Intel Core i9-12900K

[lexluthermiester]

That's misrepresenting my argument. I challenged you to check properly before accusing W1z of a bad review, I didn't outright say that you were wrong, hence no bet to lose.

I've been watching this exchange, fevgatos seems to be baiting you. They're being a bit subtle about it, but it's definitely some baiting. If they want to be ignorant and fail to see what is staring them in the face let them, but don't let them rope you into to their sad nonsense.

 

[fevgatos]

I've been watching this exchange, fevgatos seems to be baiting you. They're being a bit subtle about it, but it's definitely some baiting. If they want to be ignorant and fail to see what is staring them in the face let them, but don't let them rope you into to their sad nonsense.

Posting some actual numbers that seem to contradict the review is baiting nowadays? Okay

 

[qubit]

I've been watching this exchange, fevgatos seems to be baiting you. They're being a bit subtle about it, but it's definitely some baiting. If they want to be ignorant and fail to see what is staring them in the face let them, but don't let them rope you into to their sad nonsense.

I think you're right, Lex.

Anyway, I've set him straight on it now and he's not come back to me even though he's seen my post, so I call that a result lol.

 

[lexluthermiester]

so I call that a result lol.

Right? Very telling..

 

[fevgatos]

I think you're right, Lex.

Anyway, I've set him straight on it now and he's not come back to me even though he's seen my post, so I call that a result lol.

Whats there to come back to? You said we should wait for wiz to explain, what more do you want me to reply to that? LOL

 

[qubit]

Oh dear.

 

[lexluthermiester]

You said we should wait for wiz to explain

He's not going to bother as he is not in the wrong. The problem is your understanding of facts and the context thereof. It's as simple as that. We're done here.

 

[R0H1T]

fev's been baiting almost everyone here! Let's look at the blender workloads ~

See anything different aside from the different chips themselves

Ironically the Igor's lab review also shows that if the workload lasts longer 5950x will be better!

 

[fevgatos]

fev's been baiting almost everyone here! Let's look at the blender workloads ~

See anything different aside from the different chips themselves

Ironically the Igor's lab review also shows that if the workload lasts longer 5950x will be better!

Different chips cause one 12900 to outperform the 12700k while the other one barely wins the 12600k? LOL

He's not going to bother as he is not in the wrong. The problem is your understanding of facts and the context thereof. It's as simple as that. We're done here.

What facts are there to understand. My tests show completely different results. Igors test show completely different results. What facts are there that i don't understand? Can you mention any?

 

[R0H1T]

Let's try again, this time slowly ~

Blender is one of the few professional-grade rendering programs out there that is both free and open source. That fact alone helped build a strong community around the software, making it a highly popular benchmark program due to its ease of use as well. For our testing, we're using the Blender "BMW 27" benchmark scene with Blender v2.92.

Intel Core i9-12900K Alder Lake Tested at Power Limits between 50 W and 241 W

We test Intel's Core i9-12900K at various TDP levels all the way down to 50 W to determine how much efficiency is really in the new Alder Lake core, and how these power limits affect performance. Competing with the efficiency of AMD's Zen 3 Ryzen lineup is just two settings changes away.

www.techpowerup.com

This is the one Igor's lab used ~

 

[fevgatos]

Let's try again, this time slowly ~

Intel Core i9-12900K Alder Lake Tested at Power Limits between 50 W and 241 W

We test Intel's Core i9-12900K at various TDP levels all the way down to 50 W to determine how much efficiency is really in the new Alder Lake core, and how these power limits affect performance. Competing with the efficiency of AMD's Zen 3 Ryzen lineup is just two settings changes away.

www.techpowerup.com

Dude, do you realize that the TPU's result show the 12600k BEATING or being equal to the 12900k at the same wattage in the MT workloads? You do realize that cannot happen, RIGHT? I really have no idea why you keep insisting when you are obviously wrong.

 

[R0H1T]

Same wattage, where? Just for the blender workload the 12600k is beating the i9 with PL1/2 limited to 100W, you do realize that chip isn't going to run full throttle at those limits right? I'm assuming 12600k isn't power limited. The issue is likely because of big little(cores) & maybe even thread director!

Oh look the guys at AT also making Intel look bad

 

[fevgatos]

Same wattage, where? Just for the blender workload the 12600k is beating the i9 with PL1/2 limited to 100W, you do realize that chip isn't going to run full throttle at those limits right? I'm assuming 12600k isn't power limited.

Since he measured power consumption in cinebench, it is easier if you just compare cinebench. So the 12900k at 125w draws more power than the 12600k and basically performs identically. Which just CANNOT be the case. I know cause, besides everything else, I tested it in 2 different motherboard, the 12900k gets over 23500 at 125watts

 

[R0H1T]

You don't know much about PC then, depending on how the OS scheduler is assigning threads to the cores a task on the 20c/40t big little chip can take 2x as long as a regular 10c/20t big core chip. This is also one of the reasons why the initial Dozer chips were super bad, worse than they actually performed after the Windows updates!

 

[fevgatos]

You don't know much about PC then, depending on how the OS scheduler is assigning threads to the cores a task on the 20c/40t big little chip can take 2x as long as a regular 10c/20t big core chip. This is also one of the reasons why the initial Dozer chips were super bad, worse than they actually performed after the Windows updates!

No you don't know much about PC then, the biggest chip won't be consuming more power while performing less or equal in MT workloads. Besides igor's lab, we are 3 people that tested the 12900k in 4 different motherboards, all with the same results. 18500 to 19200 at 75W in CBR23 and 23500 to 24500 at 125w. And im talking about DEFAULT settings, no undervolting or anything of the likes.

Also, there is this review, again from TPU

Intel Core i9-12900K E-Cores Only Performance Review

With Alder Lake, Intel is betting big on hybrid CPU core configurations. The Core i9-12900K has eight P(erformance) cores and eight E(fficient) cores. We were curious and tested the processor running the E-Cores only to see how well they perform against architectures like Zen 2, Zen 3, Skylake...

www.techpowerup.com

Here it shows that the Ecores on their own outperform the 8+8 configuration at the same wattage! It also shows the 8p cores with ht OFF also outperform the 8+8 configuration at the same wattage. That obviously cannot be the case. It's blatantly false.

 

[Dr. Dro]

No you don't know much about PC then, the biggest chip won't be consuming more power while performing less or equal in MT workloads. Besides igor's lab, we are 3 people that tested the 12900k in 4 different motherboards, all with the same results. 18500 to 19200 at 75W in CBR23 and 23500 to 24500 at 125w. And im talking about DEFAULT settings, no undervolting or anything of the likes.

Also, there is this review, again from TPU

Intel Core i9-12900K E-Cores Only Performance Review

With Alder Lake, Intel is betting big on hybrid CPU core configurations. The Core i9-12900K has eight P(erformance) cores and eight E(fficient) cores. We were curious and tested the processor running the E-Cores only to see how well they perform against architectures like Zen 2, Zen 3, Skylake...

www.techpowerup.com

Here it shows that the Ecores on their own outperform the 8+8 configuration at the same wattage! It also shows the 8p cores with ht OFF also outperform the 8+8 configuration at the same wattage. That obviously cannot be the case. It's blatantly false.

It should not be surprising that a processor with fewer cores and threads would behave better under a strict power budget. This is why i've also never believed in traditional mobile processors with 8 or more cores, I was actually even a bit distrustful of the 5600H when I bought my laptop - turns out 6 Zen 3 cores is about fine for a 45W processor without sacrificing frequency, if it had a lower design power such as, say, 25W, it would be better for most workloads you'd do on a laptop that it had no more than four cores. I hate throttling that much!

In the review you linked, the E-cores show the precise reason they exist - high power efficiency, with the P-cores all but disabled and drawing no power, they'll show exactly the aforementioned behavior, they're going to run at their optimal frequencies, throttle free, and produce the best possible benchmark scores. In case you didn't figure out by now, this is what the i9-12900KS does to achieve its high performance, they just throw power limits right off the window and say screw efficiency, letting it chug as much juice as it can to put out performance indiscriminately. That's also why you found no powerlimited benchmarks for the 12900KS, it makes that SKU pointless, it's binned for clocks and clocks alone.

 

[fevgatos]

It should not be surprising that a processor with fewer cores and threads would behave better under a strict power budget. This is why i've also never believed in traditional mobile processors with 8 or more cores, I was actually even a bit distrustful of the 5600H when I bought my laptop - turns out 6 Zen 3 cores is about fine for a 45W processor without sacrificing frequency, if it had a lower design power such as, say, 25W, it would be better for most workloads you'd do on a laptop that it had no more than four cores. I hate throttling that much!

In the review you linked, the E-cores show the precise reason they exist - high power efficiency, with the P-cores all but disabled and drawing no power, they'll show exactly the aforementioned behavior, they're going to run at their optimal frequencies, throttle free, and produce the best possible benchmark scores. In case you didn't figure out by now, this is what the i9-12900KS does to achieve its high performance, they just throw power limits right off the window and say screw efficiency, letting it chug as much juice as it can to put out performance indiscriminately. That's also why you found no powerlimited benchmarks for the 12900KS, it makes that SKU pointless, it's binned for clocks and clocks alone.

Actually, that is definitely not the case. More cores perform better at the same wattage in MT workloads.

E cores are NOT more efficient than P cores actually. I dont know why people say that, it is not the case. Testing shows they are less efficient than P cores. Even intel admits so in their own slides

 

[Dr. Dro]

Actually, that is definitely not the case. More cores perform better at the same wattage in MT workloads.

E cores are NOT more efficient than P cores actually. I dont know why people say that, it is not the case. Testing shows they are less efficient than P cores. Even intel admits so in their own slides

No, not always. More execution units at lower frequencies do not always beat less execution units at higher frequencies, that depends on the nature of the workload and also on the instructions per clock rate of the microarchitecture. Over time, this can grow to a disparity that can become quite extreme, as i'm going to show you:

For example, my 18-core, 36-thread Xeon E5-4669 v3 running at 2.4 GHz (and this is on DDR4-2133 CAS 9 memory, about as tweaked as you can do!) scores around 12100 points in Cinebench R23:

You will find this score trivial to match with an affordable consumer-grade 6-core processor today:

The reason is obvious, too, just compare my meager 550 points that are so bad that it's off scale even against low-budget modern CPUs:

End of the day, I don't believe w1zz's testing is incorrect, it simply reflects the experience early adopters had at the time, kinks and all. It doesn't make your processor less valuable or less enjoyable to use, nor his testing methodology flawed. So why be bothered by it? Just enjoy your i9, man. Be thankful that you could afford one, a lot of people want one and can't have them

 

[lexluthermiester]

E cores are NOT more efficient than P cores actually.

See, THAT is an example of baiting. You make a totally preposterous and incorrect statement just to get a reaction out of people. Knock it off.

 

[fevgatos]

See, THAT is an example of baiting. You make a totally preposterous and incorrect statement just to get a reaction out of people. Knock it off.

What do you mean? E cores ARE way less efficient than the P cores in most of their performance / watt curve. They only get more efficient when you drop them down to less than 3-4 watts per core, which is not a thing you do on a desktop CPU.

Do you think that Intel themselves are baiting? According to their very own graphs, they claim P cores ARE indeed more efficient than E cores. Which turns out to be true, and you would already know that if you had the CPU.

Here is the graph from Intel

And here are some graphs from a review that actually tested this exact thing, efficiency of P vs E cores. Pay attention to the graph, E cores are more efficient when they are running like below 2ghz - while the P cores are at like 1ghz. In most other scenarios they lose in terms of efficiency

So, you are basically saying taht Intel is baiting, and they are making a totally preposterous and incorrect statement. About their own CPU. Do you mind if I throw the whole sentence back at you, it seems like you are the one baiting

 

[95Viper]

Stop the arguing.
You made your points... now move on!