Given the lack of unified memory and the inferior process node, the expectation is that this would be much less power efficient (and the article points this out too).
I don't know. It does not feel impressive to me that the absolute leader in CPU for the past decades latest unreleased cpu beats what I can buy in a store today by a sliver on one benchmark (with all the doubts about power consumption).
The big elephant in the room for Intel is the x86 instruction set. It's a technical debt that is not addressed in anyway as far as I can see.
Until then, we'll have to choose between the Apple way or the inefficient way.
I still think it's not clear how relevant the whole x86 story is. The closest comparison I can find is the apple A13 bionic vs. the amd 5850u pro - both on TSMC's N7p process I believe (but I'm no expert in all the various flavors). The A13 has a 6W TDP; the 5850u 15W (and I can't find a good benchmark source with actual powerconsumption with benchmarking runs, so TDP is the best I have). And the TDP can be scaled down or up at least on the AMD side.
Geekbench scores:
A13: 1336 single, 3187 multi
5850u: 1414 single, 8140 multi
And there are lots of tricks apple uses to keep power consumption down _in addition to_ using arm; and that's partially why it does better than e.g. qualcomm. But many of those tricks aren't somehow arm or apple exclusive; they're just a question of (expensive) engineering or die size - things that aren't fundamental to intels usage of x86.
I mean, it might well be that x86 hits some ceiling well before arm's ISA does, but I don't see actual evidence of that being a huge factor yet. At the very least, it feels like it's clear the process node helps, as does all kinds of other engineering, but that the impact of the ISA is not as obvious.
Course, I might just be missing the obvious evidence, sure ;-).
Essentially this, there are just more cores and more memory.(however the bandwidth aspect is the most interesting one imo).
2022 will be the year of big surprises: m2, amd's new series, riscv, whatever intel comes up with, and why not: even other players making an appearance(especially in the server space)
I feel like this isn't mentioned enough. Right as Intel and AMD get out their M1 competitors a year later, the M2 is probably going to be coming out. Apple doesn't really pre-announce so it's not in people's heads.
Not only because M1 is at TSMC "5nm" (which is probably equivalent to Intel 7nm) but also because this happens every time intel gets kicked to the curb.
"Guys, don't go AMD, the new Intel CPU will be faster"
When the product arrives it usually comes with a huge trade-off such as thermals and power consumption and doesn't do the performance originally claimed anyway.
This has happened after every Ryzen launch, suspiciously only a day or so after each event. It's so obvious and I don't think I'm going to fall for that again.
No more promises: Goodbye Intel, I'll see you when your product can be in my hands and do what you claim.
> this happens every time intel gets kicked to the curb.
>
> "Guys, don't go AMD, the new Intel CPU will be faster"
To be fair, for years and years, new Intel CPUs were faster/better/whatever. What you remember as "kicked to the curb" was really something more like an occasionally great product in a sea of mediocrity where the industry was having terrible trouble competing. Someone would have a chip out with great reviews, and then the new Intel architecture would ship and everyone would forget about it.
The current environment where they lag in almost all the high end product spaces is very much a novelty.
> Someone would have a chip out with great reviews, and then the new Intel architecture would ship and everyone would forget about it.
What time period or space are you talking about here? Because that did not happen in the last 10 years for consumer and server processors. The last decade saw AMD with the FX processors failing to be competitive, at first single threaded, over time also multi-threaded and in performance per watt. Then after many years Ryzen came out and was at first just surprisingly competitive, then plainly better in every metric. And that's where we are now. The whole back-and-forth you are describing did not happen at all, and there were only AMD and Intel anyway, not "someone" or multiple players.
I remember being in an intel engineering meeting, and some guy was showing a roadmap of how the next generation the chip was going to allow our solution to be almost on par with Nvidia's solution. That was at a time where the "next generation" chip was 2 years late, 2 years away at best, and we were comparing to what some consumer could run on their nvidia device that they had in their computer already. I'm not sure if everyone in the meeting viewed it as badly as I did, but my take-away from the meeting was "If everything goes to plan, and nvidia stop producing chips in 2-3 years time we'll be a little bit worse than them". Strange times. I guess the good news is Intel don't just do this to their customers.
Are we talking about Intel's plans for a standalone GPU?
Intel hadn't shipped a dedicated GPU in >20 years (until the low-end DG1 last year) and their integrated graphics were anemic at best for most of that time period.
"2-4 years behind Nvidia" would be pretty impressive of a leap to have been aiming for in some meeting a few years ago.
I don't expect to be purchasing their DG2/"Arc Alchemist" when it launches in a few months, but if it's remotely competitive with mid-range current generation cards (as has been rumored/claimed), that'll be pretty impressive to my mind, to basically enter a completely new market segment for them and have a product that's even worth discussing.
“…that'll be pretty impressive to my mind, to basically enter a completely new market segment for them and have a product that's even worth discussing.”
Sounds pretty normal to me. Why would a company enter a new market if they don’t have a product people want?
I can’t think of any company that launched a new product by announcing “We don’t have anything competitive.”
At the very least, have the new Intel CPUs fixed the speculative execution vulnerabilities or are they still manufacturing those? Eg SPOILER, Spectre, etc?
My guess is they've fixed in hardware known issues that previously got patched with performance-robbing microcode updates. So the bulk of that claimed 15% IPC boost might be from that, with some new stuff sprinkled in.
I have to believe that there's a marketing strategy at work here: leak details in highly controlled environments where you "beat" the competition to imply that your next product is better across the board. In the months that pass before your product rolls out, allow the internet/shadow marketing agents to perpetuate the idea that the next release is going to beat everything else.
Roll your product out with a huge array of confusing names like i9-115479k that nobody can really follow. Nobody has the money to buy all of these products, so nobody can truly claim that none of your products are better than the competition, even if none of them offer a better price:performance:consumption ratio.
And even if somebody does call you out on it... well, don't worry! The next release is going to be even better. Just you wait.
All too frequently we find tech companies comparing their unreleased products against shipping products.
When Apple moved from Motorola to IBM PPC: Motorola soon to release super fast chip updates!
When Apple moved from IBM PPC to intel: IBM soon to release super fast chip updates!
When Apple moved from Intel to M1: Intel soon to release super fast chip updates!
These companies need to release this press coverage - they’re publicly traded and need this.
The fact though is that right now people are using the Max, of course I’d expect some unreleased future chip to be faster.
However even the slower M1’s are still attractive because of the myriad of other benefits they bring: true mobility chips, huge battery life, quiet operation: the Air doesn’t even have a fan, boots/wakes/video switches are seamless, and the special sauce: the chips merge general purpose computing with application specific computing: apple will always be able to generate a tailored product for their audiences.
And it should go without saying: the pro desktops are certain to have compelling performance, and we can expect M2, M3 and so on to continually advance from where we are today.
Moving away from intel wasn't about performance alone, it was about taking control of their own product’s direction.
I don't think we need to be skeptical because of the lack of information given.
The article just says some future chip, with unknown memory, at some unspecified power level and cooling solution, will be able to eke out a benchmark win against the part Apple is shipping today in laptops.
It wouldn't make a difference in the comparison against M1 Max, but the benchmark was also done on Windows 11, which introduced bugs for AMD scheduling. Fixed now, but maybe the AMD results have been skewed because of this. Overall, this is the least scientific and realistic benchmark article ever - until people get their hands on the chips and test them properly everything is speculation. Also - competing for absolute benchmark rankings, when you still need to put a discrete GPU in an intel laptop to even connect a 4k screen is meaningless - people should stop making purchasing decisions just based on benchmarks.
> you still need to put a discrete GPU in an intel laptop to even connect a 4k screen
What do you mean by this? I have an 8th gen i5u in an HP laptop at work, and it's been driving an external 4k@60Hz screen for years. It has no dedicated GPU at all, just the integrated UHD 620 or something like that.
It can even daisy chain two external displays, in addition to the integrated panel. Not sure if it can drive two external 4k like this, I only have FHD + 4K to test.
I somehow cannot make same setup HP work with 4k@60Hz at all - 4k@30Hz works, but not 60Hz. Would you be kind an post screenshot (link to an imgur.com) showing that this indeed, is possible, and I shall have hope?
Also, what monitors do you have, sorry?
The display is an LG, can't remember the model, but it worked with multiple other displays, HP, Dell, etc.
Your issue may be related to the connection. The limitation of 30 Hz makes me think you're connecting via HDMI. If that's all you have, you're SoL.
4k@60 requires HDMI 2, but Intel CPUs only recently have got that. 8th gen certainly doesn't have it.
You need DisplayPort for this to work. Laptops don't often have DP connectors, but they may have USB-C and/or Thunderbolt with DP alt-mode. This is what I use, hence the DP-xxx output name.
> The Apple M1 Max is a fantastic chip and is going to push both AMD and Intel to perform at their very best. And while I doubt x86 architecture will ever beat ARM on power efficiency, absolute power (within roughly comparable battery life) is another matter altogether.
I'm not sure I can follow this line of reasoning. If ARM chips will always be more efficient, then how can you continue stating comparable battery life.
Moreover, there is no mention of TDP. Past flagship Intel processors have been fighting thermal throttling on mobile devices, and it's already been demonstrated that the M1 Max chips don't run into throttling even under load.
It's interesting to see first benchmarks of the new Alder Lakes chips, but we will see how well they work in real products, i.e. with thermal and other constraints.
Doesn't 2x efficiency translate into 2x performance if one doubles cores? Well, maybe 1.8-1.9x due to various overhead, and the package gets more complicated the more you stuff into it.
Let’s just go back to “portable computers” like the good ok’ Compaq 8086! Has a nice, well constructed handle for carrying around, only weighs 13kg (very luggable, Intel could partner with HP and could call it the “lovable luggable”). Think of the possibilities, at that point you can put an actual box fan in there so you can have a 1500W PSU to power your new Intel chip that just uses the entire 14nm Intel die at this point. Another bonus is that with such a high power usage in these winter months it’ll have the same heating efficiency as running a space heater in your office! It’s a win/win for everyone.
Copying my answer [1] from the Alder Lake Preview.
1. Base Power and Turbo Power. Finally a simple term so we dont have to constantly argue about TDP and PL1 / PL2. Which still happens fairly often on HN. And some of us have been ranting about this for almost a decade.
2. + 19% IPC improvement vs 11900K which is basically an Icelake. Or ~10% improvement over TigerLake ( Which was never made available on Desktop )
3. PCI-E 5 and DDR5. Intel went from falling behind in PCI-E offering to leaping ahead of AMD. We should expect PCI-E 5.0 SSD shipping soon after ( 14 - 16GBps )
4. Chipset is on 14nm. No USB 4 or Thunderbolt 4. But integrated WiFi 6E MAC.
5. For those interested, this put Alder Lake, on a Desktop Platform, with Intel 7nm High Performance Node, at roughly 20-25W per Performance Core. With a ( non-verified ) Geekbench Score fairly similar ( ~6% faster ) to M1 HP Core. The Apple M1, on a TSMC 5nm Low Power node, at roughly 5W per core.
6. It will be interesting to see how the new efficiency cores perform. Which is basically a new generation Atom Core. These were previously scheduled for a Graviton like 64 - 128 Core chip on server. Not sure if that is still the case with Intel's chiplet strategy or what they called Tiles.
7. Worth wondering, Alder Lake was originally scheduled on Intel 7nm or what is now called Intel 4/3nm in 2019. What would happen had Intel not been so badly managed? But if that didn't happen, Pat Gelsinger may never be back at Intel.
8. AMD Zen 4, also with DDR5 and PCI-E 5.0 will be coming in late 2022. Depending on Intel's pace of execution, which seems to be getting better and better every time Pat Gelsinger provides an update, AMD may have some tough competition.
> 3. PCI-E 5 and DDR5. Intel went from falling behind in PCI-E offering to leaping ahead of AMD. We should expect PCI-E 5.0 SSD shipping soon after ( 14 - 16GBps )
The claim I've heard is that intel 7 (previously 10nm) is approximately as good as tsmc's 7nm (but which 7nm?). It sounds like most reviewers seem willing to accept that it's at least a good enough match that you're unlikely to do better.
I mean, those literal nm measurements had no reasonably obvious basis in measurement for either intel or TSMC even before the intel rebrand, so...
It'd be nice if they advertised with a few real benchmarks (e.g. logic density + sram density, some kind of clockspeed metric, some kind of powerleakage metric)... but as is, it's just a branch anyhow.
Whether it "is" a 10nm process depends on what your definition of "is" is ;-). It's certainly not the case that a transistor is 10nm large, not even close.
Definitely, no argument here. I meant Intel 7 is the previous Intel's 10nm. It is confusing in my parent thread where the commenter uses Intel 7nm, which is not a thing in any case.
Of course, this is probably all intentional from Intel's point of view (letting people say "Intel 7nm" rather than "Intel 7").
Better single thread performance is always a nice win because it is immediately perceivable on unmodified software.
I believe Intel is still counting on using x4 PCIE4 for SSDs. I wonder if the PCI-E 5 x16 makes any difference for the GPUs?
DDR5 may bring in nice capacity boosts. However, I read somewhere that at lower capacities it is almost indistinguishable perf wise from high-end DDR4.
The key to M1's attraction is the cohesiveness of its implementation. Apple can make MacOS tightly knit to the heterogenous core types and caches. and essentially gate keep applications to behave optimally. From Intel's announcements, it looks like Intel came up with Thread Director to _advice_ Windows on how to schedule stuff and leave it to MS to implement things. Not sure how that synergy is going to take shape. Perhaps Linux will come out and eat this.
if AMD can improve on Zen 3 and come up with Zen 4, they already provide much more PCI-E lanes and ECC on DDR4/DDR5 memory support, with better than intel multi-threading perf. Competition sure is nice!
I think your data is not correct. H-series CPUs from Intel always had 35W-45W TDP. Now, you write here that a single core will consume 20-25W... That will make TDP of 200W when all cores are running. That doesn't make any sense at all.
I don’t know about the 20-25W figure, but I do know that Intel’s TDP figures are not maximum TDP like Apple’s are. Intel chips are well known for blowing through that TDP under load, whereas the M1* figures are absolute, as far as I have read.
In short, 35-45W is a pipe-dream if you’re pushing the chip. I personally doubt it gets near 200W though…
Intel has always used TDP as normal TDP under baseline usage. They have now introduced a term called Turbo Power or previously known as PL2, when you max out the CPU to better reflect on true TDP.
And I should have wrote it better. It was compared to i9-12900K on Desktop which has a Turbo Power of 240W.
But on a single core basis, the difference should be minimal. The Mobile CPU will still uses ~20W max assuming information and Geekbench number being equal. We will have to wait for final review to confirm.
Marketing BS, PR BS, etc. In the long run, Apple will dominate the laptop market, because there's a substantial tech advantage that Intel can't easily mitigate/reduce.
Don't know what do you mean by "dominate the laptop market" because in terms of market share it certainly isn't.
Also, contrary to popular belief the A14 and it's child/sibling the M1 isn't made of magic and the "double in every stuff" each generation train has long gone away ( see A15 ). The M1 is great, performance per watt is really great, but the "tech advantages" Apple has are due to Intel's horrible track record as a tech company in the last decade.
I think the M1 was another "AMD moment" for Intel, I'm mildly optimistic they are starting to get their shit together. While I'm mildly certain Apple is starting to suffer from the same rot Intel has been suffering.
> In the long run, Apple will dominate the laptop market, because there's a substantial tech advantage that Intel can't easily mitigate/reduce
Tech advantage is not what drives sales in laptop market. Price does. Unless Apple comes up with a $600 entry-level MacBook SE, to think that "Apple will dominate the laptop" is just absurd.
> In the long run, Apple will dominate the laptop market, because there's a substantial tech advantage
That's a very short-sighted view. The fact of the matter is that there's no offer from Apple in the $800 segment.
There's also a lack of 2-in-1 devices (and arguably the iPad family cannot fill this niche as long as iPadOS remains dumbed-down).
So unless the entire Laptop market shifts to $1000+ devices exclusively, there's no way this will happen.
Unfortunately I cannot provide you with hard sales data, since the relevant market reports cost upwards $2.500, but their overall market share is about 10% [0] so go figure.
Agreed. I bought my laptop four years ago for £600. It has a 3200x1800 bright and clear display, 8GB RAM, is thin and light, the battery lasts six hours, and I really don't care that the CPU is an Atom, because I'm always logged into a heavier system over the network.
Apple isn't in my section of the market. I appreciate the technological advances that they are making - they're good. But it's not for me.
> And while I doubt x86 architecture will ever beat ARM on power efficiency, absolute power (within roughly comparable battery life) is another matter altogether.
This is a bit confusing, are they suggesting that laptops with this Alder Lake Mobility CPU will have "rougly comparable battery life" to M1 Max Macbook Pros, with better performance, despite having worse power efficiency?
That's true, but do they have much headroom to do that without increasing the watt-hours? I know max size Macbook Pros have been at the TSA flight limit on that metric for ages and that's one reason it never increases.
There's a 100Wh restriction on laptop batteries imposed by civil aviation, which the 16" MacBook Pro hits, so this is not a problem that can simply be solved with thicker chassis and larger batteries either.
> There's a 100Wh restriction on laptop batteries imposed by civil aviation
This made me curious so I looked it up and found this bit on FAA's web site:
> Size limits: Lithium metal (non-rechargeable) batteries are limited to 2 grams of lithium per battery. Lithium ion (rechargeable) batteries are limited to a rating of 100 watt hours (Wh) per battery. These limits allow for nearly all types of lithium batteries used by the average person in their electronic devices. With airline approval, passengers may also carry up to two spare larger lithium ion batteries (101–160 Wh) or Lithium metal batteries (2-8 grams). This size covers the larger after-market extended-life laptop computer batteries and some larger batteries used in professional audio/visual equipment.
Can we see the size of the laptops it will be installed in?
Will it be able to reach the same performance while on battery, like the new MacBook Pros?
C’mon, I understand Intel is in a push to counter the Intel-is-screwed narrative, but their comms department should try to do better than this old school benchmark “leaks” post, with cherry picked comparisons and such.
Not to mention power comparisons are not entirely fair considering the node disadvantage Intel has. I would be excited if Intel can show higher clicks and overall IPC with reasonable excess power attributable to node difference.
It's absolutely fair. Try to find an Intel CPU on the market without a node disadvantage. Doesn't exist. We can only (and it only makes sense to) compare what exists now, what consumers are able to buy right now. If Intel magically comes up with a CPU that runs on a similar node, then sure, but until then, the reality is that they're on the market at a node disadvantage. That's not something you can just wave away as "oh, well, Apple has a better node". All that matters is what devices you can buy and what their performance/power metrics are actually like.
Agreed. The argument seems to be I’ll be happy with Intel if IPC etc is competitive as I want the raw power and don’t care about battery life etc. Which might just be true for a very small section of the laptop buying market but leads to a much less useful product for the very large majority.
The point of fairness is not to give participation trophies to a company because the others had better tools. If a product is worse, then it is worse. Warm feelings because Intel does not have access to the best technology is not going to make any laptop run cooler.
I can understand the “they’re an underdog, so they deserve some money even if their products are not the absolute best”, but casting Intel as the poor underdog being steamrolled by their competition because they cannot get the latest and greatest is strange, to say the least.
I should have clarified this earlier. Personally I'm interested in how competitive Intel will be in the future (their stock took a massive dive recently).
Right now, it seems to me that Intel is behind both on the node and on the architecture vs M1 (and AMD). I expect them to eventually fix the node (or just buy from TSMC) but if they don't fix the latter that's a big deal.
From the POV of buying a CPU/laptop today, absolutely, power comparison and performance both are essential.
You've got a point, but the discussions are most of the time not about the theoretical characteristics of an architecture on a process node that it will never use. It's people beating each other on the head about which mega-corp has CPUs a few percents faster than the other.
From a historical point of view, a systematic shift to hotter, more power-hungry chips was never a good sign (see for example the various Pentium 4s and the successive incarnations of Bulldozer and Piledriver). I hope it's a stopgap and that they have a new Core 2 masterpiece being readied, but let's not kid ourselves: it does not look good. the Physics of heat transfer put a limit to the gains in performance you can achieve if you don't also increase efficiency. It's cold comfort to know that your CPU would be wonderful if cooled with liquid nitrogen when your desktop throttles under load.
To be strictly fair, the AMD Ryzen 7 PRO 5850U is a 10-25W TDP depending on manufacturer configuration, not nailed to 15W. Do cpubenchmark verify that it's running at a 15W TDP when coalescing the numbers?
There is no guarantee for that, but usually 5850U is used in ultra-thin laptops and they don't have enough cooling for more than 15W... For example, CPU in Thinkpad X1 Extreme GEN 4 is limited 27 WATT (at least models without vapor chamber) and it's a laptop with quite a good cooling and larger size.
That's significantly less true than it was in the past. I think it's pretty easy to argue that the browser is the most important desktop app now and all modern browsers are multiprocess / multithreaded.
Beyond that, almost everything that does signal processing of some shape or another (audio, video, whatever) is multithreaded these days, so that one's out too. Compiling code is multi-process too.
I actually struggle to think of important work loads that are single-threaded these days. Obviously most simple apps are, but mainly because they don't use enough CPU to bother optimizing. We're more than a decade into multi-core systems being the norm and most CPU-heavy apps have taken advantage of that.
Sure, but there are very few UIs that require 100% of a modern CPU core. Exceptions exist, but they're just that, exceptions.
I'd go one step further and say that if modern apps are doing that that they should move the CPU heavy tasks out of the message thread. (And most have, again, since we're like 10-15 years into multi-core systems being the default.)
Not an «apple fanboy». You should justify that TDP is the measure we are looking for, instead of just the available measure. TDP is a maximum, and an average one for that matter.
Those measures should use something like `powerstat` results standardizing other conditions - in fact, they should use benchmarking (just like the numerator of the numbers you report does, but not the denominator). This system here has a TDP of 6W, but uses 3.3W during light use, and peaks around 7.5W.
Sure, there could be a literal interpretation of "Performance/Power" which somehow could link speed and TDP (benchmarking at the same stated TDP), but (°) it would be imprecise as TDP is an estimate (so consumption should be benchmarked) and (°) it would not represent some more useful scenarios like "which system allows squeezing more working time out of a battery" (which is not represented by considering maximums - especially given technologies like big.LITTLE).
Similarly, Intel often states TDPs of 25W, 45W, 65W, etc... while peaking actual power usage at nearly double those numbers during "power boost."
So unfortunately TDP has basically become a marketing term and we're back to literally measuring power usage instead. Which of course gets complicated when you consider that different machines have wildly different heat management systems.
"TDP" as such is going away for Alder Lake. Intel is now using a much clearer base power and turbo power number. So for example, the new i5-12600K has a Base Speed of 3700MHz @ 125W and Turbo Speed of 4900MHz @ 150W (if and when thermals allow it) [0]. The old marketing would've just called this a TDP of 125W.
M1 Max operates with a peak of 50W.
With a top performance difference of let’s say 20% (unattainable by Intel), I care more about having a computer that runs at 50W tops than a water boiler at 150W.
It’s fascinating to me that society was willing to accept laptops that could not safely be used on a human’s lap for so long. My ‘old’ 16” MacBook Pro could probably cook an egg.
I struggled to understand that you compute score/TDP
> Waiting for apple fanboys telling me something like "but TDP is not a good measure"...
Not an Apple fan, but TDP is not even a measurement, it's a specification of the heat dissipation required for that CPU. So in theory it corresponds to the maximum power dissipated by the CPU under load, but in practice two machines with the same TDP may be equipped with differently dimensioned heatsinks which will influence your metric.
You should consider using a real average power consumption measurement to establish your metric.
> So in theory it corresponds to the maximum power dissipated by the CPU under load
I think this isn't quite right - TDP is the amount of heat the system needs to be able to dissipate in order to prevent thermal throttling at a given base speed. If the system can dissipate more (and provide more power), chips can clock higher to a turbo or boost speed.
Even if true I'm done with Intel. I'd rather give my money to AMD at this point. Team Red moving forward. All those years with 4 cores maximum is over.
>So someone sent you some bullshit in the mail and you jizzed all over yourself. It means nothing unless you tested it. Anyone can make a graph and fill it with whatever number they choose. Intel has been making up numbers for more than a decade.
HN's obsession for power per watt is hilarious. Just because you can push out X TFlops at 20W does not mean you can push out 5X TFlops at 100W. It doesn't work this way.
Despite my disappointment in Intel in the past years, I can only be happy about them pushing PCIE5 and DDR5. This confirms my next PC upgrade, although likely on a future Ryzen will be a spicy one.
For a laptop, sure, but that's not what's interesting me in a desktop, which is what people are jerking themselves off over with the potential of the M1 and M2. For many workloads, power consumption isn't an issue. I want my CPU to draw 150W and burn as hard as possible to make my build finish as fast as possible.
The State of California last summer sent out news they were considering banning or restricting the sales of desktop computer intended for gaming, out of power consumption considerations. And there is an eletric power crises in many areas of the world currently.
Please, do not get your news from Reason magazine. California Title 20 efficiency code for appliances includes formulas for how much energy a computer may consume in a year. They don't ban gaming systems or any other system. I actually really appreciate how these regulations are written because they favor computers that do more with their energy input, since they allow a larger energy budget for powerful computers, scaling up based on system memory, storage, and bandwidth.
Ah, yes, the evil gaming PCs with their 650-800W power supplies must be banned.
I don't see California banning keeping neon signs on at night when buildings and shops are closed. Their threat is pure posturing and pretending that the problem lies on individuals when the most egregious power wastes are done in companies.
With soldered on, unreplaceable, unupgradeable LPDDR5, which is not DDR5. There's massive differences in the specs.
I could have bought SoCs way before Apple decided it was a perfect way to get people to have to buy brand new hardware every time. They're not new technology.
LPDDR is soldered down because it's faster that way. There is no such thing as LPDDR on a stick. All 11th-generation Intel parts with LPDDR are also soldered on, because that's the only way to build them. I guarantee you that all Alder Lake mobile designs will have soldered RAM.
Your original comment is false, as Apple's M1 uses LPDDR5, something that has about half the clock rate and transfer rate of current DDR4. It is not DDR5 as is commonly expected when talking about DDR5, and is merely Apple's marketing Kool-aid.
Additionally, Tiger Lake is merely LPDDR-compatible, not forced. You can have a Tiger Lake with regular DDR, and upgrades are easy to do. So, once again, _some_ Alder Lake designs will have soldered LPDDR5, which is low grade DDR4.
You are seriously confused. The LPDDR4X is faster, not slower, than DDR4. You can equip Tiger Lake with DDR4, but it sucks, which is why the Framework laptop has worse performance and less battery life than all of its competitors.
Specs wise, LPDDR4X would be indeed, able to beat out an average DDR4-2800 under certain circumstances. Those circumstances are quad channel (that the M1 does not have) and 3733 MHz (which the M1 does have). That of course goes out the window if you spec some DDR4-3200 and quad channel it too, which will blow LPDDR4X out of the water.
And of course, once again, you make this a laptop thing when the original post is about people believing the M1/M2 would scale linearly with power usage and be a strong contender to desktop CPUs. So, no, there's no DDR5 in the M1, its LPDDR4X is not faster than DDR4, and please stop trying to force feed me the Apple kool aid, this has never been about battery life.
I’m just going to say: after the monopoly-induced stagnation of the 2010s, finally we’re back to something looking like competition-driven Moore’s Law-level performance increases.
I don't really think it is fair to describe the situation where there are two major competitors and one of them ships abject garbage for ten years as a "monopoly". Can we blame Intel for the junk AMD made starting from Barcelona all the way up to and including Naples?
Nah, this is one of the last pushes before true stagnation due to physics.
The next 5-10 years will have rapid innovation where you'll see ever growing caches and/or unified caches, higher bit interconnects (a la 512bit ddr5 bus on the m1 max), and a bigger emphasis on heterogeneous computing.
Nano ribbons and better gaa should give another boost in 3-5years or so but we're rapidly approaching an end to what we can do.
After reading the comments here: I want everyone to know that wccftech is tabloids, not reporting. Go there if you're the Men in Black looking for alien leads or if you're just bored and are curious about the (incredibly dubious side of the) hardware rumor mill. If you don't believe me, just click around a bit.
125 comments
[ 6.6 ms ] story [ 207 ms ] threadStill, if ever they manage to at least be competitive on the process node front, that might make things interesting...
As is, while technically impressive they even managed this, it's not like the difference is large enough to be impactful.
The big elephant in the room for Intel is the x86 instruction set. It's a technical debt that is not addressed in anyway as far as I can see. Until then, we'll have to choose between the Apple way or the inefficient way.
All those caveats aside:
- https://www.cpu-monkey.com/en/cpu-amd_ryzen_7_pro_5850u-1905 - https://www.cpu-monkey.com/en/cpu-apple_a13_bionic-1654
Geekbench scores: A13: 1336 single, 3187 multi 5850u: 1414 single, 8140 multi
And there are lots of tricks apple uses to keep power consumption down _in addition to_ using arm; and that's partially why it does better than e.g. qualcomm. But many of those tricks aren't somehow arm or apple exclusive; they're just a question of (expensive) engineering or die size - things that aren't fundamental to intels usage of x86.
I mean, it might well be that x86 hits some ceiling well before arm's ISA does, but I don't see actual evidence of that being a huge factor yet. At the very least, it feels like it's clear the process node helps, as does all kinds of other engineering, but that the impact of the ISA is not as obvious.
Course, I might just be missing the obvious evidence, sure ;-).
The M1 was released in Nov of 2020, if rumors are true, we're due an M2 in Feb of 2022.
Not only because M1 is at TSMC "5nm" (which is probably equivalent to Intel 7nm) but also because this happens every time intel gets kicked to the curb.
"Guys, don't go AMD, the new Intel CPU will be faster"
When the product arrives it usually comes with a huge trade-off such as thermals and power consumption and doesn't do the performance originally claimed anyway.
This has happened after every Ryzen launch, suspiciously only a day or so after each event. It's so obvious and I don't think I'm going to fall for that again.
No more promises: Goodbye Intel, I'll see you when your product can be in my hands and do what you claim.
To be fair, for years and years, new Intel CPUs were faster/better/whatever. What you remember as "kicked to the curb" was really something more like an occasionally great product in a sea of mediocrity where the industry was having terrible trouble competing. Someone would have a chip out with great reviews, and then the new Intel architecture would ship and everyone would forget about it.
The current environment where they lag in almost all the high end product spaces is very much a novelty.
What time period or space are you talking about here? Because that did not happen in the last 10 years for consumer and server processors. The last decade saw AMD with the FX processors failing to be competitive, at first single threaded, over time also multi-threaded and in performance per watt. Then after many years Ryzen came out and was at first just surprisingly competitive, then plainly better in every metric. And that's where we are now. The whole back-and-forth you are describing did not happen at all, and there were only AMD and Intel anyway, not "someone" or multiple players.
Intel hadn't shipped a dedicated GPU in >20 years (until the low-end DG1 last year) and their integrated graphics were anemic at best for most of that time period.
"2-4 years behind Nvidia" would be pretty impressive of a leap to have been aiming for in some meeting a few years ago.
I don't expect to be purchasing their DG2/"Arc Alchemist" when it launches in a few months, but if it's remotely competitive with mid-range current generation cards (as has been rumored/claimed), that'll be pretty impressive to my mind, to basically enter a completely new market segment for them and have a product that's even worth discussing.
Sounds pretty normal to me. Why would a company enter a new market if they don’t have a product people want?
I can’t think of any company that launched a new product by announcing “We don’t have anything competitive.”
Am I misunderstanding your comment? (Or reality.)
Roll your product out with a huge array of confusing names like i9-115479k that nobody can really follow. Nobody has the money to buy all of these products, so nobody can truly claim that none of your products are better than the competition, even if none of them offer a better price:performance:consumption ratio.
And even if somebody does call you out on it... well, don't worry! The next release is going to be even better. Just you wait.
When Apple moved from Motorola to IBM PPC: Motorola soon to release super fast chip updates!
When Apple moved from IBM PPC to intel: IBM soon to release super fast chip updates!
When Apple moved from Intel to M1: Intel soon to release super fast chip updates!
These companies need to release this press coverage - they’re publicly traded and need this.
The fact though is that right now people are using the Max, of course I’d expect some unreleased future chip to be faster.
However even the slower M1’s are still attractive because of the myriad of other benefits they bring: true mobility chips, huge battery life, quiet operation: the Air doesn’t even have a fan, boots/wakes/video switches are seamless, and the special sauce: the chips merge general purpose computing with application specific computing: apple will always be able to generate a tailored product for their audiences.
And it should go without saying: the pro desktops are certain to have compelling performance, and we can expect M2, M3 and so on to continually advance from where we are today.
Moving away from intel wasn't about performance alone, it was about taking control of their own product’s direction.
The article just says some future chip, with unknown memory, at some unspecified power level and cooling solution, will be able to eke out a benchmark win against the part Apple is shipping today in laptops.
Imagine if that wasn't true. THAT would be news.
What do you mean by this? I have an 8th gen i5u in an HP laptop at work, and it's been driving an external 4k@60Hz screen for years. It has no dedicated GPU at all, just the integrated UHD 620 or something like that.
It can even daisy chain two external displays, in addition to the integrated panel. Not sure if it can drive two external 4k like this, I only have FHD + 4K to test.
The display is an LG, can't remember the model, but it worked with multiple other displays, HP, Dell, etc.
Your issue may be related to the connection. The limitation of 30 Hz makes me think you're connecting via HDMI. If that's all you have, you're SoL.
4k@60 requires HDMI 2, but Intel CPUs only recently have got that. 8th gen certainly doesn't have it.
You need DisplayPort for this to work. Laptops don't often have DP connectors, but they may have USB-C and/or Thunderbolt with DP alt-mode. This is what I use, hence the DP-xxx output name.
If you look closely, there's the DP logo next to the USB-C connector for my particular model: https://wu-tc-shop.com/WebRoot/Store11/Shops/a31626df-5333-4...
I'm not sure what you're on about in regards to 4k needing a dGPU. That's not true.
What do you mean by this? I started using 4k screen with my shitty employer-issued Intel-based Lenovo laptop more than 4 years ago.
I'm not sure I can follow this line of reasoning. If ARM chips will always be more efficient, then how can you continue stating comparable battery life.
Moreover, there is no mention of TDP. Past flagship Intel processors have been fighting thermal throttling on mobile devices, and it's already been demonstrated that the M1 Max chips don't run into throttling even under load.
It's interesting to see first benchmarks of the new Alder Lakes chips, but we will see how well they work in real products, i.e. with thermal and other constraints.
Hey it’s also 3% faster than the M1 Max!
1. Base Power and Turbo Power. Finally a simple term so we dont have to constantly argue about TDP and PL1 / PL2. Which still happens fairly often on HN. And some of us have been ranting about this for almost a decade.
2. + 19% IPC improvement vs 11900K which is basically an Icelake. Or ~10% improvement over TigerLake ( Which was never made available on Desktop )
3. PCI-E 5 and DDR5. Intel went from falling behind in PCI-E offering to leaping ahead of AMD. We should expect PCI-E 5.0 SSD shipping soon after ( 14 - 16GBps )
4. Chipset is on 14nm. No USB 4 or Thunderbolt 4. But integrated WiFi 6E MAC.
5. For those interested, this put Alder Lake, on a Desktop Platform, with Intel 7nm High Performance Node, at roughly 20-25W per Performance Core. With a ( non-verified ) Geekbench Score fairly similar ( ~6% faster ) to M1 HP Core. The Apple M1, on a TSMC 5nm Low Power node, at roughly 5W per core.
6. It will be interesting to see how the new efficiency cores perform. Which is basically a new generation Atom Core. These were previously scheduled for a Graviton like 64 - 128 Core chip on server. Not sure if that is still the case with Intel's chiplet strategy or what they called Tiles.
7. Worth wondering, Alder Lake was originally scheduled on Intel 7nm or what is now called Intel 4/3nm in 2019. What would happen had Intel not been so badly managed? But if that didn't happen, Pat Gelsinger may never be back at Intel.
8. AMD Zen 4, also with DDR5 and PCI-E 5.0 will be coming in late 2022. Depending on Intel's pace of execution, which seems to be getting better and better every time Pat Gelsinger provides an update, AMD may have some tough competition.
[1] https://news.ycombinator.com/item?id=29017221
The x4 lanes usually used for m.2 SSDs are PCIe4.
I mean, those literal nm measurements had no reasonably obvious basis in measurement for either intel or TSMC even before the intel rebrand, so...
It'd be nice if they advertised with a few real benchmarks (e.g. logic density + sram density, some kind of clockspeed metric, some kind of powerleakage metric)... but as is, it's just a branch anyhow.
Whether it "is" a 10nm process depends on what your definition of "is" is ;-). It's certainly not the case that a transistor is 10nm large, not even close.
Of course, this is probably all intentional from Intel's point of view (letting people say "Intel 7nm" rather than "Intel 7").
I believe Intel is still counting on using x4 PCIE4 for SSDs. I wonder if the PCI-E 5 x16 makes any difference for the GPUs?
DDR5 may bring in nice capacity boosts. However, I read somewhere that at lower capacities it is almost indistinguishable perf wise from high-end DDR4.
The key to M1's attraction is the cohesiveness of its implementation. Apple can make MacOS tightly knit to the heterogenous core types and caches. and essentially gate keep applications to behave optimally. From Intel's announcements, it looks like Intel came up with Thread Director to _advice_ Windows on how to schedule stuff and leave it to MS to implement things. Not sure how that synergy is going to take shape. Perhaps Linux will come out and eat this.
if AMD can improve on Zen 3 and come up with Zen 4, they already provide much more PCI-E lanes and ECC on DDR4/DDR5 memory support, with better than intel multi-threading perf. Competition sure is nice!
In short, 35-45W is a pipe-dream if you’re pushing the chip. I personally doubt it gets near 200W though…
Intel has always used TDP as normal TDP under baseline usage. They have now introduced a term called Turbo Power or previously known as PL2, when you max out the CPU to better reflect on true TDP.
And I should have wrote it better. It was compared to i9-12900K on Desktop which has a Turbo Power of 240W.
But on a single core basis, the difference should be minimal. The Mobile CPU will still uses ~20W max assuming information and Geekbench number being equal. We will have to wait for final review to confirm.
Also, contrary to popular belief the A14 and it's child/sibling the M1 isn't made of magic and the "double in every stuff" each generation train has long gone away ( see A15 ). The M1 is great, performance per watt is really great, but the "tech advantages" Apple has are due to Intel's horrible track record as a tech company in the last decade.
I think the M1 was another "AMD moment" for Intel, I'm mildly optimistic they are starting to get their shit together. While I'm mildly certain Apple is starting to suffer from the same rot Intel has been suffering.
Tech advantage is not what drives sales in laptop market. Price does. Unless Apple comes up with a $600 entry-level MacBook SE, to think that "Apple will dominate the laptop" is just absurd.
That's a very short-sighted view. The fact of the matter is that there's no offer from Apple in the $800 segment.
There's also a lack of 2-in-1 devices (and arguably the iPad family cannot fill this niche as long as iPadOS remains dumbed-down).
So unless the entire Laptop market shifts to $1000+ devices exclusively, there's no way this will happen.
Unfortunately I cannot provide you with hard sales data, since the relevant market reports cost upwards $2.500, but their overall market share is about 10% [0] so go figure.
[0] https://bit.ly/3pNAamq
Apple isn't in my section of the market. I appreciate the technological advances that they are making - they're good. But it's not for me.
This is a bit confusing, are they suggesting that laptops with this Alder Lake Mobility CPU will have "rougly comparable battery life" to M1 Max Macbook Pros, with better performance, despite having worse power efficiency?
This made me curious so I looked it up and found this bit on FAA's web site:
> Size limits: Lithium metal (non-rechargeable) batteries are limited to 2 grams of lithium per battery. Lithium ion (rechargeable) batteries are limited to a rating of 100 watt hours (Wh) per battery. These limits allow for nearly all types of lithium batteries used by the average person in their electronic devices. With airline approval, passengers may also carry up to two spare larger lithium ion batteries (101–160 Wh) or Lithium metal batteries (2-8 grams). This size covers the larger after-market extended-life laptop computer batteries and some larger batteries used in professional audio/visual equipment.
https://www.faa.gov/hazmat/packsafe/more_info/?hazmat=7
Can we have Alder lake’s power per watt metrics?
Can we see the size of the laptops it will be installed in?
Will it be able to reach the same performance while on battery, like the new MacBook Pros?
C’mon, I understand Intel is in a push to counter the Intel-is-screwed narrative, but their comms department should try to do better than this old school benchmark “leaks” post, with cherry picked comparisons and such.
I can understand the “they’re an underdog, so they deserve some money even if their products are not the absolute best”, but casting Intel as the poor underdog being steamrolled by their competition because they cannot get the latest and greatest is strange, to say the least.
Right now, it seems to me that Intel is behind both on the node and on the architecture vs M1 (and AMD). I expect them to eventually fix the node (or just buy from TSMC) but if they don't fix the latter that's a big deal.
From the POV of buying a CPU/laptop today, absolutely, power comparison and performance both are essential.
I’m an Intel bear on their existing markets - mainly because even if they achieve parity with TSMC the lead is gone and others will take market share.
I’m more positive on the Intel GPU business though given the potential huge demand for ML compute.
From a historical point of view, a systematic shift to hotter, more power-hungry chips was never a good sign (see for example the various Pentium 4s and the successive incarnations of Bulldozer and Piledriver). I hope it's a stopgap and that they have a new Core 2 masterpiece being readied, but let's not kid ourselves: it does not look good. the Physics of heat transfer put a limit to the gains in performance you can achieve if you don't also increase efficiency. It's cold comfort to know that your CPU would be wonderful if cooled with liquid nitrogen when your desktop throttles under load.
https://www.cpubenchmark.net/compare/AMD-Ryzen-7-PRO-5850U-v...
AMD 5850U 1253 scores vs Apple M1 1004 scores
Waiting for apple fanboys telling me something like "but TDP is not a good measure"...
I think this might be the comparison you were after? https://www.cpubenchmark.net/compare/AMD-Ryzen-7-PRO-5850U-v...
Beyond that, almost everything that does signal processing of some shape or another (audio, video, whatever) is multithreaded these days, so that one's out too. Compiling code is multi-process too.
I actually struggle to think of important work loads that are single-threaded these days. Obviously most simple apps are, but mainly because they don't use enough CPU to bother optimizing. We're more than a decade into multi-core systems being the norm and most CPU-heavy apps have taken advantage of that.
I'd go one step further and say that if modern apps are doing that that they should move the CPU heavy tasks out of the message thread. (And most have, again, since we're like 10-15 years into multi-core systems being the default.)
Those measures should use something like `powerstat` results standardizing other conditions - in fact, they should use benchmarking (just like the numerator of the numbers you report does, but not the denominator). This system here has a TDP of 6W, but uses 3.3W during light use, and peaks around 7.5W.
Sure, there could be a literal interpretation of "Performance/Power" which somehow could link speed and TDP (benchmarking at the same stated TDP), but (°) it would be imprecise as TDP is an estimate (so consumption should be benchmarked) and (°) it would not represent some more useful scenarios like "which system allows squeezing more working time out of a battery" (which is not represented by considering maximums - especially given technologies like big.LITTLE).
So unfortunately TDP has basically become a marketing term and we're back to literally measuring power usage instead. Which of course gets complicated when you consider that different machines have wildly different heat management systems.
[0] https://www.anandtech.com/show/16959/intel-innovation-alder-...
> Waiting for apple fanboys telling me something like "but TDP is not a good measure"...
Not an Apple fan, but TDP is not even a measurement, it's a specification of the heat dissipation required for that CPU. So in theory it corresponds to the maximum power dissipated by the CPU under load, but in practice two machines with the same TDP may be equipped with differently dimensioned heatsinks which will influence your metric. You should consider using a real average power consumption measurement to establish your metric.
I think this isn't quite right - TDP is the amount of heat the system needs to be able to dissipate in order to prevent thermal throttling at a given base speed. If the system can dissipate more (and provide more power), chips can clock higher to a turbo or boost speed.
>So someone sent you some bullshit in the mail and you jizzed all over yourself. It means nothing unless you tested it. Anyone can make a graph and fill it with whatever number they choose. Intel has been making up numbers for more than a decade.
Despite my disappointment in Intel in the past years, I can only be happy about them pushing PCIE5 and DDR5. This confirms my next PC upgrade, although likely on a future Ryzen will be a spicy one.
But it does mean you can push out X TFlops at 20W for 5x longer given the same battery than you can at 100W, no?
Economy of consumption remains a "hot" topic.
https://energycodeace.com/site/custom/public/reference-ace-t...
I don't see California banning keeping neon signs on at night when buildings and shops are closed. Their threat is pure posturing and pretending that the problem lies on individuals when the most egregious power wastes are done in companies.
I could have bought SoCs way before Apple decided it was a perfect way to get people to have to buy brand new hardware every time. They're not new technology.
Additionally, Tiger Lake is merely LPDDR-compatible, not forced. You can have a Tiger Lake with regular DDR, and upgrades are easy to do. So, once again, _some_ Alder Lake designs will have soldered LPDDR5, which is low grade DDR4.
This is still not DDR5.
Specs wise, LPDDR4X would be indeed, able to beat out an average DDR4-2800 under certain circumstances. Those circumstances are quad channel (that the M1 does not have) and 3733 MHz (which the M1 does have). That of course goes out the window if you spec some DDR4-3200 and quad channel it too, which will blow LPDDR4X out of the water.
And of course, once again, you make this a laptop thing when the original post is about people believing the M1/M2 would scale linearly with power usage and be a strong contender to desktop CPUs. So, no, there's no DDR5 in the M1, its LPDDR4X is not faster than DDR4, and please stop trying to force feed me the Apple kool aid, this has never been about battery life.
The next 5-10 years will have rapid innovation where you'll see ever growing caches and/or unified caches, higher bit interconnects (a la 512bit ddr5 bus on the m1 max), and a bigger emphasis on heterogeneous computing.
Nano ribbons and better gaa should give another boost in 3-5years or so but we're rapidly approaching an end to what we can do.