dgianstefaniPlus the idle or low load draw is much better than Zen, which never goes below 30-40 W, depending on your IF/RAM settings. It's only balls to the wall, all 24 cores loaded etc, that the 300 W+ numbers people like to throw about wantonly show up. But then it's also faster (and often running cooler) than a 7950X/X3D in that case too, so...

Yeah, and I'm pretty confident that a 7950x (cause the 7950x 3d can't anyways) will need a heck of a lot more power than the 14900k to hit 43k in cbr23.

So the same price as the 13900KS. Not too bad. Except this time around it's not exactly a new product. Oh well.

Dr. DroSo the same price as the 13900KS. Not too bad. Except this time around it's not exactly a new product. Oh well.

Hope someone reviews it properly before it's too late and you can't buy one anymore. Happened with the 13900ks, until I could find some decent reviews (iso wattage) it went poof.

fevgatosHope someone reviews it properly before it's too late and you can't buy one anymore. Happened with the 13900ks, until I could find some decent reviews (iso wattage) it went poof.

If you need a review first, these aren't for you ;)

bugIf you need a review first, these aren't for you ;)

I share his interest in the binning quality of this processor. I'd like to see how it fares against the standard 13 and 14th K as well as the 13900KS

dgianstefaniPlus the idle or low load draw is much better than Zen, which never goes below 30-40 W, depending on your IF/RAM settings. It's only balls to the wall, all 24 cores loaded etc, that the 300 W+ numbers people like to throw about wantonly show up. But then it's also faster (and often running cooler) than a 7950X/X3D in that case too, so...

Both the 14900k and 7950X have similar idle consumption:

Both the 14900k and 7950X have similar idle consumption:

"idle consumption"

Links a picture of ST 100% load power consumption -

The chart you're referencing shows that under single core full load, the 14900K and 7950X/X3D have similar power draw, with the Zen chips consuming a bit more.

The "idle" stat I mentioned is related to when the CPU is not working at 100% load, ST or MT. In this case, Zen idles at 30-40 W and Intel is sub 5 W.

If you meant to demonstrate how Intel has superior ST efficiency, then that's certainly true, but Intel also has significantly superior idle consumption, which is actually what most CPUs do most of the time, unless you run 24/7 benchmarks.


You can check this by running some monitoring software and simply observing the total CPU power (not the core power or any of the other breakdown stats), in fact the cores on Zen typically use less than 10 W, often just 3-4 W. Unfortunately the non monolithic architecture means the IO die, which is also made on an older, less efficient process, is always active. Intel has a monolithic design so it's much easier to idle efficiently.

bugIf you need a review first, these aren't for you ;)

In what way? If they are just 5% better binned than the average K then these are useless. How will you know about the bin quality without an actual review?

dgianstefani"idle consumption"

Links a picture of ST 100% load power consumption -

The chart you're referencing shows that under single core full load, the 14900K and 7950X/X3D have similar power draw, with the Zen chips consuming a bit more.

The "idle" stat I mentioned is related to when the CPU is not working at 100% load, ST or MT. In this case, Zen idles at 30-40 W and Intel is sub 5 W.

Then link the review showing a 14900K getting 5w on idle. Otherwise that which is submitted without evidence can be dismissed without evidence.

dgianstefaniIf you meant to demonstrate how Intel has superior ST efficiency, then that's certainly true, but Intel also has significantly superior idle consumption, which is actually what most CPUs do most of the time, unless you run 24/7 benchmarks.

Except if you don't cherry pick and look at the whole chart you can see the answer is a lot more nuanced:



To say that Intel has superior ST efficiency is factually incorrect. It really depends on the CPU models we are comparing.

ST efficiency isn't that important either, most workloads are going to utilize more than a single thread of which Ryzen is vastly more efficient in.

I don't really see the point of your argument, in essence you are saying if you purchase the 24-core 14900K and only use a single thread you'll have slightly better efficiency. That's great unless you ignore the obvious waste of money that is. Who is buying a 24-core CPU to only be running a single thread most of the time? It's nonsense, yes the 14900K is more efficient if you only run a single thread which is precisely the least likely scenario for people buying this CPU.

bugI know people think peak power draw is what happens most of the time. But the reality is:

Less than 50W between stock 14900K and stock 7950X. It's still a 40% difference, just not as big, in absolute numbers, as some people think.
And while I like my CPUs well behaved (I run my 12600k power constrained), I imagine that wouldn't be a concern for someone buying the absolute highest end.

It's actually a 35% difference between the 14900K and 7950X. Mind you that's a generous comparison to Intel, if you compare to the 7950X3D the 14900K consumes 130% more power.

fevgatosIn what way? If they are just 5% better binned than the average K then these are useless. How will you know about the bin quality without an actual review?

Bin quality will vary between chips, as with all CPUs. The only guarantee is that the KS will run 6.2 GHz below a certain voltage Intel has decided is appropriate. In general, this means the silicon is better quality, but it doesn't guarantee that every KS has a better OC potential than a random K series, or a better memory controller etc., it's just more likely to be the case.

evernessinceThen link the review showing a 14900K getting 5w on idle. Otherwise that which is submitted without evidence can be dismissed without evidence.



Except if you don't cherry pick and look at the whole chart you can see the answer is a lot more nuanced:



To say that Intel has superior ST efficiency is factually incorrect. It really depends on the CPU models we are comparing.

ST efficiency isn't that important either, most workloads are going to utilize more than a single thread of which Ryzen is vastly more efficient in.

I don't really see the point of your argument, in essence you are saying if you purchase the 24-core 14900K and only use a single thread you'll have slightly better efficiency. That's great unless you ignore the obvious waste of money that is. Who is buying a 24-core CPU to only be running a single thread most of the time? It's nonsense, yes the 14900K is more efficient if you only run a single thread which is precisely the least likely scenario for people buying this CPU.



It's actually a 35% difference between the 14900K and 7950X. Mind you that's a generous comparison to Intel, if you compare to the 7950X3D the 14900K consumes 130% more power.

You'll have slightly better per core efficiency (what ST is) but much better idle efficiency, which, like I said, is what the CPU is doing most of the time. Load a new webpage - straight to idle, open your mail - straight to idle. Etc.

You're talking about cherry picking results, while failing to mention that in the instance where the Intel CPU is using "35%" more power (in 100% all core synthetic load, at a whole 42 W more than it's competitor, the 7950X), it's also doing more work, faster.

Since you misquoting TPU's test of 100% single core load as "idle" power apparently puts the burden of proof on me, as requested, here's a review that compares CPUs at idle/load power.

hothardware.com/reviews/amd-ryzen-5-7600-ryzen-7-7700-and-ryzen-9-7900-65w-review?page=3

As you can see, the 13900K system idles at 67 W, whereas the otherwise identical 7950X system idles at 107 W. Even the six core 7600X still idles at 89 W, with single core load being just 18 W higher, lending further evidence to my statements.

You can also, as I suggested, simply observe your systems power draw to reach the same understanding.



Not only that, but since hothardware.com tests using manufacturer spec settings, the Zen 4 parts used 5200 MT memory while the 13th gen Intel parts used 5600 MT memory, so if you used 6000 MT "sweetspot" RAM in the same Zen 4 build, the powerdraw would be yet higher, since IF/IO die would be consuming more power.

More evidence if you still don't want to believe me. Identical systems other than CPU, and mobo/RAM, for obvious reasons, since you can't plop a 13900K into an X670 motherboard.

Here's the image in a better format, they use .webp for some reason.


Nuanced?

evernessinceExcept if you don't cherry pick and look at the whole chart you can see the answer is a lot more nuanced:



To say that Intel has superior ST efficiency is factually incorrect. It really depends on the CPU models we are comparing.

Nuanced? As in a two generation old 12600K is more efficient than the entire Zen 4 lineup besides the 7800X3D?

Doesn't seem very nuanced to me.

That 7800X3D (an undervolted 7700X with some cache on top) is still beaten by a last generation 13400 by the way (an undervolted 12600K)

By 154.8 to 106.2 too, a significant lead.

So what exactly is "factually incorrect" about the statement that Intel has superior ST efficiency?

fevgatosThis is exactly what it is. A cpu binned for power efficiency. Lock it to the same clockspeeds as a 13900k for example and it will drop power draw probably by around 100 watts.

Do you have one in hand and have you tested that to make a claim that it will use 100W(?!) less power at the same clocks, because that is an astounding number.

This is for that customer who wants " The Most Expensive CPU available".
Trust me, they're out there.

Dr_b_Do you have one in hand and have you tested that to make a claim that it will use 100W(?!) less power at the same clocks, because that is an astounding number.

I compared a 14900k to a 13900k, both were pretty terrible bins, the 14900k can match the clockspeeds at 70w less power (280w vs 350w). Those numbers are not undervolted, just using the built in VID.

evernessinceThen link the review showing a 14900K getting 5w on idle. Otherwise that which is submitted without evidence can be dismissed without evidence.



Except if you don't cherry pick and look at the whole chart you can see the answer is a lot more nuanced:



To say that Intel has superior ST efficiency is factually incorrect. It really depends on the CPU models we are comparing.

ST efficiency isn't that important either, most workloads are going to utilize more than a single thread of which Ryzen is vastly more efficient in.

I don't really see the point of your argument, in essence you are saying if you purchase the 24-core 14900K and only use a single thread you'll have slightly better efficiency. That's great unless you ignore the obvious waste of money that is. Who is buying a 24-core CPU to only be running a single thread most of the time? It's nonsense, yes the 14900K is more efficient if you only run a single thread which is precisely the least likely scenario for people buying this CPU.



It's actually a 35% difference between the 14900K and 7950X. Mind you that's a generous comparison to Intel, if you compare to the 7950X3D the 14900K consumes 130% more power.

Although I kinda agree that ST efficiency isn't very important, the graph you posted shows an incredible efficiency advantage for the Intel cpus. The context you are missing is that the Intel CPUs are more efficient while being FASTER. In order to make eg. the 7950x as fast as the 14900k in ST workloads, the gap in efficiency would skyrocket.


Regarding idle / browsing the web etc. my experience from a 7950x is that even just browsing youtube comments power draw hits 45w average and it peaks as high as 65 (!!!!). A 12900k browsing the web while watching 2 videos (youtube and twitch) draws 7 watts. The difference is pretty insane.

fevgatosI compared a 14900k to a 13900k, both were pretty terrible bins, the 14900k can match the clockspeeds at 70w less power (280w vs 350w). Those numbers are not undervolted, just using the built in VID.


Although I kinda agree that ST efficiency isn't very important, the graph you posted shows an incredible efficiency advantage for the Intel cpus. The context you are missing is that the Intel CPUs are more efficient while being FASTER. In order to make eg. the 7950x as fast as the 14900k in ST workloads, the gap in efficiency would skyrocket.


Regarding idle / browsing the web etc. my experience from a 7950x is that even just browsing youtube comments power draw hits 45w average and it peaks as high as 65 (!!!!). A 12900k browsing the web while watching 2 videos (youtube and twitch) draws 7 watts. The difference is pretty insane.

Race to idle is what every modern CPU does. So disregarding per core and idle efficiency is silly, when otherwise comparable platforms have a 40-50 W difference between them in the state they'll most commonly be in.

dgianstefaniYou'll have slightly better per core efficiency (what ST is) but much better idle efficiency, which, like I said, is what the CPU is doing most of the time. Load a new webpage - straight to idle, open your mail - straight to idle. Etc.

No, Single-threaded efficiency is specifically the efficiency of the processor in a single threaded application. The name should make that obvious.

dgianstefaniYou're talking about cherry picking results, while failing to mention that in the instance where the Intel CPU is using "35%" more power (in 100% all core synthetic load, at a whole 42 W more than it's competitor, the 7950X), it's also doing more work, faster.

The 14900K competes against the 7950X3D, of which the 14900K uses 130% more power to achieve 4% higher performance. That number is indeed cherrypicked in favor of Intel, thanks for pointing that out.

dgianstefaniSince you misquoting TPU's test of 100% single core load as "idle" power apparently puts the burden of proof on me, as requested, here's a review that compares CPUs at idle/load power.

lol, burden on proof is always on the one making a claim.

dgianstefaniAs you can see, the 13900K system idles at 67 W, whereas the otherwise identical 7950X system idles at 107 W. Even the six core 7600X still idles at 89 W, with single core load being just 18 W higher, lending further evidence to my statements.

You can also, as I suggested, simply observe your systems power draw to reach the same understanding.



Not only that, but since hothardware.com tests using manufacturer spec settings, the Zen 4 parts used 5200 MT memory while the 13th gen Intel parts used 5600 MT memory, so if you used 6000 MT "sweetspot" RAM in the same Zen 4 build, the powerdraw would be yet higher, since IF/IO die would be consuming more power.

I see four things in the above chart:

1) An utter lack of evidence to support your sub 5w claim. There's nothing on here that isolates specifically to CPU power only nor is any figure alluding to 5w provided. I can safely assume you pulled that one out of your rear as usual.

2) You couldn't be bothered to get a chart with 14900K, which is the CPU in question. The chart is also missing the 7950X3D / 7800X3D, which as we are all aware are the more efficient CPUs of AMD's current lineup.

3) You can pretty clearly see Intel and AMD trading on Idle power consumption in that graph. The 5800X3D is in fact the most efficient of the bunch at idle.

4) You provided a chart with total system power consumption without addressing the fact that such a chart includes variables like fans, motherboard, ect. A high end motherboard can significantly add to idle power consumption for example, especially a 2-chip solution like the X670E chipset. I have a 7800X3D with a X670 motherboard with 4 NVMe drives, 2 of which are 15.36TB 10w enterprise drives (these drives do not enter sleed states), one of which is the power hungry T700 4TB (7w), and the last of which is a renegade fury 4TB (3.5w). My watt meter measures total idle power consumption at 89w. If I were to eliminate power hungry SSDs that brings my system idle power down to 59.5w, which is coincidentally close to the 5800X3D and better than the Intel systems. I bet I can shave off even more by removing some fans and doing to a single chip motherboard like B class. Suffice it to say, point to this chart as if it only indicates CPU power consumption is highly misleading.

evernessinceNo, Single-threaded efficiency is specifically the efficiency of the processor in a single threaded application. The name should make that obvious.



The 14900K competes against the 7950X3D, of which the 14900K uses 130% more power to achieve 4% higher performance. That number is indeed cherrypicked in favor of Intel, thanks for pointing that out.



lol, burden on proof is always on the one making a claim.


I see four things in the above chart:

1) An utter lack of evidence to support your sub 5w claim. There's nothing on here that isolates specifically to CPU power only nor is any figure alluding to 5w provided. I can safely assume you pulled that one out of your rear as usual.

2) You couldn't be bothered to get a chart with 14900K, which is the CPU in question. The chart is also missing the 7950X3D / 7800X3D, which as we are all aware are the more efficient CPUs of AMD's current lineup.

3) You can pretty clearly see Intel and AMD trading on Idle power consumption in that graph. The 5800X3D is in fact the most efficient of the bunch at idle.

4) You provided a chart with total system power consumption without addressing the fact that such a chart includes variables like fans, motherboard, ect. A high end motherboard can significantly add to idle power consumption for example, especially a 2-chip solution like the X670E chipset. I have a 7800X3D with a X670 motherboard with 4 NVMe drives, 2 of which are 15.36TB 10w enterprise drives (these drives do not enter sleed states), one of which is the power hungry T700 4TB (7w), and the last of which is a renegade fury 4TB (3.5w). My watt meter measures total idle power consumption at 89w. If I were to eliminate power hungry SSDs that brings my system idle power down to 59.5w, which is coincidentally close to the 5800X3D and better than the Intel systems. I bet I can shave off even more by removing some fans and doing to a single chip motherboard like B class. Suffice it to say, point to this chart as if it only indicates CPU power consumption is highly misleading.

Damn.

He's saying that because I used a 13900K (physically the same silicon as a 14900K) it doesn't matter.

Cute how suddenly you're comparing it against the X3D chip, since that supports your argument more (big power number is bigger) which is slower than the 7950X in applications as a trade off for being faster in gaming.

Lmao. Gonna end it here, pointless showing evidence and simple facts to people with their heads in the sand.

Chatted to W1z and next CPU test bench will have idle power measurements. Guess we'll know then who's "pulling that out of their rear, as usual".

dgianstefaniRace to idle is what every modern CPU does. So disregarding per core and idle efficiency is silly, when otherwise comparable platforms have a 40-50 W difference between them in the state they'll most commonly be in.

Yes but the charts are mp3 encoding which I assume puts a lot more load in the memory subsystem than browsing the web for example, so mp3 encoding isn't really indicative of actual real world application power draw.

BTW, here is a chart including 14th gen, system power at idle from the wall

fevgatosYes but the charts are mp3 encoding which I assume puts a lot more load in the memory subsystem than browsing the web for example, so mp3 encoding isn't really indicative of actual real world application power draw.

BTW, here is a chart including 14th gen, system power at idle from the wall

The charts Evernessince posted are indeed (mistakenly) full single thread load, as TPU doesn't currently test idle draw for CPUs (this will change in the next test suite).

Hence why I'm using other reviews (as requested) to compare full system power draw, since the only differences in those test systems are the CPUs, this is appropriate.

Maybe after 20 years it will be interesting to test fastest i9.

How much transistors inside?

ReadlightMaybe after 20 years it will be interesting to test fastest i9.

How much transistors inside?

Same as 14900K, which is same as 13900K (26 billion). It's just process refinements the past year and a bit, binning, and different microcode.

Claimed Ryzen 7000 power draw is all over the place in reviews. Some say they are similar, with Ryzen overall a bit higher than 13th and 14th Gen Core:

CPU only:



Whole system:



What I've found it most/all non-K Intel CPUs can idle lower than 5W but different circumstances in the OS can change that to 15-20W at idle.

For example: Dell Optiplex 9020 i7-4790.

Intel iGPU drivers: 15W idle
Nvidia drivers: 15W idle
AMD drivers: 4W idle

These are GPU drivers so WTF?

dgianstefaniThe charts Evernessince posted are indeed (mistakenly) full single thread load, as TPU doesn't currently test idle draw for CPUs (this will change in the next test suite).

Hence why I'm using other reviews (as requested) to compare full system power draw, since the only differences in those test systems are the CPUs, this is appropriate.

Agree with all you stated, full power draw benchmarks have been quite a red herring for years. An extreme minority of users do things with their desktops where that might be relevant.

Meanwhile virtually everyone's PC spends a lot of time at or near idle.

That said, AMDs chipsets / motherboards are one of the bigger culprits behind their system idle power draws.

Vortez.net is the only site I've seen incorporate this into their motherboard reviews, but they are a small site with a limited sample. From their data though, it's clear that AMD mobo/cpu combination idle is notably higher.

Specifically, Z690 idle power draws ranged from a low of 72W to a high of 80W.
X670E power draws ranged from a low of 88W to a high of 102W.

From their data, AMDs idle power draw is ranging from a low of +10% (highest idle intel vs lowest idle AMD) to +41.7% (lowest idle Intel vs highest idle AMD) vs Intel.

X670E (idle is the gray bar) :


Z690:


@Lew Zealand The above is why they are all over the board. It depends on what motherboard they used. However the overall pattern of high idle power usage is pretty clear.

RandallFlaggAgree with all you stated, full power draw benchmarks have been quite a red herring for years. An extreme minority of users do things with their desktops where that might be relevant.

Meanwhile virtually everyone's PC spends a lot of time at or near idle.

That said, AMDs chipsets / motherboards are one of the bigger culprits behind their system idle power draws.

Vortez.net is the only site I've seen incorporate this into their motherboard reviews, but they are a small site with a limited sample. From their data though, it's clear that AMD mobo/cpu combination idle is notably higher.

Specifically, Z690 idle power draws ranged from a low of 72W to a high of 80W.
X670E power draws ranged from a low of 88W to a high of 102W.

From their data, AMDs idle power draw is ranging from a low of +10% (highest idle intel vs lowest idle AMD) to +41.7% (lowest idle Intel vs highest idle AMD) vs Intel.

X670E (idle is the gray bar) :


Z690:


@Lew Zealand The above is why they are all over the board. It depends on what motherboard they used. However the overall pattern of high idle power usage is pretty clear.

What a nice post. So many signs of actually understanding the technical details of what is being talked about unlike some others. I salute your reading comprehension. Refreshing.

(not being sarcastic).

It's true that the chipset/motherboard influences things, but I'd say the chiplet architecture with a separate IO die on a less efficent older process is more influential.

dgianstefaniWhat a nice post. So many signs of actually understanding the technical details of what is being talked about unlike some others. I salute your reading comprehension. Refreshing.

(not being sarcastic).

It's true that the chipset/motherboard influences things, but I'd say the chiplet architecture with a separate IO die on a less efficent older process is more influential.

What is sad is that, if we look back in time, Z490 motherboard had idle draws of ~55-60W. The X570 (Zen 3) was around 80-95W.

That is 45%-58% more power draw, doing what PCs do (nothing) 95% of the time.

You have to go back to Zen 1+ and X470 to see AMD match up vs Intel on idle power draw. From what data they have a Zen 1+ (2XXX series) on an X470 had around 10% less idle power draw than equivalent Intel.

IIRC those were the last monolithic AMD desktop CPUs.

RandallFlaggWhat is sad is that, if we look back in time, X490 motherboard had idle draws of ~55-60W. The X570 (Zen 3) was around 80-95W.

That is 45%-58% more power draw, doing what PCs do (nothing) 95% of the time.

You have to go back to Zen 1+ and X470 to see AMD match up vs Intel on idle power draw. From what data they have a Zen 1+ (2XXX series) on an X470 had around 10% less idle power draw than equivalent Intel.

IIRC those were the last monolithic AMD desktop CPUs.

Old Zen also had much slower RAM support. As RAM speeds have scaled, so too have Infinity Fabric speeds, and thus power draw increases. This is one of the main reasons Zen 4 uses so much more idle power than Zen 3 (which already wasn't that impressive). DDR5 RAM speeds and faster PCIe etc force a higher clocked IO die.