News Intel issues official statement on Core K-series crashes: stick to Intel's official power profiles

[TerryLaze]

I never made any assertions for the above conspiratorial claims. I focused my writings on my claim that the default profiles were used to make Intel look as good comparatively as possible at "default" settings.

For what reasons did they do it then?
Other than conspiracy it can only be bias which would be as much an reason as conspiracy, or complete incompetence which would also be the same.

Because Intel has no official default power profile it's very difficult to get anything but essentially OCed CPUs with varying amounts of performance boosting settings based on motherboard selection.

We had an official profile from day one, pl1=pl2=253W was plastered all about, reviewers deliberately ignored that and searched for mobos with the highest settings...because clicks.

With in mind, the so called 'default' profiles are all that there is to compare against AMD, thus giving Intel an optics win in reviews because their 'default' CPUs are winning by X%. If there were an actual Intel prescribed power/behavior profile, you could have much better comparisons with AMD. Winning by a higher percentage is always better optically. Intel prefers this strategy, but it comes at a cost sometimes and it's time to pay. This practice led directly to the issues in the article. Fly too close to the sun and eventually your wax wings will melt.

The default profile would have been pl1=pl2=253W , reviewers where and still are completely free to use this setting, and with that setting intel would have won enough benchmarks to not care about another 1-2% in general benchmarks, especially for 40-50% more power.
And they would also be stable.

 

[helper800]

For what reasons did they do it then?
Other than conspiracy it can only be bias which would be as much an reason as conspiracy, or complete incompetence which would also be the same.

We had an official profile from day one, pl1=pl2=253W was plastered all about, reviewers deliberately ignored that and searched for mobos with the highest settings...because clicks.

The default profile would have been pl1=pl2=253W , reviewers where and still are completely free to use this setting, and with that setting intel would have won enough benchmarks to not care about another 1-2% in general benchmarks, especially for 40-50% more power.
And they would also be stable.

I gave a multiple reasons in the quoted sections, please read them again. It's better for Intel to win with as large of margins as possible or lose with as small of margins as possible, so every % of performance matters.

Intel has no officially endorsed default profile given to the motherboard manufacturers that can be set in BIOS. Intel has given motherboard manufacturers carte blanche with their 'default' performance profiles' settings. All we have are guidelines for pl1 and pl2 power and that's pretty much it. There are 30 other different settings that have an effect on power, boosting behavior, et cetera that they have not said anything about that is crucial to what performance profiles are.

 

[Lucky_SLS]

For what reasons did they do it then?
Other than conspiracy it can only be bias which would be as much an reason as conspiracy, or complete incompetence which would also be the same.

We had an official profile from day one, pl1=pl2=253W was plastered all about, reviewers deliberately ignored that and searched for mobos with the highest settings...because clicks.

The default profile would have been pl1=pl2=253W , reviewers where and still are completely free to use this setting, and with that setting intel would have won enough benchmarks to not care about another 1-2% in general benchmarks, especially for 40-50% more power.
And they would also be stable.

https://www.techpowerup.com/review/intel-core-i7-14700k/18.html

Reviewers are just lazy/not through sometimes. The above linked review clearly states the performance figures for all 3 OC states.

Intel did not say to limit the CPU at 253W. It says to include the default profiles and enable them if the mobo supports it.

The only hard limit it wants to impose is ICCmax 400A.

Intel is fine with running the CPU at 350W if the mobo supports it. And it may have well been the official guideline from Intel to reviewers to OC the chip in a high-end board and show the best results possible. Thats the entire point of having an overclockable chip.

The problem of instability comes when mobo manufacturers include those "limits removed" OC profiles without QA testing them for stability in low-end boards.

 

[bit_user]

Intel did not benefit from reviews being done with unlimited power

TechPowerUp tested both ways, so we can actually see they did!

Source: https://www.techpowerup.com/review/intel-core-i9-13900k/22.html
​

That's a 6.1% performance improvement between Stock and Power Limits Removed.

if they were done with enforced power limits Intel would be winning on every price segment they competed at on ST, MT, gaming performance and efficiency. You cannot possibly disagree with that, it's literally just facts.

Here's the efficiency ranking and.... Nope.

​

Reminder: higher is better. So, look towards the bottom of the graph for better efficiency!

It's a cpu gaming review, of course it's done at 720p. 7800x 3d isn't missing, it didn't exist on zen 4 / 13th gen release.

I didn't mean they willfully omitted it from the review, I was simply observing that it's not included in the graph. Here's a newer ranking that includes it.

Source: https://www.techpowerup.com/review/amd-ryzen-7-7800x3d/27.html
​

You're welcome.
: )

 

[bit_user]

Are you saying that if you lock the 13600k to 125w it would lose to the 7600x? It's leading by 34%, by limiting it to 125w it would now lead by 30%, big whoop.
Are you saying that if you lock the 13700k to 125w it would lose to the 7700x? It's leading by 35%, by power limiting it to 125w it would lead by 28%, big whoop.

You're the only one saying this. No one else would match up those specific models against each other, because it doesn't make sense.

Since I've already demonstrated how they match based on street price, let's look at another aspect. Here, we look at the thread counts:

Intel CPU Model	Threads	AMD CPU Model
	12​	R5 7600X
i5-13400	16​	R7 7700X
i5-13600K	20​
i7-13700K	24​	R9 7900X
i9-13900K	32​	R9 7950X

 

[TheHerald]

You're the only one saying this. No one else would match up those specific models against each other, because it doesn't make sense.

You are the only one saying this. Let me quote a bunch of reviews (anandtech, techpowerup, techradar)

"Across the general lineup of performance tests I threw at this chip, it pretty much came out on top in every one of them, beating out the competing AMD Ryzen 5 7600X"
"Intel is pricing the Core i5-13600K at USD $320, a similar price to that of the i5-12600K, and $20 higher than the competing Ryzen 5 7600X"

I can keep going for ever. You are a reviewer. A cpu comes out with the same name, same price as another CPU that came out 15 days earlier. You don't consider them competitors? What the hell man

Number of cores is a silly way to compare products. Do you do that with GPU as well? Does the 1200$ 7900xtx compare with the 599$ 4070 because they have the same amount of cores?

 

[TheHerald]

TechPowerUp tested both ways, so we can actually see they did!

​

Source: https://www.techpowerup.com/review/intel-core-i9-13900k/22.html​

And seeing as the 13900k competitor is siting at 28.6k points, yeah, they really needed that boost my man.

The rest of the chips on your efficiency graph aren't limited btw, so...??

 

[bit_user]

"Intel is pricing the Core i5-13600K at USD $320, a similar price to that of the i5-12600K, and $20 higher than the competing Ryzen 5 7600X"[/I][/I]

They're talking about the MSRPs, which were a fiction. Street price reflects the actual matchups.

I can keep going for ever.

Yes, you can quote launch reviews citing MSRP. That's about it. As I said, it's irrelevant.

A cpu comes out with the same name,

Nope. This is another point you're confused by. AMD has a long-established naming convention for their CPUs, which they've followed since Zen 2.

AMD CPU Model (last 3 digits)	Number of Cores	Number of Threads
600​	6​	12​
700​	8​	16​
800​	8​	16​
900​	12​	24​
950​	16​	32​

AMD just followed their own naming convention. I expect Intel also focused on the relation of its model numbers to its prior generation, rather than specifically targeting them at AMD.

Number of cores is a silly way to compare products.

That's why I compared them based on number of threads.

Do you do that with GPU as well?

I do look at what Nvidia likes to call a "core", but it's less telling because unlike CPUs, GPUs don't even share a common ISA. Hence, what core-count and theoretical FLOPS means for real world performance is a bit more disconnected.

Does the 1200$ 7900xtx compare with the 599$ 4070 because they have the same amount of cores?

I'm not sure what you mean by that question. Are you referring to specific boards implementing each GPU? Anyway, we don't need to get further sidetracked on a whole GPU tangent. Best to forget it.

The rest of the chips on your efficiency graph aren't limited btw, so...??

They are. Those scores are with stock settings.

 

[thestryker]

You are the only one saying this. Let me quote a bunch of reviews (anandtech, techpowerup, techradar)

"Across the general lineup of performance tests I threw at this chip, it pretty much came out on top in every one of them, beating out the competing AMD Ryzen 5 7600X"
"Intel is pricing the Core i5-13600K at USD $320, a similar price to that of the i5-12600K, and $20 higher than the competing Ryzen 5 7600X"

I can keep going for ever. You are a reviewer. A cpu comes out with the same name, same price as another CPU that came out 15 days earlier. You don't consider them competitors? What the hell man

I'm curious what your obsession with historical pricing is. At any given time the only pricing that matters is whatever the current pricing is. The 7600X was poorly priced at $300 when it launched, but that didn't last long. Pretty sure most reviewers commented on the pricing as well given 12th Gen dropping. Then there's the fact that in less than 2 months from launch it was available for $250.

 

[TheHerald]

They're talking about the MSRPs, where were a fiction. Street price reflects the actual matchups.

But THAT'S what the discussion is about. Reviews are done with MSRPs in mind, not with prices a month or 2 years down the line. So the question is - did reviews done with no power limits benefit Intel.

Nope. This is another point you're confused by. AMD has a long-established naming convention for their CPUs, which they've followed since Zen 2.

Yes, AMD has an established naming scheme that they followed since Zen 1, which was fully based on Intel's naming scheme.

That's why I compared them based on number of threads.

Number of threads is also silly. Who cares with how many threads or cores it manages X performance? Price and the performance are important, how it gets there is irrelevant.

They are. Those scores are with stock settings.

The 13600k has a PL1 of 125 watts, the review wasn't done with a PL1 of 125w. The 13600k doesn't score 24k at 125w, but around 21-22k.

 

[TheHerald]

I'm curious what your obsession with historical pricing is. At any given time the only pricing that matters is whatever the current pricing is. The 7600X was poorly priced at $300 when it launched, but that didn't last long. Pretty sure most reviewers commented on the pricing as well given 12th Gen dropping. Then there's the fact that in less than 2 months from launch it was available for $250.

Since the discussion is whether Intel benefitted from reviews being done without power limits, of course MSRPs are important. How else would you conclude anything about reviews and who benefitted from power unlimited if you don't take into account the price at the time?

Regardless, and although it's true that amd dropped prices quickly, zen 4 was still more expensive due to more expensive mobos (RPL had access to z690 and b660) and DDR5, while RPL could use DDR4 as well.

 

[bit_user]

AMD has an established naming scheme that they followed since Zen 1, which was fully based on Intel's naming scheme.

Whatever inspired Ryzen's naming scheme, Zen 2 established a straight-forward relationship between their model numbers and the chiplet & core counts. AMD has kept to this exact scheme, completely independent of what Intel has been doing.

I believe that Intel, as I said, is more interested in making sure that each CPU in their product lineup outperforms the comparably-named chip from the previous generation, or else they know a lot of clickbait articles would seize on it and claim that Intel's performance is regressing or only advancing by some tiny amount.

It's too hard to try and tweak your product numbering to exactly match up with the competition. That's why I believe street price, performance, and power are the best ways to do that. Specs are also a relevant point of comparison, but only to the extent they predict performance. Therefore, performance is a better indicator.

Number of threads is also silly. Who cares with how many threads or cores it manages X performance?

As I said above, specs are relevant to the extent they predict performance. The performance of threads is roughly comparable between the two companies' CPUs. I don't mean across different generations or ISAs, but rather when comparing similar and contemporary products.

Since the discussion is whether Intel benefitted from reviews being done without power limits, of course MSRPs are important.

That's a non-sequitur.

 

[TheHerald]

Whatever inspired Ryzen's naming scheme, Zen 2 established a straight-forward relationship between their model numbers and the chiplet & core counts. AMD has kept to this exact scheme, completely independent of what Intel has been doing.

I believe that Intel, as I said, is more interested in making sure that each CPU in their product lineup outperforms the comparably-named chip from the previous generation, or else they know a lot of clickbait articles would seize on it and claim that Intel's performance is regressing or only advancing by some tiny amount.

It's too hard to try and tweak your product numbering to exactly match up with the competition. That's why I believe street price, performance, and power are the best ways to do that. Specs are also a relevant point of comparison, but only to the extent they predict performance. Therefore, performance is a better indicator.


As I said above, specs are relevant to the extent they predict performance. The performance of threads is roughly comparable between the two companies' CPUs. I don't mean across different generations or ISAs, but rather when comparing similar and contemporary products.


That's a non-sequitur.

How the hell is it a non sequitur. I made the argument that reviews being done with no limits doesnt benefit intel, since even with power limits they would top the charts at every price point. Every price point that reviews are done with, which is day one price / msrp whatever you wanna call it. Street prices 2 months or 2 years down the line are completely irrelevant to the argument.

And that applies equally to amd. Pushing twice the power on zen 4 vs zen 3 didn't do them any favors either, since the last 100 watts practically offer single digits performance increase.

It makes both amd and intel cpus look way worse than they actually are, perfectly demonstrated by the graph you posted, with the 5950x being on the top of the efficiency chart when in reality, it's really far behind both 7950x and 13900k.

 

[TheHerald]

Whatever inspired Ryzen's naming scheme, Zen 2 established a straight-forward relationship between their model numbers and the chiplet & core counts. AMD has kept to this exact scheme, completely independent of what Intel has been doing.

I believe that Intel, as I said, is more interested in making sure that each CPU in their product lineup outperforms the comparably-named chip from the previous generation, or else they know a lot of clickbait articles would seize on it and claim that Intel's performance is regressing or only advancing by some tiny amount.

It's too hard to try and tweak your product numbering to exactly match up with the competition. That's why I believe street price, performance, and power are the best ways to do that. Specs are also a relevant point of comparison, but only to the extent they predict performance. Therefore, performance is a better indicator.

If we remove the generations from the product naming schemes we have

An r5 600x on a x370 board vs
An i5 600k on a z370 board.

You still don't see it?

 

[bit_user]

If we remove the generations from the product naming schemes we have

An r5 600x on a x370 board vs
An i5 600k on a z370 board.

You still don't see it?

The similarity might have originally followed Intel's scheme, but they clearly aren't trying to calibrate each generation to match up their products by number. You're reading too much into it. Such a superficial similarity does not imply equivalence, nor even correspondence.

 

[TheHerald]

The similarity might have originally followed Intel's scheme, but they clearly aren't trying to calibrate each generation to match up their products by number. You're reading too much into it. Such a superficial similarity does not imply equivalence, nor even correspondence.

Well, now that Intel allegedly is going to change their naming scheme, let's see what amd does.

 

[thestryker]

Since the discussion is whether Intel benefitted from reviews being done without power limits, of course MSRPs are important. How else would you conclude anything about reviews and who benefitted from power unlimited if you don't take into account the price at the time?

Regardless, and although it's true that amd dropped prices quickly, zen 4 was still more expensive due to more expensive mobos (RPL had access to z690 and b660) and DDR5, while RPL could use DDR4 as well.

I guess I don't view reviews in a vacuum, but rather as a reference point for performance comparison at any time. As for whether or not Intel benefits from unlimited power of course they do because it separates them further for most reviews. Now they certainly would have been at the top without unlimited power, but they still wouldn't have won on overall efficiency which has been shown already using TPU graphs (though it would have been much closer).

I do not think that most reviewers do their due diligence with regards to Intel power levels and I've mentioned that before on this forum. I'm fine with reviewers doing unlimited power/board defaults, but without any testing of Intel recommended PL2 they do a big disservice due to lacking context.

 

[TerryLaze]

I gave a multiple reasons in the quoted sections, please read them again. It's better for Intel to win with as large of margins as possible or lose with as small of margins as possible, so every % of performance matters.

I understand the reasons you give for intel to want that, you haven't given one single reason for reviewers to do this for intel.
Your argument is basically that every review works for intel.
But every review also bashes intel for high power draw and high costs due to that, so that doesn't make any sense.

TechPowerUp tested both ways, so we can actually see they did!

​

Source: https://www.techpowerup.com/review/intel-core-i9-13900k/22.html​

​

That's a 6.1% performance improvement between Stock and Power Limits Removed.

Yeah, a whopping 6.1% if you look at a ( one single) best case scenario...in a complete benchmark with several things tested it would amount to nothing, as techpowerups own results show.
This is from your link and shows the overall result, it's a 1,7% difference, below even run variance, and 3% max difference compared to the highest settings.

Not worth it to be known as hot and inefficient since the stock 13900k is already 4% faster than the 7950x so why would intel bother with making their CPU look hotter and more power hungry than they need to be?

 

[-Fran-]

Round and round we go and no one will ever let go!

Regards.

 

[TheHerald]

Round and round we go and no one will ever let go!

Regards.

Yeah, when it comes to Intel cpus unless everyone agrees that they are inefficient pieces of crap and amd is light-years ahead, people will keep arguing. The same things happens across every forum on planet earth

 

[helper800]

Yeah, when it comes to Intel cpus unless everyone agrees that they are inefficient pieces of crap and amd is light-years ahead, people will keep arguing. The same things happens across every forum on planet earth

As I have denoted previously, I like and even prefer several Intel products at several price points. I recommend them all the time for builds in the systems section. Personally, I think it comes down to the answer to this question; Is it better to beat your competitor by 3-4% or 5-8%? IMO, the clear answer is it's better to win by more than less of an advantage at 5-8%. If your opinion is a win is a win and it does not matter because x, y, z, then you are not looking at the bigger picture business wise. We can all have differences of opinion. I can agree to disagree. Good luck everyone!

 

[TheHerald]

As I have denoted previously, I like and even prefer several Intel products at several price points. I recommend them all the time for builds in the systems section. Personally, I think it comes down to the answer to this question; Is it better to beat your competitor by 3-4% or 5-8%? IMO, the clear answer is it's better to win by more than less of an advantage at 5-8%. If your opinion is a win is a win and it does not matter because x, y, z, then you are not looking at the bigger picture business wise. We can all have differences of opinion. I can agree to disagree. Good luck everyone!

That wasn't the reality of the situation though. Reality was, intel was already beating the competion by 25-30%, and removing the power limit would just turn that into 28-33%, a meaningless improvement. And for that meaningless improvement, almost if not every reviewer basically ignored the huge performance lead and focused on fires, temperatures, power draws etc. Both the reviews and the comments about 13th gen were all negative, based solely on the power draw figures.

 

[helper800]

That wasn't the reality of the situation though. Reality was, intel was already beating the competion by 25-30%, and removing the power limit would just turn that into 28-33%, a meaningless improvement. And for that meaningless improvement, almost if not every reviewer basically ignored the huge performance lead and focused on fires, temperatures, power draws etc. Both the reviews and the comments about 13th gen were all negative, based solely on the power draw figures.

Your comparisons are meaningless as others have pointed out. The reality is that if you compare similar products from both companies the performance differences are in the mid to low single digits. The only reason you are comparing these products with such a performance disparity is because Intel undercut AMD's CPU prices so such comparisons can be erroneously made, to again, point to as massive performance wins as possible. Winning by more is always better, you will never convince me otherwise.

 

[NinoPino]

That wasn't the reality of the situation though. Reality was, intel was already beating the competion by 25-30%, and removing the power limit would just turn that into 28-33%, a meaningless improvement. And for that meaningless improvement, almost if not every reviewer basically ignored the huge performance lead and focused on fires, temperatures, power draws etc. Both the reviews and the comments about 13th gen were all negative, based solely on the power draw figures.

How not agree, as you wrote Intel outperform AMD by a miserable 33%, but I rebate, why not 40%...
https://www.phoronix.com/benchmark/...test-results-result-composite-ici1ici1lb.svgz

And with 30% less power usage...
https://www.phoronix.com/benchmark/...arks/cpu-power-consumption-monitor-ptssm.svgz

Source : Phoronix

 

[TheHerald]

Your comparisons are meaningless as others have pointed out. The reality is that if you compare similar products from both companies the performance differences are in the mid to low single digits. The only reason you are comparing these products with such a performance disparity is because Intel undercut AMD's CPU prices so such comparisons can be erroneously made, to again, point to as massive performance wins as possible. Winning by more is always better, you will never convince me otherwise.

I did compare similar products. The i5 13600k, on the 20th of October of 2022 that the reviews were conducted, was competing to the r5 7600x that was released 2 weeks earlier. No? Reviewers themselves agree with me, ive already quoted a number of them