Question Alder Lake - Official Thread

[Hulk]

With the release of Alder Lake less than a week away and the "Lakes" thread having turned into a nightmare to navigate I thought it might be a good time to start a discussion thread solely for Alder Lake.

 

[DrMrLordX]

Those Alder Lake-N scores are a little fishy. The ST score I can believe, but shouldn't it be more around 5k for MT, assuming it is on par with old Skylake-family cores?

 

[dark zero]

Those Alder Lake-N scores are a little fishy. The ST score I can believe, but shouldn't it be more around 5k for MT, assuming it is on par with old Skylake-family cores?

Is pending to see the power draw. Seems that it migth go higher than expecting.

 

[LightningZ71]

I expect the MT scores to take a hit with hard power limits in many really low specced laptops. This may be the saving grace for mendocino. Having only 4 cores to power, it can give more power to each at the same total wattage limit. I suspect that N6 is a superior node to Intel7 when it comes to lower power operation, which should favor AMD further.

What we don't know is how the reduced floating point performance will affect things. We know that Mendocino was specifically noted by AMD to have a somewhat hobbled FPU unit (its likely shared with the console CCX designs) and this may show up in some benchmarks.

I suspect that Mendocino may gove a better user experience by some mior degree, but that Adl-n is superior on really taxing MT stuff when it has sufficient wattage limits.

 

[Exist50]

Having only 4 cores to power, it can give more power to each at the same total wattage limit

Generally, you want the opposite. For a MT load, the best efficiency is achieved by spreading out the load among more cores. For single or lightly threaded, you can always idle the unused cores.

 

[LightningZ71]

True, and we've seen both vendors do that in the past. For this situation though, I think that there's a little more to it. The gracemont cores are optimized for their single threaded existence. They have what is essentially a pair of tightly coupled CCXs with a shared L2 coupled with a shared small L3. It is my opinion that there is congestion overhead for the gracemont quads in heavy MT loads that heavily use the L2 that doesn't favor them in benchmarks and in pure throughput loads like handbrake, blender and C23. Each of the cores in Mendocino has a private L2, though the total L2 for the CCX is similar to the gracemont quad, its just not shared.

As for power, as we have seen, the V/F curve is linear to a point, then exponential or worse. If they were on the same exact process, and the Zen2 cores required so much frequency to achieve the same throughput as the 8 gracemont cores that they were well into the exponential area of the V/F curve, then mendocino would have a problem. However, they aren't on the same process, and TSMCs N6 likely has a better B/F curve than the ADL variant of Intel7. This means that mandocino can comfortably run at higher clock speeds before it has to deal with a runaway power curve. That means that it doesn't have to pay the overhead for running twice as many cores and having all the L2s lit all the time for any of the cores in ADLn to talk to the L3. Ot is my opinion that the hracemont quads are going to be somewhat wasteful in power in high MT workloads as they constantly block on L2 access and are unable to do useful work while also being unable to go to sleep as they have to be ready to go for when the L2 becomes available. With SMT, Zen2 cores can block on one thread while still crunching on another, producing more useful work over time. They are also not going to block on their own L2 access either.

I expect that, overall, mendocino will be able to lead ADLn in non floating point ST throughput and generally keep up with ADLn on MT tasks while staying in similar power limits.

 

[igor_kavinski]

Are you a human?

www.newegg.com

ECC support for your Alder Lake CPU is now a reality.

 

[Markfw]

Are you a human?

www.newegg.com

ECC support for your Alder Lake CPU is now a reality.

I don't think so, for 2 reason, what they says on 12th version and 125 watt TDP

12th Generation Intel® Core™ i3/i5/i7/i9 Processors, Single Socket LGA-1700 supported, CPU TDP supports Up to 125W TDP

 

[Abwx]

Generally, you want the opposite. For a MT load, the best efficiency is achieved by spreading out the load among more cores.

With SMT a high IPC 4C/8T is more efficient while providing better ST perf at low threads counts, even at same power/core it should do better, higher IPC get you lower frequency and hence better efficency at same ST perf.

 

[stebler]

From the X13s NBC review; the difference between ADL-Y, ADL-U and ADL-P is massive: Y is competing with ARM chips, P is losing to Broadwell. There is also a lot of variance between implementations of the same CPU.

 

[mikk]

Alder Lake-N TDP:

https://twitter.com/x/status/1592466828357701634

 

[igor_kavinski]

Alder Lake-N TDP:

https://twitter.com/x/status/1592466828357701634

BRING THESE TO MARKET ALREADY, INTEL!

 

[moinmoin]

Alder Lake-N TDP:

https://twitter.com/x/status/1592466828357701634

About time. Those are bad names though.

 

[LightningZ71]

I have to agree about the poor naming. To dare call the N300/305 a Core i3 without having a single P core is beyond laughable, it's deliberately deceptive. I don't doubt that it will be sufficient for 99% of what could be thrown at it, but, that's just bad karma...

 

[Doug S]

I imagine these won't be available in socketed form so it won't matter for DIY trying to build fanless systems, and in a prepackaged system will probably be marked up enough by the OEM that there's no point to considering them vs underclocking/undervolting a higher end socketed SKU.

 

[DrMrLordX]

I imagine these won't be available in socketed form so it won't matter for DIY trying to build fanless systems, and in a prepackaged system will probably be marked up enough by the OEM that there's no point to considering them vs underclocking/undervolting a higher end socketed SKU.

If history is any indicator, there should be cheap SFF PCs based on Alder Lake-N once Intel makes it available to OEMs:

https://www.amazon.com/Mini-PC-Processor-Computer-Ethernet/dp/B09SZ5SSS5

And there are cheaper ones than that available. That just popped up instantly on a Google search for N6005 desktop PC.

 

[hemedans]

I imagine these won't be available in socketed form so it won't matter for DIY trying to build fanless systems, and in a prepackaged system will probably be marked up enough by the OEM that there's no point to considering them vs underclocking/undervolting a higher end socketed SKU.

If history is any indicator, there should be cheap SFF PCs based on Alder Lake-N once Intel makes it available to OEMs:

https://www.amazon.com/Mini-PC-Processor-Computer-Ethernet/dp/B09SZ5SSS5

And there are cheaper ones than that available. That just popped up instantly on a Google search for N6005 desktop PC.

all pentium/celeron silver are available in single board pc, like this

ASUS N5105I-IM-A: A fanless single-board computer with an Intel Jasper Lake processor

ASUS has introduced a single-board computer with an Intel Jasper Lake processor. The N5105I-IM-A follows the Mini-ITX form factor and is fanless thanks to a large copper heatsink.

www.notebookcheck.net

 

[phillyman36]

Sorry wrong thread

 

[Hulk]

Can someone tell me the benefit of the 8 extra E cores on the 13900k vs 13700k? Are all 16 E cores used for system resources and background tasks when gaming? What apps will actually use all the cores? I use i dont think Handbrake uses more than 6 to 8 cores so at what point are they help in non benchmarking use? Is there a list of software some can tell me the 8 extra cores are utilized? Real world usage.

Handbrake will use all cores with x265 codec. When I render in Vegas Pro all cores are utilized. DxOPureRaw uses all cores. These are a few I know from personal experience. And of course whenever you are multitasking the E's are on the background task(s) so the P's are available for the foreground application.

 

[phillyman36]

Handbrake will use all cores with x265 codec. When I render in Vegas Pro all cores are utilized. DxOPureRaw uses all cores. These are a few I know from personal experience. And of course whenever you are multitasking the E's are on the background task(s) so the P's are available for the foreground application.

Thanks for the reply hulk realized I put in the wrong thread and now posted in the right one.

 

[mderbarimdiger]

Hi All

I am new here , i wonder if anyone can help me maximize the utilization of my CPT , i am using PowerBi to crunch some numbers, i am loading a 3GB file, and it takes about 3 minutes to complete loading, i wonder why i am not able to utilize my full memory and CPU power in order to speed up the load time.

I am using a ASUS zenbook, set the fan mode for performance (which is the setting for maximizing CPU) my laptop is plugged in to power, what else can i do in order to speed up the loading process ?

 

[igor_kavinski]

what else can i do in order to speed up the loading process ?

Try disabling the E-cores first. See if that improves the loading time.

Next try disabling hyperthreading to see if that has any impact.

Finally, I see that you have some RAM available. Create a RAM drive of about 3 GB and put the 3 GB file there. See if loading the file from there is quicker. Afterwards, you can turn on compression on the RAM drive and see if that also improves the loading time.

 

[mderbarimdiger]

Thanks @igor_kavinski

i am a novice user, will have to figure out how to disable E cores, hyperthreading, let me try to do that.

 

[Exist50]

Hi All

I am new here , i wonder if anyone can help me maximize the utilization of my CPT , i am using PowerBi to crunch some numbers, i am loading a 3GB file, and it takes about 3 minutes to complete loading, i wonder why i am not able to utilize my full memory and CPU power in order to speed up the load time.

I am using a ASUS zenbook, set the fan mode for performance (which is the setting for maximizing CPU) my laptop is plugged in to power, what else can i do in order to speed up the loading process ?

Especially given that you're using a laptop, I doubt you'll see any benefit from disabling cores or hyperthreading. Will probably hurt, if anything.

 

[mderbarimdiger]

@Exist50 so what is your recommendation ?

is it that a laptop will always have a challenge to harvest the full power of a CPU?

I am now in search of a new desktop, wondering what specs i have to look for in order to make use of a good CPU.

 

[igor_kavinski]

i am a novice user, will have to figure out how to disable E cores, hyperthreading, let me try to do that.

Restart your laptop. When the screen goes blank, start pressing Esc (if it's an HP laptop) or F2 (Dell) or Delete key (almost any other brand). You will enter the BIOS. From there, you can search for an option to disable E-cores or hyperthreading.