Question Anandtech.com article on 13900k and 7950x power scaling

[adamge]

A Lighter Touch: Exploring CPU Power Scaling On Core i9-13900K and Ryzen 9 7950X

www.anandtech.com

It's an interesting article, but IMO severely crippled by the reliance on "configured" power limit rather than measuring power usage. It's even more frustrating when one of the datapoints is measured power usage, so the author has access to this data.

In this scenario where configured power limits are basically hand waving guidelines, measured power usage (specifically, average power usage by the CPU during the benchmark run) is the only useful, meaningful, and concrete way to analyze the performance results relative to the power usage.

 

[Markfw]

Gotcha, you just dont like the numbers

No, thats not it, and not wehat I said, but I am done arguing with a troll.

 

[coercitiv]

I don't know about idle power but intel beating amd in mixed workloads isn't anything new, the same was true between ALD and zen 3.

The amd part was drawing way more power. If you actually do the math 22911 / 71 = 322pts vs 346 pts for the amd part. So the AMD part is 7.4% more efficient at 65w in heavy MT

Somebody doesn't understand the difference between peak power and average power, while also using one software for the performance benchmark and another for energy consumption. Meanwhile, in the world where people use the same workload to calculate efficiency, this stuff happens: the Intel part loses in efficiency at stock settings, let alone at 65W where performance relative to 7950X drops even more.

Spoiler: Cinebench nT Efficiency

I should note that Techpowerup uses physical power measurements in their reviews, with 30 readings per second straight from the 12-pin connector. Here's how the 7950X and 13900K behave in their application benchmark suite:

Spoiler: 13900K ~170W across mixed workloads.

Spoiler: 7950X ~125W across the same mix of workloads.

So the 7950X uses 25% less power while delivering 5% less performance in mixed workloads, all measured accurately via hardware at the 12-pin connector.

 

[Just Benching]

Somebody doesn't understand the difference between peak power and average power, while also using one software for the performance benchmark and another for energy consumption. Meanwhile, in the world where people use the same workload to calculate efficiency, this stuff happens: the Intel part loses in efficiency at stock settings, let alone at 65W where performance relative to 7950X drops even more.

Spoiler: Cinebench nT Efficiency

View attachment 75225

I should note that Techpowerup uses physical power measurements in their reviews, with 30 readings per second straight from the 12-pin connector. Here's how the 7950X and 13900K behave in their application benchmark suite:

Spoiler: 13900K ~170W across mixed workloads.

View attachment 75226

Spoiler: 7950X ~125W across the same mix of workloads.

View attachment 75227

So the 7950X uses 25% less power while delivering 5% less performance in mixed workloads, all measured accurately via hardware at the 12-pin connector.

View attachment 75228

I dont understand why in a thread about anandtechs results, we are talking about other reviews.

According to anandtechs numbers, the difference in efficiency is miniscule. Sure their numbers arent perfect since they meaure peak power, but when peak power is so much higher than the ppt limit, its safe to assume average draw is also higher.

Even if you think the power draw numbers from anandtechs are completely wrong, great, how can you compare efficiency then from this review? You can't.

Now regarding the tpup numbers, tpus stock result is basically 280+ watts for the 13900k. That's not iso wattage though. Still, if you want to accept these numbers, you need to also accept that rpl is not only much faster in single and mixed workloads, it also consumes WAY WAY less power. If someone decides to run these with same wattage, the difference will be catastrophically huge, the 7950x would need to consume almost twice as much for same performance. 👀

 

[Just Benching]

We also have to accept that the 13900k is 14% faster in games, again according to the tpup results. We cant just cherry pick what we like from a review and ignore the rest

 

[coercitiv]

I dont understand why in a thread about anandtechs results, we are talking about other reviews.

You had no problem talking about another review in this thread about Anandtech results. What happened since yesterday?

But let's not linger on this too much, tell us more about how you managed to calculate Cinebench efficiency using yCruncher PEAK power readings.

 

[Just Benching]

You had no problem talking about another review in this thread about Anandtech results. What happened since yesterday?

But let's not linger on this too much, tell us more about how you managed to calculate Cinebench efficiency using yCruncher PEAK power readings.

That review wasnt talking about mt efficiency but about mixed usage. Which is not what anandtech was testing. You brought tpus results to disprove what anandtech found.

Regarding your question, the same way people managed to calculate whos more efficient between the two when all they had was peak power draw. You didnt seem to be bothered by that

Anyways, since you seem to trust tpup, we do agree that in light workloads the 13900k is VASTLY more efficient than the 7950x,right? Its also plenty faster in games as well. Am i right?

 

[DrMrLordX]

I dont understand why in a thread about anandtechs results, we are talking about other reviews.

Because you are misrepresenting data.

 

[Hitman928]

Anyways, since you seem to trust tpup, we do agree that in light workloads the 13900k is VASTLY more efficient than the 7950x,right? Its also plenty faster in games as well. Am i right?

Nope and nope.

 

[Just Benching]

Because you are misrepresenting data.

So the ppt limited zen 4 didnt consume vastly more than the limit it was set, it was all my misrepresentation.

I mean come on.....

 

[Just Benching]

Nope and nope.

Then why bother posting a review you dont agree with? That seems weird

 

[Hitman928]

Then why bother posting a review you dont agree with? That seems weird

Single threaded does not equal lightly threaded in 2023. Additionally, if single threaded or very lightly threaded performance and efficiency are key driving factors for someone, they wouldn't be looking at any of these CPUs to begin with.

 

[Just Benching]

Single threaded does not equal lightly threaded in 2023. Additionally, if single threaded or very lightly threaded performance and efficiency are key driving factors for someone, they wouldn't be looking at any of these CPUs to begin with.

So you agree that the 7950x gulps way more power than the 13900k in st workloads, but you dont think its an important metric. Don't want to misinterpret what you said, so please feel free to correct me

 

[igor_kavinski]

Oh man, it's gonna be a 1000 pages till Zen 5 release if this keeps up...

 

[Hitman928]

So you agree that the 7950x gulps way more power than the 13900k in st workloads, but you dont think its an important metric. Don't want to misinterpret what you said, so please feel free to correct me

Way more is an exaggeration. Yes, there is a power penalty for AMD having a non-monolithic design that mostly shows up at idle or when using only 1-2 cores. This has been known for quite a long time, you are not bringing new information to the table. We are talking like, 10-11 watts of delta between the two. That's an LED light bulb worth of power, it's not a significant amount overall and again, we are talking about CPUs with 32 threads. This is like comparing the fuel efficiency of two 18 wheel trucks while they are idling or completely unloaded. No one cares who wins. The idle/light core usage is far more important in laptops and SFF systems, where AMD has monolithic designs to fight that battle.

 

[Just Benching]

Way more is an exaggeration. Yes, there is a power penalty for AMD having a non-monolithic design that mostly shows up at idle or when using only 1-2 cores. This has been known for quite a long time, you are not bringing new information to the table. We are talking like, 10-11 watts of delta between the two. That's an LED light bulb worth of power, it's not a significant amount overall and again, we are talking about CPUs with 32 threads. This is like comparing the fuel efficiency of two 18 wheel trucks while they are idling or completely unloaded. No one cares who wins. The idle/light core usage is far more important in laptops and SFF systems, where AMD has monolithic designs to fight that battle.

I don't think way more is an exaggeration. The techpowerup showed a 52% efficiency delta, and that's while the 13900k was faster. At similar performance testing that difference could easily climb to 100%, meaning that basically the amd part would use double the power for similar performance.

Now sure, you wanna call that insignificant, go ahead, but then don't tell me that the 10-15% efficiency difference in heavy MT workloads is significant because thats double standards

 

[Markfw]

I don't think way more is an exaggeration. The techpowerup showed a 52% efficiency delta, and that's while the 13900k was faster. At similar performance testing that difference could easily climb to 100%, meaning that basically the amd part would use double the power for similar performance.

Now sure, you wanna call that insignificant, go ahead, but then don't tell me that the 10-15% efficiency difference in heavy MT workloads is significant because thats double standards

The only plce that occurs is at 1-2 cores out of 32. And as said, the wattage is small. In A:LL other scenarios and workloads, the 7950x is much more effecient. Why do you persist in lying and trolling ?

 

[Just Benching]

The only plce that occurs is at 1-2 cores out of 32. And as said, the wattage is small. In A:LL other scenarios and workloads, the 7950x is much more effecient. Why do you persist in lying and trolling ?

So you admit that what I said is true, but then call me a lier and a troll. Makes sense

Sorry, come back with something more than personal attacks (this is what, the 3rd time?) if you want a reply. This is embarrassing, calm down, it's just a CPU.

 

[igor_kavinski]

it's just a CPU.

What do you gain by convincing everyone that Alder or Raptor Lake is more efficient?

 

[Markfw]

So you admit that what I said is true, but then call me a lier and a troll. Makes sense

Sorry, come back with something more than personal attacks (this is what, the 3rd time?) if you want a reply. This is embarrassing, calm down, it's just a CPU.

No, you mainttain its more effcient in all scenarios, and even above said the change could be as high as 100%. I call them as I see them, as you maintain this, even though every other poster here has proven you wrong. To continue this activity is called trolling.

 

[Just Benching]

What do you gain by convincing everyone that Alder or Raptor Lake is more efficient?

Never said it is, and I don't think ill gain anything, unless you plan on getting me something

 

[Just Benching]

No, you mainttain its more effcient in all scenarios

Never said that. That's you putting words in my mouth. It's called lying. The thing you accused me of.

I've said, 2 or 3 times, maybe more, that in full heavy MT the 7950x is more efficient. Want me to quote myself?


My very first post in this thread

The 7950x is indeed more efficient in very heavy MT workloads

LOL

 

[Carfax83]

I don't think this video has been posted yet, but PC World did a 13900K power consumption review. They tested total power consumption and didn't do performance per watt testing, which is a shame because that would have been more impactful to this debate.

Basically though, they corroborate what @Herald and I have been saying, that in light to moderate CPU usage, RPL has an advantage over Zen 4 while in heavy thread usage, Zen 4 has the advantage. The browsing power usage was the most eye opening, as the 7950x looked like it was drawing on average about 120w, while RPL was about 80w.

 

[Markfw]

I don't think this video has been posted yet, but PC World did a 13900K power consumption review. They tested total power consumption and didn't do performance per watt testing, which is a shame because that would have been more impactful to this debate.

Basically though, they corroborate what @Herald and I have been saying, that in light to moderate CPU usage, RPL has an advantage over Zen 4 while in heavy thread usage, Zen 4 has the advantage. The browsing power usage was the most eye opening, as the 7950x looked like it was drawing on average about 120w, while RPL was about 80w.

In EVERY scenario I saw, the 13900k was consuming a LOT more power(like as much as 100 watt or more than the 7950x), except this one test. Way to cherry pick.

 

[Just Benching]

I don't think this video has been posted yet, but PC World did a 13900K power consumption review. They tested total power consumption and didn't do performance per watt testing, which is a shame because that would have been more impactful to this debate.

Basically though, they corroborate what @Herald and I have been saying, that in light to moderate CPU usage, RPL has an advantage over Zen 4 while in heavy thread usage, Zen 4 has the advantage. The browsing power usage was the most eye opening, as the 7950x looked like it was drawing on average about 120w, while RPL was about 80w.

You can compare efficiency, the 13900k at 85w consumes the same amount of power as the 7950x at 65w in their review, and the scores are pretty similar. The 7950x has an advantage, but its around 10%. Thats in cinebench r23.

 

[Just Benching]

In EVERY scenario I saw, the 13900k was consuming a LOT more power(like as much as 100 watt or more than the 7950x), except this one test. Way to cherry pick.

You mean when power unlimited? Of course, if you ask the cpu to draw 350 watts, it will draw 350w, why is that a surprise? I mean the 7950x draws 200 more watts than the 13900T, does that tell us anything useful? Not really.

In the isowattage testing towards the end of the video you can see that even in heavy mt workloads the difference in efficiency is less than 10%