • 2 weeks ago
      Anonymous

      Zen4 has excellent inter-core latency. Zen1-3 were garbage but Zen4 is a great architecture. Not as good as ADL/RPL but AMD is getting there, it'll be funny if they achieve uarch parity with Intel by the time they lose node superiority via TSMC because everything points to Intel leapfrogging TSMC with 18A. Assuming they don't frick up and delay it but it's Intel we're talking about so they probably will. Their foundries are cursed.

      • 2 weeks ago
        Anonymous

        Bro thinks he's Ian Cutress

        • 2 weeks ago
          Anonymous

          >Ian Cutress
          The king of midwits

  31. 2 weeks ago
    Anonymous

    Does anyone actually buy AMD to care?

  32. 2 weeks ago
    Anonymous

    I didn't read the post. What did they do now? 30 minute long POST?

  33. 2 weeks ago
    Anonymous

    >add moar cores
    stfu gay games are probably programmed to use like 3 of them

    • 2 weeks ago
      Anonymous

      bro just stepped out of 2014

      • 2 weeks ago
        Anonymous

        Pajeet coders send their regards

  34. 2 weeks ago
    Anonymous

    gotta wait for another gen when normies will finally sell their 5800x3d/5900x so i can upgrade my am4 build

  35. 2 weeks ago
    Anonymous

    Sold my AMD stock and bought into qualcomm a cool minute ago.

    ARM is the future.

    • 2 weeks ago
      Anonymous

      >Sold my AMD stock
      nice play moron

  36. 2 weeks ago
    Anonymous

    I've been out of the PC building game for a long time. Did they ever fix the problem with AM4 where you needed to ebay an old CPU so you can flash the bios to accept your "compatible" brand new CPU?
    I can't be fricked trawling through motherboard listings trying to tell the difference between 9000 series "compatible" AM5s and 9000 series "will actually work out of the box" AM5s.

    • 2 weeks ago
      Anonymous

      any decent board has usb flashing now

  37. 2 weeks ago
    Anonymous

    >No core upgrade
    prepare your anus

Your email address will not be published. Required fields are marked *