News Desktop CPU Sales Lowest in 30 Years, AMD Gains Market Share Anyway

[TerryLaze]

Another thing: benchmarks are tricky, because they typically have a limited run duration. During that time, the Intel CPU can flex its 117 PL2 threshold to juice the results beyond what you'd see in a sustained compute job.

But those benchmarks also only show the pl2 power draw and not the power after TAU, so that's fair, they show the result you get from using that amount of power.
Also PL2 is the amount of power an intel CPU can sustain indefinitely so if you put TAU to unlimited or put PL1= PL2 performance won't change no matter how long a job you run as long as cooling is sustained.

Except for where I highlighted the massive efficiency win of Cezanne, where the 5600G showed a 17% advantage over your best example of the i5-12400.

He was talking about the heat and not the efficiency.
And AMDs most efficient CPU the 5950x runs more than 10 degrees hotter than what you consider intels least efficient CPU the 12900k, when both run at the same W (123 for the 5950x and 125 for the 12900k)
Also as seen the 12900k (8/8+8) at 125W is much more efficient than the 5900x (12/24) , at least in cb23, while still behind of the 5950x (16/32) per thread the 12900k is still the most efficient CPU you can get.

Having an intel CPU running at full power all the time while having an AMD CPU
running at the most efficient point (the one AMD themselves set) is much more cherry-picking than anything else.

https://www.hardwareluxx.de/index.p...-desktop-cpus-alder-lake-im-test.html?start=8

 

[bit_user]

He was talking about the heat and not the efficiency.

Not really. Cryo's original statement was about efficiency, and that's what I took issue with. Cryo then veered off into talking about heat, but the theme of the exchange was efficiency.

And AMDs most efficient CPU the 5950x runs more than 10 degrees hotter than what you consider intels least efficient CPU the 12900k, when both run at the same W

I don't really care about your artificial scenarios. I quoted the benchmarks as Toms ran them and they used the stock settings. They note:

"Our tests use the default lifted PL1 and PL2 restrictions. Almost all enthusiast-class motherboards come with similar settings, so this reflects the out-of-box experience with a high-end motherboard."​

If you want to take issue with that, then your issue is with the motherboard makers and the flexibility & guidance Intel gave them. But you don't get to have it both ways - comparing performance at unrestricted PL settings, while comparing efficiency at restricted, non-default settings.

Anyway, go back and read my posts. They have all the data that's needed, here.

And I don't care about the OC or PBO results. I'm comparing out-of-the-box behavior of both AMD and Intel.

 

[TerryLaze]

If you want to take issue with that, then your issue is with the motherboard makers and the flexibility & guidance Intel gave them. But you don't get to have it both ways - comparing performance at unrestricted PL settings, while comparing efficiency at restricted, non-default settings.

Where did I do that?!
The numbers from Hardwareluxx show both maximum power at 240W and the performance from that as well as restricted power at 125W and the performance from that.

You are the one that want's to have it both ways, taking a benchmark that has overblown settings for intel and the most efficient settings for AMD and trying to make it look as if that's what makes a good review...

 

[bit_user]

You are the one that want's to have it both ways, taking a benchmark that has overblown settings for intel and the most efficient settings for AMD and trying to make it look as if that's what makes a good review...

It's the out-of-the-box settings that Toms used. I think that's fair enough.

Again, your issue is with Intel & motherboard makers, for such aggressive out-of-the-box settings. Don't get it twisted.

 

[TerryLaze]

It's the out-of-the-box settings that Toms used. I think that's fair enough.

Again, your issue is with Intel & motherboard makers, for such aggressive out-of-the-box settings. Don't get it twisted.

You can't take a benchmark that has one cpu at max power and one cpu at minimum power and use it to compare efficiency...that's just not how it works.

Out-of-the-box is just another way to say "too stupid to bend finger to hit del key to get into bios" you can use that as an excuse to say how stupid people are for using a CPU out-of-the-box but you can't blame the CPU for efficiency if the problem is the settings and not the CPU.

 

[-Fran-]

You can't take a benchmark that has one cpu at max power and one cpu at minimum power and use it to compare efficiency...that's just not how it works.

Out-of-the-box is just another way to say "too stupid to bend finger to hit del key to get into bios" you can use that as an excuse to say how stupid people are for using a CPU out-of-the-box but you can't blame the CPU for efficiency if the problem is the settings and not the CPU.

You're not wrong, but you're also not considering this is something Intel is allowing motherboard makers to do and even in their review guidelines, since they care about the pure performance of the platform at the expense of efficiency for the consumer sector. That's just the best way to interpret this.

Several review sites have talked about this same exact issue and have taken stances, but the "out of the box" methodology is the fairest of them all, since that is exactly what 90% of consumers will experience.

If you want to check one of the biggest talks about this, from what I remember, it was in Anandtech when Dr. Ian C. asked, point blank, Intel about this and they were OK with how these settings were used:

https://www.anandtech.com/show/1458...ng-an-interview-with-intel-fellow-guy-therien

Regards.

 

[bit_user]

You can't take a benchmark that has one cpu at max power and one cpu at minimum power and use it to compare efficiency...that's just not how it works.

Out-of-the-box is just another way to say "too stupid to bend finger to hit del key to get into bios" you can use that as an excuse to say how stupid people are for using a CPU out-of-the-box but you can't blame the CPU for efficiency if the problem is the settings and not the CPU.

Again, you can argue with me until your fingers fall off, but that changes nothing. Out-of-the-box is how most people use this stuff, it's how Toms tested it, and that's why it's the case I care about.

And contrary to your claims about them using aggressive motherboards, they state:

"We're sticking with our standard policy of allowing the motherboard to exceed Intel's recommended power limits, provided the chip remains within warrantied operating conditions. Our tests use the default lifted PL1 and PL2 restrictions. Almost all enthusiast-class motherboards come with similar settings, so this reflects the out-of-box experience with a high-end motherboard."​

Rather than blame Toms for choosing a bad motherboard, I think the burden of proof is on you to demonstrate that their motherboard's behavior was uncharacteristically egregious.

Furthermore, Intel could've limited these parameters, if they wanted.

 

[TerryLaze]

Several review sites have talked about this same exact issue and have taken stances, but the "out of the box" methodology is the fairest of them all, since that is exactly what 90% of consumers will experience.

Intel is allowing it because if intel would not allow using a safe (for the CPU) power level people like you would react even worse to that.
And the part i quoted is just none sense because 90% of people don't even know what the things in the multithreaded benchmarks even are let alone what they do.

They (review sites) are just using out of the box as an excuse to show an extremely high power usage that 99% of people will never see in their life because they will never use any of the software shown that can draw that much power.

Again, you can argue with me until your fingers fall off, but that changes nothing. Out-of-the-box is how most people use this stuff, it's how Toms tested it, and that's why it's the case I care about.

And contrary to your claims about them using aggressive motherboards, they state:

"We're sticking with our standard policy of allowing the motherboard to exceed Intel's recommended power limits, provided the chip remains within warrantied operating conditions. Our tests use the default lifted PL1 and PL2 restrictions. Almost all enthusiast-class motherboards come with similar settings, so this reflects the out-of-box experience with a high-end motherboard."​

Rather than blame Toms for choosing a bad motherboard, I think the burden of proof is on you to demonstrate that their motherboard's behavior was uncharacteristically egregious.

Furthermore, Intel could've limited these parameters, if they wanted.

People that use distributed computing are smart enough to not use default settings there is nothing you can tell me that will change that, stupid people do not use any DC.

"provided the chip remains within warrantied operating conditions. Our tests use the default lifted PL1 and PL2 restrictions."
This is contradictory since a lifted PL2 restriction is very much against warranty and will void it immediately.

 

[-Fran-]

Intel is allowing it because if intel would not allow using a safe (for the CPU) power level people like you would react even worse to that.
And the part i quoted is just none sense because 90% of people don't even know what the things in the multithreaded benchmarks even are let alone what they do.

They (review sites) are just using out of the box as an excuse to show an extremely high power usage that 99% of people will never see in their life because they will never use any of the software shown that can draw that much power.


People that use distributed computing are smart enough to not use default settings there is nothing you can tell me that will change that, stupid people do not use any DC.

"provided the chip remains within warrantied operating conditions. Our tests use the default lifted PL1 and PL2 restrictions."
This is contradictory since a lifted PL2 restriction is very much against warranty and will void it immediately.

"People like you".

Ok...

Regards?

 

[TerryLaze]

"People like you".

Ok...

Regards?

Yes people like him that complain about a company releasing an unlocked overclockable CPU and allowing it to use a lot of power under warranty...if they complain about something that makes complete sense they would loose their minds if a company would release a CPU that is highly overclockable but they would only allow very low stock settings and would void warranty even if you would increase power only a tiny bit.

 

[bit_user]

Yes people like him that complain about a company releasing an unlocked overclockable CPU and allowing it to use a lot of power under warranty...if they complain about something that makes complete sense they would loose their minds if a company would release a CPU that is highly overclockable but they would only allow very low stock settings and would void warranty even if you would increase power only a tiny bit.

Eh, leaving aside who you originally meant in "people like you" or are now talking about, let's address this latter point.

So, what's troubling to see is CPUs pushed well past their peak efficiency point, to eke out a few more % in a transparent attempt to snag a few more benchmark wins. I have no fundamental problem with overclocking, and then by extension it's hardly a bad thing to make some form of it easier, safer, and not violate the warranty.

With that said, there's some kind of disconnect when a CPU can be advertised as having one power dissipation level and you plop it in a motherboard and it comes up doing something else by default. We all know Intel indeed has leverage over its motherboard partners, so Intel isn't blameless in this whole affair.

I like AMD's approach, where you have to manually enable PBO. Otherwise, the behavior is quite reasonable. If Intel would do the same, I think that would be ideal.

 

[TerryLaze]

So, what's troubling to see is CPUs pushed well past their peak efficiency point, to eke out a few more % in a transparent attempt to snag a few more benchmark wins. I have no fundamental problem with overclocking, and then by extension it's hardly a bad thing to make some form of it easier, safer, and not violate the warranty.

That logic would stand if intel would force every review site to use a certain mobo (one that pushes the CPU to the edge) ,if there is proof of that I would like to see it.
If review sites are free to use whatever mobo they like then it's just them doing a terrible job reviewing CPUs.

 

[bit_user]

If review sites are free to use whatever mobo they like then it's just them doing a terrible job reviewing CPUs.

Again, the burden of proof is on you to show that motherboards which don't juice the power limits aren't the norm. If they are, then it's reasonable for review sites to use them.

 

[TerryLaze]

Again, the burden of proof is on you to show that motherboards which don't juice the power limits aren't the norm. If they are, then it's reasonable for review sites to use them.

It IS reasonable for them to use them...as an additional data point, "see what can happen if you don't choose your mobo wisely" , not use it as the only data point "this is the only thing possible"

Also what do you consider norm when talking about mobos? I'm pretty sure there are more locked mobos out there than there are z mobos.
And then z mobos are there for overclocking so the norm for z mobos is overclocking which many people that buy K CPUs don't want, they just want the extra clocks without wanting the CPU to use any overclock.

If a review just uses a z mobo , and claims norm or out of the box, they should then also disclaim it as an overclock review or they should also use a high end locked mobo and show both results.

Benchmarks are made because there is no norm, so everything has to be benchmarked on many GPUs or CPUs or ssds or rams or what have you.

If there where any norm we would have the MSX 2022 right now, and everybody would know that each one of them has the same performance in everything you run on it.