Discussion Comet Lake Intel's new Core i9-10900K runs at over 90C, even with liquid cooling TweakTown

[RetroZombie]

So 95c with a 280mm cooler and 95c vs 77c in CBR20 under similar conditions?

That's why i asked if the guy is a fraud.
And the so called mr cheap builds uses aio system? Something is not right that's for sure.

 

[DrMrLordX]

So 95c with a 280mm cooler and 95c vs 77c in CBR20 under similar conditions?

@UsandThem just gave you data showing that's not the case. Even with the crappy stock cooler, that 3600 wouldn't go past 84C.

@RetroZombie

The source seems highly questionable.

 

[Zucker2k]

@UsandThem just gave you data showing that's not the case. Even with the crappy stock cooler, that 3600 wouldn't go past 84C.

@RetroZombie

The source seems highly questionable.

I already acknowledged that. However, it's difficult to comparenthe testing methodology of these systems, duration of loads, and setups. The reviewer was fair and neutral in his conclusions so I wouldn't rush to call him a fraud. What I took away from all the reviews with regard to cooling is that these 10th Gen chips are surprising everyone with how cool they run if they're not being blasted with AVX based stress apps to push them to consume all their max power budget. None of these chips hit max tdp under under stock boost conditions that I saw in any review, even with air coolers, eg. Anandtech and Phoronix.

 

[Magic Carpet]

Aida FPU power tests I don't really care about, regular loads like compressing stuff, gaming, streaming, all done at the same time, would be far more useful.

 

[Ranulf]

Having watched TechYesCity for a couple of years now, he does not strike me as a fraud. I'd guess he just screwed up his data slide.

 

[Markfw]

I already acknowledged that. However, it's difficult to compare many of the methodology of these systems, duration of loads, and setups. The reviewer was fair and neutral in his conclusions so I wouldn't rush to call him a fraud. What I took away from all the reviews with regard to cooling is that these 10th Gen chips are surprising everyone with how cool they run if they're not being blasted with AVX based stress apps to push them to consume all their max power budget. None of these chips hit max tdp under under stock boost conditions that I saw in any review, even with air coolers, eg. Anandtech and Phoronix.

You need to just believe what Tweaktown, Anandtech, Phoronix and (I think its called) servethehome. Everybody says the same thing. If you are running a benchmark that runs more than 56 ms and they have it configured for pl2, it will take about 250 watts. Its a power hog accept that, and it runs hot. If you are doing a game, thats entirely game dependent, so I don't know how that will work out, but for any productivity more than 56ms (meaning anything, but a short benchmark) it will take power and run hot. Stop trying to finf the one or two situations where it migh take 125 or less. You spend all you time trying to prove it takes the same or less than Ryzen. Just the 7nm vs 14 nm is a huge problem.

Edit: I just reread what I quoted that "everyone" was suprised at how they cool. Maybe you can twist the facts so you have convinced youirself, but nobody here (except you and one or two others) are convinced. And I have read all the replies. I am GOING to buy one, and use the same cooler as my 3900x's. Let me know when you can find other to buy for $530, not paying the ebay $750. And I also need to buy a coolermaster 240 AIO (or the corsair 240 mm aio)

Right now my 3900x is running 99^% load @ 72c, 24/7. And its 77f ambient, so not cool.

 

[Ajay]

You need to just believe what Tweaktown, Anandtech, Phoronix and (I think its called) servethehome. Everybody says the same thing. If you are running a benchmark that runs more than 56 ms and they have it configured for pl2, it will take about 250 watts. Its a power hog accept that, and it runs hot. If you are doing a game, thats entirely game dependent, so I don't know how that will work out, but for any productivity more than 56ms (meaning anything, but a short benchmark) it will take power and run hot. Stop trying to finf the one or two situations where it migh take 125 or less. You spend all you time trying to prove it takes the same or less than Ryzen. Just the 7nm vs 14 nm is a huge problem.

Well, having a mother board that sets Tau for PL2 for more than 56 seconds is going against Intel specs. It's supposed to drop back to PL1 (125W) after that 56 second period. OFC, that would kill performance in productivity apps - so all the mobo makers cheat; probably with Intel pretending it doesn't notice. Anyway, there is no real reason for the 10900K - it's just a marketing tool. If you're and Intel gamer, buy a 10700K and be done. If you're an AMD gamer, buy a 3700X and be done.

 

[jpiniero]

Well, having a mother board that sets Tau for PL2 for more than 56 seconds is going against Intel specs. It's supposed to drop back to PL1 (125W) after that 56 second period. OFC, that would kill performance in productivity apps - so all the mobo makers cheat; probably with Intel pretending it doesn't notice.

Is it really cheating if Intel actively encourages board makers to do it?

You need to just believe what Tweaktown, Anandtech, Phoronix and (I think its called) servethehome. Everybody says the same thing. If you are running a benchmark that runs more than 56 ms and they have it configured for pl2, it will take about 250 watts.

Such is the price of 5 Ghz.

 

[Topweasel]

Well, having a mother board that sets Tau for PL2 for more than 56 seconds is going against Intel specs. It's supposed to drop back to PL1 (125W) after that 56 second period. OFC, that would kill performance in productivity apps - so all the mobo makers cheat; probably with Intel pretending it doesn't notice. Anyway, there is no real reason for the 10900K - it's just a marketing tool. If you're and Intel gamer, buy a 10700K and be done. If you're an AMD gamer, buy a 3700X and be done.

Intel doesn't care, in fact they would rather you ran on optimized boards. The whole trick with the 10900k and really going back to the 8700k. Is that they are playing a tight rope game requiring to have reasonable settings, have a stated TDP, have a boost time limit, and so on. Even the K sku's sell a lot in OEM systems. They need to have technical information to give to them and have functions in the chipset to allow those restrictions to be followed. For example the PL2 There was a big debate last year, Intel's documentation suggests PL2 should be set to *1.5 TDP (or was it 1.25). Anyways they had the mobo guys set the defualts on the 9900k to over 200w (think the same 241w its set to now which is looking more and more like set to off value). We look at it now confident that is what Intel wants it to be for the 10900k, but you had people fighting that was the case for the 9900k.

Intel doesn't care if they cheat if they do then its looks better for them. But this chip. It's boosts it's time. It's all designed primarily to look great for reviews. If you enable the "cheats" it just looks that much better. And for the retail enthusiast its probably the intended settings for it.

 

[Markfw]

Intel doesn't care, in fact they would rather you ran on optimized boards. The whole trick with the 10900k and really going back to the 8700k. Is that they are playing a tight rope game requiring to have reasonable settings, have a stated TDP, have a boost time limit, and so on. Even the K sku's sell a lot in OEM systems. They need to have technical information to give to them and have functions in the chipset to allow those restrictions to be followed. For example the PL2 There was a big debate last year, Intel's documentation suggests PL2 should be set to *1.5 TDP (or was it 1.25). Anyways they had the mobo guys set the defualts on the 9900k to over 200w (think the same 241w its set to now which is looking more and more like set to off value). We look at it now confident that is what Intel wants it to be for the 10900k, but you had people fighting that was the case for the 9900k.

Intel doesn't care if they cheat if they do then its looks better for them. But this chip. It's boosts it's time. It's all designed primarily to look great for reviews. If you enable the "cheats" it just looks that much better. And for the retail enthusiast its probably the intended settings for it.

Only one problem. It will not give you the sustained productivity like the 3900x and the 3950x, but the people its targeteted for won't see that until its too late.

But Intel sure does not care about the customer.

 

[Ajay]

Is it really cheating if Intel actively encourages board makers to do it?

The motherboard makers are fighting each other. Intel knows this will happen. I don't have it in writing that Intel is encouraging this behaviour, but they certainly are looking the other way - they know most gamers will just look at top frame rates, not power consumption.

 

[StrangerGuy]

I don't like how this industry is digging deeper into the hole of fuzzy "standards" like XMP' where there is stealth overvolting of the CPU uncore which was already scummy enough in my books.

 

[Tarkin77]

Techyescity testet with an $300 Asrock z490 Board... combined with DDR4 3200 (both systems i think)

so... cost of Motherboard, CPU, Ram, Cooler for the Intel System was somehing like USD 800?

What was the configuration and cost of the AMD testsystem?

not telling what exact configurations where used is misleading IMO!!

(and if he would have used a cheaper motherboard option for the intel system, he would not been able to run the memory at 3.200)

Fo example: in his testing, the 10400 was about 15% fast in SOTTR.

But over at computerbase.de, it only was 7% faster (CB used 2.666 Memory speed)

 

[Thunder 57]

Forget 380 Watts, when did 250W become acceptable. Everybody mocked the FX-9590. At least the Intel chip has performance to go with it, but its performance per watt is crap compared to AMD.

 

[Markfw]

Techyescity testet with an $300 Asrock z490 Board... combined with DDR4 3200 (both systems i think)

so... cost of Motherboard, CPU, Ram, Cooler for the Intel System was somehing like USD 800?

What was the configuration and cost of the AMD testsystem?

not telling what exact configurations where used is misleading IMO!!

(and if he would have used a cheaper motherboard option for the intel system, he would not been able to run the memory at 3.200)

Fo example: in his testing, the 10400 was about 15% fast in SOTTR.

But over at computerbase.de, it only was 7% faster (CB used 2.666 Memory speed)

Thats only one game. BUT this is exactly why I ignore places like this, when they don't release all the setup, and make it fair.

 

[chrisjames61]

Techyescity testet with an $300 Asrock z490 Board... combined with DDR4 3200 (both systems i think)

so... cost of Motherboard, CPU, Ram, Cooler for the Intel System was somehing like USD 800?

What was the configuration and cost of the AMD testsystem?

not telling what exact configurations where used is misleading IMO!!

(and if he would have used a cheaper motherboard option for the intel system, he would not been able to run the memory at 3.200)

Fo example: in his testing, the 10400 was about 15% fast in SOTTR.

But over at computerbase.de, it only was 7% faster (CB used 2.666 Memory speed)

In his defense he did not recommend the i9-10900K as it barely beat the AMD cpu in gaming but got trounced everywhere else.

 

[chrisjames61]

No, this is wrong. Intel wins 900p and 720p gaming as well .

On a more serious note, for productivity I wouldn't say it's a 100% clear cut victory for AMD. Many productivity flows rely on a mix of single/lightly threaded loads and multi-threaded loads. So there are some cases where a 9900K or 10900K may make sense, but I agree that generally speaking for productivity, AMD is the higher performer.

AMD the "higher" performer in productivity is the understatement of the year. You also are overlooking multiple apps running concurrently.

 

[Magic Carpet]

We haven't had this in a long time. Can't wait for Zen 3 and Rocket Lake S. In the meantime, the new i5 seems to be most reasonable SKU all things considered, if going Intel today.

 

[RetroZombie]

In the meantime, the new i5 seems to be most reasonable SKU all things considered, if going Intel today.

According to the hardware unboxed it seams the case, until you factor the following:

- intel i5 10600K is the best of the i5 and still doesn't beat the ryzen 5 3600​

- the total price of the intel system is 70% higher than the amd system​

- intel much more power hungry system add the problem of components wear out sooner, and HU forgot to put into those 70% costs the fact you need a better psu to handle the extra 2X peak TDP​

- the 3700X is much better and total cost of ownership is much lower​

If the best of the intel 10th gen cpu is junk what to say of the other models, there's literally no reason to consider any of their processors.

If you like to pay a premium for something that is actually worst, just choose any of the intel 10th gen cpus.

Edit:
From that old review the Pentium 4 was actually competitive against amd cpu's in applications out side of gaming, but in games it simply gets a beating, the memory latency is the key here 44ns on the amd systems and ~75ns on intels.

 

[UsandThem]

According to the hardware unboxed it seams the case, until you factor the following:

- intel i5 10600K is the best of the i5 and still doesn't beat the ryzen 5 3600​

- the total price of the intel system is 70% higher than the amd system​

- intel much more power hungry system add the problem of components wear out sooner, and HU forgot to put into those 70% costs the fact you need a better psu to handle the extra 2X peak TDP​

- the 3700X is much better and total cost of ownership is much lower​

If the best of the intel 10th gen cpu is junk what to say of the other models, there's literally no reason to consider any of their processors.

If you like to pay a premium for something that is actually worst, just choose any of the intel 10th gen cpus.

Junk? C'mon.

The 10th gen has some major downsides, but saying it's junk is going way too far, and is not truthful. Plus, I don't see how the 10600k would come out to be 70% higher in price. The CPU is $262 and the motherboards start at $150.

 

[RetroZombie]

The CPU is $262 and the motherboards start at $150.

According to HU you can build the amd system with just the price of the intel cpu.
Intel Core i5 10600K vs. Ryzen 5 3600 & R7 3700X

Junk is a little strong but then again after 2.5 years, what did intel bring new into the market outside of new motherboards?

 

[UsandThem]

According to HU you can build the amd system with just the price of the intel cpu.
Intel Core i5 10600K vs. Ryzen 5 3600 & R7 3700X

In that video, they say the 10600k is Intel's best value gaming CPU, which is so close in performance to the new i7 and i9, that those CPUs over $400 aren't worth it.

I didn't see a pricing breakdown, but there's no way the i5 system is 70% more than an AMD build.

Junk is a little strong but then again after 2.5 years, what did intel bring new into the market outside of new motherboards?

New? Absolutely nothing.

However, despite them being finely tuned Sky Lake CPUs, they are (and have been) competitive, winning in some ways, while losing in others. By you saying that Intel CPUs are junk, that's not any better than the anti-AMD crowd claiming AMD processors aren't good for things like gaming. It's just hyperbole in an effort to make only one side to look good, and the other side to look like an inferior product.

 

[RetroZombie]

By you saying that Intel CPUs are junk, that's not any better than the anti-AMD crowd claiming AMD processors aren't good for things like gaming.

You are right, junk is not the correct word, not overpriced too, not sure, now it's the same old stuff and still paper launched.... I'm an hardware enthusiast and very disappointed with this 'new' intel line up.

 

[RetroZombie]

In that video, they say the 10600k is Intel's best value gaming CPU, which is so close in performance to the new i7 and i9, that those CPUs over $400 aren't worth it.

"...factoring all that in you're looking at having to pay at least $500 for the 10600K with a decent entry level Z490 motherboard and a budget tower cooler the ryzen 5 3600 on the other hand that'll set you back $290 with something like the MSI B450 tomahawk max and coincidentally that's about what you'd pay for just the 10600K so from an upgraders perspective you're paying a little over 70% more for the Intel option and under realistic gaming conditions in most games that'll net you single-digit performance gains if you're not overclocking then when it comes to productivity tasks the Intel CPU is generally a little bit slower though it is close enough to call a tie except for the fact that the 10600 K use loads more power to achieve comparable performance..."

 

[DrMrLordX]

In terms of what makes the most sense to buy, find the new product in the Comet Lake-S lineup that is most-heavily discounted versus the old Coffee Lake product it replaces performance-wise and look at that. Whether it's the 10400, 10400F, 10600k, or anything else, you're looking at the price first.

And for people who have been waiting to buy Coffee Lake until later, it might make things easier for them (so they don't feel compelled to wait until October instead). That is probably the most-positive thing that can be said about Comet Lake-S.