I9 9900k Official Reviews from Anandtech, Tomshardware. Add your own links to others !

[DrMrLordX]

I saw a system power that looked reasonable even at 5ghz but can't remember where

Probably the Blender power numbers from the AT review (initial version and revised):

https://www.anandtech.com/show/13400/intel-9th-gen-core-i9-9900k-i7-9700k-i5-9600k-review/22

197W is not that bad, certainly not 221-240W bad. Temps are a little ugly though.

 

[coercitiv]

Honest question*: is TDP any problem at all if user is not doing any rendering and DC computing? This back and forth about TDP in enthusiast forum is a bit like reading proceedings of "Bees against honey" association. As long as people have selection of chips, mbs, coolers and BIOS settings there should be no problem with TDP at all for enthusiast chips like 9900K or 2700x stock or oced.

No, it's no problem at all, and I hope it won't ever become a problem. I hope we won't ever have to discuss stock settings for top mainstream, as performance matters more for these enthusiast parts.

If we draw this line in the sand we better not conveniently wander back when it suits a cause.

 

[PotatoWithEarsOnSide]

Yet 50% of K CPU owners never overclock it.
Thanks to motherboard makers they no longer get that choice, and get a CPU that consequently operates way outside of spec.

 

[Zucker2k]

In practice a heavy AVX load under a 95W TDP adherence will yield sustained clocks of 4.2GHz, which isn't that bad if you consider a 8700K can only do 4.3GHz under the same power envelope and has 33% fewer cores.

At default TDP the 9900K is actually a very power efficient CPU (better than 2700X performance for 10% lower power draw) but the problem is that Intel wants a bigger performance lead due to the much higher price, plus mobo manufacturers want to look 'as good' as their competitors that also pegs the 9900K at full 4.7GHz boost, so you end up with the current situation...

So Intel found over 25% more power efficiency on the same node without losing clocks. How did this go unnoticed? This is essentially a "tick" and "tock" combo release, to borrow the old Intel terms. That's freaking nuts!!
It's even outperforming the 8700k in a laptop in preliminary cpu tests.

 

[PotatoWithEarsOnSide]

Are we sure that the 8700k was tested in the same manner?
Given the fact that motherboards all pushing CPUs out of sec has only been brought to attention with the 9900k, its quite possible that much before it simply was not recorded correctly.
I'm not suggesting that reviewers redo all prior reviews, just that we should take any comparisons with a grain of salt.
That being said, at 95W the 9900k most certainly is the most efficient CPU on the market.

 

[alcoholbob]

So Intel found over 25% more power efficiency on the same node without losing clocks. How did this go unnoticed? This is essentially a "tick" and "tock" combo release, to borrow the old Intel terms. That's freaking nuts!!
It's even outperforming the 8700k in a laptop in preliminary cpu tests.

Seems like these chips are binned or of a higher quality? 9900K hits 5GHz at 1.25V standard it seems. Whereas that would be a golden sample 8700K, or a typical 8086K. Most 8700Ks need something closer to 1.35V to hit 5GHz. I'm pretty sure that's where most of the gains are, it's simply better binned chips.

 

[coercitiv]

So Intel found over 25% more power efficiency on the same node without losing clocks. How did this go unnoticed? This is essentially a "tick" and "tock" combo release, to borrow the old Intel terms. That's freaking nuts!!

Such a pity this 25% increase in power efficiency only bought the 9600K a 200Mhz MT jump over the 8600K.

 

[rbk123]

Seems like these chips are binned or of a higher quality?

Hence the low availability.

 

[LTC8K6]

Such a pity this 25% increase in power efficiency only bought the 9600K a 200Mhz MT jump over the 8600K.

Currently, Newegg has the 9600K priced below RCP.

 

[Markfw]

Are we sure that the 8700k was tested in the same manner?
Given the fact that motherboards all pushing CPUs out of sec has only been brought to attention with the 9900k, its quite possible that much before it simply was not recorded correctly.
I'm not suggesting that reviewers redo all prior reviews, just that we should take any comparisons with a grain of salt.
That being said, at 95W the 9900k most certainly is the most efficient CPU on the market.

Well, not sure about that. My 2700x was "factory" set to 1.35 vcore and was running hot on my x470 Taichi. After ignoring that and manually setting it to 1.05 vcore, it ran way smoother and cooler and I am sure way less wattage. If they try the same tests on both, the results may surprise you. I don't have to time to see how low it can go.

 

[epsilon84]

Well, not sure about that. My 2700x was "factory" set to 1.35 vcore and was running hot on my x470 Taichi. After ignoring that and manually setting it to 1.05 vcore, it ran way smoother and cooler and I am sure way less wattage. If they try the same tests on both, the results may surprise you. I don't have to time to see how low it can go.

Manual undervolting isn't exactly the same as following TDP guidelines though. Most CPUs have a certain tolerance to undervolting and can obviously be more efficient as a result of that.

Btw, I still see that you're going with the '221W power!' line in the OP? Think that's been debunked as erroneous, Anandtech updated their power figures long ago.

 

[Markfw]

Manual undervolting isn't exactly the same as following TDP guidelines though. Most CPUs have a certain tolerance to undervolting and can obviously be more efficient as a result of that.

Btw, I still see that you're going with the '221W power!' line in the OP? Think that's been debunked as erroneous, Anandtech updated their power figures long ago.

1.35 is OVERvolting. Not sure what factory vcore is, since my motherboard was going 1.35, but I am sure its closer to 1.05 or less.

As for the 221 watt, I have not had time to update the OP, sorry. When I do get time, I will go look it up. I am sure everybody knows whats up currently.

Edit: I made this thread from my hospital bed 2 days after having my cancerous bladder removed while I was on pain killers. I am still recovering from that surgery. Sorry my editing of the OP does not meet your approval, I would like to see you do better in my shoes.

 

[epsilon84]

1.35 is OVERvolting. Not sure what factory vcore is, since my motherboard was going 1.35, but I am sure its closer to 1.05 or less.

As for the 221 watt, I have not had time to update the OP, sorry. When I do get time, I will go look it up. I am sure everybody knows whats up currently.

Edit: I made this thread from my hospital bed 2 days after having my cancerous bladder removed while I was on pain killers. I am still recovering from that surgery. Sorry my editing of the OP does not meet your approval, I would like to see you do better in my shoes.

Mate, no need to get all defensive, I actually just opened the first page by mistake instead of coming to the last page and realised it still said 220W, just thought for the sake of accuracy you can make it '160W power!' I'm sure it will still make for a similar point it's a hot chip regardless especially since 90% of mobos choose to ignore the TDP guidelines.

As for overvolting, some mobos are prone to that and I don't know why. I don't think a 2700X defaults to 1.05V does it? I could be wrong though

PS. Wish you all the best with your cancer treatment, please don't take any potential differences in opinion on a tech forum as a personal attack.

 

[epsilon84]

Just on the issue of different mobos exceeding TDP guidance, I wonder if it would be possible to have different power level settings in the BIOS?

Surely that can be done. At the very least we should be able to configure the chip as a true 95W part, plus 'unlimited TDP' and perhaps an intermediate TDP like 125W. That would be ideal

 

[ozzy702]

Just on the issue of different mobos exceeding TDP guidance, I wonder if it would be possible to have different power level settings in the BIOS?

Surely that can be done. At the very least we should be able to configure the chip as a true 95W part, plus 'unlimited TDP' and perhaps an intermediate TDP like 125W. That would be ideal

Agreed. I'd love to be able to set TDP limits in 25W increments.

 

[epsilon84]

Agreed. I'd love to be able to set TDP limits in 25W increments.

That plus cooling guidance for each TDP level. Probably too much to ask but something along the lines of:

95W = standard air cooling
125W = strong air cooling or 240mm AIO
Unlimited = 280mm AIO or better

 

[coercitiv]

Just on the issue of different mobos exceeding TDP guidance, I wonder if it would be possible to have different power level settings in the BIOS?

Surely that can be done. At the very least we should be able to configure the chip as a true 95W part, plus 'unlimited TDP' and perhaps an intermediate TDP like 125W. That would be ideal

Agreed. I'd love to be able to set TDP limits in 25W increments.

How about 1W increments? How about you just enter BIOS and do it!

You're wishing for something that has been available on the Intel platform for many years.

 

[Zucker2k]

1.35 is OVERvolting. Not sure what factory vcore is, since my motherboard was going 1.35, but I am sure its closer to 1.05 or less.

As for the 221 watt, I have not had time to update the OP, sorry. When I do get time, I will go look it up. I am sure everybody knows whats up currently.

Edit: I made this thread from my hospital bed 2 days after having my cancerous bladder removed while I was on pain killers. I am still recovering from that surgery. Sorry my editing of the OP does not meet your approval, I would like to see you do better in my shoes.

Hindsight is 2020, and I know you love tech but in the physical state you were in when you created this thread, you should've left it alone to more able members. I remember not too long ago when the AMD camp was in arms against @sweeper2 for the same reasons. In my frank opinion, you were only two clicks away from editing out that error but you rather chose to respond to @epsilon84. Several members have reminded you to edit the power consumption number and you've promised to but it seems this is (correctly) not a priority for you; hence the reason why I believe you should've left the thread creation alone. Thanks, and Speedy recovery @Markfw.

 

[ozzy702]

How about 1W increments? How about you just enter BIOS and do it!

You're wishing for something that has been available on the Intel platform for many years.

Where? I certainly haven't seen it in any of the Intel BIOS that I've used. Sure I can set multiplier and voltage (with several options on how to set voltage), but that's not the same as just plugging in a wattage number and the CPU receiving a stable clockspeed and voltage that directly corresponds to that TDP.

 

[Bouowmx]

Back to first introduction of Turbo Boost 2.0: DP67BG.
Open Performance tuning guide, page 18.
Power limits right under voltage override.

Today with Intel XTU, one can change power limits on the fly, while booted into Windows.

 

[ozzy702]

Back to first introduction of Turbo Boost 2.0: DP67BG.
Open Performance tuning guide, page 18.
Power limits right under voltage override.

Today with Intel XTU, one can change power limits on the fly, while booted into Windows.

That's pretty cool. I'll have to take a look. I've never played with the software.

 

[Phynaz]

As for the 221 watt, I have not had time to update the OP, sorry. When I do get time, I will go look it up. I am sure everybody knows whats up currently.

But yet you've had time to post 9 replies. I don't think it's a matter of time, I think it's a matter of agenda.

I remember when posting false information was against the rules of this forum.





Member callouts are not allowed.


esquared
Anandtech Forum Director

 

[coercitiv]

That's pretty cool. I'll have to take a look. I've never played with the software.

The software is nice for tests, but considering you probably have premium Z boards under those 8700K and 7700K, both boards should have options for configuring TDP. I can't tell you the exact naming of the settings as they likely come with different names based on mb brand, but they should be something like "Long term power limit" which corresponds to PL1 and "Short term power draw" which corresponds to PL2. They should be found in the advanced / overclocking section, since they're mainly used to enable overclocking rather than keep power consumption low.

Based on a few tests on a MSI Z170 board the minimum achievable PL1 limit for a 4C Skylake is around 25W IIRC, anyhing lower than that is a hit or miss (in the sense that setting 20W instead of 25W will get you 25W anyway).

 

[Markfw]

But yet you've had time to post 9 replies. I don't think it's a matter of time, I think it's a matter of agenda.

I remember when posting false information was against the rules of this forum.

Depending on the motherboard, and bios settings (even when not overclocked) its still true. There are articles and discussions about this all over this forum, so its not false information.

And I have no agenda.

 

[Zucker2k]

Depending on the motherboard, and bios settings (even when not overclocked) its still true. There are articles and discussions about this all over this forum, so its not false information.

And I have no agenda.

The 221 watts was in reference to AnandTech. Anandtech's figure has been revised so what justification do you have to insist on that figure? This is pure trolling, and you should know better.


Member callouts not allowed.


esquared
Anandtech Forum Director