Question Raptor Lake - Official Thread

[Hulk]

Since we already have the first Raptor Lake leak I'm thinking it should have it's own thread.
What do we know so far?
From Anandtech's Intel Process Roadmap articles from July:

Built on Intel 7 with upgraded FinFET
10-15% PPW (performance-per-watt)
Last non-tiled consumer CPU as Meteor Lake will be tiled

I'm guessing this will be a minor update to ADL with just a few microarchitecture changes to the cores. The larger change will be the new process refinement allowing 8+16 at the top of the stack.

Will it work with current z690 motherboards? If yes then that could be a major selling point for people to move to ADL rather than wait.

 

[nicalandia]

Intel Core i9-13900K Raptor Lake CPU Offers Same Performance As Core i9-12900K With "Unlimited Power" at Just 80W

Intel Core i9-13900K Raptor Lake CPU shows the same performance as Core i9-12900K Alder Lake while consuming just 80W of power.

wccftech.com

Tempting but the holiday price will determine my purchase between the 7050X and 13900K at low wattage.

Get a 13900 At 65Watts it's doing the same as the 12900K....

 

[nicalandia]

Intel Core i5 13400, 13500 and 13600 share the same die and Cache System Configuration. 1.25 MiB L2$ Per P Core and 2 MiB L2$ Per e-Core Cluster

Here is the sample where Geekbench5 Reads the P-Core System Cache (1.25 MiB L2$)

ASUS System Product Name vs Founder Predator PO5-650 - Geekbench

Here is the sample where Geekbench5 Reads the e-Core System Cache(2.00 MiB L2$)

ASUS System Product Name vs Acer Predator PO5-650 - Geekbench

The 13600K, 13700K and 13900K share the same die and Cache System Configuration.
2.00 MiB L2$ Per P core and 4 MiB L2$ Per e-Core Cluster

ASUS System Product Name vs ASUS System Product Name - Geekbench

 

[DrMrLordX]

Tempting but the holiday price will determine my purchase between the 7050X and 13900K at low wattage. Intel still runs hotter by 15 degrees Celsius, so a deeper discount is called for before I jump to buy the 13900K.

Remember the temp isn't really that important, unless you just want it low as some kind of a temperature e-peen kind of thing. They're both designed to run "hot" and be okay with it over the warranty period. They will not burn up. And you can always underclock/undervolt to make either CPU fit in with just about any compatible cooler you throw at them. Just be sure to look at power vs. performance to make your decision, since they don't necessarily scale the same way with power.

 

[Kocicak]

Intel still runs hotter by 15 degrees Celsius, so a deeper discount is called for before I jump to buy the 13900K.

What? I was able to cool 13900K at 350W with a little 240 AIO (barely, larger radiator would be better, but who would want to run it at this insane power draw long term anyway), and at 195W it holds below 90°C after half an hour of Cinebench with a small Freezer 34 air cooler in a poorly ventilated case. It is very easy to cool.

 

[ondma]

This guy has the hardware, and is using an RTX 4090 so there is no GPU bottlenecking like there was with the RTX 3000 series, even at 1440p max settings. As I predicted, Raptor Lake is the faster gaming CPU, and will be even faster when DDR5 7000+ kits become widely available later this month. These benchmarks were done with DDR5 6200, which is more advantageous to Zen 4 than it is to Raptor Lake.

Couldn't do embedded links because he blocked that feature from being used for his videos as he wants the clicks. If you do watch the videos please leave a like:

7700x vs 13700K

7950x vs 13900K

5800x 3D vs 13600K

Not much going on in those videos. Be interesting to see scenes with more characters on screen and more going on. That said, Techspot reviews show 7xxx trading blows or only slightly ahead of 12900k, so if Intel can eke out a few percent improvement with RL over AL, they should be very competitive or slightly ahead of regular Zen 4. 7xxx 3D will almost certainly have the gaming crown when it comes out though.

 

[Kaluan]

I have no idea. Perhaps it's because the CPUs have not been officially released yet and he wants to be informative. Does it matter? It's not like he's an official source or something. He can put whatever he wants in his video description.



Because he's giving an early glimpse of the 13th gen gaming performance and his benchmark method was actually decent. There are other Raptor Lake vs Zen 4 comparisons on YouTube already with the RTX 4090 but they were testing at 4K......which for a CPU performance preview is nonsensical; even with an RTX 4090.

I think we can all appreciate proper benchmarking. These videos are far superior to that trash from that idiot framechaser you posted a few pages back at least.

Bruh, no offense, but that "test" is wildly fishy.

For starters, why does it only show the 7700X clock to only 5350MHz? It's pretty wideley known that 7700X boosts to 5550MHz basically sustained, in gaming, check any respectable outlet, or just people testing it with overlays on.
Second, why 6200MTs RAM? And how is it advantageous for AMD more than Intel lol
AMD said 2000:3000:3000 (timings aside) is a sweet spot for a reason, 6200 ain't really a thing, unless you make sure IF also runs at 2066MHz. So something like 3:1:1, otherwise there is a slight performance regression.
His supposed 5800X3D runs at 4450MHz or less in RDR2 (from review data, this is the expect clock)... in his 13600K video... but at 4550MHz in his orignal 5800X3D vs 5800X video from months back... yeah, I'm pretty sure this is just another one of those clickbait make numbers up as you go channel.
In the description of the same 13600K vs 5800X3D video, he writes "5 13600k + rtx 4090 vs r5 7600x + rtx 4090 test at 1440p max settings" ahahahaha, I think he/they themselves got lost in the BS they're trying to spin.

I think they just make up crap as they go, casually correcting it (ie making it more beliavable) as time goes in new videos.

The fact that you can't embed them is also bizzare. Most big channels don't do that, but this 1K peepsqueak thinks it's a good idea for their bottom end lol

I wish there'd be a rule against posting clickbait side-by-side "comparison" shovelware videos on this forum.

 

[Carfax83]

Bruh, no offense, but that "test" is wildly fishy.

For starters, why does it only show the 7700X clock to only 5350MHz? It's pretty wideley known that 7700X boosts to 5550MHz basically sustained, in gaming, check any respectable outlet, or just people testing it with overlays on.

Is that all core boost or single core boost? According to the 7700x info page on AMD's website, the 7700x's max boost single core clock speed is 5.4ghz:

AMD 7700x

Usually a CPU will have higher single core boost than all core boost so his numbers makes sense if you ask me. The only thing weird I thought was his AMD CPUs running a bit cool compared to what I've heard but I don't know what his testing condition is.

Second, why 6200MTs RAM? And how is it advantageous for AMD more than Intel lol
AMD said 2000:3000:3000 (timings aside) is a sweet spot for a reason, 6200 ain't really a thing, unless you make sure IF also runs at 2066MHz. So something like 3:1:1, otherwise there is a slight performance regression.

I'm assuming he would know how to configure the Zen 4 CPUs properly but I have no way of knowing admittedly.

His supposed 5800X3D runs at 4450MHz or less in RDR2 (from review data, this is the expect clock)... in his 13600K video... but at 4550MHz in his orignal 5800X3D vs 5800X video from months back... yeah, I'm pretty sure this is just another one of those clickbait make numbers up as you go channel.

A 100hz difference isn't large at all, plus in that previous comparison he was using an RTX 3090 which would have been more GPU limited than the 4090 and likely allowed for a higher CPU boost clock speed. Or he may have swapped to a different cooling solution.

As for his channel being clickbait, I'm not sure. He's been around posting videos for a year with all sorts of comparisons and not just elite level hardware. Usually clickbait channels crop up before hardware releases only and are shortlived. You see that a lot with mobile phones in particular.

Also, I don't know why people are surprised that Raptor Lake would pull ahead of Zen 4 in gaming. Depending on configuration, Zen 4 was barely ahead of Alder Lake and that was with a clock speed advantage.

Now that the clock speeds are more evenly matched and Raptor Lake has other enhancements, it's no surprise that it would surpass Zen 4 in games.

I wish there'd be a rule against posting clickbait side-by-side "comparison" shovelware videos on this forum.

You're probably just saying that because they show Raptor Lake winning, which as I said before, should surprise no one given what we knew about Alder Lake.

 

[Carfax83]

7xxx 3D will almost certainly have the gaming crown when it comes out though.

More than likely I'd say, but Raptor Lake with high speed DDR5 7000 and above may be harder to take down than initially thought.

 

[nicalandia]

More than likely I'd say, but Raptor Lake with high speed DDR5 7000 and above may be harder to take down than initially thought.

7800X3D just wont care about RAM Speeds

 

[Markfw]

More than likely I'd say, but Raptor Lake with high speed DDR5 7000 and above may be harder to take down than initially thought.

I think in 2 days we will know a lot more. I am waiting for official reviews. And anything over ~253 watts that Raptor lake is running at, I will ignore, as thats the most the "official" chip should be using. These 350 watt unlimited benchmarks are crap IMO. You can't sustain that even with an AIO.

So we will See in 2 days..

 

[Carfax83]

7800X3D just wont care about RAM Speeds

I'll remember this comment when the 7800X3D launches and hope that reviewers use the stock memory configuration of DDR5 5200 against Raptor Lake with DDR5 7600

Ram speed definitely helps, especially in games because the datasets are so large that the CPU goes to system memory often to retrieve data.

Cache is effective but it's not the be all end all.

 

[Harry_Wild]

Get a 13900 At 65Watts it's doing the same as the 12900K....

Day after Thanksgivings !

 

[nicalandia]

I'll remember this comment when the 7800X3D launches and hope that reviewers use the stock memory configuration of DDR5 5200 against Raptor Lake with DDR5 7600

Just like stock 5800X3D 3200 did not care much about DDR5 6400 12900KS?

This time around the Reviewers will do a DDR5 6000(Sweet Spot) vs Tune to moon Raptor Lake with DDR5 8000 and it wont matter much...

You can bookmark this post...!

 

[Kocicak]

I have put the AIO back on the CPU and spent some time with playing with Intel extreme tuning utility, here are my OC results (only of P cores). I ended up setting max turbo for the best two cores to 60 and 59 to all other. When I tested all core performance, it could run at 57 multiplier with higher voltage offset.

So far it seems that all P cores can be completely stable at 5900 MHz. I have not tested it thoroughly yet.

I have seen all P cores boost up to 6100 MHz, not stable, I tried allocating 6100 only to the best two P cores, not stable (passed cca half of 1T run), but I did not mess with voltages much. I believe 6100 could be achieved with some more tweaking.

Here are my results for 1, 2, 4, 8 and all threads load:



Some screenshots:

 

[nicalandia]

I have put the AIO back on the CPU and spent some time with playing with Intel extreme tuning utility, here are my OC results. I ended up setting max turbo for the best two cores to 60 and 59 to all other. When I tested all core performance, it could run at 57 multiplier with higher voltage offset.

So far it seems that all cores can be completely stable at 5900 MHz. I have not tested it thoroughly yet.

I have seen all cores boost up to 6100 MHz, not stable, I tried allocating 6100 only to the best two cores, not stable (passed cca half of 1T run), but I did not mess with voltages much. I believe 6100 could be achieved with some more tweaking.

Here are my results for 1, 2, 4, 8 and all cores load:

View attachment 69381

Some screenshots:

View attachment 69382

View attachment 69383

View attachment 69384

Impressive...! Can you Run Geekbench5 and see if your MT Score has improved? It would appear that GB5 Just don't like your DDR4 RAM

 

[Kocicak]

I am not doing MT at 360W again... You can see it has minimal benefit anyway.

 

[Carfax83]

Just like stock 5800X3D 3200 did not care much about DDR5 6400 12900KS?

The 5800X3D with DDR4 3800 was only 1% faster than the regular 12900K with DDR5 6400 in HWU's 40 game review.

But as everyone knows, Alder Lake had some glaring weaknesses that have been fixed with Raptor Lake; namely the slow ring bus, plus a more efficient memory controller, higher clock speeds, more cache, better prefetch etc. These enhancements will mesh well with high speed DDR5 in a way that Alder Lake never could.

You can bookmark this post...!

I think I will. Will definitely be interesting to be sure.

 

[Kosusko]

Sure... I just don't see the sense of using one pcore- and all the rest e-core, instead of a much cheaper4 p-core and some e-cores/. Its your money though

Talking power efficiency, to aggressively save power – you could use the large number of Little Atom cores (16!) to handle just about any workload – and keep the big Cores parked. Effectively, unless gaming/benchmarking/etc. – you’d just be using an ultra efficient many threaded Atom system – but still be able to crank it up when needed.
source: https://www.sisoftware.co.uk/2022/10/15/intel-13th-gen-core-raptorlake-i9-13900ks/

Simply a paradigm shift.

Isn't better to wait for a Raptor Lake -N or something like that?

Alder Lake-N isn't even here yet. So how long should I wait for Raptor Lake-N?
16c .LITTLE "Raptormont" E-cores will do just fine. If it is necessary to have at least one big. Raptor Cove P-core turned on, I have no other choice. We will see.

 

[controlflow]

This guy has the hardware, and is using an RTX 4090 so there is no GPU bottlenecking like there was with the RTX 3000 series, even at 1440p max settings. As I predicted, Raptor Lake is the faster gaming CPU, and will be even faster when DDR5 7000+ kits become widely available later this month. These benchmarks were done with DDR5 6200, which is more advantageous to Zen 4 than it is to Raptor Lake.

Couldn't do embedded links because he blocked that feature from being used for his videos as he wants the clicks. If you do watch the videos please leave a like:

7700x vs 13700K

7950x vs 13900K

5800x 3D vs 13600K

Pretty good showing from Raptor Lake if these are accurate. Similar power use while gaming to Zen 4 but a bit better performance.

Never heard of this channel though so will take with a grain of salt for now.

 

[DrMrLordX]

Effectively, unless gaming/benchmarking/etc. – you’d just be using an ultra efficient many threaded Atom system

That's cool and all but as @Kocicak has demonstrated you can just set a 13900k to 65W (or some other arbitrary low power target) and go to town. There's no need to focus solely on the Gracemont+ cores as nobody has really demonstrated that they're any more power-efficient than Raptor Cove at completing tasks (Gracemont was shown to be no more efficient than Golden Cove). The only advantage to Gracemont/Gracemont+ is area efficiency.

– but still be able to crank it up when needed

Can you re-enable the 7 disabled Raptor Cove cores without rebooting the entire system?

 

[Falkentyne]

I have put the AIO back on the CPU and spent some time with playing with Intel extreme tuning utility, here are my OC results (only of P cores). I ended up setting max turbo for the best two cores to 60 and 59 to all other. When I tested all core performance, it could run at 57 multiplier with higher voltage offset.

So far it seems that all cores can be completely stable at 5900 MHz. I have not tested it thoroughly yet.

I have seen all cores boost up to 6100 MHz, not stable, I tried allocating 6100 only to the best two cores, not stable (passed cca half of 1T run), but I did not mess with voltages much. I believe 6100 could be achieved with some more tweaking.

Here are my results for 1, 2, 4, 8 and all threads load:

View attachment 69381

Some screenshots:

View attachment 69382

View attachment 69383

View attachment 69384

That's a VERY low 5900 mhz score.
I get that same R23 score at 5.6 ghz (with static set vcore) and stock e cores and ring (4.3 and 4.5 ghz).
You're either throttling or something is clipping your performance, because your original 5.5. ghz score is correct.
I also saw in your screenshot that one of your P cores dropped down to 4300 mhz.
Might want to fix that.

 

[alexruiz]

Is that all core boost or single core boost? According to the 7700x info page on AMD's website, the 7700x's max boost single core clock speed is 5.4ghz:

AMD 7700x

That shows how little you know about AMD CPUs.

Anyone who has built a PC with Ryzen knows that in real life, the single core boost is 50MHz to 150MHz higher than the official spec.
In the 7700X it happens to be 150MHz.

 

[Carfax83]

That shows how little you know about AMD CPUs.

Anyone who has built a PC with Ryzen knows that in real life, the single core boost is 50MHz to 150MHz higher than the official spec.
In the 7700X it happens to be 150MHz.

Perhaps for single core boost it may hit 5.5ghz, but all core boost? I checked a few reviews of the 7700x doing Cinebench and the all core boost was around 5.1ghz. The in game boost from his channel was around 5.34ghz so that looks accurate to me as games will not stress the CPU like Cinebench would.

 

[alexruiz]

Perhaps for single core boost it may hit 5.5ghz, but all core boost? I checked a few reviews of the 7700x doing Cinebench and the all core boost was around 5.1ghz. The in game boost from his channel was around 5.34ghz so that looks accurate to me as games will not stress the CPU like Cinebench would.

For the 7700X, all core boost, even when doing less strenuous stuff than cinebench, is around 5.15GHz.
The 5.35GHz number you are getting is single core boost, and somebody who is getting 5.35GHz single core boost on a 7700x has absolutely no clue on how to set up a Ryzen.

In the end, however, only 2 more days for more reputable reviewers to put their numbers
We will see.

 

[Kocicak]

That's a VERY low 5900 mhz score.

What score? 4 and 8T are at that frequency, 32T score is at 5700 MHz. I also have DDR4 memory. All I changed is multiplier and main voltage offset, nothing else.