Please can your share the Voltage/Frequency chart?I've already mapped the minimum V/F curve for this chip in the past, but I added 25mV margin (1.070 >> 1.095V) for 24/7 reliability for this one.
~105W average in non-256b workloads.
https://techreport.com/review/34192/intel-core-i9-9900k-cpu-reviewed/13
Better performance/watt than 2700X:
For our testing, we use POV-Ray as our load generator then take the register values for CPU power. This software method, for most platforms, includes the power split between the cores, the DRAM, and the package power. Most users cite this method as not being fully accurate, however compared to system testing it provides a good number without losses, and it forms the basis of the power values used inside the processor for its various functions.
Starting with the easy one, maximum CPU power draw.
That is the most bogus efficiency comparison I have ever seen!So, Intel raised the performance bar again, but that was to be expected. I'm especially happy to see they didn't trip a breaker while achieving it. The icing on the cake is actually seeing the 9900k @4.7Ghz drawing essentially the same power as the 2700x @4.2Ghz. I got into trouble for pointing out how inefficient the 2700x is when it launched so I feel vindicated about that. Overall, great job by Intel. Personally, I would've loved to see the 9900k at $399, but hey, it's Intel. Top of the line performance doesn't come cheap after all.
This sums the 9900k up, it is plain ridiculous to label this processor a 95w CPU, even taking into account some of the over zealous power numbers (from two reviews I've read)...if you love gaming at the best FPS...buy an 8700k.
It looks like the game "pulls"the power from the CPUs. As if Ryzen if it has more IPC uses the same power. Strange.Technically, that is not true. Gaming hardly stresses a 9900K, in fact power consumption in games is practically identical to a 8700K:
If and when games can fully load an 8C/16T CPU at 100% usage (or close to), then we might start seeing some of the thermal struggles exhibited in other MT apps. That's probably not going to happen in the useful lifespan of the 9900K.
If you own a GTX 2080 Ti and want the absolute best fps? A 9900K is it. The 8700K and in particular 9700K come mighty close though, and is better value. But my point is that you won't hit the thermal issues in gaming, not with a 70W load.
Maybe Mark can update his OP now? It's not 221W under load... it's interesting how nobody thought the Anandtech numbers were off, but instead thought the other reviewers showing 'proper' power numbers were the 'outliers'.
You know what is interesting. Either Intel has succeeded in another refinement of the almost perfect 14nm process they have. Or other numbers are strange as well.
The 8700K draws 150 watts. If we take ball park numbers, that is 25watts per core.
The 9900K has 2 more cores but only draws ~9 watts per core extra.
The 9900K draws about 21 watts per core now.
There are so many variables here that it is hard to see what is the reality.
What kind of software, ddr4 memory specs, type of motherboard, type of heatsink, process refinement, core clocks, core voltage, i/o voltage. Memory voltage.
Variation from die to die.
For a proper analysis, every added variable must be controlled and kept the same and logged as well to be able to see a trend.
Otherwise it remains difficult to see what is the culprit.
For Blender yes, but if we take Pov-Ray the Ryzen 2700X has better perf/watt than 9900K
Pov-Ray - From AT review
Core i9 9900K = 5542 / 168.48W = 32.89
Ryzen R7 2700X = 4024 / 117.18W = 34.34
So, it depends on the application.
https://www.anandtech.com/show/13400/intel-9th-gen-core-i9-9900k-i7-9700k-i5-9600k-review/21
Lastly is a look at the AC system power consumption collected over the wide variety of benchmarks carried out. The Core i9 9900K with ASUS PRIME Z370-A and RX Vega 64 had an average AC power draw of 224 Watts and a peak of 386 Watts. The Ryzen 7 2700X with ASUS CROSSHAIR VII HERO motherboard and RX Vega 64 had a average power draw of 254 Watts and a peak of 420 Watts.
The Core i9 9900K was being tested with an Arctic Freezer 12 cooler since no stock heatsink is bundled. The Core i9 9900K had an average temperature under load of 44 degrees and a peak of 64 degrees while at idle was about 30 degrees. The reference AMD Ryzen 7 2700X results are with its stock heatsink.
Right, so tdp should only be calculated off of gaming tests? Nobody uses these things for anything else?Technically, that is not true. Gaming hardly stresses a 9900K, in fact power consumption in games is practically identical to a 8700K:
If and when games can fully load an 8C/16T CPU at 100% usage (or close to), then we might start seeing some of the thermal struggles exhibited in other MT apps. That's probably not going to happen in the useful lifespan of the 9900K.
If you own a GTX 2080 Ti and want the absolute best fps? A 9900K is it. The 8700K and in particular 9700K come mighty close though, and is better value. But my point is that you won't hit the thermal issues in gaming, not with a 70W load.
Right, so tdp should only be calculated off of gaming tests? Nobody uses these things for anything else?
Tdp used to..and should...be a number that indicated what kind of cooling you will need whilst also roughly indicating the power consumption so you can plan your PSU and system requirements out...all worst case.
Take the intel 6900k...it was a 140w rated tdp processor...in typical use including gaming it didn't use near 140w...but using AVX2 heavy apps it could max out at that...so it was correctly labelled and a consumer could plan out his/her build appropriately.
Sometimes in the past you could coax certain processors to exceed tdp rating slightly, some intel processors, ryzen 1800x and 2700x certainly...(should have been 105w and 140w tdp respectfully)..but this 9900k has taken things to the next level of absurdity.
Calculating tdp based on base clock is bordering on deciet, there is no other way to put it.
Let's not make excuses for them by cherry picking apps that show it in it's best light (such as games), 2700x mostly sticks to it's tdp rating but can be stress tested to exceed it's 105w tdp.
9900k is way beyond its 95w tdp, to get anywhere near its touted potential you need a very needy water cooler and exceed it tdp by at least 50%...it is becoming a joke now.
To be fair, I find those 8700K figures hard to believe as well, as I own one and I'm pretty sure it's not pulling 150W at stock, especially since I'm running it on a modest CM Hyper 212+ HSF that would struggle with a true 150W load.
FWIW, with my 8700K at stock the 'package power' in XTU remains at ~95W, when overclocked to 5.0GHz it's at ~140W.
My suspicion is that the 8700K numbers there suffer from the same overvoltage scenario as the 9900K did previously. I bet if they reran it on the Z390 motherboard you'll get closer to 100W rather than 150W.
Show me where the 9900k draws a maximum of 140w in a AVX2 stress test and I will agree with that.So if Intel had rated the 9900K as a 140W TDP chip, you would have no complaints? Is that the crux of the issue here, the 95W TDP?
So in summary, the numbers are all over the place.
Is it any wonder why Intel expected to get away with the PT FUD?
The man on the street buys a 9900k, plugs it into his existing motherboard, and either it fails to boot or it gives worse performance than what he was replacing. It seems as though without exactly the right components you'll never get this CPU running at the performance level that it is being advertised as. As for price comparisons, there's no point in even comparing it to a 8700k since what you'll need under the hood is going to make the overall system cost a good 50% more even if both builds have a 2080ti.
The 9900k is simply a ridiculous CPU.
For pure gaming performance you'll be wanting any one of the 9700k, 9600k, 8700k, 8086k, or the 2700x. Much of a muchness depending on your own preferences and/or upgrade mentality.
Any reviewer that actively advises that you buy a 9900k has truly lost their mind. It's a bloody unicorn.
Show me where the 9900k draws a maximum of 140w in a AVX2 stress test and I will agree with that.
Well Tom's did a prime 95 stress test using AVX..https://www.gamersnexus.net/hwrevie...paste-delid-gaming-benchmarks-vs-2700x/page-2
Not exactly a 'stress test' but Blender is pretty intensive in its own right:
With a rise of 37 degree above ambient on a 280mm AIO:
https://www.gamersnexus.net/images/media/2018/cpus/9900k/intel-9900k-soldered-thermals.png
This would equal 58 degrees load in a 21C room.
LinkFor our testing, we use POV-Ray as our load generator then take the register values for CPU power. This software method, for most platforms, includes the power split between the cores, the DRAM, and the package power. Most users cite this method as not being fully accurate, however compared to system testing it provides a good number without losses, and it forms the basis of the power values used inside the processor for its various functions.
Well Tom's did a prime 95 stress test using AVX..
"But power becomes more of an issue in some productivity applications because a constant load on all cores at high clock rates is almost too much. And to be clear, the Core i9-9900K gets super hot faced with Prime95 and AVX instructions (205W stock, 250W overclocked), exceeding the specified TDP."
https://www.tomshardware.com/reviews/intel-core-i9-9900k-9th-gen-cpu,5847-11.html
So according to them 205w should be it's tdp..if the power numbers are not off then that is the correct tdp for this processor...205w!...more than double it tdp rating.
But in fairness the average consumer would get away with a 140w tdp rating, which is still absurdly past it's tdp rating, and I wouldn't agree even with that, more inline with Tom's.
TR 1950X is $680 on Amazon and Newegg, i9 9900K is $580 but not available on neither major retailers.Also, @mikk and @epsilon84 called how competitive the 9900k at 5GHz was going to trade blows with the TR 1920X in multithreaded scenarios.
Remember when I kept nagging you about the power consumption at 4.7Ghz and beyond? Remember when both you and @JoeRambo told me neither TDP or power consumption would be an issue for enthusiasts. Well, here's the thing about enthusiasts: once they expect something to happen, it better happen the way they've been made to expect it.What did people really expect adding an extra 33% cores + 9% frequency to a 8700K would do to power draw?
TR 1950X is $680 on Amazon and Newegg, i9 9900K is $580 but not available on neither major retailers.
Remember the mantra: once you factor in the rest of the system the price difference is peanuts.
I think you are understating the issue, some consumers are going to be spending $500+ on this CPU, buying a respectable 120w air cooler and getting a big shock when some of the tasks are not getting full performance they expected.I think a 140W TDP would have sufficed, AFAIK Intel has never stated TDP as a 'worst case' measurement.
So I agree with you that a 95W TDP is a joke, but if that is the biggest issue you have with the chip, then it's not half as bad as half the posters here are claiming it to be. Especially since its 'overall' performance is more in line with 10C/12C HEDT CPUs. https://www.computerbase.de/2018-10/intel-core-i9-9900k-i7-9700k-cpu-test/2/
Its clearly a tier about the 2700X/9700K/8700K so I think the higher power consumption is justified. What did people really expect adding an extra 33% cores + 9% frequency to a 8700K would do to power draw?
Excuse me, where's the overclock?That is the most bogus efficiency comparison I have ever seen!
Take one processor overclocked to within an inch of its life, then compare it to your preferred processor using it's stock turbo...ok then!.