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.

 

[Doug S]

A non-zero percentage of the input power onto the processor "system" is converted to useful work.the rest is dissipated primarily as heat but is also emitted in various forms of electromagnetic radiation.

Any "useful work" the CPU does also results in heat. Even electromagnetic radiation mostly results in dissipated heat inside the room where the PC resides. Only that radiation which escapes the room (as opposed to being blocked by the PC's case, the walls, you, etc.) ends up outside the room.

While the amount of power input to the PC isn't dissipated 100% inside the room, it is so close to 100% that it is not worth considering the tiny percentage that escapes it.

 

[Kocicak]

I updated the MB bios to 20b, the power unlimited run (it really seems it is the default setting) draw 344W, reached 91°C and CNB score was 40 858.

The next run reached 94°C and the score was lower, it probably started to throttle. The third run was 97°C and score dropped even more.

It seems that the new bios really squeezes the last drop from the CPU performance, my little AIO is probably not up to the task. I need to check what happens at lower power draw limits.

 

[nicalandia]

I updated the MB bios to 20b, the power unlimited run (it really seems it is the default setting) draw 344W, reached 91°C and CNB score was 40 858.

The next run reached 94°C and the score was lower, it probably started to throttle. The third run was 97°C and score dropped even more.

That is the normal behavior of Alder Lake, it looks good on Reviews, for example on Cinebench R23 The MT Score for 12900K at stock is higher than stock 5950X, but if you put both CPUs on a 10 minute loop... The 5950X ends up faster because the 12900K drops the performance.

Something similar will happen to 13900K vs 7950X, Somewhat similar MT Scores at stock, but at 10 minute loop will see the 7950X as the clear superior.



On that 10 minute loop a stock 7950X is 61.5% Higher performance from stock 12900K. Will a stock 13900K beat that on a 10 minute loop??

 

[Markfw]

That is the normal behavior of Alder Lake, it looks good on Reviews, for example on Cinebench R23 The MT Score for 12900K at stock is higher than stock 5950X, but if you put both CPUs on a 10 minute loop... The 5950X ends up faster because the 12900K drops the performance.

Something similar will happen to 13900K vs 7950X, Somewhat similar MT Scores at stock, but at 10 minute loop will see the 7950X as the clear superior.

View attachment 69127

I have been saying this since Alder Lake came out, and everyone says "well who uses that much that long ? And so what you lose a little speed when you use it that long ". But a lot of people DO use it for that long. Me included.

Edit: your attachment does not show in your post, but it shows in my quote. Odd....

 

[mikk]

I know, I've been saying that all day long.. OG Gracemont on Raptor Lake CPU means a rebranded Alder Lake..

Raptormont Cores have 4MiB of L2
View attachment 69125

13900 no 4MB, why?

https://meterpreter.org/wp-content/uploads/2022/08/Intel_Core_i9-13900-768x741.jpg

 

[nicalandia]

13900 no 4MB, why?

The reading is fine, look at the L1 Data and L1 instruction reading they are Raptor Cove. The L2$ is 2MiB for P Cores.

 

[Kocicak]

After I updated the BIOS to 20b and increased the RAM frequency to 3600 MHz, here are the results at different power limits. I also recorded frequency of P and E cores and total PC power draw.



At 35W it performs nearly as good as my 12600K performed at 116W.

Here are the Geekbench results at 160W limit:

Gigabyte Technology Co., Ltd. Z690 GAMING X DDR4 - Geekbench

Benchmark results for a Gigabyte Technology Co., Ltd. Z690 GAMING X DDR4 with an Intel Core i9-13900K processor.

browser.geekbench.com

... at 250W limit:

Gigabyte Technology Co., Ltd. Z690 GAMING X DDR4 - Geekbench

Benchmark results for a Gigabyte Technology Co., Ltd. Z690 GAMING X DDR4 with an Intel Core i9-13900K processor.

browser.geekbench.com

and at unlimited power:

Gigabyte Technology Co., Ltd. Z690 GAMING X DDR4 - Geekbench

Benchmark results for a Gigabyte Technology Co., Ltd. Z690 GAMING X DDR4 with an Intel Core i9-13900K processor.

browser.geekbench.com

I think I am done with testing.

 

[nicalandia]

After I updated the BIOS to 20b and increased the RAM frequency to 3600 MHz, here are the results at different power limits.

and at unlimited power:

Gigabyte Technology Co., Ltd. Z690 GAMING X DDR4 - Geekbench

Benchmark results for a Gigabyte Technology Co., Ltd. Z690 GAMING X DDR4 with an Intel Core i9-13900K processor.

browser.geekbench.com

I think I am done with testing.

Thanks for sharing and all of your hard work Your MT scores are quite low even at Unlimited Power. Someone said it was due to DDR4, but I am not sure.

Can you please post a screenshot of a CPUZ?

 

[Abwx]

After I updated the BIOS to 20b and increased the RAM frequency to 3600 MHz, here are the results at different power limits. I also recorded frequency of P and E cores and total PC power draw.

View attachment 69131

At 35W it performs nearly as good as my 12600K performed at 116W.

Here are the Geekbench results at 160W limit:

A bios update wont improve efficency by 10%, if your chip previously made 32352 pts@100W it wont make 4.4% better at same power because of a better bios, 4.4% require in principle at least 9-10% more power.

Besides if a CPU use 160W then it will amount to at least 200W on the main, to wich idle power should be added, in our case that would do 200 + 42 = 242W, yet you measure something like 218W, 10% deviation is a lot...

 

[Kocicak]

Here it is:

 

[Kocicak]

Abwx: I increased frequency of memory and also ran the Cinebench 3 times and chose the best result. I do not have any test bench with optimised operating system, I am running this on my main computer and there is stuff happening in the background which affects results. All the previous results I ran just once. However now I ran the geekbench only once too.

Another thing to note is that I had previously problems with the system freezing on this motherboard (rev 1.0), the actual revision is 1.1 (God knows what they changed), this CPU is still not on the official supported CPU list of the motherboard. So things may not be working 100% correctly on my motherboard and a new BIOS may change a lot of things.

And this CPU is not supposed to be sold to end customers yet too. I just bought it as a normal customer, I have no support from Gigabyte or Intel at all. I believe reviewers may have different bioses than those officially released.

And also I have the last version of Windows that may be buggy by itself.

 

[BorisTheBlade82]

@Kocicak
Would you be so nice and give this a try: http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=thread...-efficiency-of-x86-cpu-architectures.2597905/

It could give us quite some more insights into performance efficiency of RTL.

 

[DreadBelch]

Does anyone expect RPL to sell out on launch day? Did ADL sell out at launch?

 

[DrMrLordX]

I think I am done with testing.

Thanks for your testing. It will make a good compare/contrast with DDR5-equipped systems people bench here later.

Does anyone expect RPL to sell out on launch day? Did ADL sell out at launch?

Given the way Raphael has been selling, probably not.

 

[Kocicak]

Would you be so nice and give this a try: http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=thread...-efficiency-of-x86-cpu-architectures.2597905/

It could give us quite some more insights into performance efficiency of RTL.

HEY! Put the warning that multithread test runs multiple times in the main instructions and information text! I run it first at unlimited power and prayed that it ends and finally had to shut it down when I smelled things burning!!!! You nearly killed my computer!

EDIT: Now i see it is there but nobody can see such important information inside a block of text!

 

[Markfw]

HEY! Put the warning that multithread test runs multiple times in the main instructions and information text! I run it first at unlimited power and prayed that it ends and finally had to shut it down when I smelled things burning!!!! You nearly killed my computer!

EDIT: Now i see it is there but nobody can see such important information inside a block of text!

Exactly why I won't touch raptor lake. Mine runs 100% load 24/7. At stock (230 watt) it had no problem, At 142 watt less power usage.

 

[poke01]

Given the way Raphael has been selling, probably not.

The thing is RPL will work on DDR4 and ADL mobos. Zen 4 does not support DDR4 or AM4.

 

[BorisTheBlade82]

HEY! Put the warning that multithread test runs multiple times in the main instructions and information text! I run it first at unlimited power and prayed that it ends and finally had to shut it down when I smelled things burning!!!! You nearly killed my computer!

EDIT: Now i see it is there but nobody can see such important information inside a block of text!

Wow, erm, sorry for that. Usually there are protection mechanisms in place for thermal throttling. Is this due to you having a beta BIOS?

 

[DrMrLordX]

The thing is RPL will work on DDR4 and ADL mobos. Zen 4 does not support DDR4 or AM4.

True. You're also going to have a few Alder Lake owners that will use it as a drop-in upgrade. But if Raptor Lake can't sell Z790 etc. then that amounts to a similar problem.

 

[Kocicak]

I posted the results of 13900K capped at 250, 160, 100 and 65W in this thread:
http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=thread...-efficiency-of-x86-cpu-architectures.2597905/

It seems to be less efficient than AMD 7950X but not by much.

EDIT: I added results at 65W and there 7950X is more efficient more noticeably.

 

[Kocicak]

Exactly why I won't touch raptor lake. Mine runs 100% load 24/7. At stock (230 watt) it had no problem, At 142 watt less power usage.

It is pretty efficient at lower power, I just do not understand why some motherboard manufacturers let it loose drawing as much power as it can as a default setting!

 

[DrMrLordX]

It is pretty efficient at lower power, I just do not understand why some motherboard manufacturs let it loose drawing as much power as it can as a default setting!

Same reason why some OEMs allowed extended or unlimited tau on Comet Lake, Rocket Lake, and Alder Lake boards.

 

[coercitiv]

It is pretty efficient at lower power, I just do not understand why some motherboard manufacturs let it loose drawing as much power as it can as a default setting!

Same reason why some OEMs allowed extended or unlimited tau on Comet Lake, Rocket Lake, and Alder Lake boards.

This has been a trend for many years now, ever since Coffee Lake. Actually this may have been a thing even earlier, but nobody really bothered with it until the hexa-cores arrived and were able to use more than 95W while boosting.

 

[Exist50]

This has been a trend for many years now, ever since Coffee Lake. Actually this may have been a thing even earlier, but nobody really bothered with it until the hexa-cores arrived and were able to use more than 95W while boosting.

It's basically just a slightly different implementation of MCE, which has been a thing for about a decade now. Shouldn't really come as a surprise. https://www.anandtech.com/show/6214/multicore-enhancement-the-debate-about-free-mhz

 

[inf64]

It will be interesting to see what will the big review websites do:
a) use the real stock TDP settings (253W) for Turbo mode on Raptor Lake, or
b) use the "default" unlimited TDP mode that each motherboard maker decides to set in the BIOS.