News Detailed specs for dozen Intel Arrow Lake desktop CPUs leaks ahead of touted October 10 launch date

[DS426]

TDP's are back to where they should be, so that's a good sign. Of course, grain of salt here and not all SKU's are listed (maybe the top two?), but it shows me what Intel is targeting.

Also, those worried about AMD and/or Intel going eco: money talks and competition means having competitive performance, so I don't see anyone backtracking extremely quickly -- if anything, power consumption would get paired down over time, which is fine and really a good thing, so long as raw performance is still improving gen-on-gen as well. At least until we have fusion, then they can go bonkers and make heat dissipation the biggest challenge, lol.

 

[jeremyj_83]

TDP's are back to where they should be, so that's a good sign. Of course, grain of salt here and not all SKU's are listed (maybe the top two?), but it shows me what Intel is targeting.

Also, those worried about AMD and/or Intel going eco: money talks and competition means having competitive performance, so I don't see anyone backtracking extremely quickly -- if anything, power consumption would get paired down over time, which is fine and really a good thing, so long as raw performance is still improving gen-on-gen as well. At least until we have fusion, then they can go bonkers and make heat dissipation the biggest challenge, lol.

Intel's TDP is based off of all cores active at base clock. That doesn't mean anything as even something like the 14900T is a 35W TDP but the boost power is 106W so IMO it is a 106W TDP CPU. Therefore without knowing the boost power we really don't have any idea what the power draw will look like for Arrow Lake.

 

[bit_user]

TDP's are back to where they should be, so that's a good sign.

They're specified the same as Gen 14 (Raptor Lake Refresh). K-series get 125W, "no suffix" i5, i7, and i9's get 65W, and T-series get 35W. No change in specs.

Also, those worried about AMD and/or Intel going eco: money talks and competition means having competitive performance, so I don't see anyone backtracking extremely quickly -- if anything, power consumption would get paired down over time, which is fine and really a good thing,

Well, following Sandybridge, Intel started to make some progress. Then, they slowly backtracked and reverted to 95 W by the time of Coffee Lake. With Comet Lake, they introduced the now standard 125W TDP of the K-series. Prior to Alder Lake, they didn't include Max Boost Power (PL2) in their official specs.

Launch Date	Model	TDP (PL1) Watts	Max Boost Power (PL2) Watts
2011 Q1	i7-2600K	95
2012 Q2	i7-3770K	77
2013 Q2	i7-4770K	84
2015 Q3	i7-6700K	91
2017 Q1	i7-7700K	91
2017 Q4	i7-8700K	95
2018 Q4	i9-9900K	95
2020 Q2	i9-10900K	125
2021 Q1	i9-11900K	125
2021 Q4	i9-12900K	125	241
2022 Q4	i9-13900K	125	253
2023 Q4	i9-14900K	125	253

At least until we have fusion, then they can go bonkers and make heat dissipation the biggest challenge, lol.

Well, a lot of US household outlets are only rated for 15 Amps, which puts 1600 W basically at the max (input should be higher, since PSUs are rated based on output power, and that's before allowing for brief excursions). I'm not sure exactly what spec systems people are using them for (dual-RTX 4090?), but 1600W PSUs are reasonably common, at this point.

 

[bit_user]

Intel's TDP is based off of all cores active at base clock. That doesn't mean anything as even something like the 14900T is a 35W TDP but the boost power is 106W so IMO it is a 106W TDP CPU. Therefore without knowing the boost power we really don't have any idea what the power draw will look like for Arrow Lake.

Assuming people are running at stock power limits & Tau, then you can't actually take PL2 and treat it as the CPU's dissipation.

The way boost works is by applying an Exponential-Weighted Moving Average (EWMA) filter to the power utilization. The CPU is allowed to boost up to PL2, until this "average" power exceeds PL1, at which point PL1 is enforced. The shape of this filter is determined by the time constant Tau, but people mistakenly infer Tau as a strict time limit - it's not. If you're using only a little bit more than PL1, you can boost a lot longer than if you're burning the max PL2 power.

Of course, people can tinker with their power limits, even increasing PL2, and/or extending the Tau parameter to effectively let them boost forever, which some gaming motherboards have started doing by default (at least, until the whole Raptor Lake degradation scandal blew up). At that point, we're no longer talking about stock settings, though. The above statements only apply to stock settings.

 

[jeremyj_83]

Assuming people are running at stock power limits & Tau, then you can't actually take PL2 and treat it as the CPU's dissipation.

Don't forget that until the Intel CPUs started dying the default setting from motherboard makers, and Intel by design, was to run unlimited Tau.

 

[bit_user]

Don't forget that until the Intel CPUs started dying the default setting from motherboard makers, and Intel by design, was to run unlimited Tau.

I even mentioned that. However, the two aren't definitely connected.

If you look at what types of benchmarks perform worse after Intel's microcode-based mitigation, they tend to be the single-threaded ones. Those don't run anywhere close even to PL1.

 

[DasLooney]

These tech sites are awful. Why aren't they mentioning the likelihood that these chips will have the same stability and chip damaging flaws of the last 2 generation of chips? I mean come on guys!

 

[TheHerald]

These tech sites are awful. Why aren't they mentioning the likelihood that these chips will have the same stability and chip damaging flaws of the last 2 generation of chips? I mean come on guys!

Because they won't have them. Its already fixed in the ones that did have them

That's like saying the upcoming zen 5 3ds will be self immolating. It's silly.