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

[Bobaganoosh]

I'm on an EVGA z690 Classified that hasn't had a BIOS update in quite some time (thanks for riding into the sunset, EVGA /s). I likely won't get the microcode update. Also, I've been running a launch-day 13900k at the same settings for a couple years now and so far *knocks on wood*, it's been fine. I can't set max VID on this board without doing an override voltage that would force higher than normal all-core voltage, but not enough voltage for boosting.

My settings, for reference, but they're probably not helpful to anybody with another OEM's motherboard:
I have it on adaptive "auto", but I tweaked the V/F points at 5100, 5400, and 5700 to -40. If I leave the BIOS at default, it'll boost voltages higher, but for some reason this setting seems to work. Their latest BIOS release sort of...barely...has functioning V/F point control, so shifting these three points down dropped the whole curve a bit. Their BIOS doesn't tell you the actual voltage points, it just lets you put in offsets. Also, if you put in too big of a global offset it just gets ignored, but manually setting "adaptive" and "auto" on that actually drops the voltage from leaving voltage mode as "auto" instead of "adaptive" (*shrugs*). I also have it set up to boost to 6GHz on 1-4 active cores, then 5-8 active are set to 5.8, 5.7, 5.6, 5.5GHz. e-Cores are at 4.6GHz. Ring 4.6GHz. FLL Override Mode 1. Vdroop on -75% (min droop). PWM on 600. Admittedly I got the FLL, Vdroop, and PWM settings from Loomi's suggestions on Youtube for these series of boards and all I've done to verify them is try a few other settings and confirm that these are more stable.

So the end result:
I don't have a good screenshot of HWiNFO64, but VIDs end up in the 1.25-1.325ish range for most tasks, especially large workloads that use lots of cores. If i run lots of low-thread workloads I'll see it peak around ~1.4-1.417V at 6GHz.

Also, there was an older microcode that ran lower voltages, and thus temps were lower...I'm currently thinking about doing some testing to see if I can run that one. On that one I just had the V/F points mentioned above at -25. The BIOS update thread on their forums indicated some stability improvements with the newer microcode so I've just been using it (for...over a year?...I don't remember when that version came out). I did see that thread referenced in GN's video actually, but they seemed to have the information backwards because the newer microcode increased the voltage (and heat).
Benchmate capture from pushing it a little (I think I may have had 5.6 or 5.7GHz at 8-cores, other settings above same) shows 1.37V on P-cores, but I don't know if that's max-VID:

Really pushing it a while back I did this CB23 run, with P-core voltage showing 1.42V.

For most gaming workloads though, just on my 24/7 settings mentioned earlier, it usually runs around 115-140W and it mostly stays below 1.4V (that being the max number in HWiNFO).

 

[ShrimpBrime]

Got the new board in. Set up. Water cooled with repaired waterblock.

14700K runs exactly the same as it did on Z690 MSI and Z790 MSI and B660-G boards at defaults. Also OC is exactly the same a the MSI Z690 and Z790 boards.

No degradation so far.

Static v-core and clocks, nothing hiding here. VID is over-ridden due to user set v-core.

It truly takes a bit of extra v-core to run the E-cores at 4.7ghz and higher. But the results are unchanged as far as I can tell.

This is the shipped bios for the board, I have not updated it and I see no need to do so at this time. My vote remains the same.

EDIT additional comment. This is W7 upgraded to W10. Not a direct W10 install.

 

[iameatingjam]

Wouldn't that be funny if these cpus in EVGA boards ended up out lasting all the others? Whens the last time you got an upgrade? Does it even offficially support 14th gen? I wonder if evga will ever rise from the grave. Being the only seemingly honest actor in diy market and dying, does not make a good case for caring for your customers to the other companies

I only have a crappy binned 14700kf, the max it ever got in cb2024 was 2042. It can't maintain max boost during all core loads like cinebench, not cause of thermals, but power I guess. But my first 14700k was able to though, and also died within a month, but it got 2070 and change in cb24.

My current chip with the new restrictions gets 1942. Exactly 5% less than with no limits. Well whatever. I'm not a benchmark chaser really. Just want my cpu to be fast and stable. And if these new settings make it less likely to become unstable, then thats a good thing. Even with some tweaking, I can only get my visible vcore peaks down to 1.41. Still better than then 1.488 they had before. Its nice to know vid transients aren't going higher than 1.5 behind my back. Just wish I could get it down to 1.4 like some peeps seem to be able to ( in that case I can only hit max boost at idle basically). Oh well. 1.5 is still lower than intel's 1.55, without me losing much perfomance (like 1%). Which tells me, there's even worse bins out there, lol!

Got the new board in. Set up. Water cooled with repaired waterblock.

14700K runs exactly the same as it did on Z690 MSI and Z790 MSI and B660-G boards at defaults. Also OC is exactly the same a the MSI Z690 and Z790 boards.

No degradation so far.

Static v-core and clocks, nothing hiding here. VID is over-ridden due to user set v-core.

It truly takes a bit of extra v-core to run the E-cores at 4.7ghz and higher. But the results are unchanged as far as I can tell.

This is the shipped bios for the board, I have not updated it and I see no need to do so at this time. My vote remains the same.


View attachment 359704

Didn't you used to have a 13700k? What happened to that? Or am I just misremembering? Either way you seem to have better luck than me getting well binned, stable CPUs thats for sure.

 

[ShrimpBrime]

Didn't you used to have a 13700k? What happened to that? Or am I just misremembering? Either way you seem to have better luck than me getting well binned, stable CPUs thats for sure.

I sent it to a friend in NY that needed something a little more powerful for 3D benchmarking. I also shipped him my old Z690 MSI D4 gaming and a set of memory as well. @Mr.Scott has had no issues so far that I know of. And he's got it benching around the exact same clocks and voltage I used it at. Has been DryIced several times, And I'm certain he's running a water chiller and benching between 5.9 and 6.0ghz P-Cores.

My B660-G board seems to really hammer the VID. I see between 1.5 and 1.55v in bios. But the board really wasn't or isn't equipped to handle an i7 in my opinion. Also running on the release Bios and firmware. So I expect it hot. Like we've talked, 253w is a lot of power.

I believe you just need a proper Z series motherboard and you'll have a better experience.

 

[chrcoluk]

I noticed on all of buildzoid's tests the hwinfo max reading isnt far off the oscillator, he even showed hwinfo for periods of time on his most recent videos, and the difference wasnt more than about 30mv or so.

 

[Bobaganoosh]

Wouldn't that be funny if these cpus in EVGA boards ended up out lasting all the others? Whens the last time you got an upgrade? Does it even offficially support 14th gen? I wonder if evga will ever rise from the grave. Being the only seemingly honest actor in diy market and dying, does not make a good case for caring for your customers to the other companies

Yeah, it's kind of interesting, because I don't think all the V/F points or adaptive voltage works the way intel really intended, but the way EVGA has it running under the hood seems like it may actually be safer than what Intel "intended" lol. I also have no access to the per-core TVB settings. The last BIOS update for my board was March-2023. It does not offer official 14th gen support. 14900k works...sort of, but the V/F tables aren't right and from what I've heard it's a bit wonky unless you manually set everything...even then...lol. They did release a beta BIOS for z790 boards that supports the 14th gen range, but I don't think they did anything other than add the skus to the BIOS's boot table so it recognizes them. Because it's basically just more 13th gen parts, the ones with the same P/E-core counts seem to boot on the z690 boards, but the 14700k for example does not. I've heard mixed results on stability and functionality from the few people who tried 14th gen on these boards.

As for failures, there are a couple threads on the EVGA forums where people have had this intel problem (no idea what their settings were though). There are also several, even OC'd, that are running just fine on early 13900k's and ks's.

 

[iameatingjam]

I sent it to a friend in NY that needed something a little more powerful for 3D benchmarking. I also shipped him my old Z690 MSI D4 gaming and a set of memory as well. @Mr.Scott has had no issues so far that I know of. And he's got it benching around the exact same clocks and voltage I used it at. Has been DryIced several times, And I'm certain he's running a water chiller and benching between 5.9 and 6.0ghz P-Cores.

My B660-G board seems to really hammer the VID. I see between 1.5 and 1.55v in bios. But the board really wasn't or isn't equipped to handle an i7 in my opinion. Also running on the release Bios and firmware. So I expect it hot. Like we've talked, 253w is a lot of power.

I believe you just need a proper Z series motherboard and you'll have a better experience.

Perhaps, but this is a higher end board, as far as b boards go. Its not like a budget b board is what I'm trying to say. I was looking at the cheapest z motherboards and they often had worse features than my board, like the same vrm or less m.2 slots. And my mobo officially supports DDR5-7600 memory. When I was looking for a cheap z board, most of them wouldn't even officially accept memory as fast as this board, sometimes not even going higher than 6800.

Which was kinda what ended up turning me off from the whole idea. ( this was back before I knew how bad running fast memory was for your cpu - and was planning to upgrade at some point). Thats no longer the case but still I feel like I'd be better off skating along with what I have now. Because if I want a better motherboard I really need like a good, decent non-budget z board and then we're talking over $350+(CAD) after taxes.

And what do I get from that money? Maybe slightly lower average voltages, which might bring my anxiety down a bit. But will it extend the life of the cpu? Who knows? Intel says the spikes are the problem, not the average. And my averages are fine. Intel's probably not above stretching the truth, but like we know whats really going on in these billion transitor sand boxes. I got my 4.5 years of warrranty left too.
Might it improve my cinebench score by 5%? Maybe? But hardly matters when I game at 4k60.

Seems like a big expense for not much payoff.

To me makes more sense to skate by with what I got until a real upgrade rears its head in a few years.

And yeah if I go intel ( which I probably wont honestly, but we'll see whats avaliable) I'll definitely get a z board. I had a b board on my 11600k, never stopped me from undervolting so it never occured to me intel would block undervolting. Nice of gigabyte to provide a workaround but 104 is getting pretty damn old now. And that thing is bugged to all hell. I'm convinced of it.

I'm workin with what I got.

 

[ShrimpBrime]

Perhaps, but this is higher end board, as far as b boards go. Its not like a budget b board is what I'm trying to say. I was looking at the cheapest z motherboards and they often had worse features than my board, like the same vrm or less m.2 slots. And my mobo officially supports DDR5-7600 memory. When I was looking for a cheap z board, most of them wouldn't even officially accept memory as fast as this board, sometimes not even going higher than 6800.

Which was kinda what ended up turning me off from the whole idea. ( this was back before I knew how bad running fast memory was for your cpu - and was planning to upgrade at some point). Thats no longer the case but still I feel like I'd be better off skating along with what I have now. Because if I want a better motherboard I really need like a good, decent non-budget z board and then we're talking over $350+(CAD) after taxes.

And what do I get from that money? Maybe slightly lower average voltages, which might bring my anxiety down a bit. But will it extend the life of the cpu? Who knows? Intel says the spikes are the problem, not the average. And my averages are fine. Intel's probably not above stretching the truth, but like we know whats really going on in these billion transitor sand boxes. I got my 4.5 years of warrranty left too.
Might it improve my cinebench score by 5%? Maybe? But hardly matters when I game at 4k60.

Seems like a big expense for not much payoff.

To me makes more sense to skate by with what I got until a real upgrade rears its head in a few years.

And yeah if I go intel ( which I probably wont honestly, but we'll see whats avaliable) I'll definitely get a z board. I had a b board on my 11600k, never stopped me from undervolting so it never occured to me intel would block undervolting. Nice of gigabyte to provide a workaround but 104 is getting pretty damn old now. And that thing is bugged to all hell. I'm convinced of it.

I'm workin with what I got.

The problem is worrying over the life of a cpu. You had 1 cpu die already.
Cause unknown.
Have a 14700KF, but think it's a bad bin, but only a B series to test with.
And with that board, no OC for an unlocked chip. Which means you're missing performance.

The 14700KF should be able to do 5.6ghz all P-cores without additional voltage if it's a normal bin chip. But you'll never be able to test that theory.

 

[iameatingjam]

The problem is worrying over the life of a cpu. You had 1 cpu die already.
Cause unknown.
Have a 14700KF, but think it's a bad bin, but only a B series to test with.
And with that board, no OC for an unlocked chip. Which means you're missing performance.

The 14700KF should be able to do 5.6ghz all P-cores without additional voltage if it's a normal bin chip. But you'll never be able to test that theory.

The reason I got the 14700kf was because there was no non-k at the time, and even if there was, I probably still would have bought the kf because its usually around the same price as the non-k, while having 200mhz higher all core clocks. And if my gpu fails, thats what my 12100 onboard is for.

And yeah no, overclocking is the last thing on mind right now, given the current situation. I turned by turbo 3 boost off, and my single thread score decreased by like 0.12% like within the margin of error. I don't need 5.6 all core. Honestly, anything past 5ghz is inefficient and past the point of diminishing returns anyway. Still, I wanted 5.5 and it was kinda cool to see at first. But higher than that? I just don't care. Maybe when the software I want to run, doesn't run at the level I want it to, then I'll think about upgrading.

 

[ShrimpBrime]

The reason I got the 14700kf was because there was no non-k at the time, and even if there was, I probably still would have bought the kf because its usually around the same price as the non-k, while having 200mhz higher all core clocks. And if my gpu fails, thats what my 12100 onboard is for.

And yeah no, overclocking is the last thing on mind right now, given the current situation. I turned by turbo 3 boost off, and my single thread score decreased by like 0.12% like within the margin of error. I don't need 5.6 all core. Honestly, anything past 5ghz is inefficient and past the point of diminishing returns anyway. Still, I wanted 5.5 and it was kinda cool to see at first. But higher than that? I just don't care. Maybe when the software I want to run, doesn't run at the level I want it to, then I'll think about upgrading.

What you want and need differs from facts unfortunately.

So the base clock on my 14700K is 3.4ghz P-cores. EVERYTHING ELSE above that frequency should be considered an overclock. Period. Boost is an OC feature.

 

[iameatingjam]

What you want and need differs from facts unfortunately.

So the base clock on my 14700K is 3.4ghz P-cores. EVERYTHING ELSE above that frequency should be considered an overclock. Period. Boost is an OC feature.

Yeah pretty much I mean I explained turbos to a newbie the other day as 'basically a cpu overclocking itself when it has the headroom to do so.' Though amongst people here when I say overclock I assume they are going to think I mean beyond the max turbo written on the box.

 

[ShrimpBrime]

Yeah pretty much I mean I explained turbos to a newbie the other day as 'basically a cpu overclocking itself when it has the headroom to do so.' Though amongst people here when I say overclock I assume they are going to think I mean beyond the max turbo written on the box.

That's the problem. Just like AMD, boosted to the max. There's nothing left to overclock.

So you want the performance of overclock, without overclocking. And that's what everyone gets these days.

So in that respect, buying a Z series board for processors you can't OC because of max VID and max thermals from the box makes most OC features useless.

So now we have everyone trying to de-tune or decrease v-core or VID which often times decreases performance. But we aren't looking at the problem straight forward.

The motherboards come with boost enabled. It might have been wiser of Intel to have the boards default with boost disabled or have a range of different boost clocks the user could set if in case of having not enough cooling for example.

 

[agent_x007]

You don't have to disable turbo, just drop less core multipliers to all core turbo value.
Few hundred MHz saved that way should allow decreasing max. VID of all CPUs.

Point is : Windows scheduler is too dumb to effective use under all core turbo feature under "auto" settings (example : AMD recent issues with "forcing" higher clocked CCX to be used during games, and my gut feeling is Intel's Arrow Lake CPUs will also face similar issues at the beginning due to E-cores being closer in performance and capabilities to full P-cores and lack off extra threads).
By default, Windows always tries to use all cores equally after all - to the point of user having to artificially "max out" single core OR using specific kinds of workloads, to get to those "max. turbo" states.
It's nice to play around in benchmarks (that don't push all cores at the time), but how often that extra performance makes a meaningful difference in usual workloads ?

 

[ShrimpBrime]

You don't have to disable turbo, just drop less core multipliers to all core turbo value.
Few hundred MHz saved that way should allow decreasing max. VID of all CPUs.

Point is : Windows scheduler is too dumb to effective use under all core turbo feature under "auto" settings (example : AMD recent issues with "forcing" higher clocked CCX to be used during games, and my gut feeling is Intel's Arrow Lake CPUs will also face similar issues at the beginning due to E-cores being closer in performance and capabilities to full P-cores and lack off extra threads).
By default, Windows always tries to use all cores equally after all - to the point of user having to artificially "max out" single core OR using specific kinds of workloads, to get to those "max. turbo" states.
It's nice to play around in benchmarks (that don't push all cores at the time), but how often that extra performance makes a meaningful difference in usual workloads ?

That extra performance meaningful work loads.

OK, so what are you comparing the boost too? Everyone else's similar boost clocks, or are we measuring base performance and efficiency and then compare to the boost. On person's desired workloads that matter to them. Or you. Or myself.

So I enjoy the hobby from the point of view, max performance period. Even when unstable. But I also know exactly how to make a cpu stable. And spend enough time on a platform, you learn all the little quirks.

Like my NV 4070 S. If I manually set a higher than 80% fan duty cycle, it will lower the clocks as soon as I hit apply. This is because it makes wattage space to run the fans that high, over 80%.

I'm not a fan of e-cores and scheduling. W11 is better than 10. My current version of W10 seems to think all the fkn cores are P cores. Which is fine, but makes the cpu just be at max all the time lol. And I don't mind it.

AMD, since like forever now. Set P95 small fft unticked. Load it up. Find FIT v-core by waiting for temps to level out. The v-core in HWinfo64 is what you use for overclocking or PBO blasting.

Intel you OC via positive offset. Not supposed to run static v-core like myself, but then again I want the processor to do what I want it to do and when.

I've found getting under 100w was fairly easy, simply set multi to like x36 all the P,E abcde fg cores and v-core auto, wattage fully unlocked. It's like 1.15v with a 5% swing.

You're correct this works well. Up to a certain multiplier. Then it bumps to 1.24v at like 4.3ghz all core. (I could be mistaken this actual multi could be 4ghz. I'd have to go do it all again, but you get the point.)

So, E-cores are expensive to run. They maxed these out as well. Most people can't squeeze 4.6ghz without realizing it'll take 1.40+ v-core to do it. I happily could set all 20 cores to 4.7ghz, but it won't be cool and it might sound effecient, it is totally the opposite.

And then realize, you could just shut off the e-cores and improve P-core performance with the wattage headroom. Easily 5.8ghz with adequate cooling and be no more than 250w any ways. Right around 1.395v I recon would get the job done. Give or take depending on llc. I run level 6 or 7 on Asus and level 2 on MSI.

Or let's under-volt on an air cooler.

No. The cpu wants 1.40 to 1.50v. I don't care who runs what VID. Just accept the failures will happen to me first, and I'll be the first to tell you.

My rough figures for DryIce 14700K.
As Follows.
1.35v to 1.45 5.6ghz to 5.8ghz Water
1.40 to 1.50 5.8ghz to 6.10ghz
1.55v to 1.60v 6.1ghz To 6.3ghz
1.60v to 1.65v 6.4ghz To 6.8ghz
6.9ghz to 7.0ghz, core reduction to 2 cores takes 1.70+v.
All core. At 6.4ghz e-cores off.

EDIT: it might be important to say I use this setup every day. I'm not gaming like everyone else all the time, but the above screen shots are what I leave bios set at (this week anyway). So it's not like I fire it up occasionally. Every day. Just like everyone else. I just take the waterblock off and put a dice pot on it about once a month or so.

I have seen all cores 6.6ghz dry ice come to positive temp at 26c. When I caught it, had to cut the benchmark off. Lol. Now THAT is scary.

And all the gamers worried about pathetic degradation. Just get bigger cooling. It's fine, 2x 360 rads and full copper waterblock. People just let the poor cpu bake to death, wonder why it gets unstable? Well you better cool them VRMs actively if I was to bring any advice to the table concerning Intel 13 and 14th gen processors.

 

[RootinTootinPootin]

You capped the vid of your 13900ks to 1.32v? You can reach maximum boost with that? Are you sure we're talking about the same thing? You set the ia voltage limit at 1.32? So its impossible for transients to pass that? Thats amazing if I'm understanding you right. Thats must be some incredible silicon.

Yes, I have an SP112 chip with P cores at SP117 rating, I only need around 1.348v according to BIOS prediction for my 6ghz. and thats my daily, I don't have issues of sort while gaming or doing anything, I also have my water temps around 23-25c, at the end of the day temps are still META for clocking higher without feeding too much voltage.

furthermore, I have never updated my BIOS since the 11F microcode, so all the nerfing and stuff isn't there.

My 5.7ghz all core just needs to maintain 1.25v vcore and around 1.157v on vid.

 

[iameatingjam]

Yes, I have an SP112 chip with P cores at SP117 rating, I only need around 1.348v according to BIOS prediction for my 6ghz. and thats my daily, I don't have issues of sort while gaming or doing anything, I also have my water temps around 23-25c, at the end of the day temps are still META for clocking higher without feeding too much voltage.

furthermore, I have never updated my BIOS since the 11F microcode, so all the nerfing and stuff isn't there.

My 5.7ghz all core just needs to maintain 1.25v vcore and around 1.157v on vid.

Those nerfs, are the vid limits! When my vcore was maxing out at 1.34 as seen in hwinfo, I still had a performance penalty for setting a vid limit at 1.5 presumably due to transients, not necessarily picked up by software, only about 1% though. I didn't see any difference at all with the intel default of 1.55. However, the way you have yours set up, with vcore being higher than vid (from llc I'm guessing? - idk know how asus boards work) I think it would matter even less

Anyway I did hear about there being something wrong with the ucode 125 update but only on asus boards. I thought all that was fixed with 129?.... cause any well behaving cpu shouldn't be affected by this more than a % unless they are way too overclocked, or already degraded to the point where its rma time, or voltages were higher than you thought they were (but again if your vcore is higher than vid I think it throws this whole idea out the window anyway) or asus is just being asus and did something else to f up their bios not to do with the ucode.

 

[RootinTootinPootin]

I think we are talking about different things. Those nerfs, are the vid limits! For example even when my vcore was maxing out at 1.34 as seen in hwinfo, I still had a performance penalty for setting a vid limit at 1.5 presumably due to transients, not necessarily picked up by software, only about 1% though. I didn't see any difference at all with the intel default of 1.55. However, the way you have yours set up, with vcore being higher than vid (from llc I'm guessing? - idk know how asus boards work) I think it would matter even less

I already tested countless times already, I don't see any difference on my end, does the score above I posted can't be replicated? I already mentioned my settings, 1.32 limit on my VID and 1.3v on vcore..

here's my other chip, only needs 1.135v on vid and 1.28v vcore to maintain 5.8ghz cheesecakes!! this chip walks in the park with 44k CBR23 easily..its on an Aorus Z790 Pro X with 2x24gb DD5 8000MT's CL36

 

[chrcoluk]

I think we are talking about different things. Those nerfs, are the vid limits! For example even when my vcore was maxing out at 1.34 as seen in hwinfo, I still had a performance penalty for setting a vid limit at 1.5 presumably due to transients, not necessarily picked up by software, only about 1% though. I didn't see any difference at all with the intel default of 1.55. However, the way you have yours set up, with vcore being higher than vid (from llc I'm guessing? - idk know how asus boards work) I think it would matter even less

Anyway I did hear about there being something wrong with the ucode 125 update but only on asus boards. I thought all that was fixed with 129?.... cause any well behaving cpu shouldn't be affected by this more than a % unless they are way too overclocked, or already degraded to the point where its rma time, or voltages were higher than you thought they were (but again if your vcore is higher than vid I think it throws this whole idea out the window anyway) or asus is just being asus and did something else to f up their bios not to do with the ucode.

My opinion that 1% (within margin of error) isnt transients, BZ showed the oscillator, only peaked about 20-30mv above the hwinfo readings. Plus transients are not circa 200mv above prior loads. Which would be the case if you was hitting 1.55 from a voltage load of 1.34. Jay2cents data showed the same thing regarding the spikes, nowhere near 200mv.

The older bios had more things different to the latest versions, load line, wattage limits, tjmax (in the case of asrock, wtf was you doing asrock), eTVB, CEP. Plus 1% is pretty much an irrelevant difference.

 

[iameatingjam]

isnt transients,

You could be right. I don't know what would explain the performance difference in that case though. And you are comparing a vcore and a vid there ( which I did state....) my vid is generally higher than vcore. Though I have seen other data that suggests inconsistency (like when my voltage limit was set at 1.4 and yet highest reported vid was 1.38/137 even though it was clearly not getting enough power to do what it was trying to do, and would normally request more).

But I didn't have a super high polling rate or anything though ( I feel like it would eat into processor power and skew results- wouldn't it?). Plus isn't there some add on to hwinfo that makes it more sensitive to transients? I could be mistaken but I just thought I remembered hearing about something like that.... Anyway I'm just saying I'm only on default settings so I might not be catching everything other people are catching.

(within margin of error)

That was my first thought too... which is why I ran the test many times, both ways, so its repeatable. If its not transients, its affecting performance somehow, at least in that test (passmark). That much I am pretty sure of. Can't be positive of anything in this world. But I was pretty thorough on that one. Made sure every unnecessary process was shut down, including AV each and every time. Monitoring tools would have been open but that happened every time so it should be consistent. Plus it was a little more than 1%, I was just rounding down a fair bit, probably to make myself feel better

I just ran cinebench and my highest recorded vid is 1.47 - which really makes me think it would have gone higher if allowed and my readings are just off).

The older bios had more things different to the latest versions, load line, wattage limits, tjmax (in the case of asrock, wtf was you doing asrock), eTVB, CEP. Plus 1% is pretty much an irrelevant difference.

It is indeed, which is why I decided to keep the setting on 1.5. Not sure what you're getting at in the part before that though? That there could be valid reasons to not upgrade? Well, in that case, fair enough, but CEP can be turned back on if you want it, most of those things are configurable except eTBV I don't know much about that. Still, I'd rather have the peace of mind but you're right its each person's prerogative. I'm not trying to force anybody. Sorry if it came off that way.

I already tested countless times already, I don't see any difference on my end, does the score above I posted can't be replicated? I already mentioned my settings, 1.32 limit on my VID and 1.3v on vcore..

View attachment 359766

here's my other chip, only needs 1.135v on vid and 1.28v vcore to maintain 5.8ghz cheesecakes!! this chip walks in the park with 44k CBR23 easily..its on an Aorus Z790 Pro X with 2x24gb DD5 8000MT's CL36

I'm sorry I didn't mean to call you liar or anything its just there was already some miscommunication/confusion earlier in the thread and I was thinking I wasn't clear enough in the original post. But anyway yeah I don't think any of this should really affect you if vcore is above vid. Though you might want to upgrade for other reasons. You never know what intel's fixing behind the scenes but is too embarrassed to admit.

 

[chrcoluk]

Well you capped to 1.55 now anyway, and still within 1%, seems a win to me regardless. I guess you are still undecided on longer term day to day config?

 

[iameatingjam]

Well you capped to 1.55 now anyway, and still within 1%, seems a win to me regardless. I guess you are still undecided on longer term day to day config?

No sorry, 1.55 has 0% effect as far as I can tell. it was 1.5 that was the 1-2% change. And idk I think I'm wasting brain capacity worrying about it.

However, I was thinking... maybe I should go back to 104 and undervolt, I mean even though its the old ucode its still the new bios ( I was thinking) so I could still access things like the intel profile and the voltage caps. So I tried it, and it doesn't work anymore. 104 is installed and undervolt protection is off but it acts like CEP is on, so more undervolt = less performance. Even though ucode 104 was before CEP even existed (I think.... since normally I can turn it off, but when I go to ucode 104 the option vanishes).

Anyway I can confirm that it worked just a couple months ago. So intel must have done something to patch that hack out in the last couple ucodes or something. I wonder if going back bioses would fix it? I mean I could still manually apply most of the fixes.... but then... would... it even be worth it anymore? IDK. Then there's intel ME. Can't that mess with bios downgrading or something?

Anyway I think I did my part, I did what intel asked and then a little more. So my vcore goes up to 1.4 sometimes. Whatever.

 

[ShrimpBrime]

I already tested countless times already, I don't see any difference on my end, does the score above I posted can't be replicated? I already mentioned my settings, 1.32 limit on my VID and 1.3v on vcore..

View attachment 359766

here's my other chip, only needs 1.135v on vid and 1.28v vcore to maintain 5.8ghz cheesecakes!! this chip walks in the park with 44k CBR23 easily..its on an Aorus Z790 Pro X with 2x24gb DD5 8000MT's CL36

44k from this? It's a stock score with overclock??

Am I missing something?

Can you post the score with F6 benchmate at the end of the screen shot please?

 

[chrcoluk]

No sorry, 1.55 has 0% effect as far as I can tell. it was 1.5 that was the 1-2% change. And idk I think I'm wasting brain capacity worrying about it.

However, I was thinking... maybe I should go back to 104 and undervolt, I mean even though its the old ucode its still the new bios ( I was thinking) so I could still access things like the intel profile and the voltage caps. So I tried it, and it doesn't work anymore. 104 is installed and undervolt protection is off but it acts like CEP is on, so more undervolt = less performance. Even though ucode 104 was before CEP even existed (I think.... since normally I can turn it off, but when I go to ucode 104 the option vanishes).

Anyway I can confirm that it worked just a couple months ago. So intel must have done something to patch that hack out in the last couple ucodes or something. I wonder if going back bioses would fix it? I mean I could still manually apply most of the fixes.... but then... would... it even be worth it anymore? IDK. Then there's intel ME. Can't that mess with bios downgrading or something?

Anyway I think I did my part, I did what intel asked and then a little more. So my vcore goes up to 1.4 sometimes. Whatever.

BZ explained with CEP on you need to undervolt directly on the VID level, on ASRock e.g. thats in the FIVR section, not on the main CPU voltage section (the main CPU voltage section is a VRM adjustment). So try to see if you can find another voltage setting, although I wonder if B chipsets are capable of going this deep on the voltage configuration. At least you have a gigabyte board though so if needed you could disable CEP (if they let you do that on a B chipset).

 

[RootinTootinPootin]

I don't think any of this should really affect you if vcore is above vid

as I have said, I already tested the other way around, it just makes things worse for me, even auto sends it to 1.4v even my VID is still low. There's also no mention of such rule that the VID should be equal to vcore. never heard of that mention.

44k from this? It's a stock score with overclock??

wdym?

 

[iameatingjam]

BZ explained with CEP on you need to undervolt directly on the VID level, on ASRock e.g. thats in the FIVR section, not on the main CPU voltage section (the main CPU voltage section is a VRM adjustment). So try to see if you can find another voltage setting, although I wonder if B chipsets are capable of going this deep on the voltage configuration. At least you have a gigabyte board though so if needed you could disable CEP (if they let you do that on a B chipset).

Yeah, I can turn CEP off, but I still have undervolt protection on the new ucode or I could select the old ucode and it would be the exact opposite, undervolt protection gets turned off and CEP is on ( Well I assume - I don't know if it was cep I just know it was acting like CEP). Besides thats what my 90 ac loadline setting basically is, a small undervolt, which would be considered on the VID level wouldn't it? However like I said before I can't go much further without losing performance ( undervolt protection?).

I think to get it so they are both off I would have to at least go back a couple bioses. And I feel like its just not worth my time. The last time I had it working was the bios right before they first added the intel profiles, that I remember. But my undervolt got unstable and my machine was acting REALLY weird. Lots of things were happening, for example, at one point it looked like somebody was doodling all over my windows log in screen with paint. Thats when I said you know what maybe I should update my bios and ucode LOL.

as I have said, I already tested the other way around, it just makes things worse for me, even auto sends it to 1.4v even my VID is still low. There's also no mention of such rule that the VID should be equal to vcore. never heard of that mention.

I wasn't trying to say that what you were doing it wrong. I just meant that specific feature of the update wouldn't have much effect on you, thats all. I'm not telling you how to live your life. You do you.