Raptor Lake Settings Survey and Spreadsheet "Database"

FlyingScot

Well-known member
Joined
Apr 30, 2024
Messages
1,378
Welcome to the Raptor Lake Settings Survey and Spreadsheet "Database"

While no two systems (or CPUs) are the same, it is my hope that you will find this centralized information useful when comparing your own specific setup to those who have the same CPU - and maybe similar hardware. In addition to this objective, I have also provided an opportunity for Raptor Lake owners to periodically submit additional entries to the survey database when anything changes. If degradation becomes more of a widespread issue than originally anticipated then these periodic "field reports" should allow us to see these trends as they unfold.

Instructions for Survey Submissions
Once you're satisfied with your system's stability and performance, please follow the instructions below to prepare for your first submission to the survey database.

Please download the attached Survey Guide.pdf. This guide has been designed to give each survey respondent a chance to gather all the relevant information ahead of time. This is an important step. The last two pages of the guide describes how you can make immediate use of the survey results, either before or after you complete your online entry.

In addition to the many questions in the survey, you will find a section (Section 2) that asks for performance related data. I have selected Cinebench R23 for this purpose. However, Cinebench R23 is only one small snapshot of performance and I realize will not necessarily be indicative of gaming performance, etc. In fact, as our recent testing has demonstrated, some "undervolting approaches" are much better suited to all-core CPU workloads (like Cinebench R23) than those favoring lighter, non all-core workloads, such as gaming.

[1] Use this link to access the Online Survey
NOTE: Survey responses cannot be edited once they are submitted. However, I can still assist you in making changes (see NOTEs below). Survey responses can also be deleted (by me) if you wish.
:stop: Please don’t forget to enter your Forum name (if you’re a member) to help others link you to your settings. But for privacy reasons, please don’t use your real name or email address, etc.

[2] Use this link to access the Online Spreadsheet "database".
NOTE: Pink cells indicate data that I have corrected on behalf of the respondents.
NOTE: To read extra long “User Comments”, double-click on them to pop up a scrollable box.

[3] The analytics webpage <<coming soon>>

If you are very new to the topic of Raptor Lake tuning then my guide might help you get started.
Guide: "Almost" everything you need to know about Raptor Lake Voltage/Power/Temperature Tuning

If you’re already quite comfortable with undervolting via the Intel Loadline settings (LLC, AC_LL, DC_LL) then you should find Vassil_V’s case study quite interesting.
Different undervolting methods with IA CEP enabled, and how they compare to Lite Load
You will also find his latest “favorite" settings in the Survey database.
 

Attachments

  • Survey Guide (v5).pdf
    986.2 KB · Views: 175
Last edited:
So we have an update, I went ahead and took the benchmark a little more seriously this time. Did it first thing on a cold boot with icue closed and used msi center to free up as much memory as possible. I measured the temp and it is 70F right on the nose. 21C. The Score is: 42,442 @ 1.258V, 246.2W, 248.2A with a max temp of 83C. The Vid/Vcore delta is 8mv Vcore below Vid.

I think the only way to get it better is to kill HWinfo and go into task manager and kill everything else. Which I don't really feel like doing. @FlyingScot would you like to update the spreadsheet or should I just post another one? I'm good either way, doesn't matter to me. If the other entry is edited, the answer under Q for max cpu temp is wrong, it should be 100C. I misunderstood that particular question. I guess since the ambient temp was lower, and an exact known quantity this time, I should probably make another one?

Up to you man.

Screenshot 2025-01-19 042832.png
Screenshot 2025-01-19 054226.png
Screenshot 2025-01-19 061022.png
Screenshot 2025-01-19 061506.png
 
Last edited:
So we have an update, I went ahead and took the benchmark a little more seriously this time. Did it first thing on a cold boot with icue closed and used msi center to free up as much memory as possible. I measured the temp and it is 70F right on the nose. 21C. The Score is: 42,442 @ 1.258V, 246.2W, 248.2A with a max temp of 83C. The Vid/Vcore delta is 8mv Vcore below Vid.

I think the only way to get it better is to kill HWinfo and go into task manager and kill everything else. Which I don't really feel like doing. @FlyingScot would you like to update the spreadsheet or should I just post another one? I'm good either way, doesn't matter to me. If the other entry is edited, the answer under Q for max cpu temp is wrong, it should be 100C. I misunderstood that particular question. I guess since the ambient temp was lower, and an exact known quantity this time, I should probably make another one?

Up to you man.
Sorry, I haven’t had time yet to fully absorb your previous posts. But in quickly scanning your notes, all that information you provided looks terrific. I’m actually working on a comprehensive stability testing guide for beginners (in my spare time). I‘d like to steal some of your stability comments, if you don’t mind. Some of it was confirmation of what I already knew, but some of it is more granular in nature and quite interesting. I’m also not done with my main UV guide yet, so your other feedback might also be useful. And I still have a lot of loadline tests I’d like to get to in order to fill in some gaps in my main UV guide, as well as some holes in my own understanding. Again, your experiences are very useful as you have taken things further than 90% of people. By the way, if you want to go even deeper then I can recommend the series of Skatter videos and companion articles on 13th/14th gen, assuming you haven’t already reviewed them.

In regards to your new results, I think there are enough changes that a new entry would be more appropriate. You could always add another URL to point to a post in this thread if you want to add more visual context. I will however update your max temp to 100C on your latest entry, which is the Auto default setting - if I understand you correctly.

Again, thanks for sharing all this information with the forum. I know a lot of people who come here just want quick and easy results. But perhaps some of them will return in the future to delve a little deeper. After all, PC tuning can become a hobby all of its own, especially when you get into areas like latency and memory tuning.
 
Last edited:
Sorry, I haven’t had time yet to fully absorb your previous posts. But in quickly scanning your notes, all that information you provided looks terrific. I’m actually working on a comprehensive stability testing guide for beginners (in my spare time). I‘d like to steal some of your stability comments, if you don’t mind. Some of it was confirmation of what I already knew, but some of it is more granular in nature and quite interesting. I’m also not done with my main UV guide yet, so your other feedback might also be useful. And I still have a lot of loadline tests I’d like to get to in order to fill in some gaps in my main UV guide, as well as some holes in my owner understanding. Again, your experiences are very useful as you have taken things further than 90% of people. By the way, if you want to go even deeper then I can recommend the series of Skatter videos and companion articles on 13th/14th gen, assuming you haven’t already reviewed them.

In regards to your new results, I think there are enough changes that a new entry would be more appropriate. You could always add another URL to point to a post in this thread if you want to add more visual context. I will however update your max temp to 100C on your latest entry, which is the Auto default setting - if I understand you correctly.

Again, thanks for sharing all this information with the forum. I know a lot of people who come here just want quick and easy results. But perhaps some of them will return in the future to delve a little deeper. After all, PC tuning can become a hobby all of its own, especially when you get into areas like latency and memory tuning.
Yes sir, I kind of figured it should also be a new entry.

Of course I don't mind, you can use whatever you want to. The 100c auto setting, you did understand that correctly. Although I did not leave it at auto, I physically punched in 100c. Not that it matters, that one doesn't adjust itself. But with other settings, I can't make up my mind if they adjust on the fly or not. Particularly lite load settings, They do seem to sometimes. So there's a number of settings I will not leave on auto, simply for that reason. Not positive about that though, not yet. What do you think, any insight?

Now that you mention gaps in loadline, I'm right there with ya. I'm starting to narrow it down, have a pretty good idea of most of it. But its all theory and conjecture on my part so far. I have some gaps myself. Ill look forward to your guide on the matter. If i have any great epiphanies on the matter ill let you know. We seem to be on our own with liteload. There is no concrete, authoritative info on the matter. Not easily found anyway, I'm sure its out there, but I haven't seen it yet. Although the setting itself is not proprietary, the engineering and implementation behind those settings definitely are proprietary. So that doesn't help


I'm very familiar with voltage regulators, as well as general theory behind integrated circuits, their components and purpose. But this is new, I've not dealt with or even heard of anything this complicated. Also the cpu is where i always stopped, I still cant wrap my head around the numbers and scales were really talking about here. Once we understand it, it'll be simple. But getting there will not be, that's for sure.

I will gladly consume any info you can point in my direction. Articles and videos, bring it on.
 
I have several paragraphs written up on the subject, but not anything that could be shared yet. Not even close actually. So far all I'm going to say that I'm cofident about is LLC and liteload are independant of eachother. Although they work in tandem, they are not symbiotic, so to speak. The LLC is VRM controlled only, most likely based on intrernal maps that i uses, similar to a vf curve, only based on current output thats its measuring. And its not watching or adjusting to anything liteload related in real time. (Pure speculation on my part.) Hence why it will change along with liteload if you make a change to the lite load.

I belive it has to do with the speed the cpu is running at. If things are being calculated at 6 BILLION times per second, theres just nothing that can come close to keeping up with that. Even another cpu like the VRM has. So complete non-educated guess is liteload is the cpu's way of adjusting the vrm more granularly and on a much faster time scale.

LLC is just a blanket, "this amount of current is flowing, add this amount of voltage on top of whats there." The liteload would be the way the cpu can add adjustments on top of that, either by adding input to the vrm's processor itself, or, by controlling additional transistors inside the banks, or packs or whatever or whatever they're called.


Dont use any of that, I have no idea what im talking about. Its all speculation at this point.
 
Dammit, I screwed it up again, because of course i did. TVB clipping should be custom OCTVB. If you can get around to editing that, great. If not, oh well don't sweat it. I don't think I've filled out a single one yet where I didn't make any mistakes.

Side note, Is that a Miata? you a glutton for punishment or something? Those are hard enough to work on as is, but a race car? Dang dude. Probably hauls [***CENSORED***] though.

I actually worked on one that was a track car, just like that not too long ago. It wouldn't go on a 4 post without ripping the bumper off, and to get it on a two post I had to jack it up with a floor jack just to get the arms of the lift underneath it. Never again, that car was no fun all the way around.
 
Last edited:
Dammit, I screwed it up again, because of course i did. TVB clipping should be custom OCTVB. If you can get around to editing that, great. If not, oh well don't sweat it. I don't think I've filled out a single one yet where I didn't make any mistakes.
No problem. Both fixes are in!
Side note, Is that a Miata? you a glutton for punishment or something? Those are hard enough to work on as is, but a race car? Dang dude. Probably hauls [***CENSORED***] though.
Yep! I started out just going to track days with the local car club like many people do. I did that for a few years until I got the chance to join a two-car SCCA racing team with a bunch of friends supporting our efforts. This was back in 2003 when things were SO MUCH cheaper. I mean, a US gallon of gasoline (petrol) was just $1.25!!!

To be honest, I was never much of a Mazda guy, or Miata guy, until I drove one on track. But once you stiffened them up with a full roll cage and added the Mazdaspeed racing kits, it was like WOW! Pulling 1+g in the corners was so much fun! There wasn't a standard roadcar, not even the Porsches, Corvettes, etc. that could beat your lap times unless the driver was very skilled.
I actually worked on one that was a track car, just like that not too long ago. It wouldn't go on a 4 post without ripping the bumper off, and to get it on a two post I had to jack it up with a floor jack just to get the arms of the lift underneath it. Never again, that car was no fun all the way around.
Lucky for me, my teammate and his friend were VERY familiar working on Miatas. Complete engine swaps were just a few hours. It does help to have small hands, though, so I know what you are saying. We also had a two-arm lift, so we never ran into the issues you did. It was definitely fun times! They were fun little cars and incredibly durable, too. They took that English Lotus design and added that Japanese engineering magic!
 
Last edited:
Cant quite make out the top, but it looks like a hard top. If so, that's kind of a rare duck. Unfortunately for me, I've worked on a lot of them, only once have I seen one that wasn't a convertible.

They're pretty solid cars, I don't see many problems with them at all, beyond the typical stuff. Belts and clutches and the usual. I don't believe I've ever seen one actually break though. Not yet anyway

I guess they partnered with a lot of companies. Don't know anything about Lotus, and wouldn't anyway. I don't pay attention at all, cars are not exactly my thing. But I do know there's a lot of Ford in them. specially the later ones. I'm guessing they were built here and they Partnered with Ford to build them or something. The PCM, all the electrical. The Clutch and master/slave cylinders. Those are all Ford parts. Not too sure about the earlier models but the later ones, I believe it was an 08 if I remember right, have Ford engines too.

That kind of thing is getting more and more common. All the newer Chrysler stuff has more Mercedes in it than Chrysler. Which sucks because I really hate German cars. Love Japanese cars, but really anything designed in Europe is a guaranteed bad day when it pulls in your bay. And I wont touch Minis at all. No thanks
 
Cant quite make out the top, but it looks like a hard top. If so, that's kind of a rare duck. Unfortunately for me, I've worked on a lot of them, only once have I seen one that wasn't a convertible.

They're pretty solid cars, I don't see many problems with them at all, beyond the typical stuff. Belts and clutches and the usual. I don't believe I've ever seen one actually break though. Not yet anyway

I guess they partnered with a lot of companies. Don't know anything about Lotus, and wouldn't anyway. I don't pay attention at all, cars are not exactly my thing. But I do know there's a lot of Ford in them. specially the later ones. I'm guessing they were built here and they Partnered with Ford to build them or something. The PCM, all the electrical. The Clutch and master/slave cylinders. Those are all Ford parts. Not too sure about the earlier models but the later ones, I believe it was an 08 if I remember right, have Ford engines too.

That kind of thing is getting more and more common. All the newer Chrysler stuff has more Mercedes in it than Chrysler. Which sucks because I really hate German cars. Love Japanese cars, but really anything designed in Europe is a guaranteed bad day when it pulls in your bay. And I wont touch Minis at all. No thanks
You sound like a man who has worked on cars way too long. I mean, work on cars long enough and you loose the will to live. ;) Personally, I think we may have peaked in the 90’s (maybe early 2000’s) when Japanese cars were affordable, reliable, and relatively easy to work on. In my opinion, it‘s been downhill ever since for the entire car industry.

My racing Miata was a 1990 NA chassis, which was the launch year + 1. It was as pure Mazda as you will ever likely come across. Hardtops were sourced separately, and very hard to find. However, I had a crew chief who could find anything - literally anything! I really don’t know how he did it. Of course, back in 2003 Google actually worked. In fact, a lot of things just worked. By the way, I also have the 2007 NC Miata chassis, so I know of what you speak regarding the Ford parts. Although, it has turned out to be a wonderfully reliable car, too. 18 years later and it still drives great! And so does my wife’s 2003 NB Miata chassis. Not daily drivers, though.
 
Last edited:
Wow, that's a lot of Miatas. Im guessing they are either red, or yellow. LOL That is all I've ever seen, except that one that was a track car. Yes, they are reliable. I do see a dis proportionally large amount of them with electrical issues. Particularly with the starting circuits. Its almost always related to the battery being in the trunk like that, there's that voltage drop issue I've talked about.

Thats about it. They really have no Achilles heel other than that. Usually corrosion in those long cables. Grounds will not tolerate resistance.

Yes, worked on cars way too long, about 20 years now. I pretty much hate it.

I love the newer Diesels, super complicated but damn do they have some power. As far as going down hill, I guess they're more or less the same. Just different, ya know? The only major issues i see are always related to inferior parts. Things like cams and lifters. They aren't hardened correctly, because they're made in China. But if any car you own is like that, Itll show itself well before 100k miles. If a newer car has made it that long, then you're good to go.

Everything is made in China, that's the real issue there. I think there's only a couple real parts makers left here in the states. They simply cant compete.
 
Wow, that's a lot of Miatas. Im guessing they are either red, or yellow.
Nope. My wife’s car is a metallic gray special edition with blue soft top. I think the best looking color scheme they ever made. Mine is British Racing Green…of course.
Yes, worked on cars way too long, about 20 years now. I pretty much hate it.
I‘m sure it gets old. And you start to ache in all the wrong places. I keep hearing there’s an exodus of mechanics from the industry. That can’t be good.
As far as going down hill, I guess they're more or less the same. Just different, ya know? The only major issues i see are always related to inferior parts. Things like cams and lifters. They aren't hardened correctly, because they're made in China. But if any car you own is like that, Itll show itself well before 100k miles. If a newer car has made it that long, then you're good to go.
That reminds me. Back in 1996, I bought a new car from a part-time car salesman at a small dealership in a small town…so I could get the best deal possible. He was an old farmer who even wore the raincoat and wellingtons to prove it. He told me that the best way to know if you’re getting a good car for the money was to divide the cost of the car by the weight of the car. It kind of stuck with me, even though it might sound silly to most. Maybe we should use the same yardstick when we buy motherboards! :-)
 
Nope. My wife’s car is a metallic gray special edition with blue soft top. I think the best looking color scheme they ever made. Mine is British Racing Green…of course.

I‘m sure it gets old. And you start to ache in all the wrong places. I keep hearing there’s an exodus of mechanics from the industry. That can’t be good.

That reminds me. Back in 1996, I bought a new car from a part-time car salesman at a small dealership in a small town…so I could get the best deal possible. He was an old farmer who even wore the raincoat and wellingtons to prove it. He told me that the best way to know if you’re getting a good car for the money was to divide the cost of the car by the weight of the car. It kind of stuck with me, even though it might sound silly to most. Maybe we should use the same yardstick when we buy motherboards! :-)
That's a new one on me.

But I do wish I would have sprung for a higher tier motherboard. One that has that sensor that reads the actual voltage in the sensing circuit, can't remember what that sensor is called. But it would make tuning a lot easier. Other than that the tomahawk is a pretty decent board.
 
But I do wish I would have sprung for a higher tier motherboard. One that has that sensor that reads the actual voltage in the sensing circuit, can't remember what that sensor is called. But it would make tuning a lot easier. Other than that the tomahawk is a pretty decent board.
I think it is a good motherboard for the money - at least by today's pricing standards. I think you might be referring to VR VOUT. If so, here's what I wrote in my UV Guide.

<snip-it>
* All about VR VOUT
Some of the more premium MSI motherboards (such as the Carbon, Ace, GodLike, etc.) expose an additional CPU voltage sensor to the HWInfo64 application called VR VOUT. The reason VR VOUT is useful is that, under load, it is the voltage as reported by the CPU itself (or close enough) - meaning ”as it enters the CPU die“ - making it far more accurate than Vcore, which comes from the motherboard PCB somewhere near the CPU socket. [EDIT: I recently read that it comes from the VRM before any vdroop occurs. But I kinda doubt that for MSI motherboards. Could be wrong, though.] Luckily, in the MSI world, it’s mainly just a nice-to-have feature because MSI super I/O monitoring has an overwhelming propensity to overestimate Vcore, which is not the case with some other motherboard manufacturers (depending on model) where Vcore might be significantly underreported. That’s when VR VOUT becomes not just a nice-to-have, but indispensable - especially when we’re trying to keep “real” CPU voltage under a certain target like we are with Raptor Lake. Tip: VR VOUT is usually very inaccurate when the system is in a low power/idle state.

Tip
: I would advise anyone with VR VOUT to only report it in addition to Vcore, and never in place of it. Otherwise, people are going to be comparing apples and oranges.
<end>
 
Last edited:
I think it is a good motherboard for the money - at least by today's pricing standards. I think you might be referring to VR VOUT. If so, here's what I wrote in my UV Guide.

<snip-it>
* All about VR VOUT
Some of the more premium MSI motherboards (such as the Carbon, Ace, GodLike, etc.) expose an additional CPU voltage sensor to the HWInfo64 application called VR VOUT. The reason VR VOUT is useful is that, under load, it is the voltage as reported by the CPU itself (or close enough) - meaning ”as it enters the CPU die“ - making it far more accurate than Vcore, which comes from the motherboard PCB somewhere near the CPU socket. [EDIT: I recently read that it comes from the VRM before any vdroop occurs. But I kinda doubt that for MSI motherboards. Could be wrong, though.] Luckily, in the MSI world, it’s mainly just a nice-to-have feature because MSI super I/O monitoring has an overwhelming propensity to overestimate Vcore, which is not the case with some other motherboard manufacturers (depending on model) where Vcore might be significantly underreported. That’s when VR VOUT becomes not just a nice-to-have, but indispensable - especially when we’re trying to keep “real” CPU voltage under a certain target like we are with Raptor Lake. Tip: VR VOUT is usually very inaccurate when the system is in a low power/idle state.

Tip
: I would advise anyone with VR VOUT to only report it in addition to Vcore, and never in place of it. Otherwise, people are going to be comparing apples and oranges.
<end>
Yep, that's definitely it. Not interested in godlike. So Carbon or ace is where I'll need to start?

We'll see. I'm definitely not unhappy with the tomahawk. It will do the job no problem. I just didn't give any thought to additional measurements. Is would love to have that extra piece of info. The board is still within return window. But Black Friday pricing Is long gone. So most likely I'll just keep it as is.
 
Yep, that's definitely it. Not interested in godlike. So Carbon or ace is where I'll need to start?

We'll see. I'm definitely not unhappy with the tomahawk. It will do the job no problem. I just didn't give any thought to additional measurements. Is would love to have that extra piece of info. The board is still within return window. But Black Friday pricing Is long gone. So most likely I'll just keep it as is.
If you can afford it, you might like the ACE. It has lots more goodies for you to connect and probe and analyze. And it comes with dual-BIOS in case you corrupt it. It's a tinker's mobo for sure.
 
If you can afford it, you might like the ACE. It has lots more goodies for you to connect and probe and analyze. And it comes with dual-BIOS in case you corrupt it. It's a tinker's mobo for sure.
Man, that thing looks awesome. I don't believe I can help myself, I'm gonna have to go get one.

Thanks for the tip
 
Here's some info that may muddy the waters a bit, but hopefully not. I can only speak from my experiences obviously, and I know less than you guys do, thats for sure.

However, The workloads that crash are usually not the ones that peg all cores and run the entire time that way. I am very curious to repeat this test with R15 and R15 extreme and see what those are doing, to which cores and how many. Particularly when compared to some configurations that I know will crash. So i'll be reverting to some previous settings and measuring them against the settings i have now and see if we can discern anything from it. If its clear as day I will report them back here.

This is Tiny Media Manager. Its just a little program that looks up all the metadata and posters, trailers etc. for your content, renames and cleans everything up for you. In this case I'm rebuilding the image cache for my movie collection. As you can see, the processor is pretty much all over the map as far as load, (stress.) And the power draw never really goes much above about half. But look at the activity, it is very dynamic. It is not a static load, not even close. This program, along with any other that builds cache and also R15 and R15 Extreme are a guaranteed way to induce a crash, and really, are really the only ones that will crash my system. Full bore stress testing will not, which is great because I don't really want to constantly hammer my cpu anyway.

Moreover, I couldn't really care less if my PC will not pass a stress test. I find it irrelevant. Will it crash in real workloads? That's what we should be asking ourselves. And i have found basically no correlation at all between stress test stability, and dynamic, real world stability; aside from R15. Which is why I will be repeating this exact setup, but with R15 as the load. I believe i will find that its not a load or voltage issue, but an issue of dynamics, or what the cores are doing, not how much, if that makes sense. We shall see.

Side note: Were using the term "dynamics" in order to keep it a little more palatable. Its actually the Pulse width, or could also be the duty cycle, since the frequency is never changing. But I'm not really the pretentious type. Besides, I'm just a mechanic after all. Don't really know what I'm talking about anyway. ;)



Screenshot 2025-01-25 143135.png
Screenshot 2025-01-25 143244.png
Screenshot 2025-01-25 143413.png
Screenshot 2025-01-25 143840.png
 
Back
Top