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.

 

[Timur Born]

Could be but unlikely. My system is lean. Quality components. Restarts only happen under high load in a variety of applications/benchmarks. I've been doing this (building PC's) for 30+ years. I'm pretty good at it. At 175W completely stable. 200W unstable. Last month 225W was stable.

Maybe you involuntarily set low Overcurrent protection in BIOS, which would cause restarts at high current.

 

[Hulk]

Maybe you involuntarily set low Overcurrent protection in BIOS, which would cause restarts at high current.

When it started acting up I reset bios to stock settings.

 

[coercitiv]

Fun fact, the only benchmarks in the Puget System suite that are affected by power limits are Cinebench nT and V-Ray. The rest are lightly threaded or short enough to take advantage of the PL2 window. It's even funnier to see Puget getting criticized for benchmarking with settings for normal people. They're not a review site for enthusiasts, but a shop for creators and engineers.

 

[Carfax83]

Puget Systems is not my favorite source for performance reviews. I don't think they are biased or anything, but they make questionable choices in their test systems and seem to run into 'bugs' more often than any other reviewer.

To be fair, as @coercitiv said, they don't really cater to hardware enthusiasts or PC gamers, so maybe they shouldn't be expected to know anything about hardware configuration and optimization.

 

[Timur Born]

When it started acting up I reset bios to stock settings.

To be sure I would change the auto settings to something manual, like "Standard" or even "High" or whatever they are called in your BIOS.

 

[Carfax83]

Could be the motherboard traces supplying the voltage/current to CPU are getting degraded.

You need to test your CPU in a different mobo. If it still can't run at 200W, yes, we can say CPU is likely degraded.

I'd be shocked if the CPU was degraded. CPU degradation typically takes several years unless you are an uber overclocker.

From my experience, spontaneous restarts are typically caused by power delivery issues, ie not enough voltage or power is being supplied. When I had aggressively undervolted my CPU, it ran perfectly stable at load with -165mV, but I would get spontaneous restarts from time to time while idling or doing some non strenuous activity like internet browsing.

That obviously indicated that the CPU voltage was dropping too low in idle, so reducing the negative offset fixed it.

 

[Timur Born]

I also saw idle instabilities in the past, but not with my 13900K yet. Doesn't mean that it cannot happen. Undervoltage protection on/off could make a difference. Memory instabilities can also cause restarts.

 

[lopri]

A hint for those doing per core OC on Intel CPUs with enabled E cores and any power plan/mode other than "Best Performance": You need to check your P core stability in combination with fully loaded E cores, because something like this can happen in practice: View attachment 74456

What is wrong with this? I see 2x P-cores under load and the rest are resting.

 

[Timur Born]

What is wrong with this? I see 2x P-cores under load and the rest are resting.

Nothing is wrong with this. E core load does not affect P core frequencies, but it does affect stability (power/heat). This scenario can happen when Integer background load is moved to E cores by Windows with any power PLAN or MODE other than "Best performance" (like compressing files in the background while a maximized foreground window is in focus).

 

[lopri]

Does it get unstable? I did not think the stability was an issue caused by E-cores.

 

[Timur Born]

Draws a *lot* more power, may stress the ring/L3 cache, creates heat (transfer) between adjacent cores 6+7 and E clusters 0+1. And tested unstable for P core configurations that were stable without stressing E cores concurrently. ;-)

 

[lopri]

I see. I pretty much used 13700K with E-cores disabled, so I did not notice such. Still even with just P-cores, I noticed Raptors are not as sensitive to background tasks as Raphaels. I lost 50 MHz out of boost on my 7700X since I installed.. something. (Currently investigating what it is)

 

[Timur Born]

Considerably increased heat when adjacent cores are fully loaded is quite a problem, because the differences are really big. Theoretically one could use cores 0+1 and then 4+5 and then E cores, but in practice my own CPU's best P cores are 2+3 and 6+7, the latter of which are adjacent to E cores and thus heat up considerably when E cores are loaded alongside P cores. My cores 2+3 are prioritized first because of that.

 

[Kocicak]

At 175W completely stable. 200W unstable. Last month 225W was stable.

It is not a power issue, it is voltage issue caused by power draw, more power draw means higher voltage droop.

Your CPU now FOR SOME REASON requires higher voltage to run properly.

If your CPU stopped being stable in stock configuration, use warranty and get rid of it.

Before you do that, it would be nice to know, what voltage your CPU now needs to run properly (what positive voltage offset you need to aply to stabilise it). As I wrote earlier, you could write your findings in the degradation thread. You may also write there, how intensively you used your CPU, and what you did with it overclocking and voltage wise.

 

[coercitiv]

One doesn't solve vdroop by applying a voltage offset. We have Load Line Calibration for that.

@Hulk assuming you already know the essentials about LLC (the link above) you should also watch this video on voltage regulation. Whether it gets you closer to solving your problem is hard to guess from where I'm standing, but you should know our two Z690 are similar in terms of pricing and probably voltage regulation properties as well. (my board is the one in the video, exhibiting the interesting behavior Buildzoid is talking about - your board should be mostly indifferent to switching frequency, as most boards are).

At the very least these two videos help refresh important info for anyone considering overclocking, undervolting, or maybe even operating "value" hardware at high power limits. They show the board power delivery can play a big role in establishing the minimum stable voltage for a given CPU frequency, and they also show that talking about undervolting results in terms of the relative offset value can sometimes be meaningless, it's always appropriate to mention the absolute value as well (usually Vcore).

This is why people who generally know what they're doing will specify their custom configs as something like 5Ghz @ 1.2V as opposed to 5Ghz @ -150mV offset. The offset can mean anything depending on their board loadline characteristics and voltage regulation.

Your board should handle the 13900K at stock settings, that we know already. If you don't feel like digging deeper into this, RMA your CPU and get a replacement. It's certainly easier than beggining the witch hunt with the motherboard.

 

[nicalandia]

PugetSystem Review of the 13900KS

Intel Core i9 13900KS Content Creation Review

Following the success of the initial launch of their 13th Gen processors, Intel is now launching the Core i9 13900KS which is the first CPU to be capable of hitting 6 GHz at stock settings. But will this make any real difference for content creation?

www.pugetsystems.com

The 7950X is faster



Many are not impressed by its gaming performance.

 

[igor_kavinski]

It's funny how the 13900KS runs out of breath in the 10 minute test. This is something Intel needs to work hard on.

 

[Kocicak]

One doesn't solve vdroop by applying a voltage offset. We have Load Line Calibration for that.

Nobody needs to solve vdroop. That is just a connection between power draw and voltage, which was the same in the beginning when the CPU worked and now, when it does not work anymore.

What changed is the voltage requirement of the CPU. If you found, that to stabilise it you now need e.g. + 30 mV offset and when it was new it had 80 mV reserve, you could conclude that it needs 110 mV more now to run properly.

I hope Hulk examined the negative voltage offset, with which the CPU could run when it was new.

 

[nicalandia]

Is this It? Is this Intel's counter to the X3D? $700 CPU?

Intel Core i9-13900KS Review: Hitting 6 GHz

The new Core i9-13900KS is very much like the 12900KS and 9900KS before it -- this is a mildly overclocked 'special edition' CPU, binned silicon meant to...

www.techspot.com

 

[Hulk]

One doesn't solve vdroop by applying a voltage offset. We have Load Line Calibration for that.

@Hulk assuming you already know the essentials about LLC (the link above) you should also watch this video on voltage regulation. Whether it gets you closer to solving your problem is hard to guess from where I'm standing, but you should know our two Z690 are similar in terms of pricing and probably voltage regulation properties as well. (my board is the one in the video, exhibiting the interesting behavior Buildzoid is talking about - your board should be mostly indifferent to switching frequency, as most boards are).

At the very least these two videos help refresh important info for anyone considering overclocking, undervolting, or maybe even operating "value" hardware at high power limits. They show the board power delivery can play a big role in establishing the minimum stable voltage for a given CPU frequency, and they also show that talking about undervolting results in terms of the relative offset value can sometimes be meaningless, it's always appropriate to mention the absolute value as well (usually Vcore).

This is why people who generally know what they're doing will specify their custom configs as something like 5Ghz @ 1.2V as opposed to 5Ghz @ -150mV offset. The offset can mean anything depending on their board loadline characteristics and voltage regulation.

Your board should handle the 13900K at stock settings, that we know already. If you don't feel like digging deeper into this, RMA your CPU and get a replacement. It's certainly easier than beggining the witch hunt with the motherboard.

Thanks for taking the time to post this and help me out. I will study up and report back.

Very informative video. I wish I knew what "Auto" actually was when it comes to LLC and other BIOS "Auto" settings.

 

[igor_kavinski]

OK, what is going on? How come Puget Systems had 13900KS in 2nd place in 10 minute test but Techspot show it in first place?

 

[Hitman928]

OK, what is going on? How come Puget Systems had 13900KS in 2nd place in 10 minute test but Techspot show it in first place?

Puget uses an air cooler that can’t handle higher power draw CPUs.

 

[nicalandia]

OK, what is going on? How come Puget Systems had 13900KS in 2nd place in 10 minute test but Techspot show it in first place?

Air Cooler vs Water Cooler. Due to how efficient AMD CPUs they don't lose much performance when cooled by air coolers.

This speaks High on the performance we can expect from either brand in Laptops where no water cooling is offered.

 

[Hitman928]

Air Cooler vs Water Cooler. Due to how efficient AMD CPUs they don't lose much performance when cooled by air coolers.

This speaks High on the performance we can expect from either brand in Laptops where no water cooling is offered.

They also turn off all boosting on the AMD platform but not the Intel platform which is why AMD has weirdly low performance in their tests but no real difference in short runs versus long runs.

 

[Hulk]

Okay, going down the rabbit hole.

Changed LLC from "Auto" to "3." Set power to 225W. As soon as I hit start for CB it restarts.
Now I change CPU Current Capability from "Auto" to "120%." I haven't run a long-term test but CB seems to run fine.

What's going on? If I set LLC back to "Auto" and it runs okay at 120% then suddenly the board can't supply the power it used to be able to or the CPU is drawing more than it used to?

Not sure exactly what "CPU Current Capability" is exactly and why you would want to limit it.

7 minutes into a CB R23 run. Max temp 80C. VID reading 1.2 in HWinfo. Frequencies of P's 5.1 to 5.2GHz, E between 4 and 4.1.