Those on Raptor Lake... What is your Vid cap? (Intel's default - or something else? - not updating at all??)

[iameatingjam]

Have you decreased it from intels default of 1.55?

I tried putting 1.4v but I couldn't reach max boost unless idle basically. Maybe if I had a better bin. Lots of people on reddit seem to be doing this.

I tried putting 1.5v and not much changed, passmark cpu test shows a 1-2% loss of performance. So I'm not sure if I should do this or just leave my fate in intel's hands (and maybe get golden sample as an rma replacement?)

 

[chrcoluk]

On my board its not configurable (and also cant be viewed), one thing I noticed with ASRock, is when I compare to buildzoid's bios it seems like about 90% of the settings are missing.

So no idea sorry.

I miss the days when board reviewers also reviewed the actual bios, so one knew what could be configured in the bios before buying the board.

 

[iameatingjam]

On my board its not configurable (and also cant be viewed), one thing I noticed with ASRock, is when I compare to buildzoid's bios it seems like about 90% of the settings are missing.

So no idea sorry.

Oh yeah, lots of people can't pick. Well shouldn't that put you in either the (1.55) or (on some old bios f these new updates).... options, lol.

​

 

[chrcoluk]

Oh yeah, lots of people can't pick. Well shouldn't that put you in either the (1.55) or (on some old bios f these new updates).... options, lol.

​

Yeah I assume if I update my bios it will be 1.55v (enforced via microcode), but on the old bios? who knows.

 

[iameatingjam]

Hmm I see what you're saying, worded a little too harsh? EDIT: Oh well I guess I can't edit it, can only add new options. Oh well.

 

[agent_x007]

CPU-z should be able to show it, you just have to adjust one value in cpuz.ini file (from "Sensor = 1" to "Sensor = 0").
If it works, you should see "Core VID", instead of "Core voltage" :

Yes, my waste of sand also likes to cook itself - but I'm locking it ~1.35V, so it's fine

 

[chrcoluk]

CPU-z should be able to show it, you just have to adjust one value in cpuz.ini file (from "Sensor = 1" to "Sensor = 0").
If it works, you should see "Core VID", instead of "Core voltage" :
View attachment 359364
Yes, my waste of sand also likes to cook itself but I'm locking it ~1.35V so it's fine

Yeah but thats not the cap, here is mine e.g.

Interesting in your newer build it says VID instead of vcore now, I will update it, but I suspect it will show the current VID instead of the max VID.

Updated it, and still shows live vcore for me.

 

[agent_x007]

I'm just locked to max. frequency here (on desktop), nothing in the background.

Main problem for this kind of thing is poll rate, or how fast programs checks voltage.
BZ videos will make it look easy with Oscilloscope, but it's REALLY hard to catch them.

 

[iameatingjam]

I'm just locked to max. frequency here (on desktop), nothing in the background.
Main problem for this kind of thing is poll rate, or how fast programs checks voltage.
BZ videos will make it look easy with Oscilloscope, but it's REALLY hard to catch them.

Right, thats kind of the whole benefit of the vid cap. You dont have to worry so much about having super accurate fast software reporting, if there's a hard limit in the cpu.

I see you're on 11th gen, so I don't think you're affected by this though if memory serves rma rates were still on the high side 11th gen though, correct me if I'm wrong.

 

[chrcoluk]

CPU-z should be able to show it, you just have to adjust one value in cpuz.ini file (from "Sensor = 1" to "Sensor = 0").
If it works, you should see "Core VID", instead of "Core voltage" :
View attachment 359364
Yes, my waste of sand also likes to cook itself - but I'm locking it ~1.35V, so it's fine

Ok switched it, but its the live VID not the VID limit.

 

[agent_x007]

Live VID is fine.
Now, with CPU-z window open, you have to put CPU under single core stress and see what happens to VID value.
If you want monitoring/log to check max/min VID - HWInfo64 is a good place to start.

PS. All core steady workload WILL NOT work for this, also with live view only - you have to do screenshot "in time".
That's why having a log and max/min values is important.

 

[chrcoluk]

Live VID is fine.
Now, with CPU-z window open, you have to put CPU under single core stress and see what happens to VID value.
If you want monitoring/log to check max/min VID - HWInfo64 is a good place to start.

PS. All core steady workload WILL NOT work for this, also with live view only - you have to do screenshot "in time".
That's why having a log and max/min values is important.

Yeah but I think we were talking about viewing the VID limit that is configured in the bios not the reported VID. It seems on some new boards, there is a limit now that can be configured.

As you said we already have tools like hwinfo to monitor the VID activity, and that can be set to a very high polling rate as well. Regardless, thanks for the tip because I do see when VID is configured in CPUZ it uses a fairly high polling rate as standard, Nice to learn something new about these tools.

 

[agent_x007]

I find it highly improbable for Intel/MB makers to actually show you in plain text what is max. VID of your CPU (at max. frequency CPU operates at). However, even if they won't provide number directly - there are ways to figure it out indirectly

 

[ShrimpBrime]

Got a new Z790 ordered up.
So on the temp board B660 with code x112 just let it run default VID. Yeah, it shows over 1.5v sometimes. Meh, no big deal.

When on a proper Z790 board, VID is left on auto and usually run about 1.4v all core looking for 5.8ghz P-core and 4.7ghz E-core. The area effecient cores really suck down the juice though.

In order to run 8P cores and no E-cores at 6.3 to 6.4ghz 16 threads will utilize between 1.475 and 1.50v. Requires better cooling than most people intend to work with.

Just to validate 7ghz on a couple cores, took about 1.70v.

Intel Core i7 14700K @ 7001.68 MHz - CPU-Z VALIDATOR

[ukvv7b] Validated Dump by ShrimpBrime (2024-05-28 18:15:17) - MB: MSI Z790 GAMING PLUS WIFI (MS-7E06) - RAM: 32768 MB

valid.x86.fr

I suppose everyone still worried......

 

[iameatingjam]

Got a new Z790 ordered up.
So on the temp board B660 with code x112 just let it run default VID. Yeah, it shows over 1.5v sometimes. Meh, no big deal.

When on a proper Z790 board, VID is left on auto and usually run about 1.4v all core looking for 5.8ghz P-core and 4.7ghz E-core. The area effecient cores really suck down the juice though.

In order to run 8P cores and no E-cores at 6.3 to 6.4ghz 16 threads will utilize between 1.475 and 1.50v. Requires better cooling than most people intend to work with.

Just to validate 7ghz on a couple cores, took about 1.70v.

Intel Core i7 14700K @ 7001.68 MHz - CPU-Z VALIDATOR

[ukvv7b] Validated Dump by ShrimpBrime (2024-05-28 18:15:17) - MB: MSI Z790 GAMING PLUS WIFI (MS-7E06) - RAM: 32768 MB

valid.x86.fr

I suppose everyone still worried......

I see. When I first unshackled my cpu. ( I had it under a million restrictions when the panic was at its max, just cause there was so much smoke, and I had no idea how much fire. I had vid limit 1.3v ecores max 3.5 pcores max 5.0. pl1/pl2 max 200. AC/DC at 70.

When the update came out, and I realized, I should be getting what I paid for, I unshackled it, but I was getting ( as software reports - so probably not that accurate, vcore up to 1.44) which I didn't like. But now. With just the AC/DC loadline reduced just a little bit, from 110 to 90, and taking that vid limit down from 1.55 to 1.5, my vcore (as reported) doesn't go above 1.4. On occasion it might go up to like 1.41. Gotta always assume transients are higher, but I can't imagine them being any higher than say 1.46 (again, transients, just to be clear - sorry, been speaking with some people with real poor reading comprehension lately - not your fault).

Anyway... For example when I set a vid limit of 1.4, my software reports no vid above 1.37/138 (again, probably because of transients) and then, there's just my vcore generally being lower than my vid by... idk it depends. Just checking hwinfo it looks like my max vcore is .3 lower than my max vid (dont ask me why, but I'm glad). I know some people like to have them like exactly the same but I don't really care so much about that.

So with my cpu firing on all cylinders, so old ucode, high iccmax, big undervolt, memory overclocked to snot with sky high trefi... passmark cpu test would get me like 57,000 and some change. Take away the memory OC and put on intel defaults it goes down to ~55,000. Then set the vid cap 1.5 and reduce the AC/DC loadline from 110 > 90 gets me ~54,500. So yeah, about 5% slower than the cpu completely unrestricted, and about 1% slower than intel's normal performance profile. Doesn't seem like such a bad deal, eh? I just hope there's no invisible degredation, even though I feel there probably is.

I mean how do you tell it apart from just not the best bin? Then again this chip has never quite been able to maintain boosts under load like my first chip. But also this one works after 8 months and that first one died almost immediately so its gotta get points for that, right?

Btw... I still have a moderately inflated trefi on my memory.... I figured that wouldn't stress the memory controller, if anything it would do the opposite, right? So I put it at 40,000 with trefix9 or w/e its called at 40 so that my memory still has acceptable latency. Plus at this low frequency its not like its gonna get too hot or anything, which asfaik is one of the main risks with increasing it ( If somebody wants to correct my assumptions on this, feel welcome).

I mean a Z board would be nice but.... I don't feel like I should be putting any more money into this platform and rather start saving for the next. I'm pretty broke

I find it highly improbable for Intel/MB makers to actually show you in plain text what is max. VID of your CPU (at max. frequency CPU operates at). However, even if they won't provide number directly - there are ways to figure it out indirectly

I mean intel just did.... mind you, they are in a bit of a bind so they have to do something:

here's the link: https://community.intel.com/t5/Proc...el-Core-13th-and-14th-Gen-Desktop/m-p/1622129

And the quote "The latest microcode update (0x129) will limit voltage requests above 1.55V as a preventative mitigation for processors not experiencing instability symptoms."

 

[ShrimpBrime]

Generally on a full load at defaults, I see a working all core voltage of about 1.288v

VID requests are for all that boosting. Single core and dual core loads on your fastest cores. The VID will be a reading for core 0 because that's the core that's selected by cpu-z default. Right click the frequency and a tab will open of all the cores. Select individual cores and that VID might change. Try selecting your fastest boosting cores. System must be at defaults I'd imagine though.

See it's not a V-core restriction. You should ()or could try) lower/limit current by the amps in my opinion. The danger isn't the v-core, it's the current and temperature. Just remember to get a wattage you need volt x amp.

So if you lower v-core, it'll want to pull more amps at the same frequency. Pretty much why you don't really loose performance. But you are making the VRMs work for it a little harder.

 

[iameatingjam]

1) Generally on a full load at defaults, I see a working all core voltage of about 1.288v

2) VID requests are for all that boosting. Single core and dual core loads on your fastest cores. The VID will be a reading for core 0 because that's the core that's selected by cpu-z default. Right click the frequency and a tab will open of all the cores. Select individual cores and that VID might change. Try selecting your fastest boosting cores. System must be at defaults I'd imagine though.

3) See it's not a V-core restriction. You should ()or could try) lower/limit current by the amps in my opinion. The danger isn't the v-core, it's the current and temperature. Just remember to get a wattage you need volt x amp.

So if you lower v-core, it'll want to pull more amps at the same frequency. Pretty much why you don't really loose performance. But you are making the VRMs work for it a little harder.

1) Yeah same here, I was stating the maximum vcores hit over say a several hour period of doing many different things. My averages are in the 1.2s and 1.3s too.

2) Yup. Though I disabled turbo 3. Forgot to mention that.

3) Limit my current even more than intel already has? Its already at 307.... Is that not low enough already for 20 cores? When before we were all running at 511?

I mean dude, nothing against you its just so many people tell me so many things. Just the other day when I was asking about if it was possible too much current killed my first cpu I got responses that "its vcore killing these cpus, not current"

Though I believe it can be either, or something else entirely. But I'm sure you get my point. Everybody's got their own take. Intel says elevated voltage is a factor so I'm doing pretty much the bare minimum to get vcore down while for the most part, following the rules. Besides, I got 4+ more years of rma baby. A small part of me wants to remove every restriction just so I can roll the lottery again. (Okay thats mostly a joke... but there's a kernel of truth in there...)

 

[ShrimpBrime]

1) Yeah same here, I was stating the maximum vcores hit over say a several hour period of doing many different things. My averages are in the 1.2s and 1.3s too.

2) Yup. Though I disabled turbo 3. Forgot to mention that.

3) Limit my current even more than intel already has? Its already at 307.... Is that not low enough already for 20 cores? When before we were all running at 511?

I mean dude, nothing against you its just so many people tell me so many things. Just the other day when I was asking about if it was possible too much current killed my first cpu I got responses that "its vcore killing these cpus, not current"

Though I believe it can be either, or something else entirely. But I'm sure you get my point. Everybody's got their own take. Intel says elevated voltage is a factor so I'm doing pretty much the bare minimum to get vcore down while for the most part, following the rules. Besides, I got 4+ more years of rma baby. A small part of me wants to remove every restriction just so I can roll the lottery again. (Okay thats mostly a joke... but there's a kernel of truth in there...)

I am not offended in any way. I look at all processors the same.
But I share with you my experience and my thoughts. You utilize that when and where you want too. Other people have valid points and perhaps different experiences than myself.

Current and temps kill processors.

You can pull 250 amps at 1.250v and would be 312 watts

Resistance increases with heat. Processors leak voltage.

The idea is to keep your Intel Processor (any processor) cool as possible what you are doing.

I cannot overclock Raptor Lake processors without utilizing certain temperatures. Even with Ice water, the cpu will not go past 6.2ghz and post windows. Ambient Loop is only good for 5.8ghz all cores.

Honestly man, no offence to anyone, but most people just don't have adequate cooling. 253w is a lot of power. a Lot of thermals to move. must be moved quickly! So you will see from searching that de-lidding is pretty popular. Water as close to the hot die as possible. 860 BTU/hr is a lot coming from an area as small as the tip of your finger.

I find most under-volters just don't have suffice cooling methods or have high ambient temps.

If you look at @Bill_Bright signature, it tells everyone exactly what kills processors. "Heat is the bane of all electronics!" It's always been true.

Should we just keep trying under volt this, tweak frequency. Try and keep performance. Most definitely! Any way you can take it. I'd say aim for all core loads of 90c or less and 70c or less for everything else (gaming) and the processor should have minimal degradation. I cannot speak here about if motherboards actually handle loads well or not. everyone has a different board and would have to research their power grid and be sure it's robust with good parts. So far I've killed a board though. Disappointing as hell. Really thought it was the cpu. Nah. It runs hot as hell like it should. Thanks Intel.

 

[iameatingjam]

Thats funny because I always found both my 14700ks not that hard to cool. Mind you, I don't stress em that hard. Lets see, a random day of doing typical computer things, max temp only got to 66c and its pretty hot out today. I only have a 240mm aio but the fans I have go really fast (3200rpm) so I have a curve like this:

The idea is, 2200 rpm might seem like its annoying for idle, but its totally true that its changes in fan speed you notice, not the fan speeds themselves. So it really doesn't take long to tune it out. So nice steady cooling mechanism, unless, temps hit 80, then it immediately goes into overdrive as you can see. I wish the bios had functionality so that once it passes that threshold it only slowly backs down. ( you can do it with fan control but I'm trying to keep the always running utilities I don't really need to a minimum- otherwise if you're stuck in place where your temps are running between 79-81 it can get really annoying - luckily that doesn't happen too often). The other benefit is, with higher fans speeds on idle, means takes longer to heat up. Though in reality thats probably only milliseconds lol.

Anyway this works really good, despite it being god awfully loud at full power.

 

[ShrimpBrime]

Thats funny because I always found both my 14700ks not that hard to cool. Mind you, I don't stress em that hard. Lets see, a random day of doing typical computer things, max temp only got to 66c and its pretty hot out today. I only have a 240mm aio but the fans I have go really fast (3200rpm) so I have a curve like this:


View attachment 359398
The idea is, 2200 rpm might seem like its annoying for idle, but its totally true that its changes in fan speed you notice, not the fan speeds themselves. So it really doesn't take long to tune it out. So nice steady cooling mechanism, unless, temps hit 80, then it immediately goes into overdrive as you can see. I wish the bios had functionality so that once it passes that threshold it only slowly backs down. ( you can do it with fan control but I'm trying to keep the always running utilities I don't really need to a minimum- otherwise if you're stuck in place where your temps are running between 79-81 it can get really annoying - luckily that doesn't happen too often). The other benefit is, with higher fans speeds on idle, means takes longer to heat up. Though in reality thats probably only milliseconds lol.

Anyway this works really good, despite it being god awfully loud at full power.

I do the same thing actually. At least on the air cooler. Goal is sub 70c package temps utilizing F@H recently.. This costs some performance though. Runs 3.8ghz 28 threads. With a working v-core of 1.154v VR_out at 43 amps reporting a package power of 94 watts. Cores are at 63c Package at 65c as I type this. Fan is at 55% duty cycle unless hits 90c. Then it can fly. But all I did was just set the cpu multipliers. I didn't under-volt, the board that on it's own. Just running a static multiplier. Then the CPU is like YAY, here's some nice low v-core and nice low temps. Obviously F@H isn't gaming, even though I'm using the Graphics card as well. And that I have tuned in about 155w instead of the blazing stock 220w I've seen it boost to. Like this, it's pretty efficient. I mean it's crunching quite a bit with a performance hit yes, but at a much much lower cost.

As example.
EDIT: This is overclocked. The base clocks are 3.4ghz...