Discussion Intel current and future Lakes & Rapids thread

[coercitiv]

Alder Lake peak current allegedly 28% higher than Rocket Lake

Just wait till we see the VRM stages for ADL-S. We know 10SF dropped voltage at ISO clocks by a nice margin, and Intel 7 is bound for another improvement as well. So power goes up (or stays the same), voltage goes down... VRM current goes UP and motherboard pricing goes BRRRR... No wonder OEMs are rather reluctant to adopt 12VO in the same time-frame.

I expect ADL-S mITX boards to be real wonders of modern engineering.

 

[mikk]

Alder Lake peak current allegedly 28% higher than Rocket Lake

Alder Lake Allegedly Commands 28% Higher Peak Current Than Rocket Lake

Cranking up the juice

www.tomshardware.com

If true, that's getting beyond ridiculous.

I'm curious how relevant is a PL4 peak? I know that PL1 is sustained (125W TDP) and PL2 (228W) is the short time boost which is sometimes a sustained setting for the higher end desktop mainboards. But PL4 for what?

 

[jpiniero]

I'm curious how relevant is a PL4 peak? I know that PL1 is sustained (125W TDP) and PL2 (228W) is the short time boost which is sometimes a sustained setting for the higher end desktop mainboards. But PL4 for what?

PSU might shut down? Sounds like you might need a new PSU to be able to handle those peaks.

 

[mikk]

PSU might shut down? Sounds like you might need a new PSU to be able to handle those peaks.

I mean from a performance standpoint. Intels PL4 turbo is extremely short (10ms for CML-S afaik), so it's completely useless in a benchmark and that's why it's all about PL1 and PL2 in reviews. Even the PL2 is very high (228W for ADL-S), isn't this enough for maximum turbo speeds on 1-2 cores? I don't see the PL4 importance.

 

[eek2121]

Alder Lake peak current allegedly 28% higher than Rocket Lake

Alder Lake Allegedly Commands 28% Higher Peak Current Than Rocket Lake

Cranking up the juice

www.tomshardware.com

If true, that's getting beyond ridiculous.

While 11600K is fine, 11700K is already pushing it for my tastes for all-core loads (non gaming) and 11900K is already rather ridiculous. Going beyond that isn't something I'd want, considering how crazy the GPUs already are at peaks-power. Now we have:


Yeah, I know it's peak power only for 10ms or so, but even "65W" CPUs will just not work with many older PSUs (even if rated for quite high watts) as over-current protection will just kick in at 30A.

So coupled with current 300W+ TBP custom GPUs you'll need a quality gold+ 1000W PSU or better for a new gaming rig. And be ready for the CPU to draw 250W, sustained, at times.

PSU might shut down? Sounds like you might need a new PSU to be able to handle those peaks.

You might need a new PSU anyway. Intel is pushing 12VO.

 

[RanFodar]

Alder Lake peak current allegedly 28% higher than Rocket Lake

Alder Lake Allegedly Commands 28% Higher Peak Current Than Rocket Lake

Cranking up the juice

www.tomshardware.com

If true, that's getting beyond ridiculous.

While it is surprising that peak currents would go up that big, @coercitiv said it correctly. With each node shrink, it is expected that it would come with less voltage at the expense of more drive current (Amps) for the transistors. Continuous current is maintained (and lowered on 35W) than 10th/11th gen so I don't think *anyone* should worry about their PSUs if you don't use your ADL processor in full load (at all times). An example of this is the RDNA GPUs, where they tend to produce higher spikes than Ampere, despite being usually more efficient.

Also, concluding that your PSU needs additional 50-100W is a bit inaccurate. You should buy one that has high grade electronics and can easily withstand spikes briefly such as the peak current.

I suspect the recommendations are for system integrators/OEMs who might need a stronger PSU. Also, are there any documents relating to the power current of Tiger Lake CPUs? Interested on what their Amps are in a 10nm SF process.

 

[JujuFish]

You might need a new PSU anyway. Intel is pushing 12VO.

While we're on the topic of new PSUs, who does good PSU reviews these days? I used to read jonnyGURU's reviews, but he can't even openly talk about PSUs anymore.

 

[coercitiv]

While we're on the topic of new PSUs, who does good PSU reviews these days? I used to read jonnyGURU's reviews, but he can't even openly talk about PSUs anymore.

Forum has a PSU section, I'm sure people active in that area will be more than happy to guide you.

 

[Bouowmx]

Does LGA 1700 have FIVR?

 

[JujuFish]

Forum has a PSU section, I'm sure people active in that area will be more than happy to guide you.

That literally does not answer my question.

Edit: I think I misunderstood your post. I thought you were saying the PSU section has reviews. You're just pointing out how my post was off topic. My bad.

 

[eek2121]

NUC 12 is interesting, if true: https://videocardz.com/newz/intel-nuc-12-enthusiast-to-feature-alder-lake-and-dg2-gpus-with-displayport-2-0-support

 

[mikk]

NUC 12 is interesting, if true: https://videocardz.com/newz/intel-nuc-12-enthusiast-to-feature-alder-lake-and-dg2-gpus-with-displayport-2-0-support

The interesting part is DP2.0 because there is no card from AMD or Nvidia at the moment with DP2.0, so it might come with DG2 wich should launch early 2022.

 

[eek2121]

The interesting part is DP2.0 because there is no card from AMD or Nvidia at the moment with DP2.0, so it might come with DG2 wich should launch early 2022.

Yes, it will come with DG2 as well as Alder Lake. It also looks to be smaller than the NUC11.

 

[Magic Carpet]

Does LGA 1700 have FIVR?

Good question, I am also interested. Tiger Lake does.

 

[jpiniero]

I mean from a performance standpoint. Intels PL4 turbo is extremely short (10ms for CML-S afaik), so it's completely useless in a benchmark and that's why it's all about PL1 and PL2 in reviews. Even the PL2 is very high (228W for ADL-S), isn't this enough for maximum turbo speeds on 1-2 cores? I don't see the PL4 importance.

If the PSU shuts down... that's kind of a problem. Could be an edge case that they found and increasing the max amp spec was their solution.

The amps issue would be at all core loads I would think.

 

[uzzi38]

I'm curious how relevant is a PL4 peak? I know that PL1 is sustained (125W TDP) and PL2 (228W) is the short time boost which is sometimes a sustained setting for the higher end desktop mainboards. But PL4 for what?

Latest LTT video: due to SFF PSUs having lower capacitance than regular desktop PSUs, they've killed 3 PSUs due to Ampere's transient peaks.

It can make a difference. Thankfully desktop CPUs are not even close to the kind of transient peaks we see on GPUs (450W+) but still, it can push a much cheaper PSU a bit too far.

 

[Magic Carpet]

Alder Lake-S Peak 45A, or we need more CPU performance.

Disappointed, this reminds me of Pentium 4 EE. They are not targeting performance per watt at all. Looks like, I will be moving to AMD later this year / early next after all. Or sooner, if I could snag 5900 OEM 65W for a good price.

 

[eek2121]

Samsung DDR5-4800: https://memory.net/store/?filter_technology=ddr5-4800


Looks like DDR5 is slowly beginning to trickle out. I expect when Alder Lake launches we will see much better availability.

 

[Magic Carpet]

@eek2121

It will be interesting to see how it scales and handles the increased power consumption of the enthusiast class modules.

Leading suppliers of enthusiast-grade memory modules, such as Corsair, have already announced their upcoming DDR5 SDRAM sticks that feature rather unprecedented data transfer rates that start at 6400 MT/s and go all the way up to 12600 MT/s. To a large degree, such extreme speeds will be enabled by on-board power management ICs (PMIC) and voltage regulating modules (VRM), which will require proper cooling, Corsair said this week.

"DDR5 conceivably could run much hotter than DDR4 [as] they have moved voltage regulation off the motherboard itself and now it is on the [module], so you actually could be pumping a lot more heat," George Makris, DIY marketing director at Corsair, said in a YouTube video from the company.

For years, many considered heat spreaders unnecessary for most DDR3 and DDR4 modules, even those intended for enthusiast-grade PCs. With DDR5, things may change as modules will get more complex. Corsair says that its upcoming cooling systems for DRAM sticks will do the job.

Source.

 

[mikk]

Disappointed, this reminds me of Pentium 4 EE. They are not targeting performance per watt at all. Looks like, I will be moving to AMD later this year / early next after all. Or sooner, if I could snag 5900 OEM 65W for a good price.

PL4 has nothing to do with performance per watt, it's way too short (10ms). Only PL1 and PL2 matter for perf/watt. There are rumours Intel target a 2x perf/watt improvement.

 

[Magic Carpet]

PL4 has nothing to do with performance per watt, it's way too short (10ms). Only PL1 and PL2 matter for perf/watt. There are rumours Intel target a 2x perf/watt improvement.

Well yeah, I meant the PL1/2 really. The power envelope is still huge, nonetheless. If it can best Zen 3's power efficiency in MT loads, then I will take my words back. Because, to the outsider such as myself, it looks like Intel has entered the clock/power race again just to edge AMD in most relevant benchmarks. Of course, there will be SKUs like 11400 with its exceptional power efficiency, but I am talking about the top parts here.

 

[jpiniero]

If it can best Zen 3's power efficiency in MT loads, then I will take my words back.

That's not gonna happen. The per core/gaming performance might be very good but part of that is being able to juice the frequencies way high.

 

[ondma]

That's not gonna happen. The per core/gaming performance might be very good but part of that is being able to juice the frequencies way high.

If the performance is there, you can manage the power usage on the desktop. Problem is, this seems like a mobile designed architecture, and high power usage is just the opposite of what you want there.

 

[jpiniero]

If the performance is there, you can manage the power usage on the desktop. Problem is, this seems like a mobile designed architecture, and high power usage is just the opposite of what you want there.

The small cores will get a lot more usage on mobile.

 

[dmens]

There are rumours Intel target a 2x perf/watt improvement.

LOL. I have a bridge to sell you. Prime location.