That's average power draw, not max. Max is obviously higher.
For instance, the 12900K may be quoted to have an average power draw of about 120W but it still draws up 240W.
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.
Emulated 12900K for mobile has better performance with low power consumption like M1 Max?
- Thread starter sunny5
- Start date
- Sort by reaction score
That's average power draw, not max. Max is obviously higher.
For instance, the 12900K may be quoted to have an average power draw of about 120W but it still draws up 240W.
Well, I asked you a source and yet, you didnt provide one. Where is the source that 12900K is slower than M1 Max in almost everything other than Cinebench? I seriously wish to know.
Well, the score and power efficiency looks very similar to M1 Max/Pro's CPU at 35W. If 12900K at 35W draw more power, than it will increase the performance. Is it not?
But you cannot provide the opposite either, and it is well-known that Apple Silicons performs bad in Cinebench. Basically you cannot draw a conclusion like that because Apple Silicon is already losing in Cinebench anyways.
Well, the proof that I ask is way more important. Where's the result that M1 Max beat 12900K? I'm really curious to know and yet, you are not showing anything. Didn't I ask first? Geekbench didnt show any difference from Cinebench already.
But anyway, I'm saying that Intel might achieve M1 Pro/Max's performance at lower wattage with 7nm and x86. Imagine if Intel mobile CPU uses 35W of power and the performance is similar to M1 Max/Pro. Then the next scenario will be like they will argue and proud that x86 isn't different from ARM in terms of power efficiency so forget about ARM and Apple Silicon! I thought Apple has advantage with performance by watt because of using ARM?
This is a different from 'an undervolted 12900K can outperform M1 Max in Cinebench consuming little bit more power'.
No, you don't get it. It's already drawing more power to achieve those scores.
Setting TDP with Intel processors does not limit power consumption:
Intel's Desktop TDPs No Longer Useful to Predict CPU Power Consumption - ExtremeTech
Intel's higher-end desktop CPU TDPs no longer communicate anything useful about the CPUs power consumption under load.
www.extremetech.com
Setting 35W TDP for Intel processors means at base clock, they are only allowed to draw that much power, but that doesn't mean they can't draw more for turbo boost. I.e.: you're only reducing the base clock. If cooling and power supply allow for it, the chip will still draw much more than the TDP you set.
I.e.: you're not looking at 12900K at 35W. If you think you are, then clearly you do not understand how Intel processors work. If the OP in Korea thought so, then they did not understand either.
No, I don't have 'tests' if that is what you want. But I will explain what am I thinking.
But they didn't.
Cinebench is different from the whole picture and that's the point. Apple Silicon constantly performs worse than x86 opponents in Cinebench since the very first M1. You don't even need to bring Alder Lake to make this claim, an 4700U or 5800U already outperforms M1 on CB23 in the same (or similar?) power envelop. But does 4700U or 5800U has better power efficiency? If you do think so, then I can understand your claim and close the conversation. We simply have different definition of "power efficiency."
But if you draw more power, doesn't it increase the performance? And when 12900K was limited to 35W, it was similar to M1 Max/Pro. Am I missing something? I mean it seems Intel can easily achieve M1 Pro/Max's power per watt with alder lake which is my point if they are planning to make a mobile version.No, you don't get it. It's already drawing more power to achieve those scores.
Setting TDP with Intel processors does not limit power consumption:
Intel's Desktop TDPs No Longer Useful to Predict CPU Power Consumption - ExtremeTech
Intel's higher-end desktop CPU TDPs no longer communicate anything useful about the CPUs power consumption under load.
Setting 35W TDP for Intel processors means at base clock, they are only allowed to draw that much power, but that doesn't mean they can't draw more for turbo boost. I.e.: you're only reducing the base clock. If cooling and power supply allow for it, the chip will still draw much more than the TDP you set.
I.e.: you're not looking at 12900K at 35W. If you think you are, then clearly you do not understand how Intel processors work. If the OP in Korea thought so, then they did not understand either.
Nope, I'm not Korean so I dont know what they are saying and yes, I'm quite confused so I'll search about it.
It is not. That's what you don't seem to understand. This 12900K was NOT limited to 35W.
12th gen mobile CPU is expected to release in 2022 so they might say that in 2022......
But 12th gen used the big and small core just like ARM for the first time so I dont think that's a fair comparison. Both 1165G7 and 5800U dont have big and small cores.
But M1 Pro/Max's maximum power is 30W and when 12900K tested at 35W, the score was somewhat similar or slightly better. I'm talking about the performance per watt.
Again, this particular 12900K was NOT tested at 35W. It was drawing much more power than that.
Plain and simple. I don't know what about that is so hard to get. I even posted an article.
If Cinebench is the only test that matters, then bigger or smaller cores matters little because all cores are supposed to be fully loaded at that time.
And even in that video, the chip was drawing more than 35W. It was at 45W constantly. That's 50% more power consumed than M1 Max.
12900K tested at 35W while the clock speed was 3.0Ghz and 2.4Ghz. This is what I've been saying since the beginning so I dont understand what you are talking about. Even if 45W is the actual power consumption, it isn't that different. 10W isn't that big difference TBH.
Are you saying that I should add 44.11W + 34.82W for CPU's power consumption?
Attachments
What? Let's get something out of the way first:
1. 45W IS the actual power consumption. There is no if.
2. 45W IS 50% more power consumed than M1 Max at 30W for the CPU.
3. That means the 12900K IS consuming 50% more power in order to gain 16% higher score over M1 Max.
Do I even have to tell you which of the chips is more efficient?
We will see.
Performance per watt is a function of power consumption, and a result from design choice. Again: you don’t need an Alder Lake CPU to beat Apple in term of “CB23 power efficiency”.
I get that but doesn't change the fact that Intel is now able to come very close to M1 Pro/Max with 7nm and x86. Is it not concerning? And M1 Pro/Max actually uses 35W according to Anandtech.
If we only look at Cinebench then AMD was able to do that 1 year ago. So what’ the real problem then?
No. Because they are actually very far. M1 does NOT use 35W for the CPU. That should include memory and such as well because Anandtech is measuring overall SoC power draw in all of their tests to account for memory power consumption as well.
Excerpt from your reactions:
Again, please don't take things out of context without understanding what all of this actually means. The bottom line is that Intel is not getting any closer. They made Alder Lake more efficient than the last generation. But they are not close at all.