News Intel Launches Raptor Lake at AMD: 24-core i9-13900K Arrives Oct 20 for $589

[Zerk2012]

You have 4 different buyers.

1. Buys Intel.
2. Buys AMD.
3. Buys what has the best performance.
4. Buys the lower priced platform (CPU, board) and could care less about a 2 or 3% performance difference.

 

[Phaaze88]

You have 4 different buyers.

1. Buys Intel.
2. Buys AMD.
3. Buys what has the best performance.
4. Buys the lower priced platform (CPU, board) and could care less about a 2 or 3% performance difference.

5. Buys what has the best overall performance for the money spent. [Mixing 3 and 4.]

 

[-Fran-]

5. Buys what has the best overall performance for the money spent. [Mixing 3 and 4.]

6. Buys whatever the store clerk tells them goes BRRRRR for the money.

Regards xD

 

[Phaaze88]

6. Buys whatever the store clerk tells them goes BRRRRR for the money.

Regards xD

Haha! Right, I forgot that was a thing.
"Intel is best for games. AMD is the cheap(er) stuff and not as good."

 

[Zerk2012]

5. Buys what has the best overall performance for the money spent. [Mixing 3 and 4.]

OK 5 then it is. EDIT OK now were @ 6.
5. would be really close to 4.

I'm 1. Never actually bought a AMD processor for me but have used a bunch of the video cards.
Bought a Dell a ton of years ago, then all home built, 2500K, to 4790K, to 10600K with absolutely no reason to upgrade yet.

The Dell lasted a long, long time till I started gaming the 2500K was the stuff at that time.

 

[BeedooX]

I think all of us are looking forward to seeing how RPL stacks up.

No, not really at all. All CPUs today are absurdly fast that you can't really go wrong - unless you need to justify to yourself that you made the best $/perf purchase. Personally, that part used to be fun, like 20 years ago - but now I'm older, it's just a case of buy what you can afford, so it'll probably be a 7950X and whatever the best, most expensive motherboard and RAM combo I can get, along with a brand new custom loop...

Even my old Threadripper 2950X CPU plays games fast enough - as long as you don't feel guilty about the cost of your PC - put your ego away, enjoy what you bought, it's plenty fast.

 

[BeedooX]

You have 4 different buyers.

1. Buys Intel.
2. Buys AMD.
3. Buys what has the best performance.
4. Buys the lower priced platform (CPU, board) and could care less about a 2 or 3% performance difference.

If you "could care less", you're saying that "2 or 3% performance difference" IS IMPORTANT!

 

[Zerk2012]

If you "could care less", you're saying that "2 or 3% performance difference" IS IMPORTANT!

No I meant what I said people will buy the lower priced platform because of the price and the few % of performance difference don't matter because they saved 100 bucks or so for a nice upgrade from what they had. When things get this competitive it's a good thing buy the lower priced platform and you still got very good stuff.

EDIT people can be funny I've seen a bunch of builds where they would but a K processor and pair it with a X10 motherboard to get the speed but save a bunch on the motherboard.

 

[ddrock]

Interesting times...

Intel says 15% single & 41% multi-thread increase over i9-12900K in SPECInt2017.

Anandtech SPECInt2017 benchmarks put the R9 7950X at an average of 15.8% single and 42.7% multi-threaded over the 12900K.

 

[spongiemaster]

No I meant what I said people will buy the lower priced platform because of the price and the few % of performance difference don't matter because they saved 100 bucks or so for a nice upgrade from what they had. When things get this competitive it's a good thing buy the lower priced platform and you still got very good stuff.

EDIT people can be funny I've seen a bunch of builds where they would but a K processor and pair it with a X10 motherboard to get the speed but save a bunch on the motherboard.

He was pointing out a grammar error you made. What you should have said is "couldn't care less." As he pointed out, if you could care less, then you think 2 or 3% is important.

 

[spongiemaster]

Interesting times...

Intel says 15% single & 41% multi-thread increase over i9-12900K in SPECInt2017.

Anandtech SPECInt2017 benchmarks put the R9 7950X at an average of 15.8% single and 42.7% multi-threaded over the 12900K.

The top end is going to be close. It looks like Intel will win by more and more as you head down the product stack. The cheaper you're going, the better Intel will be for you. Who ever thought that would be the case?

 

[KyaraM]

Other test suites showed Alder Lake leading in gaming benchmarks over Raphael, or both extremely close together in allmost all cases. Comparing all the test suites I looked at so far, including the one on this site, it's pretty much a wash already. TPU did a 12 game test and Alder Lake actually won; I find that somewhat more useful than 5 games and that's it. That means Raptor will most likely pull ahead as a whole. Raptor was also shown in leaks to provide much better 1% lows, which are also important. I really would test with more games before giving any one the gaming crown. It was the same with the 5800X3D, HWU's 50 game benchmark actually had it pretty much equal with the 12900K at only 1% better (meaning the 12900KS most likely beat it), and that is honestly a way better cross-section through existing video games than looking at only 5 games. So I'm pretty sure Raptor will pull ahead. Multi-threaded performance seems to end up rather similar if presentations are to be believed.

 

[Zerk2012]

They should (gives Intel a new idea) come out with a good binned 13600KS

 

[Giroro]

Too bad Windows 11 sucks. These mini-core Intel processors are a complete non-option until Microsoft either gives windows 11 a good GUI, or gives Windows 10 a better scheduler.

I wonder how much money Intel is losing because Microsoft is incompetent.

 

[tennis2]

Paraphrased: More games tested = better

More games tested for a composite score could dilute the average with many similar/equal games or even games where the results are the same for most/many CPUs all up and down the Intel/AMD stack (GPU-bound). Whereas fewer CURATED games (especially ones that actually scale with CPU prowess) allows the differences to show more.

 

[peachpuff]

Leadership gaming performance? I dunno about that...

 

[TerryLaze]

More games tested for a composite score could dilute the average with many similar/equal games or even games where the results are the same for most/many CPUs all up and down the Intel/AMD stack (GPU-bound).

If that represents the market or "real life" then what's the problem?

Whereas fewer CURATED games (especially ones that actually scale with CPU prowess) allows the differences to show more.

The issue is if they find 4-5 games that scale well with cache and only show you the results of those, how is that relevant to actual performance?
You have to show the good performance but you also have to show a cross section of what to expect when running a lot of different games.

 

[-Fran-]

According to Linus, they tested the i9 12900K with DDR5-4800 and loose timing while using DDR5-5600 with the i9 13900K, so definitely use a lot of salt with these numbers. Specially on gaming.

Regards.

 

[KyaraM]

More games tested for a composite score could dilute the average with many similar/equal games or even games where the results are the same for most/many CPUs all up and down the Intel/AMD stack (GPU-bound). Whereas fewer CURATED games (especially ones that actually scale with CPU prowess) allows the differences to show more.

We have a saying here. "Never trust a statistics you didn't forge yourself." Test websites are free to pick games for their benches as they like, and often include games because readers want to see them, not because they are oh-so good at showing CPU strenghts. I found three test suites that all saw variations in which CPU leads already, and only a single one addressed that it might be due to bias; all others, including Tom's, acted as if their conclusions were absolute. Running five games in a test suite and then calling it a day is statistically a horribly undersized sample. On the other hand, 50 games from the past couple years plus maybe one or two very popular older ones is much, much closer to actual rl conditions, if you like it or not. It gives a better overview, and offers more people a more accurate view on how their CPU stacks up to others in their games. It's also harder to forge results. So yeah, I know it sucks for the people doing those tests because it's a lot of work. But it is so much better and more conclusive for the customer...

 

[spongiemaster]

If that represents the market or "real life" then what's the problem?

The issue is if they find 4-5 games that scale well with cache and only show you the results of those, how is that relevant to actual performance?
You have to show the good performance but you also have to show a cross section of what to expect when running a lot of different games.

In the end does it really matter? I don't understand why AMD and Intel are so focused on gaming performance in the first place. Is any modern midrange CPU or better really handicapping gaming performance? Will any typical gamer be able to tell the difference between a 10600 and a 7950x in a blind test? It's not the old days where we were trying to reach a steady 30fps. When you're going from 330 to 355 fps it just doesn't seem to matter.

 

[jasonf2]

When did Zen 4 launch? Oh, that's right...

It doesn't matter that Zen 3 was a year old, because there was literally no update around the Alder Lake launch. The Ryzen 7 5800X3D didn't come out until May 29, 2022. Meanwhile, Zen 4 is coming out less than a month ahead of Raptor Lake. Obviously, no one can compare RPL and Zen 4 performance yet without breaking embargo, and AMD shouldn't have RPL chips for testing. But if his point is that he wants to see how RPL stacks up to the new Ryzen chips, that's entirely valid and waiting a few weeks is hardly that big of a deal.

In this game timing is everything. Intel under Gelsinger seems to be doing a great job of positioning against AMD releases to mitigate any real performance dominance. Zen 4 looks to be very, very solid, but Raptor lake isn't going to let it command the performance crown the way that Zen 3 did at launch. Either way you go you are getting solid IPC and clock boosts, not the ~10% annual float that equates to almost nothing that we typically get when one company is super dominant. Regardless, for Intel, this looks to be a stopgap continuation of Alderlake. Eyes should really be on Meteorlake. When Intel is able to integrate third party process and production and mix multiple process nodes into their packaged chips AMD, being fabless, will forever lose any ability to outsource a better production node. Intel will also gain the advantage of lengthening the use of mature nodes in tile components giving them a price advantage. Only benchmarks will tell, but the next year or two are shaping up to be really great for consumers.

 

[-Fran-]

In the end does it really matter? I don't understand why AMD and Intel are so focused on gaming performance in the first place. Is any modern midrange CPU or better really handicapping gaming performance? Will any typical gamer be able to tell the difference between a 10600 and a 7950x in a blind test? It's not the old days where we were trying to reach a steady 30fps. When you're going from 330 to 355 fps it just doesn't seem to matter.

I don't disagree with the overall premise, but "gaming" is just as ubiquitous as "driving" nowadays. People does it because modern generations are just showered with electronic games from birth. I don't think there's any other way to put it. You have "gaming" everywhere now, on every demographic and segment, so not accounting for it is bad marketing, or so I believe.

As for the second part, I absolutely agree. I do remember mentioning somewhere else that nowadays the "win"s are more technical than practical. The only silver lining of it is "future"; which is a bit cringey, but not completely false. Games should get more demanding gen over gen, but we're back to the hardware outpacing the software (CPU-wise) and most game engines are just not optimized to make full use of all the CPU power at their disposal. How can this be true? Well, you still have several big ("AAA") depending mainly on 1 core, which we can all agree in 2022 is just sad. Titles that actually go wide have magnificent scaling and they show how well they can utilize 6+ cores. This is just from the technical side and most give good enough frames for most people to enjoy and circling back to what you say, the average range (mid-tier/range) CPUs can handle those with no issues, for the most part. This being said, stutter is noticeable and 1% lows are the important metric to start using nowadays. Both Intel and AMD have their fallings here and there, depending on the title. This is to say, the only thing where a lower end CPU could show a real weakness is in 1% lows when the CPU is not scheduling correctly.

Regards.

 

[KyaraM]

In the end does it really matter? I don't understand why AMD and Intel are so focused on gaming performance in the first place. Is any modern midrange CPU or better really handicapping gaming performance? Will any typical gamer be able to tell the difference between a 10600 and a 7950x in a blind test? It's not the old days where we were trying to reach a steady 30fps. When you're going from 330 to 355 fps it just doesn't seem to matter.

Thry are so focused on it because, without offence to anyone here, many tend to just look at numbers. Higher number = better, no matter if the difference is 1 FPS or 100, or if the difference dwindles under RL conditions where not everyone uses a 3090Ti and plays at 1080p or 720p where you actually stand a chance to see a difference. Gaming is also what the biggest part of non-professional customers use their machines for. What does your average Joe/Jane playing video games on their computer 3h in the evening care for more; (theoretical) gaming performance or performance in, let's say, Blender or Adobe Lightroom? Sure, some people have other hobbies like casual video editing or music production they use their computers for. But those aren't the majority. The majority are gamers of all ages and budgets. So that's where the marketing is aimed at.

You are 100% correct that it doesn't matter at all which modern CPU you pick for your gaming-only rig, and in all fairness, even the longevity angle is only really an angle when you are environmentally conscious and want to avoid e-waste (and even then, recycling into other systems or reselling is always an option). Let's say you buy a 12100 today with a decent B660 board, and let's say it lasts you two or three years. We now buy a 12700k instead with a Z690 board because that is what people realistically do with these chips, let's say that one lasts you four or five years, or even six. The 12100 costs $100 plus 120 for the board (or less if you are fine with a H610), the 12700K costs $400 plus minimum 200 for the board unless you are fine with losing the OC capabilities. What is the better value, and if you only gamed, watched video etc, and nothing more, did it really make sense to get the 12700K? Flip that around to AMD, same thing. High-end chips only make really sense if you a) expect them to last much longer than those six years mentioned above, or b) you do other stuff with the machine, too, like code compiling, video editing etc, that profits more from a better CPU. That gets even more pronounced at higher resolutions. And yes, this also applies to 3D cache chips.

 

[cyrusfox]

These mini-core Intel processors are a complete non-option until Microsoft either gives windows 11 a good GUI, or gives Windows 10 a better scheduler.

I'm running a 12900k and 12600k on Windows 10, they work out fine. Only issue I had was with handbrake only using the e-cores. for that I installed process lasso and forced it to use all cores. Otherwise it is very speedy, this was an upgrade from a 10850k and a 11700k and was more than worth it from the 15-25% single core uplift and massive multi-core improvement. I wouldn't bother with Windows 11, I should really move to Linux and sandbox(VM) Windows anyways...

 

[RodroX]

Announcements are good, but still a long 20+ days till reviews .

On a side note, if "the 13900K will have to meet certain temperature and power conditions to reach 5.8 GHz", I wonder what will you need to reach the "6 GHz" of the yet not announced intel part ... oh well so much time left for reveiws!