14900 KS - the fastest Intel processor ever just launched

[AusWolf]

Because it is. Just like it is a good productivity CPU. If you want to claim otherwise that's your right, but this is an open forum and people will contest you on that.

They are free to do so. I stand by my opinion that if a £350 CPU with some cheap RAM gives me better gaming experience than a £550 one (that also consumes way more power), then it's better for gaming. Simple.

I do agree that an i9 has a reason to exist for people wanting to do work on it - I just think it's a waste of money and electricity for those who don't.

The height of embarrassment: to discuss the lack of efficiency of the processors you don't own, but not to criticize the efficiency of the video card you bought.

In 1440p gaming, according to the TPU review, the 7800XT beats the 4070 by only 3.7% with a 19.7% higher power consumption.

It's incredible how low you can bring the discussion. Invade Intel topics with the same impudence, although it is clear that you do not have one of the same class (alder-raptor), and some have not owned Intel in their lives, they just read other people's stories.

Who's moving the goalpost now?

If you want to know, I bought the 7800 XT because it was way cheaper than the 4070 (not to mention it has more VRAM), which makes up for its slight lack of efficiency (which is nothing compared to the 14900K vs 7800X3D). Also, I'm using it with a -10% power limit, which makes it around 3% slower. This is completely off-topic, though, so I'd stop if I were you.

Edit: Another thing is that I buy my hardware due to curiosity, not due to need. I buy whatever I think is interesting. You've been around the forum for a while, you might have known this about me.

 

[evernessince]

When both efficiency and performance were in Intel's camp, they attacked the prices. In the period 2017-2020, the multicore performance was a reason for praise for AMD owners, all of them were rendering and encoding. Now, when an i5 fights with an r7 in multicore, their "special operation" has moved to consumption and that 1FPS won in games with 30% extra price, all "just gaming".
They will always find a reason.

This is definitely a biased retelling.

When Ryzen first hit the scene Intel was still force feeding people 4-core CPUs. Ryzen hit the scene and gave people 6 and 8 core CPUs at affordable prices. It wasn't that people were suddenly encoding or rendering, it's that nearly everyone stood to benefit from the additional cores immediately. There were already games on the market that could utilize more than 4 cores and even games that didn't use more than 4 cores benefited via improved frame-times thanks to the CPU not always being pegged at 100% (which was common for Intel quad core products at the time).

On the flipside 8 cores is more than enough for running games and background apps to this day. Intel's mainstream to high end offerings run the gambit of 12 to 24 cores and in fact you are seeing zero practical benefit to those additional cores outside a few professional use cases. On top of that a good chunk of those cores are e-cores, which are useless for gaming.

There's no moving goal post as you imply, just the market clearly indicating what we all know. More than 8 cores is useless right now for the vast majority of users.

The height of embarrassment: to discuss the lack of efficiency of the processors you don't own, but not to criticize the efficiency of the video card you bought.

In 1440p gaming, according to the TPU review, the 7800XT beats the 4070 by only 3.7% with a 19.7% higher power consumption.

?? That would be completely off topic, just like your whataboutism is here. Why in the world would he start discussing his GPU out of the blue, makes no sense.

It's incredible how low you can bring the discussion. Invade Intel topics with the same impudence, although it is clear that you do not have one of the same class (alder-raptor), and some have not owned Intel in their lives, they just read other people's stories.

I sure as heck don't see you saying the same to trolls that comment on every AMD article. Or is it that you only apply this to Intel threads?

Mind you it's complete BS, someone doesn't have to own a product to comment it. In essence you are saying that you only want to listen to an echochamber.

You can argue as much as you want but the temperature readings don't lie.

I mean yeah, it tells you the temperature of the CPU and nothing more. It doesn't tell you CPU power consumption or total heat output.

It's the AMD fanboys who bring up the 7800X3D every time an Intel CPU is reviewed, conveniently because it's the only AMD part that comes close to being more efficient in both single and multi threaded than an Intel equivalent. Besides, non K series parts aren't on this chart, but you can emulate them by looking at the power limited 14900K.

View attachment 339247View attachment 339248

"Intel CPUs are more efficinet!"

*Links article with power tuned Intel processors without performance number vs untuned AMD processors.

Not even trying to hide the bias.

 

[dgianstefani]

This is definitely a biased retelling.

When Ryzen first hit the scene Intel was still force feeding people 4-core CPUs. Ryzen hit the scene and gave people 6 and 8 core CPUs at affordable prices. It wasn't that people were suddenly encoding or rendering, it's that nearly everyone stood to benefit from the additional cores immediately. There were already games on the market that could utilize more than 4 cores and even games that didn't use more than 4 cores benefited via improved frame-times thanks to the CPU not always being pegged at 100% (which was common for Intel quad core products at the time).

On the flipside 8 cores is more than enough for running games and background apps to this day. Intel's mainstream to high end offerings run the gambit of 12 to 24 cores and in fact you are seeing zero practical benefit to those additional cores outside a few professional use cases. On top of that a good chunk of those cores are e-cores, which are useless for gaming.

There's no moving goal post as you imply, just the market clearly indicating what we all know. More than 8 cores is useless right now for the vast majority of users.



?? That would be completely off topic, just like your whataboutism is here. Why in the world would he start discussing his GPU out of the blue, makes no sense.



I sure as heck don't see you saying the same to trolls that comment on every AMD article. Or is it that you only apply this to Intel threads?

Mind you it's complete BS, someone doesn't have to own a product to comment it. In essence you are saying that you only want to listen to an echochamber.



I mean yeah, it tells you the temperature of the CPU and nothing more. It doesn't tell you CPU power consumption or total heat output.



"Intel CPUs are more efficinet!"

*Links article with power tuned Intel processors without performance number vs untuned AMD processors.

Not even trying to hide the bias.

If you actually read the discussion, some... questionable claims were made about how power draw = temperature.

"bias" . Remind me what platform I'm currently using? The fact I (and others) can objectively evaluate reality rather than parroting reddit efficiency claims (by people who can't seem to grasp that consumption does not equal efficiency) is not evidence of bias, you'll need an actual argument to show that.

TPU didn't test non K high end this time around. Since the only difference is power limits, the chart literally shows both K and non K chips. We did something similar to emulate the 7800X3D before it's release by disabling one CCD of the 7950X3D, this is hardly "tuning". If objective evidence is bias, and basic power limits are "tuning", then that gives insight into your mindset. Feel free to believe whatever you want to believe, but be careful with your language.

Besides, you can make an argument that this entire power draw "discussion" is off topic. The topic is "fastest processor ever", yet some people can't seem to handle that (despite not owning or ever using the hardware being discussed) and have to nitpick at how one specific CPU (that they acknowledge isn't even a competitor) does as well in gaming while using less power, or how peak power draw (at over 40000 points in cinebench) apparently conclusively represents "efficiency". What a simplistic view. Perhaps we can do a little less insecurity over what part we bought personally, and a little more focus on the actual CPU being discussed.

I'm looking forward to revisiting the "efficiency" discussion when TPUs test bench is updated with hardware measurements of idle and low load CPU only power draw. We'll find out how Intel/AMD silicon does in that situation too and some more coffins will be nailed shut regarding "but that's not TPU data" responses to testing shown as evidence in that discussion.

 

[SPDIF]

The latest mitigations for Intel does make it slower, that's why they need 400W to make up for the decreased performance again...

The Downfall and RFDS security issue makes up for 39 percent decreased performance in the worst case...

 

[Dr. Dro]

If you actually read the discussion, some... questionable claims were made about how power draw = temperature.

"bias" . Remind me what platform I'm currently using? The fact I (and others) can objectively evaluate reality rather than parroting reddit efficiency claims (by people who can't seem to grasp that consumption does not equal efficiency) is not evidence of bias, you'll need an actual argument to show that.

TPU didn't test non K high end this time around. Since the only difference is power limits, the chart literally shows both K and non K chips. We did something similar to emulate the 7800X3D before it's release by disabling one CCD of the 7950X3D, this is hardly "tuning". If objective evidence is bias, and basic power limits are "tuning", then that gives insight into your mindset. Feel free to believe whatever you want to believe, but be careful with your language.

Besides, you can make an argument that this entire power draw "discussion" is off topic. The topic is "fastest processor ever", yet some people can't seem to handle that (despite not owning or ever using the hardware being discussed) and have to nitpick at how one specific CPU (that they acknowledge isn't even a competitor) does as well in gaming while using less power, or how peak power draw (at over 40000 points in cinebench) apparently conclusively represents "efficiency". What a simplistic view. Perhaps we can do a little less insecurity over what part we bought personally, and a little more focus on the actual CPU being discussed.

I'm looking forward to revisiting the "efficiency" discussion when TPUs test bench is updated with hardware measurements of idle and low load CPU only power draw. We'll find out how Intel/AMD silicon does in that situation too and some more coffins will be nailed shut regarding "but that's not TPU data" responses to testing shown as evidence in that discussion.

On the 14900KS review thread I asked W1zzard if he was interested on creating a mini-review that compared the 13900KS and 14900KS (possibly K variants too) at iso clocks to see which chip could exhibit the best v/f characteristics, but he did not reply

I'd really want some data on this but unless there's someone who's enough of a lunatic to upgrade from the 13900KS to the 14900KS (it'd be like upgrading from a i7-980X to a i7-990X, or a 3960X to a 3970X, absolute insanity to even consider), and that this person is fully qualified to take these measurements, I just don't think i'll come across that, and my wallet isn't deep enough to be able to take that one for the team

 

[AusWolf]

Besides, you can make an argument that this entire power draw "discussion" is off topic. The topic is "fastest processor ever", yet some people can't seem to handle that

So if a thread title makes a claim, then commenters are not allowed to examine the situation from multiple points of view?

I'm looking forward to revisiting the "efficiency" discussion when TPUs test bench is updated with hardware measurements of idle and low load CPU only power draw. We'll find out how Intel/AMD silicon does in that situation too and some more coffins will be nailed shut regarding "but that's not TPU data" responses to testing shown as evidence in that discussion.

Here you go:

 

[AusWolf]

You must have misread what I wrote, or are trying to prove something.

"idle" or "low load" ≠ maximum synthetic single core load.

Incidentally, the efficiency of the state at which most consumer CPUs around the world tend to be at most of the time, is, actually, kind of important.

It's not idle, but Cinebench 1T, or gaming in general, is definitely a low load scenario in my books.

I agree that incorporating idle power consumption tests would be nice, although a bit difficult to objectively accomplish (what is true idle, what programs do you have installed, what OS features are running, etc.).

 

[dgianstefani]

Here you go:
View attachment 339263View attachment 339264

You must have misread what I wrote, or are trying to prove something that's unrelated.

"idle" or "low load" ≠ maximum synthetic single core load.

Incidentally, the efficiency of the state at which most consumer CPUs around the world tend to be at most of the time, is, actually, kind of important.

So if a thread title makes a claim, then commenters are not allowed to examine the situation from multiple points of view?

Points of view are fine, as has been pointed out many times throughout this thread. Ignoring factual evidence that doesn't support your opinion and going on about your own CPU preferences/needs in a thread about a specific CPU (which makes a lot of sense for people who need/want both gaming and workstation performance, but don't want to drop thousands on HEDT, which will actually be slower in most cases) is another thing. You haven't used this chip, you don't have the use case for this chip, yet you feel compelled to tell everyone how bad it is because it doesn't fit your use case/budget?

It's not idle, but Cinebench 1T, or gaming in general, is definitely a low load scenario in my books.

I agree that incorporating idle power consumption tests would be nice, although a bit difficult to objectively accomplish (what is true idle, what programs do you have installed, what OS features are running, etc.).

It's not difficult at all to objectively accomplish.

Simple at the wall measurements already exist, where Zen platforms identical except for CPU/Mobo (for obvious reasons) idle at 30-40 W more than Intel platforms. The way TPU will test will be more precise, using hardware wireview mobo/CPU measurements at the power connectors for those components.

It's also interesting how none of the non K Intel Processors are in that chart. Yet the X3D chips, which are essentially underclocked, undervolted normal Zen chips with some extra cache, are.

That's why I used the single thread/multithread benchmark charts, they show how even a 12100f is ridiculously efficient compared to any Zen parts, besides the monolithic APUs which suck for gaming anyway due to gimped cache.

Even a 10400, a Skylake based part more than four years old, has better efficiency.

Now when you consider a power limited 14900K, ie a 14900 (identical silicon remember, besides VF curve out of the box), things start to become clearer.

 

[Vya Domus]

This is TPU. We should assume a higher technical dialog than the layperson.

That would be a baseless assumption unfortunately. Some time ago there was a discussion about how Nvidia GPUs run "cooler", because the temperatures reported are lower and I made the point that this doesn't mean anything because we do not know what those figures represent, what exactly they are measuring and it makes no sense to compare GPUs of different make like that but no : "smaller number on screen = cooler" was the prevailing belief.

 

[dgianstefani]

That would be a baseless assumption unfortunately. Some time ago there was a discussion about how Nvidia GPUs run "cooler", because the temperatures reported are lower and I made the point that this doesn't mean anything because we do not know what those figures represent, what exactly they are measuring and it makes no sense to compare GPUs of different make like that but no : "smaller number on screen = cooler" was the prevailing belief.

They run cooler because they're more efficient. End of story. Off topic anyway. This is a thread about the 14900KS.

 

[Vya Domus]

They run cooler because they're more efficient.

If two processors consume the same amount of power they'll dissipate the same amount of heat, meaning they're both just as cool or hot depending how you look at it, it doesn't matter if one is a million times more efficient. The measure of performance/watt is totally irrelevant in determining what runs "cooler". And it makes no difference anyway, the temperatures would differ depending on how they're sampled.

See, point taken. Never assume that.

 

[Denver]

You must have misread what I wrote, or are trying to prove something that's unrelated.

"idle" or "low load" ≠ maximum synthetic single core load.

Incidentally, the efficiency of the state at which most consumer CPUs around the world tend to be at most of the time, is, actually, kind of important.


Points of view are fine, as has been pointed out many times throughout this thread. Ignoring factual evidence that doesn't support your opinion and going on about your own CPU preferences/needs in a thread about a specific CPU (which makes a lot of sense for people who need/want both gaming and workstation performance, but don't want to drop thousands on HEDT, which will actually be slower in most cases) is another thing. You haven't used this chip, you don't have the use case for this chip, yet you feel compelled to tell everyone how bad it is because it doesn't fit your use case/budget?


It's not difficult at all to objectively accomplish.

Simple at the wall measurements already exist, where Zen platforms identical except for CPU/Mobo (for obvious reasons) idle at 30-40 W more than Intel platforms. The way TPU will test will be more precise, using hardware wireview mobo/CPU measurements at the power connectors for those components.

It's also interesting how none of the non K Intel Processors are in that chart. Yet the X3D chips, which are essentially underclocked, undervolted normal Zen chips with some extra cache, are.

That's why I used the single thread/multithread benchmark charts, they show how even a 12100f is ridiculously efficient compared to any Zen parts, besides the monolithic APUs which suck for gaming anyway due to gimped cache.

Even a 10400, a Skylake based part more than four years old, has better efficiency.

Now when you consider a power limited 14900K, ie a 14900 (identical silicon remember, besides VF curve out of the box), things start to become clearer.

I agree with this guy's hidden wisdom. Anyone recommending Quad-cores for gaming is really ridiculous, if you want to waste time looking at the loading screens (shader compilation) buy an HDD too. it's akin to handing someone a whip and suggesting they flagellate themselves.

 

[AusWolf]

Points of view are fine, as has been pointed out many times throughout this thread. Ignoring factual evidence that doesn't support your opinion and going on about your own CPU preferences/needs in a thread about a specific CPU (which makes a lot of sense for people who need/want both gaming and workstation performance, but don't want to drop thousands on HEDT, which will actually be slower in most cases) is another thing. You haven't used this chip, you don't have the use case for this chip, yet you feel compelled to tell everyone how bad it is because it doesn't fit your use case/budget?

No. I feel compelled to acknowledge that it's the fastest CPU in MT only, while its use case for a gaming only scenario is highly questionable due to its inefficiency and high price.

It's not difficult at all to objectively accomplish.

Simple at the wall measurements already exist, where Zen platforms identical except for CPU/Mobo (for obvious reasons) idle at 30-40 W more than Intel platforms. The way TPU will test will be more precise, using hardware wireview mobo/CPU measurements at the power connectors for those components.

It's also interesting how none of the non K Intel Processors are in that chart. Yet the X3D chips, which are essentially underclocked, undervolted normal Zen chips with some extra cache, are.

That's why I used the single thread/multithread benchmark charts, they show how even a 12100f is ridiculously efficient compared to any Zen parts, besides the monolithic APUs which suck for gaming anyway due to gimped cache.

Even a 10400, a Skylake based part more than four years old, has better efficiency.

Now when you consider a power limited 14900K, ie a 14900 (identical silicon remember, besides VF curve out of the box), things start to become clearer.

It's difficult to accomplish due to no clear definition on what "idle" really means.

A non-K CPU is different, because it's not an underclocked, undervolted version, just one with a lower power limit. At least my 11700 non-K is. The TPU review of the 14900K at various power limits is good for this.

 

[dgianstefani]

If two processors consume the same amount of power they'll dissipate the same amount of heat, meaning they're both just as cool or hot depending how you look at it, it doesn't matter if one is a million times more efficient. It doesn't matter anyway, the temperatures would differ depending on how they're sampled.

See, point taken. Never assume that.

It absolutely matters.

Two processors may consume the "same" amount of (peak) power, but if one of those two is 20%, 50% or whatever amount faster, it will spend less time at full power state. Hence "efficiency". AMD CPUs and GPUs are typically less efficient at partial load (where the average system resides for most of its existence outside of synthetic benchmarks), and for GPUs, both full and partial load. You can easily see this by looking at GPU power testing. Here's some nice charts for your convenience.
If you look closely, you may notice the order within them isn't the same. I wonder why?

For CPUs, to stay on topic, it's the same story. Idle/load aren't related.

AusWolf said:
A non-K CPU is different, because it's not an underclocked, undervolted version, just one with a lower power limit. At least my 11700 non-K is. The TPU review of the 14900K at various power limits is good for this.

The one I linked and people cried about "tuned" vs stock comparisons?

 

[R-T-B]

That would be a baseless assumption unfortunately. Some time ago there was a discussion about how Nvidia GPUs run "cooler", because the temperatures reported are lower and I made the point that this doesn't mean anything because we do not know what those figures represent, what exactly they are measuring and it makes no sense to compare GPUs of different make like that but no : "smaller number on screen = cooler" was the prevailing belief.

I famously remember when some users believed a lot of fans effectively made their rigs into magical air conditioners too.

Point taken.

 

[Vya Domus]

It absolutely matters.

No it doesn't if the subject at hand is which runs "cooler". Cooler/hotter isn't and has never been a measure of efficiency, the thing which generates the least amount of heat is the "coldest", that's it. That's why these types of discussions are so dumb, most people have literally no clue what they mean when they say "hotter/cooler"

 

[dgianstefani]

No it doesn't if the subject at hand is which runs "cooler". Cooler/hotter isn't and has never been a measure of efficiency, the thing which generates the least amount of heat is the "coldest", that's it. That's why these types of discussions are so dumb, most people have literally no clue what they mean when they say "hotter/cooler"

Nope, wrong again.

A stock 14900K/KS runs cooler than equivalent Zen CPUs under the same 360 mm AIO. Why? Because the bottleneck is the AIO, not the thermal transfer efficiency between the silicon and the coolant, with all the steps in between, IHS, TIM, silicon thickness etc., like it is for Zen. Hence why a 14900KS gets 85 C under full load at 375 W, yet a 7950X at 100-150 W is running at 95 C, it's "designed" temperature goal.

You're confusing wattage with temperature, like others have done. "Cooler" and "hotter" are relative terms used to describe temperature, not power usage.

I don't care about the amount of hot air going out of my case, I care about the temperature of my components.

 

[AusWolf]

The one I linked and people cried about "tuned" vs stock comparisons?

Yes, that one.

You can limit your 14900K to ridiculous power levels, like 35 W, which will throw your gaming efficiency through the roof, but your performance out of the window. At that level, you're better off with a 13400 or a 7600, which are much cheaper as well.

 

[dgianstefani]

Yes, that one.

View attachment 339271View attachment 339272

You can limit your 14900K to ridiculous power levels, like 35 W, which will throw your gaming efficiency through the roof, but your performance out of the window. At that level, you're better off with a 13400 or a 7600, which are much cheaper as well.

Ah yes, goalposts. For the man who can't/won't understand the 14900K isn't exclusively a gaming CPU, but a productivity one too. Keep linking only one half of the story though.

 

[AusWolf]

For the man who can't/won't understand the 14900K isn't exclusively a gaming CPU, but a productivity one too. Keep linking only one half of the story though.

Obviously - I thought we'd already agreed on that, so I didn't feel the need to acknowledge it again.

 

[dgianstefani]

Obviously - I thought we'd already agreed on that, so I didn't feel the need to acknowledge it again.

So why are you still trying to make arguments based exclusively off gaming performance when the discussion is about efficiency? Overall efficiency, in productivity too, not just gaming, at that.

I'm done here anyway, got better things to do and I've said all that needs to be said. You guys have fun.

 

[Vya Domus]

I don't care about the amount of hot air going out of my case, I care about the temperature of my components.

Thanks for being forward and admitting you are totally uneducated on this matter.

Two objects can emit the same amount of heat but have different temperatures. The amount of hot air going out of your case is always the same if the components consume the same amount of power, no matter what the temperatures are. Heat and power, as correctly pointed out before, are almost perfectly correlated, heat and temperature are not.

 

[Dr. Dro]

Now when you consider a power limited 14900K, ie a 14900 (identical silicon remember, besides VF curve out of the box), things start to become clearer.

This reminds me, once the microcode with the ability to disable the power excursion protection (which is a mechanism to avoid sudden reboots undervolted) rolls out to more motherboards, we may be able to get even better figures with Raptor here. Flagship Z790s got the BIOS this week, I know at least MSI plans to release it next week for the Z690's.

 

[AusWolf]

So why are you still trying to make arguments based exclusively off gaming efficiency?

I'm talking about what I think is worth talking about. The 14900K(S) is the fastest productivity CPU. Everybody knows that. What purpose would parroting it serve?

 

[Vya Domus]

The 14900K(S) is the fastest productivity CPU.

Fastest "consumerish" CPU, a threadripper will totally outclass it, in the same 400W power envelope and by a lot.