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.

 

[Harry_Wild]

Intel is planning to refresh its desktop processor product stack with new "Raptor Lake Refresh" SKUs in Q3-2023, according to a leaked roadmap. At this point it's unclear if these are just new SKUs within the 13th Gen Core desktop product stack, or if they'll form the 14th Gen Core family, much in the same way as "Coffee Lake Refresh" formed the 9th Gen Core, replacing the 8th Gen Core "Coffee Lake." At this point we don't know what constitutes "Raptor Lake Refresh," but it provides Intel's product managers with the opportunity to increase CPU core-counts across the product stack without needing a new silicon (the Raptor Lake silicon has 8 P-cores and 16 E-cores), slightly higher clock-speeds, and other improvements. We don't know if this will herald a new CPU socket or platform at this point, either.

The most interesting item in this leaked roadmap slide has to be the reference to the "mainstream workstation" segment, with products in the 250 W TDP bracket. The so-called "Sapphire Rapids 64L" could be a cut-down version of the "Sapphire Rapids" enterprise processor on a new socket, backed by the Intel W790 chipset. The "64L" part of the codename could be a reference to its PCIe Gen 5 lane count of 64, which is less than the 112 available to the full "Sapphire Rapids" silicon in its W-3400 product-stack. It's unclear if these processors feature a Core X branding like their predecessors from the "Cascade Lake-X" family, or Xeon W. Besides fewer PCIe lanes, Intel could also segment these chips with fewer DDR5 memory channels, though both the PCIe and DDR5 connectivity will be much wider than those of the "Raptor Lake-S" mainstream desktop processors.

https://www.techpowerup.com/302118/...rs-for-q3-2023-sapphire-rapids-64l-hedt-in-q1

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
I am waiting for the Entry Workstation category 65W - Raptor Lake Refresh in Q3 of 2023 to be released and then jump on the 65W Raptor Lake S CPU to finally purchase.

 

[nicalandia]

View attachment 72809

Intel is planning to refresh its desktop processor product stack with new "Raptor Lake Refresh" SKUs in Q3-2023, according to a leaked roadmap. At this point it's unclear if these are just new SKUs within the 13th Gen Core desktop product stack, or if they'll form the 14th Gen Core family, much in the same way as "Coffee Lake Refresh" formed the 9th Gen Core, replacing the 8th Gen Core "Coffee Lake." At this point we don't know what constitutes "Raptor Lake Refresh," but it provides Intel's product managers with the opportunity to increase CPU core-counts across the product stack without needing a new silicon (the Raptor Lake silicon has 8 P-cores and 16 E-cores), slightly higher clock-speeds, and other improvements.

You can bet they will be 14900 Series CPUs

 

[Hulk]

You can bet they will be 14900 Series CPUs

That will be disappointing since Raptor is really a refresh of Alder. Refresh of a refresh. Kind of like the old 14nm + + + + + + + joke. But hey Intel generation 6, 7, 8, 9, and 10 were all Skylake on the desktop so perhaps we're lucky if Golden only runs 3 generations.

 

[dullard]

We don't know if this will herald a new CPU socket or platform at this point, either.

I would hazard a guess that they are not a new CPU socket, since, well, it says so right on the image. As in, everything with a grey bar is socket compatible along that grey bar.

 

[nicalandia]

I would hazard a guess that they are not a new CPU socket, since, well, it says so right on the image. As in, everything with a grey bar is socket compatible along that grey bar.

Intel giving Early Alder Lake adopters an upgrade path..... Due to MTL-S delays ofcourse.

 

[dullard]

Intel giving Early Alder Lake adopters an upgrade path..... Due to MTL-S delays ofcourse.

No MTL-S delay has been confirmed. The only delays were rumors that have been denied. Can you please wait with your conjectures until Friday when we get an official timeline?

 

[nicalandia]

No MTL-S delay has been confirmed. The only delays were rumors that have been denied. Can you please wait with your conjectures until Friday when we get an official timeline?

According to tweets by Dr. Ian Cutress, Intel’s Isic Silas said that Raptor Lake only exists because Meteor Lake wasn’t going to be ready on time.




The delay I am talking about it's not the most recent one speculated. But this is Intel confirming that two years ago they did not have Raptor Lake in design, they thought that MTL was a early 2023 product, then they realized that it was not going to meet that timeframe so they designed Raptor Lake and now Raptor Lake Refresh until MTL is ready in mid-late 2024

 

[dullard]

According to tweets by Dr. Ian Cutress, Intel’s Isic Silas said that Raptor Lake only exists because Meteor Lake wasn’t going to be ready on time.

View attachment 72814

Yes, Raptor Lake is 2022/2023. Meteor Lake was not going to be ready by 2022. EUV machines are just now letting Intel start Intel 4 manufacturing. There was no way to launch an Intel 4 product (Meteor Lake) before they had the equipment running to do so. We all know that. Now please wait until Friday to see if there is an actual delay in Meteor Lake past their always planned end of 2023. I'm not denying a delay might be announced. But, I just don't like your non-stop claims that there is a delay just before the official announcement.

 

[nicalandia]

Yes, Raptor Lake is 2022/2023.

Hence I said that early ADL-S adopters have an upgrade path(Raptor Lake and Raptor Lake Refresh) till the end of 2023 so potentially three Generations(12th, 13th and tentatively 14th)

 

[zir_blazer]

Wasn't there an "Ocean Cove" or so that was supposed to come after Alder Lake but before Meteor Lake? I recall hearing about it circa 2018-2020 or so. Raptor Lake is supposed to be filling that gap.

 

[Hulk]

According to tweets by Dr. Ian Cutress, Intel’s Isic Silas said that Raptor Lake only exists because Meteor Lake wasn’t going to be ready on time.

View attachment 72814


The delay I am talking about it's not the most recent one speculated. But this is Intel confirming that two years ago they did not have Raptor Lake in design, they thought that MTL was a early 2023 product, then they realized that it was not going to meet that timeframe so they designed Raptor Lake and now Raptor Lake Refresh until MTL is ready in mid-late 2024

So let me get this straight. Raptor Lake was not supposed to exist even though, and this is my prediction, it will end up being more performant on the desktop than 6+16 Meteor Lake. I say this because I'm betting there will be a clock speed regression from Raptor to Meteor and Meteor is down 2 P cores.

If this is all true then Raptor Lake, like me and many youngest children, is happy accident (for Intel).

When Intel finally gets a new node going they just can't seem to let go of it. Once yields are good, clocks are up, and it's a "sure thing" why risk moving the bulk of production to a new process? I'm starting to think Intel 7 is the new Intel 14nm. Get the plusses warmed up on your keyboards when discussing Intel moving forward!

 

[dullard]

So let me get this straight. Raptor Lake was not supposed to exist even though, and this is my prediction, it will end up being more performant on the desktop than 6+16 Meteor Lake. I say this because I'm betting there will be a clock speed regression from Raptor to Meteor and Meteor is down 2 P cores.

Don't get your hopes up for Meteor Lake desktop. It will likely be available, but in limited quantities and it isn't intended to be a huge desktop game changer. All signs point to you needing to wait for Arrow Lake.

 

[TheELF]

I would hazard a guess that they are not a new CPU socket, since, well, it says so right on the image. As in, everything with a grey bar is socket compatible along that grey bar.

We did have socket compatible that needed a new chipset before, depends on how much they change it, if they add a bunch of cores to a sku it will need a new chipset, or they will push a new one anyway for the same reasons they did so in the past.

 

[Hulk]

Don't get your hopes up for Meteor Lake desktop. It will likely be available, but in limited quantities and it isn't intended to be a huge desktop game changer. All signs point to you needing to wait for Arrow Lake.

We are in agreement. As I wrote a few pages back I think Meteor Lake will be like Broadwell in that due to the newness of the process Intel won't be able to ramp up clocks so it will at first be a mobile only release. I just don't see Intel 4 hitting Raptor like 6GHz frequencies on the first go at it where yields are good enough for production.

 

[dullard]

We did have socket compatible that needed a new chipset before, depends on how much they change it, if they add a bunch of cores to a sku it will need a new chipset, or they will push a new one anyway for the same reasons they did so in the past.

? I'm confused. The leaked slide both says which chips are socket compatible AND lists the chipsets. The leaked slide could be wrong, but then we have nothing to talk about. Talking about changing sockets and changing chipsets when the only piece of information we have says they don't change is just odd.

 

[Exist50]

Wasn't there an "Ocean Cove" or so that was supposed to come after Alder Lake but before Meteor Lake? I recall hearing about it circa 2018-2020 or so. Raptor Lake is supposed to be filling that gap.

Ocean Cove is long-dead.

 

[Hulk]

My 280mm AIO came in. Once I get the desire to start tearing into the rig to get it going I'll report back on some power/temp data. Part of me really wants to check it out but a bigger part wants to collapse in front of the TV. You know Friday night and all.

 

[Carfax83]

My 280mm AIO came in. Once I get the desire to start tearing into the rig to get it going I'll report back on some power/temp data. Part of me really wants to check it out but a bigger part wants to collapse in front of the TV. You know Friday night and all.

I know the feeling. Tinkering can be pleasant when things go your way, and a total pain in the ass when things don't go as expected.

I'm curious to see how much better the AiO is compared to your previous Noctua NH-U12A. I have the same heatsink and so far, I've been very pleased with it. I have issues trusting AiOs and water cooling so I'll probably stick with air cooling.

 

[Hulk]

About a month ago the Cooler Master ML280 AIO dropped in price to $59.99. I've been wanting to see how a 280mm AIO could do with the 13900K so I bought it. It wasn't supposed to arrive until early to mid January but here it is.

First thing I did was let the 13900K go nuclear and see what this AIO could do. CB R23 score is 40,000 give or take 100 points. Package power was as high as 315 W but generally was around 300 W once it started throttling. Obviously in my 74 degree F house it ain't handling 300W.

So I backed it down to 225W PL1 with a 12 tau of 250W. One CB run scored 38,100 with a max temp of 86C. 10 minute score was 37,200 again with a max temp of 86C.

One the plus side besides handling higher clocks/power than my U12A it's also quieter. The down side is I'll need a new case to fit this thing. This is my first experience was an AIO and I have to admit it's pretty fun to mess around with. Also since it has a lot more thermal capacity than my air cooler it's very suited to bursty loads, which makes it a good fit for the 13900K. For $60 I think I'll keep both it and the air cooler.

So there it is. This particular 280mm AIO can handle perhaps 250W, but a more realistic level I think is the 225W I'm running it currently. Also I did a pretty quick install so I might be able to do better if I take more care with the set up. Might have to sit the radiator in an ice water bath tomorrow just to see how it does.

 

[Carfax83]

So I backed it down to 225W PL1 with a 12 tau of 250W. One CB run scored 38,100 with a max temp of 86C. 10 minute score was 37,200 again with a max temp of 86C.

Wait, so you set the TAU to 12? Those results are worse than what I get with my U12A Chromax.Black at 5.2ghz.

This just cements the notion that your main problem is your case. It lacks the proper ventilation to keep your components cool, even with an AiO.

An AiO won't do the job properly if your case ventilation is poor.

 

[Hulk]

Wait, so you set the TAU to 12? Those results are worse than what I get with my U12A Chromax.Black at 5.2ghz.

This just cements the notion that your main problem is your case. It lacks the proper ventilation to keep your components cool, even with an AiO.

An AiO won't do the job properly if your case ventilation is poor.

Has nothing to so with the case. I've ran my U12A with the side off the case and a powerful fan blowing into the case, like a tornado. Still can't really do more than 190 or 200W. Same with this radiator, this testing was done out of the case.

So exactly what are your CB scores at 190W, 200W, 225W?

I think the reality is your chip is better than mine.

BTW I can break 40,000 with the AIO as well.

Exactly how many watts package are you pulling at 5.2 and what are the E cores running at?

Let's get on an apple-for-apples footing here before jumping to conclusions.

 

[Carfax83]

So exactly what are your CB scores at 190W, 200W, 225W?

Just ran this 10 minute run and I got 38618. That's at max power draw of 215w (average was 184w), with the P cores at 5.2ghz and the E cores running at 4.3ghz. But looking at the core frequencies during the run, they bounce between 4.3ghz and 4.1ghz

P cores bounce between 5.2ghz and 5.1ghz. I think it's because the TAU is on auto and I don't know what value the motherboard uses.

Max package temp was 77c, with an ambient temperature of 67c. I turned the heat off in my gaming room and it's 43F outside.

I think the reality is your chip is better than mine.

I honestly don't know. I think you need to raise that offset. It will lower the temps and make your clocks more stable. I have mine at -0.125mv.

But regarding ventilation, it's not just a question of how much cool air is flowing into the case, but also how quickly warm air is evacuated.

 

[Hulk]

Just ran this 10 minute run and I got 38618. That's at max power draw of 215w (average was 184w), with the P cores at 5.2ghz and the E cores running at 4.3ghz. But looking at the core frequencies during the run, they bounce between 4.3ghz and 4.1ghz

P cores bounce between 5.2ghz and 5.1ghz. I think it's because the TAU is on auto and I don't know what value the motherboard uses.

Max package temp was 77c, with an ambient temperature of 67c. I turned the heat off in my gaming room and it's 43F outside.



I honestly don't know. I think you need to raise that offset. It will lower the temps and make your clocks more stable. I have mine at -0.125mv.

But regarding ventilation, it's not just a question of how much cool air is flowing into the case, but also how quickly warm air is evacuated.

I was running a -0.075 offset but when I uncapped power limits it was an immediate restart at that setting. Being impatient I got rid of the offset and let it go nuke. Results reported last night were stock auto settings.

I'll see if I can sneak down the Vcore a bit but I'm telling you I'm getting to know my chip and there is no way I'm going to get the frequencies you are at that offset.

I'm telling you the fan I had on the air cooler into the open case was like hurricane. There was not a cubic centimeter of warm air to be found. My house is quite hot at 74 F, my wife and girls like it warm. It's like an oven. That doesn't help underclocking either.

Your rig is better tuned than mine, no doubt, but I also think you are working with a better chip.

Your rig would absolutely scream with this ML280 AIO. At $60 it was a steal, that's nearly half the price of the Noctua we both have. Also it's quieter than the Noctua at equal heat transfer. This was my first WC experience and I was all worried about pump noise, leaking, set up issues, etc.. I don't want to jinx myself but it was really easy to get going. I think as long as you keep the pump low and radiator in/out high you avoid a lot of cavitation issues.

 

[nicalandia]

First Review of the 13400 CPU..

Intel Core i5-13400 Early Review: Up to 43% Faster in Games & 30% Faster in Productivity Compared to i5-12400 at the Cost of Increased Power Consumption & Temperatures

We got our hands on a Core i5-13400 and put it through the ringer. Here are the results in both gaming and productivity workloads.

appuals.com

 

[nicalandia]

First entry of the 13700HX and its Identical to the 12900HX. Same Core Count, Same Cache Subsystem and Same Performance

Notebook NP5x_6x_7x_SNx vs Micro-Star International Co., Ltd. Titan GT77 12UHS - Geekbench

browser.geekbench.com