Watch Intel’s “Performance Unleashed” event at 7 AM PT/10 AM ET

Comments closed
    • ronch
    • 1 year ago

    No thanks I’m going Ryzen. You’re only giving us all this because of AMD and if it weren’t for them you’d still be shoving 4C/8T down our throats for $350 until who knows when. So out of principle alone, I’m going AMD all the way until the Apocalypse.

    • anotherengineer
    • 1 year ago

    I predict a lack of supply and a large price increase.

    • christos_thski
    • 1 year ago

    The next intel event I’m hyped for is still 2 years away (and I can’t wait).

      • NTMBK
      • 1 year ago

      What, the announcement that they’re going fabless?

        • christos_thski
        • 1 year ago

        Nope, though I chuck(u)led at that.

        Arctic Sound, their discrete gpu. GPU market has been stagnating too long now, it’s in dire need of some waves.

    • techguy
    • 1 year ago

    Here’s to hoping 5.5GHz+ is the new norm thanks to the magic of soldered IHS. Also I hope it’s less than $500. That’s a bit much for a “mainstream” desktop CPU.

      • Krogoth
      • 1 year ago

      Soldering isn’t going do anything honestly. It just makes the IHS fitting at the factory more consistent. It’ll just reduce the chances of “hotspots” and ironically make it harder to delid.

      ~5.5Ghz with all cores at full load will require exceptional binning and high-end to borderline exotic cooling to keep it relatively tame.

        • Voldenuit
        • 1 year ago

        My understanding was that the tin solder has a significantly higher thermal conductivity than intel’s current TIMs.

        I’ve seen people post videos of stock 8700Ks getting to 83C on custom water loops, and dropping over 20C after delidding.

        My own experience is I jumped straight to delidding and liquid metal when I got my 8700K, and it hovered at 58C with a 12-thread load on a budget AIO.

        Ryzen doesn’t seem to have this problem, and they use solder.

          • Krogoth
          • 1 year ago

          The whole “TIM/IHS thermal issues” came almost entirely from QC issues at the factory back with Ivy Bridge and first-batch of Haswell chips. Intel’s tools and QC procedures were too used to using solder. When they did the switch over. The same standards and tools were simply inadequate.

          That’s how you got Ivy Bridge and Haswell users who got massive hotspots on stock chips. They were some people were reporting a difference of 20-25C difference from one core to another at full load. Just simply just de-lidding and reseat the IHS fixed the problem. I happened to be one of the few lucky Ivy Bridge users who didn’t have the hotspot(s) problem.

          Intel mostly fixed the issue with second-batch of Haswell a.k.a Devil’s Canyon when they updated TIM and revamp their QC and chip packaging. The hotspot problems became mostly a non-issue. People still de-lidded chips though if they wanted optimal thermal dissipation (replacing the stock TIM with a better medium).

          The real reason why Intel even switched to TIM is because of the Nvidia’s “Bumpgate” debacle. They feared that long-term thermal cycling on laptop platforms might run into reliability issues down the road where the solder would start to crack the substrate on the silicon. The bean counters loved the idea because it also happened to reduce production costs.

          Ironically, die-hard overclockers actually prefer the whole TIM/IHS setup because it makes de-lidding a lot less risky.

          The whole “return to solder” is just a marketing move on people who didn’t understand what the whole controversy was really about.

            • Voldenuit
            • 1 year ago

            [quote<]The real reason why Intel even switched to TIM is because of the Nvidia's "Bumpgate" debacle. They feared that long-term thermal cycling on laptop platforms might run into reliability issues down the road where the solder would start to crack the substrate on the silicon. The bean counters loved the idea because it also happened to reduce production costs. [/quote<] That's especially strange, because the last laptop I re-pasted (7700HQ on Asus GX501) had a naked die interface to the heatsinks. I guess most non-performance laptops might opt for the safer route, but they wouldn't be socketed BGA chips anyway. It was the bean counters. And I also had a Devil's Canyon, which dropped about 15C after delidding.

            • Krogoth
            • 1 year ago

            Nope, it it was Intel Executives being super paranoid of dealing with their own version of “Bumpgate” down the road when CPU engineers warn them of potential thermal cycling/material fatigue issues with solder.

            • techguy
            • 1 year ago

            Bumpgate? Really? Explain all the soldered Sandy Bridge era (and prior) CPUs out there that have been chugging along with a 24/7 OC since 2011 or earlier then.

            • Krogoth
            • 1 year ago

            The thermal density of newer chips were catching up (22nm and beyond) and material engineers were merely concerned about potential long-term issues with solder affecting portable platforms.

            Bumpgate debacle (Nvidia end-up playing billions for it and it damaged their good-will with OEMs) made Intel scared enough to take concerns of those engineers seriously. The bean counters were obviously on board since it reduced production costs.

            Intel wasn’t worrying about chips dying on some stupid “1337 killer OC’ing” setup. They were worrying about it affecting laptops which have taken over the mainstream computing scene. They didn’t want risk pissing-off their OEM partners and cause general public outrage with this potential long-term issue.

            Notice how Intel switch over to IHS/TIM setup not long after Bumpgate debacle was resolved? Not a coincidence.

            BTW, those crazy Sandy-Bridge OC’ed rigs don’t undergo the same thermal cycling as a laptop system or your run-of-the-mill OEM system.

            • techguy
            • 1 year ago

            1) Intel has never sold a soldered laptop CPU – if it’s naked chips you want then these ARE the droids you’re looking for
            2) I still don’t buy the “avoid bumpgate” theory – soldered CPUs have been around for a long time, and many of these chips are still in operation today, including overclocked systems
            3) Intel sold said overclocked chips to OEMs as well as enthusiasts so the idea that they have not been subjected to frequent thermal cycling is demonstrably false

            • Krogoth
            • 1 year ago

            1.) They do sell soldered laptops CPUs. It depends on the type of SKU and platform it was going for. DTR SKUs were definitely soldered.

            2.) Some Intel engineers were worried about solder being an potential issue with 22nm and beyond since the heat density was getting higher and substrates were thinner then before. They projected that thermal cycling and material fatigue could cause the solder to wrap and eventually crack the substrate = dead chip.

            3.) It is not about overclocking SKUs. It is about protecting mainstream tiers from a potential long-term product defects which will cause an massive recall and PR storm. Nvidia’s bumpgate debacle was enough to scare Intel at the time. Remember that they made these decisions back in the start of 2010s. If reducing production costs was such an pressing concerned they would have moved to TIM long ago and likely never bored with soldering in the first place.

            • techguy
            • 1 year ago

            1) correction: *sold* – Yes, I admit that 15 years ago Intel was shoving desktop P4s into DTR laptops because they literally had nothing else to compete with AMD. Not entirely relevant to the market today though, wouldn’t you say?

            2) do you have links to an actual engineer saying this, and not a marketeer?

            3)this was a typo – I meant to say “soldered chips” as in, Intel sold plenty of soldered chips to OEMs.

            • Krogoth
            • 1 year ago

            There are several articles back during the whole thermal paste fiasco (2010-2011) explaining why Intel did it in the first place. It made perfect sense from a business standpoint. The only people who had grievances over the whole affair where overclockers who got spoiled rotten by how easy it was to overclock Conroe => Sandy Bridge).

            • techguy
            • 1 year ago

            I’ve delidded numerous chips since and including Ivy Bridge, and every single one of them had both a reduction in temperature, and hit higher overclocks thanks to not thermally throttling post-delid.

            You are wrong.

            Here’s my firsthand knowledge with my own CPUs from the very timeframe you mention:
            My 3770k hit 4.7GHz at TJMax pre-delid
            My 4790k (the Devil’s Canyon CPU you cite which “fixed all these issues”) hit 4.8GHz pre-delid, again at TJMax

            Post delid:
            3770k: 4.9GHz and 72C
            4790: 5.0GHz and 77C

            All testing conducted with IntelBurnTest 2.54 max mem test and same watercooling setup.

            • Krogoth
            • 1 year ago

            Did you de-lid the chips and just reseat the IHS?

            If you just de-lid the chip and ran the silicon “naked” then obviously is going to be a noticeable reduction. If you just replaced the TIM with a superior paste then that kinda throws things out of the window.

            The reports of “hotspots” almost evaporated (only die-hard overclockers who refused to delid their chips griped about it) when Intel revamp their IHS/TIM procedure.

            As far as majority of people and Intel were concerned. The issue was fixed.

            • techguy
            • 1 year ago

            I re-seated the IHS and removed all epoxy from both PCB and IHS. The IHS was held in place only by the socket retention bracket. My results are not one-off results, there are literally thousands of people with the same experience. [url<]https://www.google.com/search?q=delid+temperature&rlz=1C1GCEB_enUS785US785&oq=delid+temperature[/url<] As for the issue being having been remediated from Intel's perspective, sure, I can agree with that. However, given the fact that overclocking headroom clearly decreased when Intel made this change to manufacturing post-Sandy Bridge, there is still an effect to consumers so it is a matter of perspective.

            • Krogoth
            • 1 year ago

            Again, you throw things out of the window with that. You are pretty much running the chip nearly “naked” that at point (hoping that weight of the heatsink/water block will keep the “loose” IHS in place). You are doing the poor’s man version of a “CPU shim”.

            Not exactly delidding and reseating IHS eh?

            • Voldenuit
            • 1 year ago

            Mine was delidded and reseated with 4 drops of superglue.

            I had researched using silicone RTV, but most overclocking forums said the superglue was easier to apply and clean up, with similar/identical results.

            • techguy
            • 1 year ago

            You’re over-complicating this.

            There’s no such thing as “nearly naked” in terms of delidding. You either have a lid, or you don’t. All of my “de-lidded” CPUs had the IHS reseated. The IHS is held in place by the adhesion of the liquid metal in addition to the clamping pressure of the CPU socket retention bracket. The only difference between my method and that of many others is the use of some kind of sealant or adhesive between the IHS and the CPU package. This has the effect of raising the contact point of the IHS with the CPU die, an outcome which is to be avoided at all costs and is in fact an essential component of this methodology, particularly with Ivy Bridge/Haswell CPUs. Even with Devil’s Canyon there was still significant gap between the underside of the IHS and the CPU die, just not as much as there was before.

    • Krogoth
    • 1 year ago

    A wild Raja appears!

    “BEHOLD THE iVEGA! IT WILL REVOLUTIONIZE GAMING GRAPHICS!”

    #PoorAMDRTG
    #PoorNavi
    #PoorNvidia
    #PoorRTX

    • Neutronbeam
    • 1 year ago

    CHUCKULA OVER-REACTION CONFIRMED!

    [well, I’M laughing.]

    • DPete27
    • 1 year ago

    Paper launch only I see.

    • chuckula
    • 1 year ago

    Sorry everybody. The event is canceled.

    Intel was arrested for violating Santa Clara’s leash ordinance.

    • JosiahBradley
    • 1 year ago

    20 minutes until massive disappointment.

Pin It on Pinterest

Share This