Review The Intel Core i9-9900KS Special Edition Review: 5.0 GHz on All the Cores, All the Time

[TJ Hooker]

On AMD it is related to “thermal circuit” shenanigans, where TDP equals (temperature difference)/(thermal resistance), and the values of (temperature difference) and (thermal resistance) are simply artificially adjusted so that the quotient gives you whatever TDP you want…

Do you have a source for this?

Edit: Think I found it https://www.gamersnexus.net/guides/...lained-deep-dive-cooler-manufacturer-opinions
I thought I remembered Ryzen chips actually abiding by their rated TDPs in terms of power draw pretty closely, but maybe that was only for Ryzen 1/2K as it seems that Ryzen 3K can exceed its stated TDP by a not insignificant amount.
Guess TDP has become largely meaningless for both Intel and AMD, if for different reasons.

 

[joeblowsmynose]

Somehow I initally quoted the wrong post ... Edited:

You can just keep it at 4Ghz and get that 127W TDP. It is up to you.

That's not the issue ...

The issue is that people will get greatly varying out of the box performance that will not be expected by the average consumer, dependent on which mobo they run with it. It's screwing over the consumers, for the sake of being able to lie about a TDP. Its disingenuous to the people who purchase this product.

And just to be clear, the difference between a mobo that follows Intel spec and one that does not, will be multitudes greater than a chip that falls 50mhz behind on a short lived burst boost - should the outcry be equally proportional to the difference in actual performance?

 

[PCWarrior]

I thought I remembered Ryzen chips actually abiding by their rated TDPs in terms of power draw pretty closely, but maybe that was only for Ryzen 1/2K as it seems that Ryzen 3K can exceed its stated TDP by a not insignificant amount. Guess TDP has become largely meaningless for both Intel and AMD, if for different reasons.

Well these definitions have been constant for both companies for a long time, it is just that they are now pushing the performance envelope so hard that they rendered TDP meaningless. I remember that for Intel, during the "modern quad-core era" (2600K-7700K), power consumption was staying below TDP/PL1 even when turboing on all cores at the all-core turbo frequency and with intense AVX2 100% workload at that. So, there was a lot of headroom for both a frequency and core count increase on 14nm but achieving an 100% core count increase and another 10-25% frequency increase on the same node and still staying at similar power consumption levels would have needed something more than a miracle and miracles don't happen. It is still mighty impressive though what Intel has managed to do on 14nm.

The issue is that people will get greatly varying out of the box performance that will not be expected by the average consumer, dependent on which mobo they run with it. It's screwing over the consumers, for the sake of being able to lie about a TDP. Its disingenuous to the people who purchase this product.

And just to be clear, the difference between a mobo that follows Intel spec and one that does not, will be multitudes greater than a chip that falls 50mhz behind on a short lived burst boost - should the outcry be equally proportional to the difference in actual performance?

Intel does have a well-defined baseline cpu behaviour set in the BIOS settings. On top of that motherboard vendors and OEMs are allowed to tweak things and achieve an even higher performance by investing in their power delivery and cooling systems and do so while still being considered that they run the cpu in spec. Besides all Z390 boards, either straight out ignore power limits or prompt you to remove them the moment you enable XMP. That creates a level playing field again. And all Z390 boards are capable of handling an indefinite all-core turbo of 5GHz on a 9900KS and certainly all the midrange to high-end boards have VRM designs capable of doing so while also ensuring high efficiency and super long-term reliability of their components. But, even if there was actually a variability in performance based on the motherboards used, that would still be no different to laptops or phones with identical processors (say Snapdragon 855) having varying performance. People will also get different performance/experience on their systems for all sorts of other things: Different coolers, different RAM speeds/sizes, different SSD speeds/types, different GPUs and the list goes on.

That’s very different from AMD promising something, failing to deliver it, admitting said failure and taking actions to (somewhat) fix it. And we know very well why AMD did what they did with their overambitious advertised turbo speeds. It was/is a clear attempt of creating the illusion that the frequency gap with Intel has closed, while in fact it pretty much remained the same. AMD cpus still cannot be clocked/overclocked to over 4.2-4.3GHz on all cores and the supposedly higher stock single-core clocks are nothing but a sham. I am willing to bet you were one of the many people who were part of the outcry when Intel demonstrated a 5GHz 28 core Xeon. Yet when AMD is at fault you complain about the outcry.

 

[bit_user]

I am willing to bet you were one of the many people who were part of the outcry when Intel demonstrated a 5GHz 28 core Xeon. Yet when AMD is at fault you complain about the outcry.

This exchange is going downhill, fast.

Why can't we simply hold each manufacturer to account, when they over-promise, under-deliver or misrepresent? Why does it have to turn into a debate over who is worse?

Yeah, AMD was too ambitious on Ryzen 3k clocks, and yes Intel should do more to educate the public on the actual dissipation of their CPUs in gaming motherboards. ...and, yes, Intel's all-core 28-core x 5 GHz demo was disgraceful in a number of ways. Bringing that up is in no way relevant or constructive.

 

[joeblowsmynose]

...

That’s very different from AMD promising something, failing to deliver it, admitting said failure and taking actions to (somewhat) fix it. And we know very well why AMD did what they did with their overambitious advertised turbo speeds. It was/is a clear attempt of creating the illusion that the frequency gap with Intel has closed, while in fact it pretty much remained the same. AMD cpus still cannot be clocked/overclocked to over 4.2-4.3GHz on all cores and the supposedly higher stock single-core clocks are nothing but a sham. I am willing to bet you were one of the many people who were part of the outcry when Intel demonstrated a 5GHz 28 core Xeon. Yet when AMD is at fault you complain about the outcry.

5 ghz all core boost all the time ... as this headline on this article advertises. Some consumer reads that buys one, with an Asus board, is not an avid OC'er and knows not how to tweak bioses, doesn't get 5ghz all core boost all the time ... notices and gets upset, then buys AMD next round.

Like you insinuated... No problem at all!, right? I mean screw those consumers for expecting advertising and reviews to not lie to them ...

... nothing to see here folks ... move along, move along ...

I suppose you thought the 5ghz 28 core "coming before the end of year" lie was pretty cool! ... you best be careful there with that one ... it was a cheap, pathetic stunt, there's no way you can argue against that without making yourself the fool.

 

[PaulAlcorn]

I just watched Steve Burkes review, and he notes some pretty interesting caveats with this chip ...

The title of this article: "5.0 GHz on All the Cores, All the Time" --- is not true if it is used with a motherboard manufacturer that stuck to Intel's TDP guidelines, like Asus did.

On Asus boards, this chip only boosts all core @5ghz for a limited time then drops back down to maintain reasonable power consumption, as per Intel's own TDP specification for this processor. So basically, Intel gave mobo makers specs to keep TDP ~127w but really it was just their way of lying about TDP but deferring that misinfo to the mobo makers. The mobo maker that actually chose not to allow a misleading TDP now gets punished ... sounds like an Intel move.

So I assume this will mean that Asus is going to be pissed with Intel since gigabyte and MSI boards will let it suck all the power it needs to maintain 5ghz -- completely disregarding the TDP is the only way it boosts at 5ghz all cores, full time.

So how this chip performs has far more to do with the mother board, than the chip. This is stupid.


As an aside question ... what's the cooling power required for OCing? The OC testing here was done using 720mms worth of radiators on a custom loop - what's next ... LN2 testing? We know the limit is somewhere between the H115i and the dual rad custom loop, but I wonder where that is. A lot of cooling for any OCing anyway it seems ... (but expected).

Intel's TDP is rated at base clock. This is a known. The base frequency of the processor is 4.0 GHz, not 5.0. Therefore, you shouldn't expect 5.0 within the TDP rating, period. Power during boost is another level above typical TDP.

Burke says you won't get 5.0 GHz within the rated TDP, which is true, and has been true of Intel processors for a very long time (I believe since the inception of the boost feature, but would have to double check - at least five years).

 

[joeblowsmynose]

Do you have a source for this?

Edit: Think I found it https://www.gamersnexus.net/guides/...lained-deep-dive-cooler-manufacturer-opinions
I thought I remembered Ryzen chips actually abiding by their rated TDPs in terms of power draw pretty closely, but maybe that was only for Ryzen 1/2K as it seems that Ryzen 3K can exceed its stated TDP by a not insignificant amount.
Guess TDP has become largely meaningless for both Intel and AMD, if for different reasons.

It appears with zen2 AMD took a page from Intel's TDP illusion show ...

But let's all remember that TDP <> power consumption, but should be roughly aligned to it, or at least represent real world cooling requirements. I dare someone to review a 9900k with a CPU cooler that can only cool 100w. I'd love to see how much throttling would happen in that case and what performance results vs 300w of cooling on the same chip.

Thanks to Intel, the number has become pretty useless in both camps ...

 

[joeblowsmynose]

Intel's TDP is rated at base clock. This is a known.

Hi Paul,

It's a known among enthusiasts only ... and ... it's useless for anything, including cooling requirements, of which the number is roughly supposed to represent. Thus the issue that few reviewers are willing to call out.

If they were willing to call it out, perhaps AMD wouldn't also be having TDPs that somewhat have the same effect for Zen2.

Just beacause soemthing has been done "for a while" doesn't make it right.

 

[PaulAlcorn]

Hi Paul,

It's a known among enthusiasts only ... and ... it's useless for anything, including cooling requirements, of which the number is roughly supposed to represent. Thus the issue that few reviewers are willing to call out.

If they were willing to call it out, perhaps AMD wouldn't also be having TDPs that somewhat have the same effect for Zen2.

Just beacause soemthing has been done "for a while" doesn't make it right.

I'm not arguing the ethics of it, just stating a fact.

 

[bit_user]

Intel's TDP is rated at base clock. This is a known.

I'm actually not sure that's accurate. What I believe is that Intel guarantees the base clock is achievable & sustainable, within the TDP. However, the TDP could leave some room for boost, above and beyond that.

You can test this with an all-core workload, and see if a mobo that properly respects PL1 and tau will sustain the CPU anywhere above base clocks.

The base frequency of the processor is 4.0 GHz, not 5.0. Therefore, you shouldn't expect 5.0 within the TDP rating, period. Power during boost is another level above typical TDP.

I'm going to side with Joe, on this one. Readers would be well-served if the piece had stated up-front that the CPU was capable of this feat, but actually achieving it required some care in selection and configuration of mobo, cooling solution, etc.

I know you know what's up, and page 2 did touch on this point, to some degree.

 

[bit_user]

I'm not arguing the ethics of it, just stating a fact.

I think he wants to see Intel & AMD get taken to task for fudging this stuff.

I don't care about that, so much as just equipping readers with the knowledge needed to understand the product and how to get the most out of it.

Thanks for the review, BTW.

 

[joeblowsmynose]

I think he wants to see Intel & AMD get taken to task for fudging this stuff.
...
just equipping readers with the knowledge needed to understand the product and how to get the most out of it.
...

This.

This is why I also think its disingenuous to post only bottlenecked CPU gaming numbers. It misleads the average buyer.

Looking into the general forums its obvious not every visitor of this site is an enthusiast that knows these things. There's one thread posted by a poor chap that spent his hard earned money upgrading his CPU instead of his GPU thinking he would get "X" amount of increased performance because of the fact that only bottlenecked CPU numbers are ever posted. He had an RX 480 or 580, so of course the GPU was the bottleneck and he got zero performance increase. Now had he spent that money on a GPU, it would have been money well spent. But unfortunately the reviews, as they are crafted, are to blame for screwing over readers like this.

 

[PCWarrior]

Why can't we simply hold each manufacturer to account, when they over-promise, under-deliver or misrepresent? Why does it have to turn into a debate over who is worse? Bringing that up is in no way relevant or constructive.

I'm always critical to any shenanigans by any manufacturer I am interested to buy products from. That's not the point however. The point is that AMD fanboys are double-standard hypocrites who only take the high-moral ground when it comes to Intel and Nvidia. Yet, when their beloved AMD (who they want to portray as a saint or a hero) is at fault or is doing the exact same thing that they had earlier criticised Intel or Nvidia of doing, they are always quick to either give them a free pass or downplay the offence or defend it or chuck it down to “well, everyone is doing the same”. I am simply pointing out that hypocrisy which quite frankly is cancerous and I (and many other people) am sick and tired of.

Intel's TDP is rated at base clock. This is a known. The base frequency of the processor is 4.0 GHz, not 5.0. Therefore, you shouldn't expect 5.0 within the TDP rating, period. Power during boost is another level above typical TDP.

Exactly. It is actually disclosed on each page of every sku on Intel’s ark database. Here is a link to the page of 9900K in Intel’s ark database. Under performance there is an entry saying TDP. Next to it there is a question mark. You click on it and it says the following:

Thermal Design Power (TDP) represents the average power, in watts, the processor dissipates when operating at Base Frequency with all cores active under an Intel-defined, high-complexity workload. Refer to Datasheet for thermal solution requirements.

It appears with zen2 AMD took a page from Intel's TDP illusion show ...
Thanks to Intel, the number has become pretty useless in both camps ...

Well if you want to talk about negative trends adopted by competitors, how about the one of not disclosing the all-core turbo but simply stating the base and single core turbo? Intel briefly stopped doing it because AMD was not doing it (never did it). Thankfully Intel adopted this negative habit only in the 8th gen. With the 9th gen Intel started disclosing these numbers again. AMD on the other hand continues to only give base and single core turbo (which is more like a wishlist at this point than an actual spec but anyway).
https://www.techpowerup.com/237740/on-intels-decision-to-no-longer-disclose-all-core-turbo?cp=2

 

[joeblowsmynose]

...
Well if you want to talk about negative trends adopted by competitors, how about the one of not disclosing the all-core turbo but simply stating the base and single core turbo? Intel briefly stopped doing it because AMD was not doing it (never did it). Thankfully Intel adopted this negative habit only in the 8th gen. With the 9th gen Intel started disclosing these numbers again. AMD on the other hand continues to only give base and single core turbo (which is more like a wishlist at this point than an actual spec but anyway).
https://www.techpowerup.com/237740/on-intels-decision-to-no-longer-disclose-all-core-turbo?cp=2

Not really the same point ...

AMD's zen2 all core boost numbers are amazingly and surprisingly pleasant and much higher (relatively) than expectation when coming from looking at Intel's chips all core boosts (aside from the 9900k variations) ... I wish they did advertise these as well because they certainly make their chips look a lot better ... if boost clocks are that important to someone to care ... it certainly is NOT about them trying to hide something that isn't up to snuff - that is for certain. AMD's all core boost clocks are good.

I think you picked not the best example ...

If AMD NEVER advertised all core boost clocks at all, then that is not the same as allowing mobo manufacture to pick and choose the TDP levels for the chips that will go into their boards, leaving a decent variation in consumer expectation (or surprising results perhaps) on performance, all so you can keep that low number in the specs ... its not the same comparison. And I won't mention anything about how Intel's advertised all core boost is not consistently sustained in many cases (PCWorld 3900x review - they note this and comment about it on the 9900k where they ran both processors through very long running heavy workloads).

Regardless, any shenanigans shouldn't be defended on either side.

 

[bit_user]

The point is that AMD fanboys are double-standard hypocrites who only take the high-moral ground when it comes to Intel and Nvidia. Yet, when their beloved AMD (who they want to portray as a saint or a hero) is at fault or is doing the exact same thing that they had earlier criticised Intel or Nvidia of doing, they are always quick to either give them a free pass or downplay the offence or defend it or chuck it down to “well, everyone is doing the same”. I am simply pointing out that hypocrisy which quite frankly is cancerous and I (and many other people) am sick and tired of.

You'll never win a war against hypocrisy. All you can really do is call a spade a spade and try not to equivocate or engage in partisanship, yourself. Also, try to have a sense of proportion.

 

[Rider0375]

I bought the 9900KS on release day. I've had 0 issues with cooling and/or it staying at 5ghz on all cores all the time.

Here's the but. I'm running it on a Gigabyte z390 Aorus Ultra. I ran the 9900k in it before, with no issues.

This is paired with a EVGA 2080 Ti.

I know all about core counts, tdp, etc, mainly all I can say about this processor is, it's super fast, if the board will keep it at 5.0 (Which the Gigabyte Aorus does. Easily. While staying cool). In games. I use a 2k 35inch widescreen. Most games have seen a 15-20fps jump, with only the processor being changed in my system.

It Idles anywhere from 18c-23c (Better than the 9900k), and the highest it's hit is 64c, on core 3, while playing Battlefield V with DXR and all other presets on Ultra. Usually cores run 58-61 while being stessed. Also FYI, im running the Corsair H115i, with Grizzly thermal paste.

Hope this helps in the conversation. Or to anyone looking at purchasing! Expect a 15-20 FPS gain, and nice and cool temps.

 

[TerryLaze]

The issue is that people will get greatly varying out of the box performance that will not be expected by the average consumer, dependent on which mobo they run with it. It's screwing over the consumers, for the sake of being able to lie about a TDP. Its disingenuous to the people who purchase this product.

If people can read they can get a mobo that says green or eco on the box and get base performance at or even below TDP or get one that says gaming or ultra super duper on the box and get max performance with max power draw.
Someone has to have no idea about anything ,or try to make a point really really hard,to spend $500+ on a CPU and not get the exact performance they want.

And just to be clear, the difference between a mobo that follows Intel spec and one that does not, will be multitudes greater than a chip that falls 50mhz behind on a short lived burst boost - should the outcry be equally proportional to the difference in actual performance?

Yeah the problem is not the 50Mhz but the fact that you only get the boost clock if actually all other cores are not doing anything which is the exact opposite of why you would get a system with that many cores.
Stating a max single core clock,with all other cores at zero,for a multitasking/threading machine,when the actual per core clock with all cores doing even a little work is at least 10% lower(and that is actual performance) is just outright lying.

 

[TJ Hooker]

Stating a max single core clock,with all other cores at zero,for a multitasking/threading machine,when the actual per core clock with all cores doing even a little work is at least 10% lower(and that is actual performance) is just outright lying.

Err, is this not what Intel does for virtually all of their CPUs (than have turbo boost)? Off the top of my head the only CPUs that have all core boost the same as the stated max boost are the 9900KS and I think the 7700K.

 

[joeblowsmynose]

If people can read they can get a mobo that says green or eco on the box and get base performance at or even below TDP or get one that says gaming or ultra super duper on the box and get max performance with max power draw.

My apologies. I was unaware that Intel's TDPs and specified boosting behaviour only represents their CPU in ECO mode ... clearly I just missed the advertising and fine print on this.

AMD has just launched a real "Eco" mode for zen2 that reduces listed TDP by ~40% from what they state their listed TDP is. (Not that that in itself is necessarily fully accurate either, but thats a separate topic).

If I take what you say at face value about Intels TDP being the "Eco" version of that TDP, then they are still being disingenuous about their TDP ... which is actually the point being made.

 

[joeblowsmynose]

Err, is this not what Intel does for virtually all of their CPUs (than have turbo boost)? Off the top of my head the only CPUs that have all core boost the same as the stated max boost are the 9900KS and I think the 7700K.

And only the 9900KS if MCE is on, which is not even the default with Asus mother boards. MCE on, renders the stated TDP incorrect. Its not really the same at all ...

So that leaves the 7700k as the sole example ...

 

[TJ Hooker]

And only the 9900KS if MCE is on, which is not even the default with Asus mother boards. MCE on, renders the stated TDP incorrect. Its not really the same at all ...

So that leaves the 7700k as the sole example ...

You don't need MCE on to get 5 GHz on all cores with a 9900KS, that's kind of the point. Otherwise it'd be the same as a 9900K.

 

[joeblowsmynose]

You don't need MCE on to get 5 GHz on all cores with a 9900KS, that's kind of the point. Otherwise it'd be the same as a 9900K.

If the 9900KS abides by its PL2 to maintain its stated TDP it won't boost to 5ghz for more than 28 seconds (intel guidance for having a 127w TDP). Its the mobo manufacturers that have to have settings pre-tweaked on their end to ensure that it boosts longer. This is what my complaint is. 7700k doesn't have this behaviour that requires the mobo settings to get the desired behavior.

View: https://youtu.be/aVLuKqfyVyw?t=178

 

[kinggremlin]

If the 9900KS abides by its PL2 to maintain its stated TDP it won't boost to 5ghz for more than 28 seconds (intel guidance for having a 127w TDP). Its the mobo manufacturers that have to have settings pre-tweaked on their end to ensure that it boosts longer. This is what my complaint is. 7700k doesn't have this behaviour that requires the mobo settings to get the desired behavior.

View: https://youtu.be/aVLuKqfyVyw?t=178

According to Intel, removing the boosting time limit and power limits does not take a CPU out of spec, nor is it considered overclocking. Any enthusiast/gamer targetted motherboard should have these limits removed by default as almost all of them do. The blame falls on the motherboard maker if they don't.

If you spend $600 on this CPU to swap out the one in your Dell small form factor PC, you're an idiot. If you bought one of these and dropped a $30 HSF on it, you're an idiot. The information is out there. Do your homework on your purchases.

 

[TJ Hooker]

If the 9900KS abides by its PL2 to maintain its stated TDP it won't boost to 5ghz for more than 28 seconds (intel guidance for having a 127w TDP). Its the mobo manufacturers that have to have settings pre-tweaked on their end to ensure that it boosts longer. This is what my complaint is. 7700k doesn't have this behaviour that requires the mobo settings to get the desired behavior.

Yes, I'm aware, athough boost duration is determined by the variable tau not PL2.

Regardless, adjustments to tau/PL2 are not the same as MCE.

 

[joeblowsmynose]

Regardless, adjustments to tau/PL2 are not the same as MCE.

Fair enough ... my point and issue is clarified below.

According to Intel, removing the boosting time limit and power limits does not take a CPU out of spec, nor is it considered overclocking. Any enthusiast/gamer targetted motherboard should have these limits removed by default as almost all of them do. The blame falls on the motherboard maker if they don't.
...

Actually, it 100% takes it out of its TDP spec of 127w. Either the TDP is not even close to what they state it is, or the CPU won't boost beyond 28 seconds - take your pick. Either way you look at it, there is a bit of a sleight of hand happening here.

Check the video I just posted above for further clarity - it starts at the spot where this is explained.