Comments Locked

164 Comments

Back to Article

  • Marlin1975 - Tuesday, May 11, 2021 - link

    So even with a higher rated TDP it still has to have a cut down gpu with only 32 execution units.
    My guess is it will still use more power and still have issues.
  • coschizza - Tuesday, May 11, 2021 - link

    only 32 execution units because you use a separate gpu to play games
  • Aloon - Thursday, May 13, 2021 - link

    Separate GPU?

    Are you taking this material on tour, when restrictions are removed :-)
  • futurepastnow - Friday, May 14, 2021 - link

    Then why even 32EU? Since models without a discrete GPU aren't meant for gaming, why not cut it down further.
  • mode_13h - Saturday, May 15, 2021 - link

    Because they're unwilling to lock out customers who have moderate uses for a GPU yet don't want a dGPU.

    However, I think a better explanation is that the chip was originally intended for desktop use, as well.
  • Spunjji - Monday, May 17, 2021 - link

    I have a different theory here - 32EU gives them a gen-on-gen improvement with room to disable some units for segmentation purposes. If they went for 24EU then they'd be selling something with similar performance to the previous generation(s), and the harvested units would show a regression in terms of GPU performance. Not a good sell for the latest and greatest.
  • mode_13h - Tuesday, May 18, 2021 - link

    Could be. We'll have to wait and see whether they actually do disable any.

    However, I think you have a good point that they want to avoid being seen to regress in any way.
  • yeeeeman - Tuesday, May 11, 2021 - link

    yes, since you don't really use the igpu, only for displaying basic stuff.
    yes, it will use more power, tsmc process is much better, that is for sure.
  • yeeeeman - Tuesday, May 11, 2021 - link

    and also, not only that tsmc process is much better, but intel needs to clock these cpus very very high in order to stay competitive. if they were at 3ghz, like say 6700hq was when skylake launched, things would be much more efficient.
  • 29a - Tuesday, May 11, 2021 - link

    I can't wait to see how these super awesome XE iGPUs perform. I'm sure they're going to be as awesome as Intel has been telling they were going to be this whole time.
  • mode_13h - Tuesday, May 11, 2021 - link

    Huh? Go read the review of the Tiger Lake U-series. It has a much bigger iGPU of the same generation (96 EU instead of 32).
  • SarahKerrigan - Tuesday, May 11, 2021 - link

    Xe has been out in TGL-U for a while. It performs well, and is generally competitive with AMD iGPUs.
  • Spunjji - Tuesday, May 11, 2021 - link

    Since they fixed a bunch of their driver bugs, it's arguably better than the AMD alternatives. Roll on RDNA 2!
  • The_Assimilator - Tuesday, May 11, 2021 - link

    So it's competitive with the bottom of the barrel! Good to know.
  • mode_13h - Tuesday, May 11, 2021 - link

    I think the cut-down iGPU is because 10 nm SuperFin die area is still scarce and expensive, plus they expect most machines using this will be equipped with a dGPU.

    Although, another possible explanation for the small iGPU is that is was originally meant to be packaged (primarily) for desktop use. I speculate about that further, below.
  • deil - Wednesday, May 12, 2021 - link

    cut down IGPU means less power and less used space, they want to profit from it so, this GPU is a compromise for current shortage that COULD run something in mid settings reliably and tiny area to make sure they can make a lot of them.

    if tech that joined intel dgpu to igpu will kick off nicely, then those gpu's will grow to make people pick their dGPU's for extra horsepower, for now they want as many out there as they can spew.
  • mode_13h - Wednesday, May 12, 2021 - link

    > if tech that joined intel dgpu to igpu will kick off nicely, then those
    > gpu's will grow to make people pick their dGPU's for extra horsepower

    It's a gimmick. Compared with a dGPU that has like 512 EUs, the 32 EU iGPU is a rounding error and probably not worth the overhead of trying to split workload between.

    In a system with an iGPU and dGPU, it makes more sense to use the iGPU for independent tasks, such as: AI, physics, or audio. This is something you can already do today, and it doesn't require the iGPU and dGPU to be the same generation or even brand!
  • Spunjji - Thursday, May 13, 2021 - link

    100% - confused by the number of people who expect Intel to crack asymmetric GPU rendering when AMD already tried and failed with their first APUs.
  • eek2121 - Wednesday, May 12, 2021 - link

    Except the die still includes the same area for a full, fat GPU, so it is most definitely NOT about yields.
  • mode_13h - Wednesday, May 12, 2021 - link

    Huh? Are you saying the die has dead silicon on it, where a larger GPU could go?
  • Spunjji - Thursday, May 13, 2021 - link

    They seem to be, which is wrong
  • timecop1818 - Wednesday, May 12, 2021 - link

    Not as many issues as the AMD competition, lol. At least Intel stuff works.
  • Spunjji - Thursday, May 13, 2021 - link

    It really didn't on launch, but okay
  • dsplover - Tuesday, May 11, 2021 - link

    Awesome news if they can supply DIY enthusiasts.
    Maybe there is hope for Intel.
  • Irata - Tuesday, May 11, 2021 - link

    With laptop CPU ?
  • mode_13h - Tuesday, May 11, 2021 - link

    Most DIY boards don't have a soldered-down CPU. That's what this is.

    Just sit tight and wait for Alder Lake.
  • Silver5urfer - Tuesday, May 11, 2021 - link

    What is ADL going to be ? A miracle out of a hat ? Nope. It will be a mess. Small x86 cores to makeup for the lost SMT performance is min spec that Intel is targeting at, then it will have DRAM and PCIe segmentation. It's a worthless stopgap product to stick to. On top expensive Mobos.

    RKL and CML are the better products overall since they are mature enough platforms out there with almost no performance loss vs Ryzen parts in most tasks unless that 16C is counted for Productivity. Then I cannot stop mentioning how X570 and Ryzen 5000 have been plagued by the stupid USB problem mess (yeah it's still happening a quick run down on forums and reddit with simple usb search will yield that mess on top a few YT channel users mentioned like HardwareCanucks and Ali)
  • Spunjji - Tuesday, May 11, 2021 - link

    Rocket Lake is even more of a joke than your compulsive ranting about Ryzen USB issues.
  • Silver5urfer - Tuesday, May 11, 2021 - link

    You can bugger of dude just leave AT why even bother with your nonsense here's your no issues damage control idiocy.

    reddit.com/r/Amd/search/?q=USB&include_over_18=on&restrict_sr=on&sort=new
  • The_Assimilator - Tuesday, May 11, 2021 - link

    Nobody cares.
  • Silver5urfer - Tuesday, May 11, 2021 - link

    Crawled out of nowhere and did post another useless reply.

    Wonder how much you do care since you signed in and replied. Also so many people on Reddit care enough to not buy AMD again due to this unstable bs problems. As a new buyer I would care enough and many people there already asking the same question to community to help with their purchase.

    You guys won't be even hired by AMD to do their PR bidding.
  • Qasar - Wednesday, May 12, 2021 - link

    i know about 10 people that are waiting for ryzen 5000 to be in stock at better prices, none of them want intels power hungry cpus, and tired of intels BS the last few years. a couple of them couldnt wait, and picked one up with an x570 board, were aware of some of the issues, but still went with AMD, with no issues,so, some may have issues, some don't. intel has their own issues over the years with their products so keep trying to make them look like a saint, cause they aren't.

    and looks like you were already hired by intel to its PR bidding.
  • Silver5urfer - Wednesday, May 12, 2021 - link

    Everyone I know build AMD machines, and are into them only but the problem is their instability.

    Why do I get this absolute trash smearing fanboy loser defense whenever I mention problem with AMD.

    "intel has their own issues over the years with their products so keep trying to make them look like a saint, cause they aren't." What the f is this, I'm talking about CML and RKL and Zen why the hell Intel past issues and etc come did I mention any past issue with AMD or Intel ??? Who tf cares about a corporate here, I'm not but your kind does it hardcore. I'm here to buy new PC hardware and use it for many years because I'm the one paying money not you and it's my responsibility to lookout to potential issues and all to be prepared well.

    And when I try to learn what the new HW is capable of I saw WHEA BSOD RMA trash problems on AMD with OCN 80 pages where the AGESA updates and RMAs fixed but the USB problem still remains to date. You seem to be fine with using a buggy crap product out of box in 2021.
  • Qasar - Wednesday, May 12, 2021 - link

    " Everyone I know build AMD machines, and are into them only but the problem is their instability.
    " i dont have any instability with the 3 AMD zen based comps here, they have been just fine since day one.
    "Why do I get this absolute trash smearing fanboy loser defense whenever I mention problem with AMD. " probably because you make it out like intel is a saint and perfect ? which they are not, as they have had their own issues. but from what i can tell you being intel fanboy, doesnt want to admit it.

    " And when I try to learn what the new HW is capable of I saw WHEA BSOD RMA trash problems on AMD with OCN 80 pages where the AGESA updates and RMAs fixed but the USB problem still remains to date. You seem to be fine with using a buggy crap product out of box in 2021. " thats find and good, but keep in mind, not every one has those issues, or any issues at all. i know of one guy that posts on another site, that complains his amd based vid card crashes 10 times aday, and when asked about what he is running, the which PSU, etc, he stops posting on that thread.
  • Silver5urfer - Wednesday, May 12, 2021 - link

    Go to comment section on page 7 and see how I praised Intel even doing so I got some backlash about calling out their BS moves on the BGA silicon crap they pushed all these years and how restrictive these TGL machines will be and how Hot they also will be, Alienware 15 this time skipped HK processor because of the Heat, you guys have no knowledge and then call out others because someone mentioned a flaw or hw existing issue on other brand. SMH I'm calling Intel as saint.
  • ottonis - Wednesday, May 12, 2021 - link

    It's fascinating that people develop sort of a relationship with the objects the own and this kind of relationship may even resemble some aspects of interpersonal relationships. For example, Greg is a good old friend of mine. If anybody of my classmates says Greg is dumb, I will defend Greg and attack those other guys - irrespective of the fact whter Greg is really "dumb" or not.

    So, "fanboyism" is a part of us humans... ;-)

    That being said, it all depends on priorities. If I have a render or video encoding job to be done within a certain deadline, then I will use the computer/CPU/GPU that will provide me the best chances to accomplish my task. That will be fast enough and that will not stand in my way by being unstable etc.
    Ironically, the most successful and most productive colleagues of me had the oldest PC machines.
  • mode_13h - Wednesday, May 12, 2021 - link

    > So, "fanboyism" is a part of us humans... ;-)

    It's probably worth distinguishing between fanboyism and tribalism. A fanboy is someone obsessed with a celebrity, brand, product, band, etc. It can take on even a sort of religious fervor. They will defend the object of their idolatry to the point of death or at least dishonor.

    Tribalism is a little bit different. It can manifest similarly, but the allegiance could be more towards the group, rather than the person, object, etc. at the center. It can be motivated out of defensiveness or a sense of comradary or identity.

    In either case, I'd much rather someone be a rabid fan of -- or join a tribe around -- something like a sports team or a GPU brand than a suicidal cult leader, a violent religious sect, or a murderous political regime. So, my advice is to take any fanboyism or relatively harmless tribalism with a grain of salt.
  • Spunjji - Thursday, May 13, 2021 - link

    @Silver5urfer - "Why do I get this absolute trash smearing fanboy loser defense whenever I mention problem with AMD."

    Nice projection. You routinely show up to whine about specific AMD issues on unrelated threads, that's why people tell you to cram it - apparently this looks to you like being constantly under assault from "fanboy losers", but from our perspective it looks like you obsessing over an issue and lots of different people taking turns to ask you to please, PLEASE stop grinding that axe.
  • vladx - Wednesday, May 12, 2021 - link

    @Silver5urfer: Don't waste your time with AMD fanatics, logic and facts are too much to process for their pea-sized brains.
  • Qasar - Wednesday, May 12, 2021 - link

    " Don't waste your time with AMD fanatics, logic and facts are too much to process for their pea-sized brains." this coming from an intel fanatic.
  • Spunjji - Thursday, May 13, 2021 - link

    @vladx - more projection. You're one of the least-informative and most reliably biased posters in these comments.
  • Spunjji - Thursday, May 13, 2021 - link

    Did I say there were "no issues"? No, I didn't. I objected to your compulsive ranting. Still do.
    Get a better hobby, go outside, touch grass - whatever you need to do. 🤷‍♂️
  • mode_13h - Tuesday, May 11, 2021 - link

    > What is ADL going to be ? A miracle out of a hat ? Nope. It will be a mess.

    At worst, it'll be Tiger Lake-H on a bigger power budget. And that's if you don't even touch the little cores.

    To be honest, I don't really care about the little cores, for desktop use. I think they're mostly an attempt to add more multithreaded performance on the cheap, because the little cores have more PPA than the big cores. Most desktop users don't run 24-thread workloads, anyhow. Maybe it's just Intel's way of trying to gin up some better multithreaded benchmark numbers.

    As for thread-scheduling, Big.LITTLE architectures have been with us for about a decade. I think there's probably a wealth of experience with the scheduling challenges they pose, although I'm sure we'll see corner cases where it causes some issue. And most of those will probably be smoothed over, after a few OS patches.
  • Silver5urfer - Wednesday, May 12, 2021 - link

    Same I also do not care for the small x86 cores, but Intel is only doing it to fab one single silicon for both Desktops and Laptops to save cash, they do this all the time as everyone knows it, and bottom barrel silicon ends up in Laptops which is why all the laptops have high stock voltage out of the box.

    Thread scheduling problems will occur on Windows no matter what, the OS is not built like that but with the bs thin and light market addiction for both OEMs / Intel / AMD / Apple / Users. M$ will make an update without a doubt as they will also benefit with this approach for their Surface BGA locked down HW. Windows x86 / Win32 never used it so it will have first wave of issues. Just like how Ryzen had with Threadripper CPUs. Smoothness with Windows 10 as a Service is questionable at it's best.

    ADL won't be a good product. People should wait for Zen 4 / Raptor Lake if they want to get DDR5 and Gen 5.
  • Spunjji - Thursday, May 13, 2021 - link

    "bottom barrel silicon ends up in Laptops which is why all the laptops have high stock voltage out of the box"

    It's basically the other way around. Bins that can run on low voltage go to laptops, bins that need higher voltage and have a higher clock headroom go to desktop. This is basic stuff.
  • Silver5urfer - Thursday, May 13, 2021 - link

    What a complete utter dumb nut. Unfit for AT as I mention always. Better silicon is called golden because it needs lower voltage than rest to hit high clockspeed be it overclock or boost clock, while low bin needs more voltage than that golden sample.

    Intel does this often with CFL 8086K, 9900KS, CML Limited edition, X299 HEDT 9990XE. While this dumb person thinks its high voltage have high clock only hahaha LOL.

    Next the laptop users scour interwebs on how to tune their processors which is seen in so many forums esp Notebookreview and there's a specific software to do that instead of garbage XTU. PLUS people always try to UV them because its high voltage out of box this has been the case since Haswell, Intel is putting better bins in laptops and using trash bin on desktop as per your retarded logic hahaha this is epic. Even worse than AMD fanboy logic.

    ZERO knowledge and has useless personal projecting issues on others. Now I'm clear its a complete waste of time to pay attention on subject with you solid room temp IQ. I kinda expected this from a wokie before only.
  • Spunjji - Thursday, May 13, 2021 - link

    @Silver5urfer - "...has useless personal projecting issues on others."
    'Nuff said.

    "Better silicon is called golden because it needs lower voltage than rest to hit high clockspeed"
    This is a huge oversimplification. Low-leakage silicon often has a lower voltage headroom than high-leakage silicon. It's not just "good silicon" and "bad silicon", different bins have different characteristics, and the batches with the best characteristics for a given platform get used for them.

    I know full well about Throttlestop, I've used it. My Skylake notebook undervolted like a peach. The thing is, talking about *individual users* undervolting their notebook CPUs has *nothing* to do with how Intel bin entire product lines.

    But sure, keep casually tossing slurs at me, that'll convince everyone!
  • Qasar - Thursday, May 13, 2021 - link

    " But sure, keep casually tossing slurs at me, that'll convince everyone! "
    i think thats all he has left. well, other then ranting.
  • ottonis - Wednesday, May 12, 2021 - link

    Nobody actually expects Alder Lake to dominate performance wise. Alder Lake is just a first step into the field of hybrid core packages with the aim to further improve on on overall work efficiency - similar to ARMs big.LITTLE concept and the same concept that Apples M1 architecture is pursuing.

    That being said, Alder Lake is just tipping the toes into these new waters, as it will be inevitable for Intel (as well as for AMD) to further optimize the per-core efficiency of their *performance*- cores if they want to compete with Apple.
    One important aspect is of course the process node, and Apple has an edge over everybody else in the x86 market, as the M1 is built in a 5nm process.
  • mode_13h - Thursday, May 13, 2021 - link

    > Nobody actually expects Alder Lake to dominate performance wise.

    I think Intel is putting the little cores in there because they think that will make it more competitive, or at least benchmark better.

    Intel makes internal research chips that never see the light of day. So, this isn't just them running a sort of experiment.

    I think it *is* something they probably wouldn't do, without competitive pressure from ARM and AMD. It's a little more adventuresome and risky than their typical MO.

    > Apple has an edge over everybody else in the x86 market, as the M1 is built in a 5nm process.

    Not just Apple, but also Qualcomm and others.
  • SaturnusDK - Tuesday, May 11, 2021 - link

    ADL is Intel betting the farm on windows schedulers to assign tasks to the correct cores. This is effectively the same bet AMD did with Bulldozer that the windows scheduler would reliably assign FP task to the iGPU or dGPU, so they could cut half the FPUs.

    A really really bad idea.
  • drothgery - Tuesday, May 11, 2021 - link

    Microsoft is a lot more likely to work on the scheduler handling big.LITTLE-like setups correctly in 2021 (when it's needed for Intel's mainstream chips -- and even with AMD having historic levels of success the overwhelming majority of x64 CPUs sold are Intel -- and for Windows on ARM) than for Bulldozer (when AMD's market share was tanking due to not being remotely competitive with Intel's chips at the time).
  • SaturnusDK - Tuesday, May 11, 2021 - link

    Before Bulldozer, AMD was the performance leader I'll remind you. Their market share was only abysmal due to Intel literally paying off OEMs and distributors to not sell AMD products. This is not hear-say or rumours, this is fact.
  • drothgery - Tuesday, May 11, 2021 - link

    It was going to be abysmal regardless; AMD had nowhere near Intel's fab capacity (this was back when AMD had their own fabs).

    And Bulldozer was contemporary with Sandy Bridge; AMD's CPUs had been far inferior to Intel's for years at that point.
  • mode_13h - Tuesday, May 11, 2021 - link

    > Before Bulldozer, AMD was the performance leader I'll remind you.

    LONG before Bulldozer. Phenom and Phenom II were never competitive, except maybe the 6-core version on well-threaded workloads.

    That said, I did like Phenom II. It was a good bargain and had tons of I/O capacity. I had a quad core.
  • bji - Tuesday, May 11, 2021 - link

    My teenage daughter is still using the 6 core Phenom II I put together in 2011. And it's working fine for her. Not saying this is a factor unique to this processor, just trying to say that I agree that the Phenom II was decent.
  • lmcd - Saturday, May 15, 2021 - link

    Phenom II was nice for its time. So nice, in fact, that Thuban outperformed the 8150 until its lacking instruction set caught up to it on 2020-era workloads.

    Doesn't change that Sandy Bridge far outperformed both of them. AMD was neither the market leader nor the performance leader.

    Also, Bulldozer was way less consistent than Alder Lake will be for scheduling. Selecting a core in Bulldozer vs Steamroller/Excavator wasn't even the same set of trade-offs! By comparison, an Atom core vs a Core core (sidenote, can we please revive the Pentium brand and say they're Atom cores vs Pentium cores) are both extremely well understood and the logic is similar to that already being done by the Windows scheduler for Snapdragon.
  • mode_13h - Sunday, May 16, 2021 - link

    > Phenom II was nice for its time. ... Doesn't change that Sandy Bridge far outperformed both

    Phenom II competed well against Core 2. However, it launched against Nehalem and right before Sandy. The thing I liked about it is that it was cheap and supported ECC. Also, the 890FX chipset had a ton of PCIe lanes and 6x SATA 3 ports. I built a fileserver with it for a little over $700 (excluding storage drives).

    > Also, Bulldozer was way less consistent than Alder Lake will be for scheduling.

    I don't know why it should be so different from scheduling CPUs that have hyperthreading.

    > Steamroller/Excavator

    It just occurred to me what comes next in that sequence: Exfoliator
    : )

    > Bulldozer

    Is it true that one reason for its low performance is that the design team switched to ASIC-type tools and design process?
  • Oxford Guy - Tuesday, May 18, 2021 - link

    Atrocious single-thread/IPC in Cinebench (which is very FPU/heavy I recall reading). Even at 5 GHz, Piledriver is slow in that test.

    Low gaming performance (Piledriver included) due to lack of FPU and extremely poor single-threaded performance. An exception is Deserts of Kharak — the only time I’ve seen Piledriver not only hold its own with Intel but top the chart (980 Ti SLI).

    Poor L3 cache, practically as slow as DRAM.

    Piledriver had broken AVX. It would run but was slower than not using it.

    Apparently, the early going for GF’s 32nm SOI was bad for power efficiency, which carried over through the first generation of PD. That meant depressed stock clocks for BD especially — plus high wattage required.

    Memory controller always delivered slow synthetic benchmarks versus Intel, with the possible exception of AIDA’s latency rating. (A Piledriver I tested at 2133 9-11-10 CR1 topped the chart in low latency but it was memory bandwidth that Intel crushed the BD family, and Phenom, with as I recall).

    Lacking in terms of Sandy’s micro-ops cache. BD and PD both had them but apparently Intel was more aggressive with it or something. An article here credited the cache with delivering a lot of performance for Intel with Sandy.

    Piledriver used the same number of transistors as Sandy Bridge E so it wasn’t a very efficient design when it came to transistor count.

    Too much reliance on automated design tools has been blamed but some saw the Bulldozer design as being AMD’s attempt to create another Pentium IV — narrow and deep, relaying on high clocks.

    AMD also seemed to pour kerosene on the fire by not only bringing out the 9000 series (using chips so leaky they would have been sent to the crusher, according to ‘The Stilt’) but trying to sell the 9590 for an extreme amount of money. Truly sleazy.

    Again, even at 5+ GHz and enough watts to heat your house, the single-thread performance (in Cinebench, at least) was too slow to be adequate. But the 9590 couldn’t really do 5 GHz. The first thing overclockers did with Piledriver is disable turbo. Even with the most efficient binning (e.g. 8370E), 4.7 GHz non-turbo is as far as one should have gone.
  • mode_13h - Thursday, May 20, 2021 - link

    @Oxford Guy, thanks tor the info.
  • SaturnusDK - Tuesday, May 11, 2021 - link

    But on the important subject of the scheduler. There's no chance that it will ever get a higher than 80% hit ratio outside a controlled test environment. ADL might look good on selected tests but in real world apps, it'll tank... badly.
  • The_Assimilator - Tuesday, May 11, 2021 - link

    And you, having access to ADL test systems, know this for a fact huh?
  • mode_13h - Tuesday, May 11, 2021 - link

    It'll probably have some early issues, but I expect a lot of those will get smoothed out, over time.
  • Kangal - Wednesday, May 12, 2021 - link

    The thing is Android experienced this scheduler problems with ARM's big.LITTLE as early as 2013 and we've had a lot of problems. However, people still didn't expect getting perfect performance from their light, slim, and portable phones. It took until 2016 for most of these issues to get ironed out by both chipset vendors and phone oems. I think it was (finally) Google's involvement in building their own big.LITTLE phone (Pixel) which finally lead to it being a seamless implementation in 2017.

    Now, I don't know if Intel is as innovative as ARM or if they're more incompetent. Maybe they (Wintel) will make the same mistakes. Maybe they'll make more. Maybe less. Or maybe completely new ones. And what about security? Anyways it is A LOT of work, and I wish them the best. HOWEVER! Consumers will NOT be as tolerant with their PCs as they were about their phones when it comes to performance and security. And, a lot of the early big.LITTLE problems were solved by the consumers basically throwing away their old devices and upgrading. That is not an option on PC. People will expect companies to send updates and fix those bugs, which will take resources out of R&D to actually make their second hybrid chipset great.
  • Oxford Guy - Tuesday, May 18, 2021 - link

    Should have been little.BIG then.
  • jeremyshaw - Tuesday, May 11, 2021 - link

    I agree.

    MSFT has had to work with big.LITTLE-eqsue setups since 2012, with the original Surface and it's 2nd gen. Both had Nvidia Tegra 4+1 core SoCs. Different switching method vs big.LITTLE, though not out of the realm different. Later on MSFT would develop the Surface X which uses Qualcomm SoCs with big.LITTLE. Intel has already launched Lakefield, their first big.LITTLE-style SoC.

    MSFT and Intel have both been working on this for years and years. MSFT has rewritten the Windows scheduler before for Intel, for much, much smaller things. MSFT even worked with a much weaker AMD during Bulldozer and during the Ryzen 1000 launch to fix scheduling for those CPUs (or has everyone forgotten about launch-era Ryzen Windows vs Linux scheduler debates?).
  • Spunjji - Thursday, May 13, 2021 - link

    Everything you've said is true, to some extent, and yet when Lakefield launched it didn't work properly and applications have had to be re-written to work across both sets of cores.

    Historically what Windows seems to do is assign a task to *either* a high-performance core *or* a low-performance one - it's not clear that it can move tasks between them, or harness all of the resources for one task.
  • mode_13h - Thursday, May 13, 2021 - link

    > when Lakefield launched it didn't work properly and applications have had to be re-written to work across both sets of cores.

    Really? Got a reference? You're not just talking about the little cores lacking AVX, are you?

    > Historically what Windows seems to do is assign a task to *either* a high-performance core *or* a low-performance one

    Based on what history? Windows hasn't run on more than a handful of Big.LITTLE CPUs, and they were all pretty niche.

    Even if you're right (and I'm questioning more how you know that, than the specific claim) that's all the more reason to expect MS to put some effort into getting it right -- Alder Lake is gonna be mainstream.
  • Spunjji - Friday, May 14, 2021 - link

    "Really? Got a reference?"
    I wish I did - I recall reading this in a review of a Lakefield device in the context of CineBench being rewritten to properly recognise the core layout, but I haven't been able to find it in the brief search I just did.

    It's actually very difficult to find Lakefield reviews in general, barring a few that popped out right around when it first launched - those mention that multi-threaded applications were only running on the small cores, and none seem to have been revisited.

    "Based on what history?"
    Exactly the limited history you refer to - Lakefield and, before it, the Snapdragon 8cx. It's not a lot.

    I also hope MS would be trying to get it right, but then they wanted Snapdragon 8cx devices to take off, yet when Lakefield rolled around the scheduler was still not working well with big.LITTLE style designs.
  • Oxford Guy - Tuesday, May 18, 2021 - link

    ‘MSFT even worked with a much weaker AMD during Bulldozer’

    Too vague to be particularly meaningful. Either Window achieved Linux-quality scheduling efficiency with BD/PD or it didn’t. Which was it?
  • mode_13h - Tuesday, May 11, 2021 - link

    The PCIe & iGPU specs make this sound more like a desktop CPU. Is there any chance it was designed as a desktop CPU and only packaged for mobile?

    I've long had the impression that H-series mobile CPUs were essentially just that: desktop CPUs repackaged for use in laptops.
  • mode_13h - Tuesday, May 11, 2021 - link

    > All Tiger Lake-H processors will support DDR4-3200, however
    > it is worth noting that these processors do not support LPDDR4.

    I take its DRAM support as yet further evidence.

    > The chipset for TGL-H is separate to the CPU ... DMI x8 link

    And this.
  • Otritus - Tuesday, May 11, 2021 - link

    Intel makes 1 die to go across their s(desktop) and h(high-end mobile) processors. Technically s chips are just mobile chips repackaged for desktop since Intel shifted to a mobile first Design philosophy with Haswell.
  • Spunjji - Tuesday, May 11, 2021 - link

    There have been some exceptions, though - most notably the 10-core Comet Lake die, which never had a place in the mobile market.
  • Spunjji - Tuesday, May 11, 2021 - link

    "I've long had the impression that H-series mobile CPUs were essentially just that: desktop CPUs repackaged for use in laptops."

    For quite some time that was true, in a sense - they use the same dies. In this case, though, I don't think there was ever a plan to bring Tiger Lake to the desktop. Whether that's because they can't make enough of them to fulfil both markets, because it doesn't scale much above 65W (a-la Broadwell), or because it's so late that it would bump into Alder Lake is anyone's guess. I'm guessing a combination of all factors.
  • drothgery - Tuesday, May 11, 2021 - link

    Though I'd bet it's mostly because they can't make enough; traditional H-series laptops (the H35 chips really should be called U35s or given some new branding) are a pretty low-volume niche.
  • flgt - Tuesday, May 11, 2021 - link

    Look like a nice boost in performance and the chips I'll most like be given at work. They are finally doing something they should have been doing for awhile which is naming and releasing their low-end consumer and workstation processors using the same methodology. These are "good enough" products and they shouldn't be wasting tons of design resources for stupid marketing tricks. They should also extend this to low end server.

    iX-11XXX - Consumer with integrated graphics
    W-11XXX - ECC, vPro, Integrated graphics
    E-11XXX - ECC, vPro or other out of band management, No integrated graphics

    Something along these lines with something similar to the chipsets. We all know they are the same dies at this point so at least they can simplify things and get product to market faster. W-1300 still hasn't released and the E series is still last generation.
  • mode_13h - Tuesday, May 11, 2021 - link

    > W-1300 still hasn't released and the E series is still last generation.

    Sorry, it's going to be Rocket Lake.
  • flgt - Tuesday, May 11, 2021 - link

    Yeah, I diverged into Rocket Lake there. I just would like them to name and release whatever "11th Gen" is for desktop/workstation/server with a clear pattern.
  • AdrianBc - Tuesday, May 11, 2021 - link

    The Rocket Lake Xeons W-13xx have been released one week ago, without much noise.

    Their specifications are at Intel Ark, together with those of the Tiger Lake H announced today.
  • SarahKerrigan - Tuesday, May 11, 2021 - link

    Losing 2/3 of the GPU cores compared to the ULV parts is not super enticing...
  • repoman27 - Tuesday, May 11, 2021 - link

    Nah, that's what DG1 and all those PCIe lanes are for. Who cares about board real estate or battery life?
  • eastcoast_pete - Tuesday, May 11, 2021 - link

    Agree. However, Intel also knows that the vast majority of actual buyers will opt for a paired dGPU; and, in that case, the die area on the CPU is better spent on compute cores and cache. Unfortunately, there isn't a big market for those of us who want a capable CPU with a decent iGPU in a mobile form factor. AMD made a similar move with Renoir and now Cézanne when they reduced the number of graphics cores compared to the previous APUs (11 to 8 or 7, from what I remember). But, they could at least make up for some of those cutbacks by clocking the Vega cores a lot higher.
  • Farfolomew - Tuesday, May 11, 2021 - link

    Yeah, sure, there's no market for a capable iGPU when dGPUs are unobtainium ... *rolls eyes*
  • Zizy - Wednesday, May 12, 2021 - link

    For laptops, iGPU vs dGPU doesn't change much - perf/mm2 is similar enough. So, you waste the same amount of wafers for better iGPU or dGPUs, meaning you didn't solve anything by making more APUs. So, Intel did the sane decision - give people/OEMs choice of picking 3050 or 3080 for the GPU (making iGPU on par with 3050 only means bigger waste for folks grabbing 3080, but it doesn't help people wanting 3050 much - it is still the same amount of silicon = same price). Sure, there are other tradeoffs with device design but lets ignore that because Si is the bottleneck right now, not boards and whatnot.

    For desktops, a lot of output goes to miners and there good APUs do make sense - for a personal user you want CPU, but for a mining farm all those extra CPUs are a waste, so the markup would be lower. Instead of paying 500 for CPU and 1k for GPU instead of 500$ because of those pesky miners, you would just pay 1k for APU because this is how much miners value its money making GPU... so you don't overpay at all.
  • mode_13h - Wednesday, May 12, 2021 - link

    > For laptops, iGPU vs dGPU doesn't change much - perf/mm2 is similar enough.

    No. The biggest thing a dGPU gets you that an iGPU cannot is memory bandwidth. That's why Intel was mucking about with eDRAM, a few generations ago, and yet it could still barely compete with an entry-level dGPU.
  • lightsout565 - Tuesday, May 11, 2021 - link

    "The other processors here follow Intel’s traditional non-overclocked schema, whereby 45 W is the standard rating and 35 W is the power-down option"

    Can someone explain this to me. Previously I thought most the H-Series chips were 45W. Now I see two base TDP's, one at 35 and one at 45. Is the TDP set by the OEM or is this how it always worked?
  • Revv233 - Tuesday, May 11, 2021 - link

    Yes
  • AdrianBc - Tuesday, May 11, 2021 - link

    The OEMs always set the TDP they want.

    For example in my old Dell Precision laptop the Skylake H CPU is configured at a 60 W TDP, despite the fact that its official TDP is 45 W.

    For that laptop, the 60-W TDP is appropriate, because it is a thick laptop, with good cooling.

    What happened lately is that Intel, both for some of the Comet Lake H CPUs of last year and for the Tiger Lake H CPUs of this year, has begun to document publicly some of the possible TDP-base clock frequency pairs.
  • repoman27 - Tuesday, May 11, 2021 - link

    Well, technically, OEMs can set various package and/or platform power limits (PLn) and the Turbo Time Parameter (Tau), which is a constant used when calculating the average power for PL1.

    The TDP in this context is a fixed specification for a given SKU, validated at manufacturing time. When executing an Intel specified near-worst-case workload, the processor will remain below the maximum junction temperature without average power dissipation exceeding TDP. In other words, the processor will be able to sustain the rated base clocks indefinitely without throttling if the thermal solution is capable of dissipating that much power.

    Configurable TDP, or cTDP, is a design option that allows dynamic configuration of the TDP based on available cooling and adjusts the base clocks accordingly. It is not a whatever the OEM wants situation and has no effect on turbo frequencies. The configurable TDPs and corresponding base clocks are specified by Intel and set at manufacturing time.
  • repoman27 - Tuesday, May 11, 2021 - link

    And although Ian's charts don't reflect this, all of the 6 and 8-core Tiger Lake-H parts have a TDP of 45 W. The Core i9-11980HK has a cTDP up of 65 W, and the rest of the SKUs have a cTDP down of 35 W. The previously released 4-core Tiger Lake-H35 parts have a TDP of 35 W and cTDP down of 28 W.
  • repoman27 - Tuesday, May 11, 2021 - link

    "Thunderbolt 4 is natively supported, although USB4 is not listed. There is USB 3 support, however Intel fails to list if this is USB 3.2 Gen 1 or Gen 2."

    I'm surprised the errors in the platform diagram slide made it through to the final version, but where it says "10x USB2, 4x USB3", that should be "10x USB 3.2 Gen 2x1 (10 Gbps)". Where it says "PCIe 3.0 x12 lanes" next to a USB logo, that should be "14x USB 2.0"; PCIe is already correctly listed on the other side of the diagram. The datasheet for the 500 Series chipsets has been available for a while now (since March).

    Thunderbolt 4 is USB4 with Thunderbolt 3 Alternate Mode, so there's no need to list it separately. TGL-H supports up to four USB4 Gen 3x2 (40Gbps) ports, just like TGL-U.
  • Linustechtips12#6900xt - Tuesday, May 11, 2021 - link

    I am wondering but past USB 3.0 other than for something like video editing which I think thunderbolt would take care of, who really needs 10+Gbps I mean really it's not like thumb drives are going to go 40Gbps anytime soon other than thunderbolt.
  • repoman27 - Tuesday, May 11, 2021 - link

    Gen 2x1 (10Gbps) makes sense and is quickly becoming the baseline for applications like USB attached storage devices, and it's relatively inexpensive to implement. I'm not sure Gen 2x2 will ever become prevalent, as any potential applications would probably be better served by USB4/Thunderbolt, which is already much more widely available.
  • mode_13h - Tuesday, May 11, 2021 - link

    eGPU docks are an important application of USB @ 40 Gbps. That's a little more than PCIe 3.0 x4, which is still not quite as much as modern games and GPUs would like, but certainly usable.
  • vladx - Wednesday, May 12, 2021 - link

    Data is only 32Gbps, the rest is encoding.
  • mode_13h - Tuesday, May 11, 2021 - link

    Also, don't forget about > 4k monitors.
  • abufrejoval - Tuesday, May 11, 2021 - link

    Some spelling mistakes make you chuckle:

    "As always with these launches, Intel provides first-party benchmark numbers. We’re loathed to mention any of them...."

    Well, we've seen a lot in comments here, but more likely you're loath to mention them.
  • Silver5urfer - Tuesday, May 11, 2021 - link

    Intel has been doing HK trash since Skylake. These are crap parts, 10nm first bottom barrel products, all the mobile Extreme died with Haswell MX being last of the true K series binned CPUs which have better voltages out of the box, and the Ivy Bridge XM processors.

    If people want to undervolt the Plundervolt issue locked out all users out, this one probably doesn't have SGX so can do it BUT there's a PCH BIOS lock which cannot be overridden and thus the PL1 and PL2 cannot be truly unlocked as these are anemic trash 65W CPUs, this applies to both AMD and Intel as they are locked down garbage processors, while the Ivy and Haswell Extreme mobile CPUs are fully unlocked rPGA socket chips.

    Alder Lake won't do anything. It's another stop gap solution and a guinea pig experiment, who wants to get into the stupid Hardware scheduler issues that windows will have with ADL on top of the useless Gracemont cores in a desktop CPU ? Laptop yeah people will love the underpowered trash for battery life. And then it will have the DDR5 segmentation with overpriced DRAM kits, if you add mobile to the equation apart from the Clevo LGA machines no Skylake machine so far was able to properly run 3200MHz DDR4 SODIMM in a laptop. Many lock them down, so all in all BGA garbage is made for garbage is garbage and lowest common denominator crap.

    The enthusiast Laptop DTR segment is only alive with Clevo since it has unlocked BIOS options with Intel. As for AMD nope, nothing at all.
  • Silver5urfer - Tuesday, May 11, 2021 - link

    I forgot to mention the pathetic heatsinks these laptops will have will be going to cook themselves out, people want silent machines with desktop class performance they will only get silent but no desktop class since OEMs won't engineer better HS solutions, nowadays the trend is Tripod heatsink trash which is horrible they cut down costs with a lesser mounting arm !! Don't beleive it ? go and see any BGA laptop teardown, even LGA machines with Clevo started doing this I guess, and they all have Unified garbage HS design on top which will radiate heat from the GPU as well and CPU as well on full load Nvidia Samsung 8nm process with Intel's high power it will be a mess. Alienware A51M series already saw smoke out due to the pathetic DGFF cards vs the superior MXM counterparts from MSI / Clevo.

    Mobile DTR market is worthless now, overpriced and underpowered and cost cutting pile of junk.
  • Spunjji - Tuesday, May 11, 2021 - link

    You seem to be angry that things are not the same as they were when you were a young lad 🧐
  • Silver5urfer - Tuesday, May 11, 2021 - link

    As usual ZERO knowledge but some useless trash posts to fill up comment section.
  • The_Assimilator - Tuesday, May 11, 2021 - link

    U mad bro? You want desktop performance, *buy a fucking desktop*.
  • Silver5urfer - Tuesday, May 11, 2021 - link

    Who the hell are you lol got hurted hard by my reply to that acct ? Is this an Alr of Spunjii ? You don't have anything to educate people or yourself why even bother loser.
  • Silver5urfer - Tuesday, May 11, 2021 - link

    "I will pay Desktop price for a laptop" "I love TDP throttling because my Fortnite runs superb" "I got new i7 bruh and look at the thin and light cuts of this machine" "I'm a proud of that" - The_Assimilator , what a mindset no wonder Intel still is in business selling garbage products.
  • Pacinamac - Wednesday, May 12, 2021 - link

    Yikes... Calm down man.
  • mode_13h - Wednesday, May 12, 2021 - link

    Agreed. You're not going to change anyone's mind by getting belligerent. Worse, it even distracts from any good points you might make.
  • Bagheera - Wednesday, May 12, 2021 - link

    Silver5urfer belongs on WCCFTech
  • Spunjji - Thursday, May 13, 2021 - link

    @Silver5urfer - " Is this an Alr of Spunjii"
    I have one account, and I use this name everywhere. Your posts are genuinely unpopular with many people. Maybe reflect on that, instead of just doubling down on the paranoid ranting?
  • lmcd - Saturday, May 15, 2021 - link

    Yea also despite differing on our evaluations of Intel and AMD products, Spunjji seems pretty reasonable to me (in addition to agreeing on Silver5urfer)
  • Spunjji - Thursday, May 13, 2021 - link

    As usual, projection.
  • shabby - Tuesday, May 11, 2021 - link

    "4.5 GHz all-core turbo"

    For how many seconds?
  • gescom - Tuesday, May 11, 2021 - link

    Well, that depends on a chosen benchmark app.
  • Spunjji - Tuesday, May 11, 2021 - link

    And the OEM's chosen TDP, and the OEM's heatsink solution...

    The error bars for compiled benchmark results for Tiger Lake on sites like Notebookcheck are already entertainingly large, I can only see that getting worse with these!
  • Someguyperson - Tuesday, May 11, 2021 - link

    Indefinitely if you can actually deal with the heat. Extrapolating the numbers a bit, the 5 GHz turbo probably uses just over 15w/core and 4.9GHz would use just under 15w/core. Moving down to 4.5GHz would then use about 13w/core. Having 4 of them active and 4 idle would take overall power consumption to around 55 watts, which would be under the 65w TDP they're going for. However, if you have 4 cores at 4.5GHz turbo and the other 4 cores at the 45w TDP limit, the overall package would be nearly 80 watts, which is insane for a laptop. You're clearly in the realm of those giant Clevo "laptops" that have full fat desktop processors in them at that point.
  • Spunjji - Thursday, May 13, 2021 - link

    I think your numbers may be a little conservative. Based on the 50W boost TDP of 4-core Tiger Lake, it's likely that 5Ghz uses closer to 20W per core.

    On the other hand, I would be surprised if ~13W per core weren't about right for 4.5Ghz. These things really go off a power cliff over 4Ghz, and more so past 4.5Ghz.
  • DannyH246 - Tuesday, May 11, 2021 - link

    hahaha another Intel is still relevant we promise marketing deck from www.IntelTech.com
  • Spunjji - Tuesday, May 11, 2021 - link

    To give them credit, these processors should actually be pretty competitive - even superior for some people's requirements (e.g. you absolutely need Thunderbolt and some high-speed storage).

    The funny bit is, their marketing is all about gaming systems, where these will probably be much of a muchness alongside AMD's offerings - Thunderbolt and PCIe 4 storage aren't really a big draw there, outside of spec-sheet wankery.
  • Bik - Tuesday, May 11, 2021 - link

    It's actually AnamdTech the way Ian puts out those question marks on everything from Intel promotion suit.
  • vladx - Wednesday, May 12, 2021 - link

    Yep AnandTech's bias toward AMD is common knowledge at this point.
  • Spunjji - Thursday, May 13, 2021 - link

    It is literally impossible to please some people.
  • Qasar - Thursday, May 13, 2021 - link

    "Yep AnandTech's bias toward AMD is common knowledge at this point. "
    just like the posts that come up when people accuse AT of being intel bias ?
  • vladx - Friday, May 14, 2021 - link

    I don't know, it's possible they had an Intel bias at some point in time but at the present the bias is clearly towards AMD.
  • Qasar - Friday, May 14, 2021 - link

    yea ok, they dont need to, they have your bias for intel, vladx.
  • Spunjji - Friday, May 14, 2021 - link

    Alternatively: both sets of bias accusations are demonstrably wrong.
  • Spunjji - Thursday, May 13, 2021 - link

    If you think justified skepticism is bias and the site only does this with AMD, then you're an idiot.
  • Yojimbo - Tuesday, May 11, 2021 - link

    How can a company with 80% of the market not be relevant?
  • usiname - Wednesday, May 12, 2021 - link

    Ask crapple
  • xpclient - Tuesday, May 11, 2021 - link

    Compare same model laptops please if possible e.g. 2021 Ryzen 5000 based ASUS A15 vs Tiger Lake H45 F15. Or A17 and F17.
  • Farfolomew - Tuesday, May 11, 2021 - link

    How could this be 20% better than a Zen3 mobile part, like Intel claims? Isn't Zen3 IPC slightly better than TGL's? I know TGL probably has a little better frequency, but surely not 20% worth. Unless they're throwing in AVX512 benchmarks ...
  • Bagheera - Tuesday, May 11, 2021 - link

    Intel numbers. 🙃
  • mode_13h - Tuesday, May 11, 2021 - link

    Usually, it comes down to cherry-picking benchmarks that are well-optimized for Intel CPUs. AMD also likes to highlight the benchmarks that make them look good. That's why we need independent tests.
  • t.s - Wednesday, May 12, 2021 - link

    Independent test? Surely we're not talking about anandtech here, right?
  • Bagheera - Wednesday, May 12, 2021 - link

    when AMD claimed 19% IPC uplift to Zen 3, independent tests verified it.

    when Intel claimed the same, it was really more like 7%.
  • mode_13h - Thursday, May 13, 2021 - link

    I wasn't trying to say that both companies' practices were identical, but AMD does have a history of highlighting performance on benchmarks where they do well (like Cinebench, for example). I just thought I should acknowledge that.
  • Qasar - Friday, May 14, 2021 - link

    " but AMD does have a history of highlighting performance on benchmarks where they do well (like Cinebench, for example " if i remember right, intel touted cinebench in its benchmarks as well when they were faster then amd, and when amd released zen and the performance difference with cinebench wasn't as large, if not slower then amd, intel started to denounce it, and downplay it
  • pavag - Tuesday, May 11, 2021 - link

    8 cores in 2021 is totally pointless, Mr Intel
  • RanFodar - Tuesday, May 11, 2021 - link

    If they reach gaming and productivity performance as with the top Ryzen 5000 laptop SKUs, you wouldn't want to say that again.
  • RealBeast - Tuesday, May 11, 2021 - link

    8 cores is more than enough for almost any non-workshop only applications. The current Adobe recommendation is 8 cores and they say that any more will not be helpful, DaVinci Resolve 8 cores is fine but it wants 2-3 high end GPUs if you are power user. Intel needs to shrink their process to reduce TDP, not add more cores at this point for general applications.
  • mode_13h - Wednesday, May 12, 2021 - link

    I could use more, but I'd be pretty satisfied with 8 good cores.
  • Pacinamac - Wednesday, May 12, 2021 - link

    This will by my next CPU for mobile video editing. Intel quick sync is a huge advantage in Adobe Premiere.
  • Techtree101 - Wednesday, May 12, 2021 - link

    If Tiger Lake-H is launching now, why is Intel planning an Alder Lake mobile launcher in Q4? Isn't that a bit soon?

    I am all for it, but just curious what the marketing thinking is here. Perhaps it is a compressed timeline they're forced to have now.
  • drothgery - Wednesday, May 12, 2021 - link

    I don't think Adler Lake mobile is confirmed for Q4, just Adler Lake desktop.
    And H-series parts have always trailed U-series parts (not usually by this much, but usually by at least a quarter); Tiger Lake U was out last year.

    But it's quite possible that if Intel had been sure of getting Adler Lake desktop out in Q4 2021 and sure they wouldn't get Rocket Lake out until effectively Q2 2021, they would have skipped Rocket Lake (and the situation with Ice Lake SP and Sapphire Rapids is similar).
  • vladx - Wednesday, May 12, 2021 - link

    Alder Lake won't compete with H-series processors, just I series.
  • vladx - Wednesday, May 12, 2021 - link

    *U series
  • Spunjji - Thursday, May 13, 2021 - link

    This product is late, Alder Lake looks to be on time.
  • mode_13h - Thursday, May 13, 2021 - link

    Isn't this basically Alder Lake, without the little cores? If so, then it makes sense that Alder Lake shouldn't be far behind.
  • Spunjji - Friday, May 14, 2021 - link

    Alder Lake has a new large core design (Golden Cove) and will be manufactured on 10nm ESF (Enhanced SuperFin) - so it's actually quite a different proposition from Tiger Lake. Intel are catching up with themselves in terms of being able to manufacture their designs, which is a good sign.
  • lmcd - Saturday, May 15, 2021 - link

    I'm still concerned though, wasn't ESF a further relaxation of density? Or is it the same density as SF?
  • mode_13h - Sunday, May 16, 2021 - link

    Eh, if ESF wasn't a net win, then Intel would just stick with SF. So, I wouldn't worry.
  • Spunjji - Monday, May 17, 2021 - link

    @mode13h - that's what I'm taking from it, too. A density reduction could improve overall production output if it increased yields, and the claims of improved performance seem plausible.
  • Spunjji - Monday, May 17, 2021 - link

    I've not seen any reliable information about the density changes between Intel's processes. My reading-between-the-lines of their publicity on SF was that it constituted a further relaxation in density over Ice Lake's 10nm(+) - otherwise they'd have boasted about density alongside all the positive characteristics they mentioned - but I could conceive of ESF modifying the transistors in a way that doesn't affect overall density much over SF.
  • zodiacfml - Wednesday, May 12, 2021 - link

    finally but a bit pointless except the highest end products, considering quad core tiger lakes already costs as much eight core Ryzen parts.
  • xpclient - Thursday, May 13, 2021 - link

    Without seeing any benchmarks yet, I'd I/O is the main upgrade here and the CPU performance gap with AMD is slightly reduced but AMD still maintains a lead easily.

    Some Tiger Lake H45 laptops are having 2 x M.2 PCIe Gen 4 SSDs, all USB 3 Gen 1 (5 Gbps) ports upgraded to USB 3 Gen 2 (10 Gbps) and 2 x Thunderbolt 4! Now that's massive.

    But these are overpriced by Intel despite AMD CPUs/APUs offering better performance for the buck. I see many Intel Tiger Lake *Quad core* laptops costing more than 8 core AMD Ryzen 5000 laptops. And Thunderbolt 3/4 peripherals and adapters still set you back by another $150-$250
  • lmcd - Saturday, May 15, 2021 - link

    Doesn't change that Thunderbolt docks are still far and away superior to USB 3 with alt mode docks. It's not even close. If I were to provision laptops for a company, there's no question I would standardize on Thunderbolt.

    Also, better cooling solutions for the high end parts, which usually comes with the H-series unless you buy MSI, should help close the gap.
  • mode_13h - Sunday, May 16, 2021 - link

    Doesn't this have USB 4? In which case, the distinction should become moot.
  • lmcd - Sunday, May 16, 2021 - link

    The point is that the "adapters and peripherals setting you back" isn't a like for like comparison. The experience with Thunderbolt docking isn't just numerically better, the plug-and-play experience is more reliable with Thunderbolt as well, to such an extent that the price comparison xpclient brought up isn't relevant.
  • Spunjji - Monday, May 17, 2021 - link

    I *have* had plug-and-play issues with Thunderbolt docking before, but it was with Dell kit, and the problems were resolved by firmware updates. Overall I'd agree the experience is superior enough to justify the extra cost, because you don't have to go looking to find out whether, for example, your USB-C 3 port supports HBR2 or HBR3 just to know what sort of monitors you can plug into it...

    I'm a little disappointed that Thunderbolt accessories haven't really dropped in price much over the years, though.
  • sandeep_r_89 - Monday, May 17, 2021 - link

    Any word on Tigerlake-H NUCs ?
  • mode_13h - Tuesday, May 18, 2021 - link

    If Intel sells them, my guess is that they'll be like the NUC9 extreme, which aren't really NUCs in any proper sense.
  • littlefooch - Monday, May 17, 2021 - link

    if all i care about is thunderbolt 4 capability and I have a Z590 board and get TB4 from an AIC, what does it matter if I use a 10th gen Intel cpu or 11th gen (with TB4)? Yes I know the 11th gen core is more wonderful than the 10th gen core cpu. I'm focused on the TB4 capability. Thoughts?
  • mode_13h - Tuesday, May 18, 2021 - link

    Well, if you want to connect a monitor via Thunderbolt, then you probably need better integration than a add-in-card would provide, since the iGPU needs to be designed to route displayport over TB.

    Other than that, my guess is that it'd work fine. I'm not speaking from any experience, however.
  • littlefooch - Wednesday, May 19, 2021 - link

    thanks for the feedback mode_13h, my focus is on thunderbolt over IP and building a TB LAN a la the Intel Thunderbolt White Paper circa 2014
  • mode_13h - Thursday, May 20, 2021 - link

    Not something I know anything about. I just have the immortal phrase ringing in my head: "Never bet against Ethernet"

    I finally got a multi-gig Ethernet switch, last year. It has 10 ports in a mix of 10, 5, 2.5, and 1 gig ports. Cost me < $300, and it's not too hot or loud. It's perfect for my current needs, since only 2 of my machines have any business @ 10 gig. 5 gig is good for another 2. The rest are happy at 1 gig, but I could slap a 2.5 gig NIC in one of them.

Log in

Don't have an account? Sign up now