Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Intel Alder Lake vs. Apple M1
- Thread starter leman
- Start date
- Sort by reaction score
That's because it has 4 more cores. M1 Max has 10 cores while 12900HK has 14 cores.
Also, 12900HK achieved that score with 100W while M1 Max used only 30W.
Working on getting hands-on Alder Lake but for AMD it matches the specs. For example, 5950x has default 105W TDP. I'm running it above base clock with core boost and it peaks out at ~124W. Will get around to to dialing it down to base clock and remeasuring TDP.
Not too much higher and still consume a lot of power than M1 Max. So doesn't really change the fact that 12900HK consume a lot of power in order to outperform M1 Max. 100W vs 30W is a joke to me.
You're falling for marketing since it doesn't jive with performance/watt results.
12900k scores 27401 on Cinebench R23 at peak 251W
5950x scores 24071 on Cinebench R23 (I also get 23794 to 24371 on air cooling at peak 123W)
So, 12900k is ~15% faster at 2x the wattage.
https://www.techspot.com/review/2351-intel-core-i9-12900k/
As @jeanlain said that’s because 5950X is simply a better, more efficient design and that Intel processor is being pushed waaay too far over on its power curve to compensate. Tests indicate that lowering wattage down to a about 150W makes only small losses in performance (like 10-15% I think, but past 150W it gets bad) and in many non-AVX that where the chip is hanging out anyway.
I linked to 3rd party analysis somewhere in this giant of a thread that estimates that Intel 10/7 is roughly TSMC 7. However, we won’t know for certain where things stand until comparable processors are produced on each. I don’t know when the first ARM cores will be rolling out on Intel fabs with IDM 2.0 or what node they’ll be on but we’ll have a better idea when they do.
Last edited:
I agree, but you suggested that the M1 Max was faster, not just more power-efficient.
It's always difficult to compare power efficiency between CPUs that have different numbers of cores. I think results from single-threaded tasks are the most informative in that regards.
The firestorm core consumes about 5W during SPECs ST, and recent intel CPU cores used to consume about 25W for a similar score. I'm curious to know how much the 12900HK consumes during this task.
"12900HK is not even close to M1 Max base on the real test. It consumed way more power so their marketing chart is totally fake. Intel Alder lake just consume too much of power so it's so stupid to even dare compare with M1 series."
Because I was talking about the power consumption. Didn't say M1 Max is faster. Also, both single and multi core performance isn't that better than M1 Max especially since 12900HK requires 100W for that score and when it dropped to 75W, the multi core performance is totally worse than M1 Max. This is why I kept mentioning the power consumption.
It's very difficult for me to extract relevant information from that link. Not just because it's not my native language, but also because that page is very poorly organised. Where can I find power-consumption for single-threaded tasks?
I think it should be stated explicitly what the purpose and impact of a relatively small TDP that Apple silicon sticks to & means in reference to Intel's larger TDP that produces better raw performance numbers.
Professionally I keep to a 65W or lower charger for Windows laptops like the Lenovo Thinkpad E Series AMD so that when a hurricane/typhoon/quake/eruption occurs the portable generated being used can easily charge & power it vs a monster machine that exceeds 140W.
Of course each has their own use case but for those unwilling to spend more than $800 for a Windows laptop you gotta be very picky.
Yeah I had the same problem. Truthfully I’m only going off of the ST PL2 which you can see stated in one on one of the pictures which was 25W and for CB23 is probably good enough on Intel processors for actual power consumed, maybe a little less on average. I believe Anandtech said ST W for the desktop i9 was 25-30W, so that tracks.
Any exclusives MS, Sony, or Nintendo have are because they are the publisher, own the dev(or IP), or have paid for timed exclusivity and the game will be released to other platforms when it is over.
In this case Apple is using option 3, which isn’t bad.
But it sounds like Nvidia's hardware is so good that they are able to lock people in with their software or vice versa. It sounds a lot like Apple.
Metal predates Vulkan, actually. Apple first shipped Metal in 2014. Vulkan was announced at GDC 2015 and the first release was early 2016.
Mantle initial release was in 2013 which then evolved into Vulkan. Remember playing Battlefield 4 which came out in 2013 and was one of the early if not first Mantle games. What's the first MacOS Metal API game? Looks like Metal API didn't come to MacOS until 10.11 which was 2015.
Last edited:
At 115watts Scaldedlake or Scalderlake gets toasty just for the cpu. Add an Nvidia 3080 and some ram and things get down right hot!
I can't wait until laptops start to ship with 4 fans and 12 heat pipes because well Scalderlake! Where are my marshmallows and a stick! Time to start roasting them over the new multicore Intel alderlake cpu!
I have read much higher wattage levels than 115 but what do I know. Guess time will tell.
I can't wait until laptops start to ship with 4 fans and 12 heat pipes because well Scalderlake! Where are my marshmallows and a stick! Time to start roasting them over the new multicore Intel alderlake cpu!
I have read much higher wattage levels than 115 but what do I know. Guess time will tell.
08-Unplugged-1.jpg
If the geekbench score of 7849 is really what an alder lake laptop yields on battery, then it is indeed a very disappointing result. That's barely better than an M1 MacBook Pro!
I'm sure other laptop design would give better scores, but that will be at the cost of lower better life.
For a given per-core power efficiency, it is the CPU with more cores that should be more power-efficient in a MT task, not the other way around! The 12900HK has 14 cores and the M1 only 8.
The 20-thread score is only 4.12x higher than using just 1 thread. That's ridiculously bad and it shows how much performance is limited by the power allocated to each core when the machine is unplugged.
The cinebench R23 score on battery (and with apparently 23W CPU package power on average) appears to be even lower than the M1 score.
06-Unplugged-1.jpg
If the geekbench score of 7849 is really what an alder lake laptop yields on battery, then it is indeed a very disappointing result. That's barely better than an M1 MacBook Pro!
I'm sure other laptop design would give better scores, but that will be at the cost of lower better life.
For a given per-core power efficiency, it is the CPU with more cores that should be more power-efficient in a MT task, not the other way around! The 12900HK has 14 cores and the M1 only 8.
The 20-thread score is only 4.12x higher than using just 1 thread. That's ridiculously bad and it shows how much performance is limited by the power allocated to each core when the machine is unplugged.
The cinebench R23 score on battery (and with apparently 23W CPU package power on average) appears to be even lower than the M1 score.
06-Unplugged-1.jpg
Last edited:
As so often, you have summarised my sentiments of the matter way better than I could do myself
Nvidia was dominating the GPU market as well being a very powerful core member of Khronos. They pledged their support for OpenCL while de-facto sabotaging it and promoting their proprietary platform instead. They essentially politically outmanoeuvred smaller companies who fully committed to OpenCL and then found themselves sitting on top of a fragmented, broken ecosystem. Nvidia's strategy here is not unlike Microsoft's.
That's just cherry-picking. Remember GeForce FX fiasco? Regardless, it is a hard fact that Nvidia is an undisputed GPU market leader and that they were the first ones to bring important technologies to the market. My argument is simply that most of their success comes from aggressive business execution, planning and marketing rather than from pure technological superiority. They invested into right things into right times and executed flawlessly to secure their market dominance.
Oh, I was not aware that Apple Arcade is exclusive. I certainly din't support it.
Just a quick though on this: I believe Apple was waiting for Vulkan but dropped out when they saw where the committee is going with it. The thing is, Apple wanted a framework that would be easy to pick up and use, and that would cater well to their GPU tech. OpenGL, while hopelessly outdated, was at least really straightforward — you could have a basic shaded triangle running in minutes as a novice learner. Vulkan on the other hand is terrible to work with, it's an ugly API that you have to micro-manage in extreme, and you can forget about using if you are not an advanced C/C++ programmer.
I cannot really read these things. Is the 30W power usage graph for one core (which I assume it is) or for multicore? Outperforming M1 by 5% at 6x power consumption is embarrassing at best, but delivering that multicore performance at 30W would be seriously impressive.
The GB5 scores are no surprise and on par with the leaks we saw so far. A bit faster than 8+2 M1 on burst workloads, probably while consuming 80-90W of power). In sustained workloads at 45W it's probably closer to 80% of M1 Pro/Max performance.
I can not read too but pics can speak for itself. Let's wait for more and actual test results as 12900HK is not officially released yet since there are some people doubting about the result. They are waiting for actual test with actual software so far...
Yeah I made a mistake … I think those were unplugged MT PL1 numbers not ST PL2. So judging by the graphs I’m not really sure, ST W might be in the same 25-30W range if the power consumption figures for R23 plugged are for ST. Which they seem to be.
I *think* the power figures are for ST.
Last edited:
Which would make sense as it’s exactly the ST power draw of previous high-end Intel mobile CPUs.