Question Raptor Lake - Official Thread

[Hulk]

Since we already have the first Raptor Lake leak I'm thinking it should have it's own thread.
What do we know so far?
From Anandtech's Intel Process Roadmap articles from July:

Built on Intel 7 with upgraded FinFET
10-15% PPW (performance-per-watt)
Last non-tiled consumer CPU as Meteor Lake will be tiled

I'm guessing this will be a minor update to ADL with just a few microarchitecture changes to the cores. The larger change will be the new process refinement allowing 8+16 at the top of the stack.

Will it work with current z690 motherboards? If yes then that could be a major selling point for people to move to ADL rather than wait.

 

[Just Benching]

It's no surprise Alder Lake wins in short bursty workloads, it has ~11% higher (avg) IPC and ~6% higher boost clock. But if you are doing anything that will stress the full chip the power goes through the roof to achieve a slightly better (overall) score vs a chip from 2020. So whoever needs application specific performance they will choose AL over Zen 3, or vice versa depending on the workload. But performance efficiency is not a stronghold of AL if you are utilizing all the threads and are not doing extreme tweaking (undervolting).

Or you just limit it to 125 or 150w and don' t have to deal with 240w in mt workloads? I mean realistically, the only people that run the 12900k at 240w are the ones that only care about performance, and those who want to complain that it's not efficient. Who the hell expects a CPU that is drawing 240w to be efficient? There is no cpu that's going to be efficient at that wattage. Heck, even zen the 7950x will lose to the 5950x in efficiency at stock cause of the increased power limit.

 

[Abwx]

. There is a graph somewehere where it shows that the 12900k consumed the least amount of power by the end of the full 146 benchmark run. So it was both the fastest and it consumed the least

That s just impossible unless the CPUs and plateform settings are wrongly done for the 5950X.

Computerbase measure single thread full system power at 105W and 101W for the 5950X and 12900K respectively.

On MT it s 199W for the 5950X and 342W for the 12900K.

No way a 12900K could be more efficient in mixed usage.

Intel Core i9-12900K, i7-12700K & i5-12600K im Test: Leistungsaufnahme, Temperaturen und OC

Intel Alder Lake im Test: Leistungsaufnahme, Temperaturen und OC / Leistungsaufnahme in Spielen / Leistungsaufnahme in Anwendungen

www.computerbase.de

 

[Just Benching]

That s just impossible unless the CPUs and plateform settings are wrongly done for the 5950X.

Computerbase measure single thread full system power at 105W and 101W for the 5950X and 12900K respectively.

On MT it s 199W for the 5950X and 342W for the 12900K.

No way a 12900K could be more efficient in mixed usage.

Intel Core i9-12900K, i7-12700K & i5-12600K im Test: Leistungsaufnahme, Temperaturen und OC

Intel Alder Lake im Test: Leistungsaufnahme, Temperaturen und OC / Leistungsaufnahme in Spielen / Leistungsaufnahme in Anwendungen

www.computerbase.de

First of all single thread isn't indicative of a mixed workload. A mixed workloads is something like an autocad which loads 3-5-6-10 cores and heavily uses cache.

Also, according to the numbers you just quoted,the 12900k consumes less while performing up to 25% faster, so how is it a surprise?

 

[inf64]

First of all single thread isn't indicative of a mixed workload. A mixed workloads is something like an autocad which loads 3-5-6-10 cores and heavily uses cache.

Also, according to the numbers you just quoted,the 12900k consumes less while performing up to 25% faster, so how is it a surprise?

It consumes less in low threaded workloads, which is no surprise due to higher performance in low thread scenarios. Overall it's barely faster versus a 2020 part and has atrocious power draw figures under full load. Here is another review from techpowerup, they used a wide variety of desktop application workloads and 12900K is slower even when paired with DDR5 6000: https://www.techpowerup.com/review/ddr5-memory-performance-scaling/2.html

 

[Just Benching]

It consumes less in low threaded workloads, which is no surprise due to higher performance in low thread scenarios. Overall it's barely faster versus a 2020 part and has atrocious power draw figures under full load. Here is another review from techpowerup, they used a wide variety of desktop application workloads and 12900K is slower even when paired with DDR5 6000: https://www.techpowerup.com/review/ddr5-memory-performance-scaling/2.html

The power draw figures are irrelevant, cause you can limit them while not sacrificing much in performance. The 7950x is now boosting to 240w as well, do you consider it atrocious? Like, who cares, you can power limit them to whatever you want, if they are faster than their replacements at same wattage - then that's what matters.

 

[Abwx]

It consumes less in low threaded workloads, which is no surprise due to higher performance in low thread scenarios.

It doesnt, full system use 105W in ST for the 5950X and 101W for the 12900K, if you add a second thread on a second core the CPU power will rise by 20W for the 5950X and 30W for the 12900K.

At 5.2GHz a GC core consume about 30W while a 5950X core use about 20W@5GHz.

 

[inf64]

The power draw figures are irrelevant, cause you can limit them while not sacrificing much in performance. The 7950x is now boosting to 240w as well, do you consider it atrocious? Like, who cares, you can power limit them to whatever you want, if they are faster than their replacements at same wattage - then that's what matters.

We don't know if Zen 4 will consume 240W (that's above the spec btw, the limit is 230W PPT). Zen 4 will have superior performance and consume less as it will race to sleep faster (which 12900K failed to do in MT tests vs 5950X).

 

[Just Benching]

We don't know if Zen 4 will consume 240W (that's above the spec btw, the limit is 230W PPT). Zen 4 will have superior performance and consume less as it will race to sleep faster (which 12900K failed to do in MT tests vs 5950X).

Ok,, 230w, do you consider it atrocious is the question? It will most likely lose to the 5950x in efficiency at stock

 

[Just Benching]

It doesnt, full system use 105W in ST for the 5950X and 101W for the 12900K, if you add a second thread on a second core the CPU power will rise by 20W for the 5950X and 30W for the 12900K.

At 5.2GHz a GC core consume about 30W while a 5950X core use about 20W@5GHz.

And that's why most comparisons are flawed. Youi need to normalize either for wattage or clockspeeds.There is no conclusion that can be reached from the numbers you just quoted. With both cpus at same clockspeeds alderlake will finish first a ST task and probably consume way less power while doing so

 

[Det0x]

@Abwx and @inf64

Why do you guys bother arguing with an account made today.. ?

 

[coercitiv]

There is no best bin for the 12900k, they all have the same VF curve. Intel cpus are quite different to amd ones, they don't have boosting algorithms.

Welcome to the forums! Not only do Alder Lake CPUs NOT have the same VF curve, that VF curve is different from one core to another within the same CPU.

Intel cpus are quite different to amd ones, they don't have boosting algorithms.

• Intel Turbo Boost 2.0 Technology
• Intel Turbo Boost Max 3.0 Technology
• Intel Adaptive Boost Technology
• Intel Thermal Velocity Boost Technology

So they do have boosting algorithms. Welcome to the forums!

 

[Just Benching]

Welcome to the forums! Not only do Alder Lake CPUs NOT have the same VF curve, that VF curve is different from one core to another within the same CPU.

View attachment 67094

• Intel Turbo Boost 2.0 Technology
• Intel Turbo Boost Max 3.0 Technology
• Intel Adaptive Boost Technology
• Intel Thermal Velocity Boost Technology

So they do have boosting algorithms. Welcome to the forums!

Nope, alderlake boosts to the same clockspeeds no matter what cooler you put at it, assuming it doesn't throttle. TVB is disabled by default. AMD on the other hand keeps boosting depending on bin and temperatures. A terrible 12900k bin gets the exact same clockspeed as the best bin, which is not the case with zen 3. The difference between VF curve on individual CPUS is irrelevant when we are talking about all core workloads, since the CPU always uses the lowest binned core for it's voltage.

I've tested 3 12900k's, they all get the exact same score with just random run variance.

 

[nicalandia]

Found this on the web..

 

[maddie]

Nope, alderlake boosts to the same clockspeeds no matter what cooler you put at it, assuming it doesn't throttle. TVB is disabled by default. AMD on the other hand keeps boosting depending on bin and temperatures. A terrible 12900k bin gets the exact same clockspeed as the best bin, which is not the case with zen 3. The difference between VF curve on individual CPUS is irrelevant when we are talking about all core workloads, since the CPU always uses the lowest binned core for it's voltage.

I've tested 3 12900k's, they all get the exact same score with just random run variance.

I have definitely jumped to an alternate reality.

 

[nicalandia]

Intel i9 1300/K/KF/T Line UP..

 

[maddie]

If the i5-13400 has the same CPIID & stepping as the higher end models, can we say its not an ADL reuse?

 

[mikk]

If the i5-13400 has the same CPIID & stepping as the higher end models, can we say its not an ADL reuse?

In this table it's clearly Raptor Lake unless the stepping info is wrong. Maybe it's a mixed use of Raptor Lake and Alder Lake chips. The 12400 was made from two chips as well, 6+0 and 8+8.

 

[Asterox]

If the i5-13400 has the same CPIID & stepping as the higher end models, can we say its not an ADL reuse?

It has the same all 6 Cores/P cores turbo as i5 12400 or 4.1ghz.Classic i5 series, if you remove K label all 6 cores/P Cores turbo drops very hard from 5.1ghz to 4.1ghz.

 

[coercitiv]

It has the same all 6 Cores/P cores turbo as i5 12400 or 4.1ghz.Classic, if you remove K label all 6 cores/P Cores turbo drops very hard from 5.1ghz to 4.1ghz.

What does that have to do with what @maddie was pointing out?

 

[eek2121]

I figured the E-cores wouldn’t improve much in terms of clocks. lIRC Wccftech claimed 4.7ghz, which is ridiculous.

IMO this reminds me of rocket lake. Intel is pushing these chips to their limits in order to try to be competitive with AMD. Very sad to see. Hopefully Meteor Lake improves the power consumption.

 

[Just Benching]

I figured the E-cores wouldn’t improve much in terms of clocks. lIRC Wccftech claimed 4.7ghz, which is ridiculous.

IMO this reminds me of rocket lake. Intel is pushing these chips to their limits in order to try to be competitive with AMD. Very sad to see. Hopefully Meteor Lake improves the power consumption.

Wait what? Amd is the one that basically almost doubles the power consumption of their CPU's, intel kept it steady. So what the heck are you talking about man?

The R5 went from 65 to 105w TDP to be able to compete with intel's i5, yet it will end up losing to last gen's i5, and not by a small margin. This years13600k is not even in the same ballpark.

So wtf, what am I missing?

 

[nicalandia]

Wait what? Amd is the one that basically almost doubles the power consumption of their CPU's.

So wtf, what am I missing?

What on Earth are you rambling about?

 

[Just Benching]

What on Earth are you rambling about?

That amd increased the power consumption of their cpu's, not intel. According to the leaks, the 7600x is pushed to 130w consumption yet it still loses to the old 12600k, and apparently its nowhere near competitive to the 13600k

 

[Zucker2k]

Lest we all forget, on embarrassingly parallel workloads, a 12900k will boost to 241w for 56 seconds only, after which it'll drop to 125w for the remainder of the run. A 5950x on the other hand, will use 142w throughout the run as long as the chip doesn't run into a thermal wall. Some overly bias sites like computerbase, with the prodding of the AMD fan base, has resorted to running rendering files or loops to extend benching time past 56 seconds and pat themselves on the back afterwards. Meanwhile, they capture power consumption in the first 56 seconds of the run (during average peak power consumption) and say 'look the 12900k is consuming 241w.'

This is why such review sites and their numbers are artificially low relative to their power consumption numbers when compared to other sites. So, unless the site is showing a graph of power consumption during a run, it should be taken with a grain of salt. Speaking of which, has AMD fixed the "no monitoring" while the computer is running full tilt directive it sent out when Zen 3 launched? I'm an advocate for giving all platforms equal treatment.
Back to power consumption: Those reviews that put out throttled results should also be given the pinch of salt treatment. If you can't be bothered to carry out a review with adequate cooling, what business do you have reviewing the chip? You can't present 'unlimited power' results and then say the chip throttled. If the chip throttled then it clearly reverted back to mitigating (default) settings.

TLDR: If you set a 12900k to 241w, TAU unlimited, and adequate cooling, it'll win most multithreaded benchmarks against a stock-boosting 5950x. Let's not act like that's what's happening in all these reviews. It's certainly not the case for computerbase.

 

[nicalandia]

That amd increased the power consumption of their cpu's, not intel. According to the leaks, the 7600x is pushed to 130w consumption yet it still loses to the old 12600k, and apparently its nowhere near competitive to the 13600k

Have you gone mad?

AMD has the lowly 7600X beating the 12900K in games, matching or exceeding the 5800X3D. The 12600K has Nothing on the 7600X.