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

[kinggremlin]

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




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.

I'll take the word Intel Fellow Guy Therien, the Chief Architect of Intel’s Client Performance Segmentation, over some guy on You Tube about what constitutes running an Intel CPU in or out of spec.

https://www.anandtech.com/show/1458...ng-an-interview-with-intel-fellow-guy-therien


"Ian Cutress: One of the things we’ve seen with the parts that we review is that we’re taking consumer or workstation level motherboards from the likes of ASUS, ASRock, and such, and they are implementing their own values for that PL2 limit and also the turbo window – they might be pushing these values up until the maximum they can go, such as a (maximum) limit of 999 W for 4096 seconds. From your opinion, does this distort how we do reviews because it necessarily means that they are running out of Intel defined spec?

Guy Therien: Even with those values, you're not running out of spec, I want to make very clear – you’re running in spec, but you are getting higher turbo duration.

We’re going to be very crisp in our definition of what the difference between in-spec and out-of-spec is. There is an overclocking 'bit'/flag on our processors. Any change that requires you to set that overclocking bit to enable overclocking is considered out-of-spec operation. So if the motherboard manufacturer leaves a processor with its regular turbo values, but states that the power limit is 999W, that does not require a change in the overclocking bit, so it is in-spec."


Again, TDP only applies to base clocks. So, if you're boosting, you aren't at base clocks and the TDP no longer matters, so you aren't out of spec. Intel does not specify TDP for boost clocks, so you can't exceed it.

 

[joeblowsmynose]

I'll take the word Intel Fellow Guy Therien, the Chief Architect of Intel’s Client Performance Segmentation, over some guy on You Tube about what constitutes running an Intel CPU in or out of spec.

https://www.anandtech.com/show/1458...ng-an-interview-with-intel-fellow-guy-therien


Ian Cutress: One of the things we’ve seen with the parts that we review is that we’re taking consumer or workstation level motherboards from the likes of ASUS, ASRock, and such, and they are implementing their own values for that PL2 limit and also the turbo window – they might be pushing these values up until the maximum they can go, such as a (maximum) limit of 999 W for 4096 seconds. From your opinion, does this distort how we do reviews because it necessarily means that they are running out of Intel defined spec?

Guy Therien: Even with those values, you're not running out of spec, I want to make very clear – you’re running in spec, but you are getting higher turbo duration.

We’re going to be very crisp in our definition of what the difference between in-spec and out-of-spec is. There is an overclocking 'bit'/flag on our processors. Any change that requires you to set that overclocking bit to enable overclocking is considered out-of-spec operation. So if the motherboard manufacturer leaves a processor with its regular turbo values, but states that the power limit is 999W, that does not require a change in the overclocking bit, so it is in-spec.

Ok I get it now ... so according to Intel allowing the motherboard manufacturers to allow the CPU to consume 999w is staying within Intel's TDP spec of 127w. Who believes this BS?

I think this is the point I have been trying to make all along ... as I said in my second post here, its just a complicated way to lie about TDP. Again, this is my point.

 

[kinggremlin]

Ok I get it now ... so according to Intel allowing the motherboard manufacturers to allow the CPU to consume 999w is staying within Intel's TDP spec of 127w. Who believes this BS?

I think this is the point I have been trying to make all along ... as I said in my second post here, its just a complicated way to lie about TDP. Again, this is my point.

TDP is only a spec for base clocks. How many times do we need to repeat this to you? If the CPU is boosting, the base clock TDP no longer applies, so it can't be exceeded.

 

[joeblowsmynose]

TDP is only a spec for base clocks. How many times do we need to repeat this to you? ....

I'm quite aware of the disingenuous TDP numbers on modern CPUs -- I'm pretty sure this is what my beef actually is here, its that and the fact that the 9900ks doesn't have consistent performance across motherboards out of the box, this is not good for consumers. This is my beef. Trying to explain this away won't change it from being the truth.

Also "tdp base clock only" appears not quite true (and who is this "we" that keeps repeating this to me?) ... did you watch the video? At ~4.4ghz all core boost (as the Asus board @stock delivers after 28seconds) the power consumption (and likely TDP at that point) is pretty much exactly 127w because they followed Intel's (apparently meaningless) TDP specifications for the CPU. At 5ghz all core boost the consumption is 200w+.

 

[TJ Hooker]

Again, TDP only applies to base clocks. So, if you're boosting, you aren't at base clocks and the TDP no longer matters, so you aren't out of spec. Intel does not specify TDP for boost clocks, so you can't exceed it.

An average power draw limit equal to the TDP is still enforced even if you're boosting. It's just that most mobo manufacturers set that averaging window as long as they can (as mentioned in your quote above), such that the CPU can exceed TDP for quite some time before eventually having to reign in power draw to stay within TDP limits.

 

[kinggremlin]

The EPA has a fuel economy rating for your car. That is the TDP rating for your vehicle. The fuel economy rating has to be accurate as Hyundai/Kia found out a few years ago. Chevy advertises their corvette can hit 200MPH. Everyone knows that you aren't going to get the advertised fuel economy when hitting 200mph. There is no fuel economy rating for your vehicle running at top speed. This is exactly the same thing for CPU's. Intel has disclaimers on the retail box and on the spec sheets on their website. Basically, your argument boils down to people being too stupid to read the packaging or manual.

 

[kinggremlin]

An average power draw limit equal to the TDP is still enforced even if you're boosting. It's just that most mobo manufacturers set that averaging window as long as they can (as mentioned in your quote above), such that the CPU can exceed TDP for quite some time before eventually having to reign in power draw to stay within TDP limits.

Sure, but if the motherboard maker sets the PL2 limit to 999W, you're never going to reach that.

 

[TJ Hooker]

Sure, but if the motherboard maker sets the PL2 limit to 999W, you're never going to reach that.

When I say TDP, I mean the CPUs rated TDP (aka PL1), i.e. 127 W for the 9900KS. According to the interview you linked earlier, the CPU can draw up to PL2 for some duration while it's turboing, but ultimately it'll claw back if necessary to keep the average power draw within stock TDP. "Average" being some exponential moving average with a length of tau. So if you were to run some intensive all core load indefinitely, the CPU will eventually reduce clock speeds no matter what you set for PL2 and tau. Based on what Guy Therien said anyway.

 

[bit_user]

If the 9900KS abides by its PL2 to maintain its stated TDP it won't boost to 5ghz for more than 28 seconds

Yes. Intel didn't promise it would sustain all-core 5.0 GHz - just that it would be attainable.

 

[kinggremlin]

So if you were to run some intensive all core load indefinitely, the CPU will eventually reduce clock speeds no matter what you set for PL2 and tau. Based on what Guy Therien said anyway.

No, because you can set both to unlimited values.
"Then we’ll look at the power that’s being consumed, and as long as it is less than your power limit PL2, you are good. So we continue on at that frequency. "

So, if you set the PL2 to 4096W, you will always be below PL2 so you will never throttle because of power usage.

"When Tau is reached, the algorithm kicks in and sees if it needs to reduce the power that’s being consumed, so it can hit can stay within the TDP limit. "

If you set Tau to unlimited (999s is the unlimited setting) you will never reach Tau. At this point, your only potential limit is thermal. So as long as you are below TJunction (100C for the 9900KS) maximum boost shouldn't drop.

 

[TJ Hooker]

If you set Tau to unlimited (999s is the unlimited setting) you will never reach Tau.

Do you have a source for 999s being equivalent to infinity for tau?

 

[joeblowsmynose]

The EPA has a fuel economy rating for your car. That is the TDP rating for your vehicle. The fuel economy rating has to be accurate as Hyundai/Kia found out a few years ago. Chevy advertises their corvette can hit 200MPH. Everyone knows that you aren't going to get the advertised fuel economy when hitting 200mph. There is no fuel economy rating for your vehicle running at top speed. This is exactly the same thing for CPU's. Intel has disclaimers on the retail box and on the spec sheets on their website. Basically, your argument boils down to people being too stupid to read the packaging or manual.

Is it unreasonable to believe that a 9900k should perform the same on all high end motherboards regardless of the vendor? I don't think this is an unreasonable expectation at all.

 

[stbean]

I recently got 2 9700k cpus and they literally ran at 5.2Ghz with just a 10 minute change to bios settings. Basically used the same settings I had on my 8700K and bumped up the frequency from 50 to 52 on all cores and manual voltage override to 1.305. They've been running them for many weeks now and I have yet to see them hit 80C when gaming, and thats about 5 or 6 hours straight a day gaming at 1440p and 2160p on the other pc. The one thing I've noticed that bothers me is their performance when recording/encoding gaming vids at high quality/high resolutions. I use OBS software to record and the result is choppy videos with lots of dropped frames compared to my 8700k recordings with hyperthreading. I think thats where the 9900KS will shine with its 16 threads.. Damn I should have waited.

 

[bit_user]

I recently got 2 9700k cpus and they literally ran at 5.2Ghz with just a 10 minute change to bios settings. Basically used the same settings I had on my 8700K and bumped up the frequency from 50 to 52 on all cores and manual voltage override to 1.305. They've been running them for many weeks now and I have yet to see them hit 80C when gaming, and thats about 5 or 6 hours straight a day gaming at 1440p and 2160p on the other pc.

Check the stepping. I wonder if they're already the same as the KS.

 

[joeblowsmynose]

I recently got 2 9700k cpus and they literally ran at 5.2Ghz with just a 10 minute change to bios settings. Basically used the same settings I had on my 8700K and bumped up the frequency from 50 to 52 on all cores and manual voltage override to 1.305. They've been running them for many weeks now and I have yet to see them hit 80C when gaming, and thats about 5 or 6 hours straight a day gaming at 1440p and 2160p on the other pc. The one thing I've noticed that bothers me is their performance when recording/encoding gaming vids at high quality/high resolutions. I use OBS software to record and the result is choppy videos with lots of dropped frames compared to my 8700k recordings with hyperthreading. I think thats where the 9900KS will shine with its 16 threads.. Damn I should have waited.

It's strongly beginning to seem like 12 threads is the new minimum for a lot of things. Gamers nexus no longer recommends any i5s at all for gaming and tout the r5 3600 as superior due to better 1% and 0.1% lows (smoother gameplay) they attribute to having 6 cores + 6 extra threads vs 6 cores without HT.

 

[kinggremlin]

Do you have a source for 999s being equivalent to infinity for tau?

Yes, and I have one that says the setting is 0, and another one that says the setting is -1. So, who knows? Everyone seems to agree there is an infinite setting while no one seems to agree what it is. I don't care what the setting is, only that it exists.

 

[kinggremlin]

Is it unreasonable to believe that a 9900k should perform the same on all high end motherboards regardless of the vendor? I don't think this is an unreasonable expectation at all.

That is a reasonable expectation, though I don't blame Intel because they don't. It's up to the motherboard manufacturer to decide default settings. I blame Asus for selling top of the line enthusiasts targeted motherboards that don't prioritize performance out of the box.

 

[joeblowsmynose]

That is a reasonable expectation, though I don't blame Intel because they don't. It's up to the motherboard manufacturer to decide default settings. I blame Asus for selling top of the line enthusiasts targeted motherboards that don't prioritize performance out of the box.

You mean you blame then for using the TDP Intel advertises to their consumers.

If the exception becomes the norm, then it'll be impossible for the average consumer to determine performance of their expensive product without a detailed knowledge of what board, what cpu, what settings, just to get advertised specifications and performance. It would be a nightmare for the average consumer to try to make sound decisions regarding performance.

We can only hope this nonsense stops.

 

[TJ Hooker]

Yes, and I have one that says the setting is 0, and another one that says the setting is -1. So, who knows? Everyone seems to agree there is an infinite setting while no one seems to agree what it is. I don't care what the setting is, only that it exists.

Err, could you share them?