Intel Core i7-11700K Review: Blasting Off with Rocket Lake

[humbug]

It's to do with the thermal density of 7nm and will only get worse as nodes shrink. The heat produced isn't huge but is so concentrated in a small area it is hard to get it away quickly enough resulting in higher temps.

With a larger process the heat is easier to transfer across a broader area of the IHS into the cooler baseplate.

The 5800X is 105W on a single chiplet.

The 5900X and 5950X are 105W over two chiplets.

Makes sense, i'm not concerned about it, seeing everyone else with a 5800X reporting the same behaviour and AMD confirming "As expected, its normal" i'm just shrugging my shoulders over it now and waving it away.

I'm guessing this is going to be a thing that's only going to get worse as transistor density increases, its going to get more and more difficult to remove the heat, even if there isn't much to remove.

 

[Minstadave]

Makes sense, i'm not concerned about it, seeing everyone else with a 5800X reporting the same behaviour and AMD confirming "As expected, its normal" i'm just shrugging my shoulders over it now and waving it away.

I'm guessing this is going to be a thing that's only going to get worse as transistor density increases, its going to get more and more difficult to remove the heat, even if there isn't much to remove.

Exactly. The chiplet model helps a bit as physically distancing the heat sources helps with heat transfer into the IHS over a monolithic die but if as dies get smaller and more dense it will only get worse.

 

[Game]

He does have a point. All the guys who game with us with AMD have some funny little issues with things. It just isn't as mature as an Intel platform YET - thats it.

A few who I play with have mentioned micro stutter mostly in warzone and battlefield all on 3 series

but this might be the case with intel as well for all I know

 

[Freddie1980]

Rocket Lake ah? More like Bluebird K7 the way that thing sunk in the gaming benchmarks. 19% IPC gain only to be undone by high core to core latency.

 

[MartinPrince]

Noctua is rated to 225W i dont think it would have fared much better.

This is my main point, the fact that you are having to 'think' whether it would have fared better or not is entirely avoidable. Which other reviewer worth their salt would use different models of heatsinks and also fans, then try to make heat comparisons. That's just inane.

Thermalright True Cooper review from 2010 and you can see it is not as good as the Noctua D14 being about 2c off.

My D14 is ~6c behind my D15 and then add in another couple of degrees that heatpipe based coolers can loose over ~12 years which is where I get my wild stab at ~10c Though I also did say that it could be the opposite way round and Thermalright True C is better than anything we have today - but then use it also on the Ryzen system!

People seem to ignore or be OK with the fact that Mr Cutress is unnecessarily using different coolers but then he makes heat related parallels, insisting the he is using 'the best air cooler'. I'm waiting for reviews where they at least have the good sense to use the same cooler on all systems.

 

[pc-guy]

This is my main point, the fact that you are having to 'think' whether it would have fared better or not is entirely avoidable. Which other review worth their salt would use different models of heatsinks and also fans, then try to make heat comparisons. That's just inane.

Thermalright True Cr review from 2010 and you can see it is not as good as the D14 being about 2c off.

My D14 is ~6c behind my D15 and then add in another couple of degrees that heatpipe based coolers can loose over ~12 years which is where I get my wild stab at ~10c Though I also did say that it could be the opposite way round and Thermalright True C is better than anything we have today - but then use it also on the Ryzen system!

People seem to ignore or be OK with the fact that Mr Cutress is unnecessarily using different coolers but then he makes heat related parallels, insisting the he is using 'the best air cooler'. I'm waiting for reviews where they at least have the good sense to use the same cooler on all systems.

I don’t think it matters much about the cooler used.

it fact is that it is 300w peak load (albeit not a load that will be realised in real life)

Don’t think any 8c chips is capable of that past and present. So that’s a record. A better cooler will mean the chip is running 90c instead 100c. Doesn’t really matter does it.

obviously they had to put something on it to prevent it from cooking itself to death. They thought it was a good idea to have that thermalright cooler. So that was their decision. They must have used that cooler on other bench to be confident of using it for this chip.

If the power consumption is the result of the new cove architecture, it really doesn’t bode well for the next generations of intel chips where the cores are packed denser on shrunk die.

 

[Grim5]

As long as its the same cooler they use for all their CPU reviews then I have absolutely no issue with their results and neither should you. If they're using a new cooler just for the 11700k review then thats dodgy

 

[MartinPrince]

I don’t think it matters much about the cooler used.

..... A better cooler will mean the chip is running 90c instead 100c. Doesn’t really matter does it.

It matters to me because I like things to be accurate. It then allows me (and others) to make a better informed decision. 10c is the difference between a system throttling and not throttling. As simple as that.

It's a basic CPU reviewing standard - use the same cooler on all systems, it makes comparisons more valid.

Seeing you missed it I'll repeat it again. He reviewed the 10850K in Jan '21 using the same Thermalright cooler and came to the conclusion that it ran hot but on the Intel system he used a poor £250 motherboard with an old bios from Jun 20 but on the Ryzen he used one of the best £700 motherboards with a bios from Dec '20.

If he'd used the equivalent £700 motherboard for Intel with a recent bios he would have probably found the temperatures dropped by 25c, which would have totally changed his conclusion - and what he wrongly inserted in the minds of his readers.
https://linustechtips.com/topic/1246523-i9-10850k-is-mega-hot/

He is the Senior CPU Editor at Anandtech, how is not aware that Intel has released a bios fix for early heat issues, that was included in September bios'?

The devil is in the detail. Like I said, I'll wait for less amateurish reviews.

 

[torwak]

Hi guys,

Do you know when OC will open pre-orders for Rocket Lake?

thanks!

 

[Grim5]

Hi guys,

Do you know when OC will open pre-orders for Rocket Lake?

thanks!

should be 15 March

 

[pc-guy]

@MartinPrince I don’t bother with reviews of CPU temps. Cos 1) I don’t have the same setup as these reviewers be it motherboard or cpu cooler or an open bench setup 2) I just want to know under load prime/Aida/Occt what sort of TDP is likely to have 3) the overclock on CPU i won’t be able repeat on my system, plus I dont want to go balls to wall with the OC anyway as I want to keep the CPU for a while rather than burn it out 4) case and ram setup 5) fan curve

There is just too many variable in a review of CPU temperature to inform me of anything useful that I can apply to my own rig.

so what cooler they use or what motherboard they use and what CPU temp end up being are irrelevant. CPU is hot comment is irrelevant, if a CPU has higher power draw, you just need to get more expensive cooler. Simple as that.

anyway, I just look for what power the CPU suck down under stress. And see if my cooler can handle it. I have H100i. I know it can do upto 175w and keep things @70c.

that’s why I am not bothered about someone’s cooler setup and to that extent their motherboard choice.

However a good motherboard offers better OC potential and faster RAM speed or better timing then that is a potential important factor determining how “fast” the CPU is. So I would be paying attention to the motherboard choice to that respect. But again if review is using god like, then that is not something I will go and use so it becomes irrelevant to me at that point.

 

[the wopster]

Hi guys,

Do you know when OC will open pre-orders for Rocket Lake?

thanks!

Why bother, the 5800x is widely available now and ****** all over intels current and future offerings, why pay a price premium for less performance and higher power consumption

 

[timorous]

This is my main point, the fact that you are having to 'think' whether it would have fared better or not is entirely avoidable. Which other reviewer worth their salt would use different models of heatsinks and also fans, then try to make heat comparisons. That's just inane.

Thermalright True Cooper review from 2010 and you can see it is not as good as the Noctua D14 being about 2c off.

My D14 is ~6c behind my D15 and then add in another couple of degrees that heatpipe based coolers can loose over ~12 years which is where I get my wild stab at ~10c Though I also did say that it could be the opposite way round and Thermalright True C is better than anything we have today - but then use it also on the Ryzen system!

People seem to ignore or be OK with the fact that Mr Cutress is unnecessarily using different coolers but then he makes heat related parallels, insisting the he is using 'the best air cooler'. I'm waiting for reviews where they at least have the good sense to use the same cooler on all systems.

That is with 2x 75CFM fans vs 3x 88CFM fans on the Noctua. With 2x 300CFM fans on both the True Copper is ahead. Ian used 2x 173CFM fans on the true copper so performance between that and a 3 fan D14 is going to be pretty similar.

On a different note https://twitter.com/9550pro/status/1369442891198763011 might explain some gaming performance regressions. According to the small print the 11700K defaults to Gear 2 with 3200 ram and the 11900K defaults to Gear 1. That would increase latency substantially and give worse gaming performance. Seems like artificial segmentation between the products to make the 11900K look better.

EDIT: Also are you comparing your Stock D15 to Stock D14 or are you making sure the fans are the same because when overlockers.com reviewed the D15 and D14 with the same fan config performance was pretty similar. Seemed like most of the gain of the D15 was the improved stock fan over the D14.

 

[pc-guy]

On a different note https://twitter.com/9550pro/status/1369442891198763011 might explain some gaming performance regressions. According to the small print the 11700K defaults to Gear 2 with 3200 ram and the 11900K defaults to Gear 1. Seems like artificial segmentation between the products to make the 11900K look better.

why am i not surprised about this at all if this is truly what intel is doing!

 

[timorous]

why am i not surprised about this at all if this is truly what intel is doing!

Just Intel doing Intel things.

HardwareLuxx have a review https://www.hardwareluxx.de/index.p...-lake-s-als-core-i7-11700k-im-vorab-test.html

Does a bit better vs the 10700K in the 4 games they test. 5.5% faster instead of the 1.3% Anandtech got but Metro is a big outlier that seems to like Rocket Lake but it is still behind the 5800X. Seems HardwareLuxx tested with 1:1 on the memory gearing so I doubt the 11900K will improve on these numbers much.

They also had it at 190W in CB R23 nT with temps under an H150i of 74 degrees.

 

[RavenXXX2]

Why the **** are they testing with turing, known to bottleneck Zen 3.

https://www.techpowerup.com/review/intel-10900k-vs-amd-5900x-gaming-performance/

 

[Harlequin]

Why the **** are they testing with turing, known to bottleneck Zen 3.

https://www.techpowerup.com/review/intel-10900k-vs-amd-5900x-gaming-performance/

Because they are testing a `slower gpu` - which it says, in the section title.

 

[timorous]

Why the **** are they testing with turing, known to bottleneck Zen 3.

https://www.techpowerup.com/review/intel-10900k-vs-amd-5900x-gaming-performance/

Their 5xxx series reviews were showing different results to other outlets (10th gen being slightly ahead at 1080p). This is what they found out, pretty interesting to be honest.

 

[MartinPrince]

@MartinPrince ...so what cooler they use or what motherboard they use and what CPU temp end up being are irrelevant. CPU is hot comment is irrelevant, if a CPU has higher power draw, you just need to get more expensive cooler. Simple as that.

anyway, I just look for what power the CPU suck down under stress. And see if my cooler can handle it. I have H100i. I know it can do upto 175w and keep things @70c.

Sorry but that is just not a sensible approach as one can not accurately gauge the heat output of a CPU from how much power it uses.

My 5800X used ~140W under stress and using a D15 the temperature goes to ~85c. My 3950X uses ~140W under stress and using exactly the same setup the temperature goes to ~61c.
Here is a test I've just done running CBr23 multi threaded for 10mins on my 3950X

People can see the folly in using only power usage to determine how hot a CPU will be.

The point is accuracy. If you don't like accuracy then that is your prerogative but if a reviewer uses massively different motherboards or unnecessarily different coolers on his system then they open themselves up to have inaccurate results and therefore false conclusions.

This is aptly demonstrated in Mr Cutress' Jan 21 reviewer of the 10850k.

 

[MartinPrince]

That is with 2x 75CFM fans vs 3x 88CFM fans on the Noctua. With 2x 300CFM fans on both the True Copper is ahead. Ian used 2x 173CFM fans on the true copper so performance between that and a 3 fan D14 is going to be pretty similar.

On a different note https://twitter.com/9550pro/status/1369442891198763011 might explain some gaming performance regressions. According to the small print the 11700K defaults to Gear 2 with 3200 ram and the 11900K defaults to Gear 1. That would increase latency substantially and give worse gaming performance. Seems like artificial segmentation between the products to make the 11900K look better.

EDIT: Also are you comparing your Stock D15 to Stock D14 or are you making sure the fans are the same because when overlockers.com reviewed the D15 and D14 with the same fan config performance was pretty similar. Seemed like most of the gain of the D15 was the improved stock fan over the D14.

Back in the '90's I used to have 300CFM fans but I appreciate my hearing too much.

My point is that a reviewer should be using exactly the same cooler on their test system. Not a better cooler on one and a worse one on the other.

Like I previously said, I could easily also argue why is he using a much better cooler on the 11700k than on the Ryzen system? The 11700K may run even hotter but because he uses a cooler from 2008, that hardly any had (only 3000 produced) then on the Ryzen he uses a 'lesser' Noctua U12s there is a totally unnecessary disparity.

Therefore I can't rely on the comparative accuracy of his results. Why not just use the U12s on both systems? How can anybody argue against that?

 

[timorous]

Back in the '90's I used to have 300CFM fans but I appreciate my hearing too much.

My point is that a reviewer should be using exactly the same cooler on their test system. Not a better cooler on one and a worse one on the other.

Like I previously said, I could easily also argue why is he using a much better cooler on the 11700k than on the Ryzen system? The 11700K may run even hotter but because he uses a cooler from 2008, that hardly any had (only 3000 produced) then on the Ryzen he uses a 'lesser' Noctua U12s there is a totally unnecessary disparity.

Therefore I can't rely on the comparative accuracy of his results. Why not just use the U12s on both systems? How can anybody argue against that?

Well he is not really testing temperatures. He mentions it because it gets pretty high under AVX 512 workloads but since they don't really have temperature graphs in a more real world workload as long as the coolers are enough to allow the CPUs to run at 'stock' settings without artificially restricting their boost behaviour then that is 'good enough' which seems okay to me. He is not exactly disadvantaging the Intel system here.