Intel Core i9-12900KS Review: Clocking in at 5.5 GHz

Strawman

Posts: 577   +294
You've described an uncommon, cherry picked scenario. One could have said the same about 5900x vs Comet Lake not all that long ago. The number of people such a scenario applies to is minimal compared to the market. People buying a 5090 are, for the most part, going to be looking at whatever CPUs are available at that time.
It's not a cherrypicked scenario. GPU's get replaced way more frequently than CPU's are. That's a fact. I for example have a 3090 and a 12900k. I plan to upgrade my 3090 with a 4090 but I don't plan to upgrade my 12900k, since the 4090 will still be the bottleneck. .


No, one couldn't have said the same about comet lake vs 5900x because there wasn't any performance difference. In fact, the 10900k was faster since it can easily support 4000+ ram kits. On the other hand, the 12900k gets lows that are higher than the 5950x's average in some games.
 

sreams

Posts: 307   +437
It's not a cherrypicked scenario. GPU's get replaced way more frequently than CPU's are. That's a fact. I for example have a 3090 and a 12900k. I plan to upgrade my 3090 with a 4090 but I don't plan to upgrade my 12900k, since the 4090 will still be the bottleneck. .


No, one couldn't have said the same about comet lake vs 5900x because there wasn't any performance difference. In fact, the 10900k was faster since it can easily support 4000+ ram kits. On the other hand, the 12900k gets lows that are higher than the 5950x's average in some games.

It is. The number of people who have the specific CPU you mentioned and will go on to upgrade to a 5090 without changing that CPU is small. We are also talking about theoretical future performance of future games and future GPUs. An article with benchmarks can only test for current performance. 1080p along with the tested hardware does not represent current performance.

As for 5900x vs Comet Lake, maybe read this article:


Specifically, the part where they say "Leading 1080p gaming" in the "For" column.
 

Strawman

Posts: 577   +294
It is. The number of people who have the specific CPU you mentioned and will go on to upgrade to a 5090 without changing that CPU is small. We are also talking about theoretical future performance of future games and future GPUs. An article with benchmarks can only test for current performance. 1080p along with the tested hardware does not represent current performance.

As for 5900x vs Comet Lake, maybe read this article:


Specifically, the part where they say "Leading 1080p gaming" in the "For" column.
I don't need to read any article. The test is done with mediocre memory for comet lake. But it doesn't matter, even in that test the difference is actually negligible. That is NOT the case with this review, were the 12900ks 1% lows are HIGHER than the 5950x's average. If you can't understand the difference then I really can't help you.

The number of people is irrelevant. Fact is, the 12900k CAN be used with a future high end card at 1440p (like lets say a 5090) while the zen 3 cpu's cant. I'm pretty sure there are plenty of people still rocking an 8700k or a 9900k, and they both work absolutely fine with new gpus (like a 3070 or a 3080). That's not the case with a Ryzen 7 1700.
 

Strawman

Posts: 577   +294
And this is how people ensure that they are always right.
Because I already read 15 reviews, and actually had the CPUs we are talking about ;)
From your article, the stock 10900k was tested with 2933 mhz ram. As I predicted...
 

sreams

Posts: 307   +437
Because I already read 15 reviews, and actually had the CPUs we are talking about ;)
From your article, the stock 10900k was tested with 2933 mhz ram. As I predicted...

Owning a CPU doesn't mean you have the experience/expertise to benchmark the same way these tech sites do.

And no... nowhere in the article does it say they ran the memory on the Intel system at 2933MHz. Under "Test Setup", it shows the memory used was 3600MHz DDR4 for both systems, and then it goes on to list the "Stock" and "OC" specifications for the motherboard (2933/4000 for Intel, 3200/4000 for AMD).

If these results are wrong, please show the other 14 reviews where Comet Lake matches or exceeds the 5900x.
 

Strawman

Posts: 577   +294
Owning a CPU doesn't mean you have the experience/expertise to benchmark the same way these tech sites do.

And no... nowhere in the article does it say they ran the memory on the Intel system at 2933MHz. Under "Test Setup", it shows the memory used was 3600MHz DDR4 for both systems, and then it goes on to list the "Stock" and "OC" specifications for the motherboard (2933/4000 for Intel, 3200/4000 for AMD).

If these results are wrong, please show the other 14 reviews where Comet Lake matches or exceeds the 5900x.
Yes, on the test setup is says that the stock results were run with memory at 2933. LOL.

I don't know what the heck you are talking about, most reviews show them pretty close. Techpowerup has the 10900k at the top for example


Anyways, it doesn't matter, the difference is really small. That is not the case with the 12900k. The 12900k beats the crap out of zen 3 in gaming.
 

sreams

Posts: 307   +437
Yes, on the test setup is says that the stock results were run with memory at 2933. LOL.

Please post the entire quote where it says this.

I don't know what the heck you are talking about, most reviews show them pretty close. Techpowerup has the 10900k at the top for example


Anyways, it doesn't matter, the difference is really small. That is not the case with the 12900k. The 12900k beats the crap out of zen 3 in gaming.

I asked for your 14 reviews. Not the one outlier you could find.
 

Neatfeatguy

Posts: 876   +1,511
Yes, on the test setup is says that the stock results were run with memory at 2933. LOL.

I don't know what the heck you are talking about, most reviews show them pretty close. Techpowerup has the 10900k at the top for example


Anyways, it doesn't matter, the difference is really small. That is not the case with the 12900k. The 12900k beats the crap out of zen 3 in gaming.

I think this might be a more relevant graph to use from TPU to show the 12900k in relative performance at 720p against the 5900x.

The 12900k was originally tested with DDR5 6000. However, TPU's scaling testing with different RAMs and speed is a better option for comparison against the 5900x.

5900X was tested with DDR4 3200.
The 12900k scaling test used DDR4 3600; while it's slightly faster than what the 5900X used, it would be a more accurate comparison of the two processors for relative gaming performance.

At 720p the 12900k, using similar RAM has about an 8% lead over the 5900x in gaming.
At 1080p it drops to about 6%.
At 1440p it drops to 5%.
At 4k it drops to 1%.

I wouldn't say that the 12900k beats the crap out of the 5900x, but it does have a small lead at 1440p and under. At 4k it's pretty much a wash.
 

Strawman

Posts: 577   +294
Please post the entire quote where it says this.



I asked for your 14 reviews. Not the one outlier you could find.
Man are you kidding me? Its your own link

2x 8GB Trident Z Royal DDR4-3600 - Stock: DDR4-2933, OC: DDR4-4000

This means the used the trident Z royal ddr3600 kit @ 2933 for the stock numbers and @ 4000 for the overclocked numbers. If you go to the performance graphs youll notice there are 2 bars for each cpu, one at stock and one oced. Are you getting it now?

You want another review? How about the one from the very site you are posting right now


1 fps difference....

I dont know why you are willing to die on this hill. 10900k vs 5950x is a wash when it comes to gaming, the higher memory speed you put on the 10900k the faster it is.
 

sreams

Posts: 307   +437
Man are you kidding me? Its your own link

2x 8GB Trident Z Royal DDR4-3600 - Stock: DDR4-2933, OC: DDR4-4000

This means the used the trident Z royal ddr3600 kit @ 2933 for the stock numbers and @ 4000 for the overclocked numbers. If you go to the performance graphs youll notice there are 2 bars for each cpu, one at stock and one oced. Are you getting it now?

So... wait a minute. You are saying that both CPUs were tested overclocked with RAM at 4000MHz, and the Ryzen's still won more times than not? Because that's what the charts show.

But it's all moot anyway. You are talking about an unknown future. By the time the 5090 comes out, the 5900x, 5950x, and 12900KS will all be years old and very much superior CPUs will be available. The only thing we can know for sure is how today's CPUs and GPUs perform today. And, as Neatfeatguy brought up... The difference at the resolutions gamers with bleeding edge systems play at is a few percent at most. Testing such systems at ultra-low resolutions is nothing more than another synthetic benchmark.

The 12900KS may perform well, but I'd be wary to hold onto one long enough to throw a 5090 into such a system. Heat and power draw may mean these things won't even last that long before failing. It reminds me of the release of the Pentium III at 1.13GHz, which, as it turned out, Intel pushed too far.
 

Strawman

Posts: 577   +294
So... wait a minute. You are saying that both CPUs were tested overclocked with RAM at 4000MHz, and the Ryzen's still won more times than not? Because that's what the charts show.

But it's all moot anyway. You are talking about an unknown future. By the time the 5090 comes out, the 5900x, 5950x, and 12900KS will all be years old and very much superior CPUs will be available. The only thing we can know for sure is how today's CPUs and GPUs perform today. And, as Neatfeatguy brought up... The difference at the resolutions gamers with bleeding edge systems play at is a few percent at most. Testing such systems at ultra-low resolutions is nothing more than another synthetic benchmark.

The 12900KS may perform well, but I'd be wary to hold onto one long enough to throw a 5090 into such a system. Heat and power draw may mean these things won't even last that long before failing. It reminds me of the release of the Pentium III at 1.13GHz, which, as it turned out, Intel pushed too far.
No, when they were both tested with 4000 there was pretty much a tie. The stock 10900k was ran with 2933, the stock zen 3 were tested with 3200. Man seriously, cant you read a graph?

Thing is, 4000 is an extreme oc for the zen 3 most cpus can't get 1 to 1 if at those speeds. For cometlake 4400 is a walk in the park. Hence my point, when you push memory to the max, cometlake is faster.

Heat and power draw in gaming is not an issue, the 12900k is way more efficient than the 5950x in gaming. So if anything you should be more concerned about the heat and power draw of zen 3 during gaming, it stinks. As shown by igorslab and derbauer, the 5950x consumes twice as much as the 12900k per fps.

You know the same was said about 8700k vs r7 1700? In resolutions people play there isnt a big difference. Well yeah sure, but the guy with the 8700k can put a 3070 or a 3080 into his system and game at 1440p. The guy with the 1700 can't, he will have major bottleneck and need an upgrade
 

sreams

Posts: 307   +437
No, when they were both tested with 4000 there was pretty much a tie. The stock 10900k was ran with 2933, the stock zen 3 were tested with 3200. Man seriously, cant you read a graph?

Thing is, 4000 is an extreme oc for the zen 3 most cpus can't get 1 to 1 if at those speeds. For cometlake 4400 is a walk in the park. Hence my point, when you push memory to the max, cometlake is faster.

Heat and power draw in gaming is not an issue, the 12900k is way more efficient than the 5950x in gaming. So if anything you should be more concerned about the heat and power draw of zen 3 during gaming, it stinks. As shown by igorslab and derbauer, the 5950x consumes twice as much as the 12900k per fps.

You know the same was said about 8700k vs r7 1700? In resolutions people play there isnt a big difference. Well yeah sure, but the guy with the 8700k can put a 3070 or a 3080 into his system and game at 1440p. The guy with the 1700 can't, he will have major bottleneck and need an upgrade
You completely misunderstand what I am saying about power draw and heat. Heat "per fps" is meaningless in terms of a life expectancy of a CPU. Absolute heat generation is what matters. If you do anything serious beyond gaming, the 12900KS is an inferno. Will it last? Nobody knows yet.
 

Strawman

Posts: 577   +294
You completely misunderstand what I am saying about power draw and heat. Heat "per fps" is meaningless in terms of a life expectancy of a CPU. Absolute heat generation is what matters. If you do anything serious beyond gaming, the 12900KS is an inferno. Will it last? Nobody knows yet.
I dont know about the ks but my 12900k sits at 75c during cbr23 10 minutes at stock with a small single tower cooler. Undervolted its at 65. Nobody is going to use the stock 240w power limits for those kinds of tasks anyways so its kinda irrelevant honestly.
 

sreams

Posts: 307   +437
I dont know about the ks but my 12900k sits at 75c during cbr23 10 minutes at stock with a small single tower cooler. Undervolted its at 65. Nobody is going to use the stock 240w power limits for those kinds of tasks anyways so its kinda irrelevant honestly.

We do know about the KS. In Techspot's testing, 5 minutes of CBR23 resulted in a temp of 102C, which is nuts.
 

Strawman

Posts: 577   +294
We do know about the KS. In Techspot's testing, 5 minutes of CBR23 resulted in a temp of 102C, which is nuts.
Yes but the same results were shown with the 12900k as well. Techpowerup had it at like 95+,and I know for a fact thats not the case even with a small air cooler

Actually, from the 12900k techspot review, the h115i ran at 100c and the msi 360 peaked at 96c. While my u12a keeps my 12900k at 75..not even highly overclocked I dont hit 100c.
 
Last edited:

sreams

Posts: 307   +437
Yes but the same results were shown with the 12900k as well. Techpowerup had it at like 95+,and I know for a fact thats not the case even with a small air cooler

Actually, from the 12900k techspot review, the h115i ran at 100c and the msi 360 peaked at 96c. While my u12a keeps my 12900k at 75..not even highly overclocked I dont hit 100c.

Which may very well suggest your setup is throttling the CPU. If every review is seeing nearly 100C, but one person in a comments section is claiming to see 25 degrees lower, we'd need to see a lot more evidence to discount all the reviews.
 

Strawman

Posts: 577   +294
Which may very well suggest your setup is throttling the CPU. If every review is seeing nearly 100C, but one person in a comments section is claiming to see 25 degrees lower, we'd need to see a lot more evidence to discount all the reviews.
Nope, no throttling, I get 27800 points in cbr23 which is pretty much what reviews get
 

sreams

Posts: 307   +437
Nope, no throttling, I get 27800 points in cbr23 which is pretty much what reviews get
And yet, your temps are not what the reviews get. So something is up with your results. Perhaps whatever tool you are using to monitor temps has a bug. Either way, your results are not consistent with what the rest of the world is seeing.
 

Strawman

Posts: 577   +294
And yet, your temps are not what the reviews get. So something is up with your results. Perhaps whatever tool you are using to monitor temps has a bug. Either way, your results are not consistent with what the rest of the world is seeing.
Or maybe its not that hot...