Unknown hardware enthusiasts have run a burn-in test on Intel's 11th-Gen Rocket Lake processor to expose its power consumption under extreme loads and compare it to its predecessors from the Comet Lake-S family. It turns out the upcoming Core i9-11900KF CPUs can get extremely hot and power hungry under extreme loads, just like their Comet Lake ancestors. Intel's upcoming eight-core Core i9-11900KF 'Rocket Lake-S' processors can purportedly heat up to 98C and pull 250W of power during stress tests. That means the chips should place well in our CPU Benchmarks Hierarchy, at least one would hope given all that power consumption, but they'll run hot just like the previous-gen Intel chips.
Although Intel's latest 10th Generation Core 'Comet Lake-S' processors are rated for a 125W TDP, they can actually suck up to 250W ~ 330W of power when they boost on all cores for up to 56 seconds, allowing them to provide their maximum potential in situations where it is actually needed.
Intel's public-facing specs list power consumption based on the default power level (PL1). There's a big difference between the default power level and an all-core turbo power level (PL2), so you'll need an advanced motherboard, a quality PSU, and a capable cooling system to tame the Comet Lake beast. That's because Intel had to increase the PL2 level on its Comet Lake CPUs in a bid to make them more competitive against AMD's Ryzen lineup.
Apparently, the same rules apply to Intel's upcoming eight-core Core i9-11900KF 'Rocket Lake-S' processors that can heat up to 98C and pull 250W of power at 1.325V Vcore when running AIDA64's FPU stress test, according to Chiphell. The test CPU was cooled down using an entry-level 360-mm closed-loop liquid cooling system. The chip's exact clocks are unknown, but based on leaks, it should run at 3.50GHz by default and boost all of its cores to 4.8 GHz for short periods.
Being manufactured using a mature 14nm process, Intel's latest enthusiast-grade processors with eight or ten cores are not exactly energy efficiency champions, which isn't surprising because this node was not developed for CPUs that combine a high frequency and a high core count.
While the Rocket Lake-S CPU is based on a new microarchitecture and has several other advantages over Comet Lake-S processors, it looks like its thermals and power consumption will be comparable to those of its predecessors, at least as far as stress tests are concerned. Meanwhile, bear in mind that stress tests do not usually reflect real-world workloads, but are meant to reveal the weaknesses of your PC build.
As Intel is getting ready to release its 11th Generation 'Rocket Lake' CPUs this April, it has already begun to send its samples to a broad audience of its clients so they could prepare for the launch. As a result, certain test results will inevitably emerge well before full-fledged final hardware reviews show up. That said, the unreleased processors' current test results should be taken with a grain of salt.
Stay on the Cutting Edge
Join the experts who read Tom's Hardware for the inside track on enthusiast PC tech news — and have for over 25 years. We'll send breaking news and in-depth reviews of CPUs, GPUs, AI, maker hardware and more straight to your inbox.
Anton Shilov is a Freelance News Writer at Tom’s Hardware US. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.
I'm surprised processors this power hungry can be sold in the great state of California :whistle:Reply
Maybe they'll bundle them with solar panels 😂mrv_co said:I'm surprised processors this power hungry can be sold in the great state of California :whistle:
So what will RL bring over current gens? I can't think of a single thing really.Reply
Productivity is surrendered to the higher core count Ryzens, so that's out of the window.
Gaming will be a smidge better, again, very hard to spot and feel a difference especially when gaming at higher than HD levels where you're most likely to be GPU bound anyway.
250W vampire TDP sucking hog - good luck benching and overclocking on smaller PSU's.
As mentioned above, a PSU upgrade might also be in order. More money down the drain.
Hot as hell, good luck gaming with a machine so loud next to you that has to be super tuned to contain noise levels.
Half the lineup consists of non K (locked) models. If AMD can offer full gens of OC'able chips (with hardly any exceptions), why can't Intel do the same?
No 12C or 16C options. Forget upgradable options down in the future.
PCIE4 coming like 2 years late? Or shall I say well done?
Backported product. Knowing that your product's full potential won't be realized.
Regressing on core count. And they wanna call their highest boosting chips "i9's" to differentiate from the slightly lower clocked i7's?
Do you also get Farcry6 free with RL as you do with new Ryzen? Or anything else seeing that Intel has promoted it as a "gaming" chip?
I can go on and on, but my brain has finally put up the white flag. Pat should have joined 5+ years ago.
They can't have 12-16C because they must clock them at 4-4.5Ghz max. 16 cores like this with 5Ghz will hit 500W and even custom loop can't cool it, but intel will never sell that low clocked CPUs. If you look at their tiger lake you will see again much more power consumption for just 4 cores compared to 8 zen 3 cores. Alder lake wont be different and that is why they will max at 8 cores again.Reply
barryv88 said:So what will RL bring over current gens? I can't think of a single thing really.
It is faster in lightly threaded workloads, and equally fast in productivity tasks (as Intel current gen). So all in all, it is faster than the current gen.
New cpus typically are in a the same performance ballpark as their predecessors (with some notable exceptions). No one is going to upgrade current gen to RL. But people moving up from a gens back will.
It's not interesting to the least. But its ok.
This looks to me to be an AVX-512 Overclocking torture test, running all 8 cores' AVX-512 units (all 8 of them) simultaneously at @ 4.8GHz for 10's of minutes. Not surprised by the power or the heat. Actually, it's impressive to be able to run all 8 AVX-512 units stable @ 4.8GHz for this long (see the graph). Side note: running AIDA64 FP only on a Cascade Lake -X results in full AVX-512 load and I assume it will be the same here.Reply
From cascade lake-X we already know that AVX-512 is notoriously power hungry and generates a lot of heat (hence the need for high AVX-512 negative offset on cascade lake -X, especially when overclocking the rest of the CPU) and that the whole CPU down clock with the AVX-512 negative offset (AVX-512 negative offset BIOS setting) when running hard AVX-512 workloads. But in this picture RL does not seam to down clock (at all?) but stays at 4.8GHz - impressive that it's even stable... - what cooling did they use? water?
We already know that it is very hard to run AVX-512 @ 4.8 on a cascade lake X or skylake X (not sure I've see anyone do AIDA64 FP on standard air cooling with reasonable temps).
So net, net they have improved AVX-512 power consumption and temps (compared to Cascade Lake X) and if they used a bit less aggressive AVX-512 overclocking, it would go down below 80C and be a good improvement over Cascade lake - X...
I was getting annoyed with the amd fanboism after amd actually passed intel in single core performance, people saying intel is getting wrecked after being surpassed once in the last decade, but now seeing this all I can say is this sucks, what hell is this crap?Reply
Less cores + same heat as previous generation for slightly better performance than amd? Not worth it. 12 core amd using half the power for similar single core performance but with 8 more threads?
This is ridiculous, I think intel needs to accept that until they shrink the process they are going to need to adopt the we aren't as good so here are some deals strategy that got amd to where they are.
I expect it'll behave similarly to Comet Lake, with its conservative power limits.
Remove the power limits, then the Rocket can take off.
Shrinking the node will have no effect on the power draw and heat of avx since any benefit would go towards running it at higher clocks, such is the way of turbo and unlocked limits.spentshells said:This is ridiculous, I think intel needs to accept that until they shrink the process they are going to need to adopt the we aren't as good so here are some deals strategy that got amd to where they are.
TerryLaze said:Shrinking the node will have no effect on the power draw and heat of avx since any benefit would go towards running it at higher clocks, such is the way of turbo and unlocked limits.
Im fine with your opinion.