Intel Core "Special Edition" Owners' Club

[Dr. Dro]

No idea, but 13/14 is the same architecture and stepping. Not sure why they would be different

The last and fastest of their kind – Intel 13th Gen Raptor Lake Binning | Part 3: i9-13900KS | Page 4 | igor´sLAB

Today is finally the day – loyal readers won’t be surprised by this –, for Intel’s last and fastest LGA1700 CPU we also have binning data for you once more. Thanks to MIFCOM, we were able to run…

www.igorslab.de

Apparently it's the case

 

[ir_cow]

I was under the impression that ASUS rating is just based on idle VF and some math. If that was the case I think if its the same architecture, it shouldn't matter the chip name.

But Igor does good research, so I guess there is more to it.

Edit: it's not written by Igor, but a XOC guy. That's good enough for me.

 

[Dr. Dro]

There's something that I wanted to ask but I'm not sure it warrants an entire thread - particularly having this one dedicated to i9 KS's

I came across this article a couple of hours ago:

Intel wants Default Settings with PL1/PL2 at 125W/188W to be implemented by motherboard vendors by the end of May - VideoCardz.com

Intel Default Profile will lock PL2 to 188W Intel wants all LGA-1700 motherboards to support Default Profile, and there is even a deadline. The 14th Gen Core K-series of unlocked desktop CPUs has been on the market for a few months now, representing a higher clocked version of the Raptor Lake...

videocardz.com

Supposedly Intel wants to adopt a new "baseline profile" for the regular i9 K CPUs that's supposed to be 125/188 W with a 249 A ICCMax, however, upon some research it seems that this power profile is a new idea from Intel and it does not match anything that is currently written in their latest revision of the processor spec datasheet available at:

13th Generation Intel® Core™ and Intel® Core™ 14th Generation Processors Datasheet, Volume 1 of 2

WW14.3 Metadata Changes only This describes all publicly disclosed specifications including electrical characteristics, mechanical, and component functionality, a list of major features, a functional description, and an architectural overview.

www.intel.com

It seems to be closest to the 181 W profile intended for S-Processor 6P+8E 125W, which correlates to the Core i5-13600K. This will obviously cause a performance regression on all SKUs, but I am especially concerned about the i9-13900KS and 14900KS SKUs as they have a different official power specification compared to the regular 13900K and 14900K processors. As you can see they are clearly denominated as S-Processor 8P+16E 150W on the datasheet, and again - the information from the above post (which I have seen in other places) is a complete and total mismatch to both the -K and -KS specifications on Intel's own documentation.

Reducing a KS SKU to this baseline wattage would cause an even more acute performance regression and simultaneously eliminate any performance gap between a regular and special edition processor, if not make the KS's slower because their clock table is more aggressive and thus, more prone to power throttling (this is a behavior generally observed on the i9-14900K as is when compared to the 13900KS, with its slightly more aggressive boost targets). I understand "who buys a i9 KS to run at stock or follow Intel spec", but it's still not a change to be taken lightly, particularly this late in Raptor Lake's lifecycle.

Intel made a huge mess here and I'm having difficulty following. I've attempted to bring it up on the Intel discord but no one replied, and there was only one person who engaged in the conversation, summing up to "I should have just bought the 13900K back then" - quite disinterested, really.

 

[BoggledBeagle]

It has already beed discussed here:

Intel Statement on Stability Issues: "Motherboard Makers to Blame"

He mentioned that all 13700K he tested were stable. They have maximal frequencies 5400/4200 MHz. I think I may increase limits for my 14900K from 5200/4200 MHz to 5300/4200 without any fear.

www.techpowerup.com

I am not sure why are you concerned about the baseline profile, there will probably be three "official profiles" to choose from and I think almost no KS owners will choose the baseline profile. Also, the chip is fully configurable and you can tune it how you want. If you will notice some instability issues, you will learn that you need to change settings somehow.

If KS chips are pushed so hard, that their average life is limited and they start failing (becoming unstable) before the warrantly period ends in large numbers, Intel will probably exchange them for K version and pay customers the difference in price, or they will just take them and return money for them?

 

[Dr. Dro]

It has already beed discussed here:

Intel Statement on Stability Issues: "Motherboard Makers to Blame"

He mentioned that all 13700K he tested were stable. They have maximal frequencies 5400/4200 MHz. I think I may increase limits for my 14900K from 5200/4200 MHz to 5300/4200 without any fear.

www.techpowerup.com

I am not sure why are you concerned about the baseline profile, there will probably be three "official profiles" to choose from and I think almost no KS owners will choose the baseline profile. Also, the chip is fully configurable and you can tune it how you want. If you will notice some instability issues, you will learn that you need to change settings somehow.

If KS chips are pushed so hard, that their average life is limited and they start failing (becoming unstable) before the warrantly period ends in large numbers, Intel will probably exchange them for K version and pay customers the difference in price, or they will just take them and return money for them?

Benchmarking results must be done at stock to have any validity, I know realistically speaking we can still run them as we've always done, the thing is that it's going to be misleading and essentially invalidate all reviews up to this point, including for the regular K's and even the i7's.

 

[BoggledBeagle]

Benchmarking results must be done at stock to have any validity, I know realistically speaking we can still run them as we've always done, the thing is that it's going to be misleading and essentially invalidate all reviews up to this point, including for the regular K's and even the i7's.

KS is a special version of the CPU and I do not think that reviewing it with baseline profile makes any sense. You need to review an extreme CPU with the "extreme profile".

The situation that the KS CPUs probably will not last too long running with this extreme profile is a very unfortunate situation, questioning the existence of these products. KS CPU without the extreme profile does not make any sense.

 

[Beginner Macro Device]

The situation that the KS CPUs probably will not last too long running with this extreme profile

Source? I'm yet to see modern CPUs prone to early death at 3-figure wattage. My homie's 6 year old 7980XE (clocked to 5+ GHz) still does its job.

 

[Dr. Dro]

KS is a special version of the CPU and I do not think that reviewing it with baseline profile makes any sense. You need to review an extreme CPU with the "extreme profile".

The situation that the KS CPUs probably will not last too long running with this extreme profile is a very unfortunate situation, questioning the existence of these products. KS CPU without the extreme profile does not make any sense.

While it is akin to the old Core 2 Extreme processors (being essentially the same as the regular CPUs but slightly jacked up), it is still a product sold with certain specifications that it must meet in mind. If the chips cannot run in their intended specification and must be revised down after some time... that is a very very bad precedent. That's why testing at official specifications is important.

 

[BoggledBeagle]

.... it is still a product sold with certain specifications that it must meet ... If the chips cannot run in their intended specification and must be revised down after some time ... that is a very very bad precedent.

What specifications must a KS CPU meet? All I can see here:



are maximal frequencies, nothing guaranteed whatsoever.

 

[Dr. Dro]

What specifications must a KS CPU meet? All I can see here:

View attachment 346489

are maximal frequencies, nothing guaranteed whatsoever.

Read the data sheet I've provided. Their electrical specification is not the same, nor are the expected operating frequencies.

 

[Super Firm Tofu]

There's something that I wanted to ask but I'm not sure it warrants an entire thread - particularly having this one dedicated to i9 KS's

I came across this article a couple of hours ago:

Intel wants Default Settings with PL1/PL2 at 125W/188W to be implemented by motherboard vendors by the end of May - VideoCardz.com

Intel Default Profile will lock PL2 to 188W Intel wants all LGA-1700 motherboards to support Default Profile, and there is even a deadline. The 14th Gen Core K-series of unlocked desktop CPUs has been on the market for a few months now, representing a higher clocked version of the Raptor Lake...

videocardz.com

Supposedly Intel wants to adopt a new "baseline profile" for the regular i9 K CPUs that's supposed to be 125/188 W with a 249 A ICCMax, however, upon some research it seems that this power profile is a new idea from Intel and it does not match anything that is currently written in their latest revision of the processor spec datasheet available at:

13th Generation Intel® Core™ and Intel® Core™ 14th Generation Processors Datasheet, Volume 1 of 2

WW14.3 Metadata Changes only This describes all publicly disclosed specifications including electrical characteristics, mechanical, and component functionality, a list of major features, a functional description, and an architectural overview.

www.intel.com

It seems to be closest to the 181 W profile intended for S-Processor 6P+8E 125W, which correlates to the Core i5-13600K. This will obviously cause a performance regression on all SKUs, but I am especially concerned about the i9-13900KS and 14900KS SKUs as they have a different official power specification compared to the regular 13900K and 14900K processors. As you can see they are clearly denominated as S-Processor 8P+16E 150W on the datasheet, and again - the information from the above post (which I have seen in other places) is a complete and total mismatch to both the -K and -KS specifications on Intel's own documentation.

Reducing a KS SKU to this baseline wattage would cause an even more acute performance regression and simultaneously eliminate any performance gap between a regular and special edition processor, if not make the KS's slower because their clock table is more aggressive and thus, more prone to power throttling (this is a behavior generally observed on the i9-14900K as is when compared to the 13900KS, with its slightly more aggressive boost targets). I understand "who buys a i9 KS to run at stock or follow Intel spec", but it's still not a change to be taken lightly, particularly this late in Raptor Lake's lifecycle.

Intel made a huge mess here and I'm having difficulty following. I've attempted to bring it up on the Intel discord but no one replied, and there was only one person who engaged in the conversation, summing up to "I should have just bought the 13900K back then" - quite disinterested, really.

It's quite the mess. At this moment 'Intel Baseline' is different between motherboard vendors. Asus is better than GB, but neither conform to any published specs. If I choose Intel Baseline on my Encore with the 14900KS, I get PL1=PL2=253W with an ICCMax of 280A. That's a lower ICCMax than published specs.

This is also the only setting where I'm 100% stable in every situation. I received a horrid sample of the 14900KS -- 102sp. At equal clocks, my 13900K requires less voltage. The IMC on it is great though. 8200MT/s out of the box with zero tweaking. The only situation where the KS is better than my 13900K is in single thread performance, but even that requires 1.5v+ to achieve.

I submitted an RMA request with Intel last night as I waited too long to build and I'm past the return window for the board, CPU, and RAM. My hope is I can get a new unopened KS and sell it. This mess is going to cost me at least $500 to get out of. I have a sneaky suspicion that Intel will say if it's stable at baseline, no RMA for me.

Outside of my own personal situation, it's a really bad look for Intel. If the gimped settings are what people are expected to run at, they should have been released that way from the beginning.

 

[Dr. Dro]

It's quite the mess. At this moment 'Intel Baseline' is different between motherboard vendors. Asus is better than GB, but neither conform to any published specs. If I choose Intel Baseline on my Encore with the 14900KS, I get PL1=PL2=253W with an ICCMax of 280A. That's a lower ICCMax than published specs.

This is also the only setting where I'm 100% stable in every situation. I received a horrid sample of the 14900KS -- 102sp. At equal clocks, my 13900K requires less voltage. The IMC on it is great though. 8200MT/s out of the box with zero tweaking. The only situation where the KS is better than my 13900K is in single thread performance, but even that requires 1.5v+ to achieve.

I submitted an RMA request with Intel last night as I waited too long to build and I'm past the return window for the board, CPU, and RAM. My hope is I can get a new unopened KS and sell it. This mess is going to cost me at least $500 to get out of. I have a sneaky suspicion that Intel will say if it's stable at baseline, no RMA for me.

Outside of my own personal situation, it's a really bad look for Intel. If the gimped settings are what people are expected to run at, they should have been released that way from the beginning.

Oh nice, that's one of the questions I've had in the open resolved.

Dang dude that really sucks to hear you managed to lose the silicon lottery on a 14900KS... My 13900KS has always been such a well behaved and cool chip, though I've never had the chance to see how far its IMC went or what is its SP rating since I'm on a MSI Z690, but no complaints so far really.

I wanted to get a board like a Strix B760i or something similar specifically to target memory tweaking but in the end of the day it's just not worth the money

 

[A Computer Guy]

There's something that I wanted to ask but I'm not sure it warrants an entire thread - particularly having this one dedicated to i9 KS's

I came across this article a couple of hours ago:

Intel wants Default Settings with PL1/PL2 at 125W/188W to be implemented by motherboard vendors by the end of May - VideoCardz.com

Intel Default Profile will lock PL2 to 188W Intel wants all LGA-1700 motherboards to support Default Profile, and there is even a deadline. The 14th Gen Core K-series of unlocked desktop CPUs has been on the market for a few months now, representing a higher clocked version of the Raptor Lake...

videocardz.com

Supposedly Intel wants to adopt a new "baseline profile" for the regular i9 K CPUs that's supposed to be 125/188 W with a 249 A ICCMax, however, upon some research it seems that this power profile is a new idea from Intel and it does not match anything that is currently written in their latest revision of the processor spec datasheet available at:

13th Generation Intel® Core™ and Intel® Core™ 14th Generation Processors Datasheet, Volume 1 of 2

WW14.3 Metadata Changes only This describes all publicly disclosed specifications including electrical characteristics, mechanical, and component functionality, a list of major features, a functional description, and an architectural overview.

www.intel.com

It seems to be closest to the 181 W profile intended for S-Processor 6P+8E 125W, which correlates to the Core i5-13600K. This will obviously cause a performance regression on all SKUs, but I am especially concerned about the i9-13900KS and 14900KS SKUs as they have a different official power specification compared to the regular 13900K and 14900K processors. As you can see they are clearly denominated as S-Processor 8P+16E 150W on the datasheet, and again - the information from the above post (which I have seen in other places) is a complete and total mismatch to both the -K and -KS specifications on Intel's own documentation.

Reducing a KS SKU to this baseline wattage would cause an even more acute performance regression and simultaneously eliminate any performance gap between a regular and special edition processor, if not make the KS's slower because their clock table is more aggressive and thus, more prone to power throttling (this is a behavior generally observed on the i9-14900K as is when compared to the 13900KS, with its slightly more aggressive boost targets). I understand "who buys a i9 KS to run at stock or follow Intel spec", but it's still not a change to be taken lightly, particularly this late in Raptor Lake's lifecycle.

Intel made a huge mess here and I'm having difficulty following. I've attempted to bring it up on the Intel discord but no one replied, and there was only one person who engaged in the conversation, summing up to "I should have just bought the 13900K back then" - quite disinterested, really.

When news of this stuff was first breaking I remembered something about Gigabyte's baseline profile being different than the others causing some confusion. According to HU Gigabyte got it right with their default profile and by May 31st the out of box experience is looking to be very different.

Gigabyte specific example profile comparison

(edit)

Sorry OP I just realized these tests were on K not KS parts. I assume the same issues apply?

 

[Dr. Dro]

When news of this stuff was first breaking I remembered something about Gigabyte's baseline profile being different than the others causing some confusion. According to HU Gigabyte got it right with their default profile and by May 31st the out of box experience is looking to be very different.

View attachment 346501

Gigabyte specific example profile comparison
View attachment 346503
(edit)

Sorry OP I just realized these tests were on K not KS parts. I assume the same issues apply?

Yeah, same issues but to a higher degree actually since the KS processors have higher power targets than K.

If the baseline on the ROG Apex (honestly the reference motherboard) is being applied as 253W flat, and this carries over to other motherboards the gap between K and KS will widen significantly, if the new 188W profile is applied to KS there will also be a significantly performance regression

 

[Dr. Dro]

Here's an update with the recommended power specification, it seems that for us KS owners, the "baseline profile" does not apply at all, and our "Performance" profile is similar to the "Extreme" profile on regular K SKUs.

 

[Enigmatic9400]

I wonder how is everyone's SP rating for their chip? I got a i9-13900KS on ebay with SP 111, P core 122, E core 89 for $430 USD + shipping & tax. It looks like the P core SP is quite high. Not sure if my chip is rare or not.

 

[Outback Bronze]

I received a horrid sample of the 14900KS -- 102sp. At equal clocks, my 13900K requires less voltage.

Do you mind posting a screen shot of your 13900KS VF curves? Would like to see it vs a 14900KS. You can post your 14900KS vf curves too if you like.

This KS is 101 and I'm quite happy with it. I do have a SP 105 KS too but the 101 seems to be undervolting better so I'm not sure everybody should be taking SP for real.

 

[Enigmatic9400]

Do you mind posting a screen shot of your 13900KS VF curves? Would like to see it vs a 14900KS. You can post your 14900KS vf curves too if you like.

This KS is 101 and I'm quite happy with it. I do have a SP 105 KS too but the 101 seems to be undervolting better so I'm not sure everybody should be taking SP for real.

View attachment 353619

I think igor's lab tested the p core sp with achieveable overclock under the same voltage and found out that there is correlation between SP and high overclock, but it is not linear, for example, some chips having a few SP worse could get equal or better result than the chip that has the higher SP value. Which may means that Intel did not bin their chip accurately.

 

[Outback Bronze]

Which may means that Intel did not bin their chip accurately.

Or the Asus SP algorithm could not be correct as well, no?

 

[Super Firm Tofu]

Do you mind posting a screen shot of your 13900KS VF curves? Would like to see it vs a 14900KS. You can post your 14900KS vf curves too if you like.

This KS is 101 and I'm quite happy with it. I do have a SP 105 KS too but the 101 seems to be undervolting better so I'm not sure everybody should be taking SP for real.

View attachment 353619

I only have the 13900K (not the KS) and it's in an MSI board that doesn't have that info available.

Here's my highly under-clocked 14900KS

 

[Dr. Dro]

I wonder how is everyone's SP rating for their chip? I got a i9-13900KS on ebay with SP 111, P core 122, E core 89 for $430 USD + shipping & tax. It looks like the P core SP is quite high. Not sure if my chip is rare or not.

Unfortunately, I do not have an ASUS motherboard, but if it is of any interest and can potentially help, my MSI CPU Force predictor gives me this score:

 

[Enigmatic9400]

Unfortunately, I do not have an ASUS motherboard, but if it is of any interest and can potentially help, my MSI CPU Force predictor gives me this score:

View attachment 353634

What is your default VF curve for this computer? For reference, a SP 111 chip has 1443mv around 6.0GHZ from igor's lab's binning spreadsheet.

 

[Dr. Dro]

What is your default VF curve for this computer? For reference, a SP 111 chip has 1443mv around 6.0GHZ from igor's lab's binning spreadsheet.

Unfortunately, MSI's solution doesn't give the VF points readout like the ASUS SP predictor does, only that final aggregate score. I've been wanting to know myself for some time now... but it's a very well behaved chip, the default voltage goes around ~1.42 for the 6 GHz P-cores before any undervolting takes place. I'm happy with it.

 

[Enigmatic9400]

Unfortunately, MSI's solution doesn't give the VF points readout like the ASUS SP predictor does, only that final aggregate score. I've been wanting to know myself for some time now... but it's a very well behaved chip, the default voltage goes around ~1.42 for the 6 GHz P-cores before any undervolting takes place. I'm happy with it.

Noice! Maybe your chip has a better bin than mine!

 

[Dr. Dro]

Noice! Maybe your chip has a better bin than mine!

It has certainly been stable with a -0.050 offset, and I'm happy with that. End of the day, that's all that matters, you picked a great chip to build your system around! Enjoy