[SOLVED] Did I get unlucky with my i9 9900k?

[EatMyPizza]

I have an i9 9900k on a Gigabyte z390 Aurus Pro wifi board. Just to hit 4.8 on all 8 cores I need V core set to 1.315. This seems insanely high to me since people are reporting 5hz with that vcore. At stock clocks, i can't pass stress testing with anything under 1.3.

Do I just have a "slow" chip?

 

[EatMyPizza]

So I've been playing around and noticed vdroop seems to be the problem. I have the vcore set to 1.315 in BIOS and under full load in prime 95, HW monitor reports only 1.17 vcore. This is with LLC set to "auto".

Setting LLC to "low" (which is a step up from auto), helps a bit, and brings the vcore under load to 1.22 and temps go up almost TEN degrees.

What the hell is up with this vdroop?

 

[EatMyPizza]

update: I changed LLC to "medium". Changed Vcore in bios to 1.27.

Now, under full load HW monitor reports 1.2 vcore, it seems stable and temps are a bit better. This is at 4.8 ghz on all cores.

 

[The Original Ralph]

What BIOS are you running? I'm on an AORUS Master, and haven't spent that much tine but have made it to
4.9 - running Prime95 vs26.6 i see temps mainly in high 80C with 2 or 3 of the cores bouncing to low 90Cs

rendering a video file with CPU showing 100% load, i'm seeing temps in low 70Cs with a few of the cores spiking to 75 or 76C

 

[EatMyPizza]

What BIOS are you running? I'm on an AORUS Master, and haven't spent that much tine but have made it to
4.9 - running Prime95 vs26.6 i see temps mainly in high 80C with 2 or 3 of the cores bouncing to low 90Cs

rendering a video file with CPU showing 100% load, i'm seeing temps in low 70Cs with a few of the cores spiking to 75 or 76C

The newest bios. f6 I believe. at 4.8 ghz my temps are now in the upper 60s, low 70s under full load prime 95 26.6 small fft. I'm trying now at 1.26 vcore now that I have LLC on medium.

The "auto" LLC setting seems to do absolutely nothing.

 

[The Original Ralph]

You're running a helluva cooler than i am

 

[EatMyPizza]

You're running a helluva cooler than i am

I'm using a h115i pro front mounted rad. I'm also 100mhz lower than you.

edit: Seems stable at 1.26 vcore with LLC on medium. Vcore drops to 1.188 under full load for a second at a time but mostly stays at 1.2 full load.

 

[The Original Ralph]

You're running a helluva cooler than i am

I'm using a h115i pro front mounted rad. I'm also 100mhz lower than you.

edit: Seems stable at 1.26 vcore with LLC on medium. Vcore drops to 1.188 under full load for a second at a time but mostly stays at 1.2 full load.

in reviews, i don't think there was much of a difference between your AIO and my noctua D15S, and you're 15-18 degrees cooler. One thing i didn't touch was LLC - don't recall even seeing it - the GUI on that BIOS ain't exactly intuitive

have you tried seeing how low you can take your Vcore?

 

[EatMyPizza]

You're running a helluva cooler than i am

I'm using a h115i pro front mounted rad. I'm also 100mhz lower than you.

edit: Seems stable at 1.26 vcore with LLC on medium. Vcore drops to 1.188 under full load for a second at a time but mostly stays at 1.2 full load.

in reviews, i don't think there was much of a difference between your AIO and my noctua D15S, and you're 15-18 degrees cooler. One thing i didn't touch was LLC - don't recall even seeing it - the GUI on that BIOS ain't exactly intuitive

have you tried seeing how low you can take your Vcore?

It's in the advanced power settings. There is a z390 gigabyte OC guide on youtube that shows you. The guy used the same board as you too.

As far as my vcore, before I started playing around with LLC, 1.315 was the lowest I could go (or so I thought) as it would drop to 1.17 under load, but after running prime 95 for longer I found it was unstable. 1.18 (reported by hw monitor) under load seems to be the lowest I can keep it. In the bios it's set to 1.26 with LLC medium to achieve this. It stays at 1.2 under load for the most part, with 1 second dips to 1.18.

 

[The Original Ralph]

thanks for that, i'll look for that youtube vid

i think part of the difference in temps though is you're using HW Monitor. Most of the RKI (reasonably knowledge individual) here on toms seem to feel that HWiNFO does a better job on temps

you might give it a try next time - if you do download it, be sure to set it to run "sensors ony"

 

[EatMyPizza]

thanks for that, i'll look for that youtube vid

i think part of the difference in temps though is you're using HW Monitor. Most of the RKI (reasonably knowledge individual) here on toms seem to feel that HWiNFO does a better job on temps

you might give it a try next time - if you do download it, be sure to set it to run "sensors ony"

I also used core temp, with very similar results, maybe like 2 degrees different here and there, but i can try hw info

 

[EatMyPizza]

thanks for that, i'll look for that youtube vid

i think part of the difference in temps though is you're using HW Monitor. Most of the RKI (reasonably knowledge individual) here on toms seem to feel that HWiNFO does a better job on temps

you might give it a try next time - if you do download it, be sure to set it to run "sensors ony"

Tried HWinfo. Showed pretty much same temps as HW monitor. low 60s for the first few min of prime 95 small fft, and after a while stays at upper 60s, low 70s. That's after a quick approx 10 min run though.

Corsair link also shows same temps.

 

[The Original Ralph]

did a quick "follow" of that video - had a few issues, but the big item, i had MCE on Auto - disabled it and damn if temps didn't drop almost 5 degrees

i need to study that and other "how to" guides more

thanks for that tip -

 

[The Original Ralph]

haven't gotten it 5.0 yet, but did drop my temps down to mid 60s with an occasional spike to 70 or 71 on one or two cores, Main drop came from pulling my GPU out and only pulled it because my cpu wouldn't run Prime's "blend" sample or HandBrake (which runs AVX instructions) unless i set the offset to 2, which would drop OC to 4.7 from 4.9.

If i left AVX at auto, it would BSOD showing conflicts with Nvidia drivers - but how the GPU being removed could drop temps over 10C ??? and it's not negative air pressure as the case cover was off since last night.

for anyone researching 9900k OCing, here's the thread on youtube i followed
https://www.youtube.com/watch?v=d8rY4TrcDXg

 

[EatMyPizza]

haven't gotten it 5.0 yet, but did drop my temps down to mid 60s with an occasional spike to 70 or 71 on one or two cores, Main drop came from pulling my GPU out and only pulled it because my cpu wouldn't run Prime's "blend" sample or HandBrake (which runs AVX instructions) unless i set the offset to 2, which would drop OC to 4.7 from 4.9.

If i left AVX at auto, it would BSOD showing conflicts with Nvidia drivers - but how the GPU being removed could drop temps over 10C ??? and it's not negative air pressure as the case cover was off since last night.

for anyone researching 9900k OCing, here's the thread on youtube i followed
https://www.youtube.com/watch?v=d8rY4TrcDXg

I don't understand how removing your gpu could effect so much in prime 95. Weird.

Also, what is your vcore at while running prime 95. For example, right now with LLC on medium, I have mine set to 1.26 in bios and it stays at 1.2 under load in prime 95 with dips to 1.188. How much Vdroop are you getting and what are your LLC settings? Im wondering if something is wrong with my mobo.

 

[jankerson]

haven't gotten it 5.0 yet, but did drop my temps down to mid 60s with an occasional spike to 70 or 71 on one or two cores, Main drop came from pulling my GPU out and only pulled it because my cpu wouldn't run Prime's "blend" sample or HandBrake (which runs AVX instructions) unless i set the offset to 2, which would drop OC to 4.7 from 4.9.

If i left AVX at auto, it would BSOD showing conflicts with Nvidia drivers - but how the GPU being removed could drop temps over 10C ??? and it's not negative air pressure as the case cover was off since last night.

for anyone researching 9900k OCing, here's the thread on youtube i followed
https://www.youtube.com/watch?v=d8rY4TrcDXg

All you should have to is this.

Reset the BIOS to Optimized defaults.

Save and Reboot, go back into the BIOS.

Enable MCE (Multicore enhancement)
Enable XMP

Save and reboot.

It will be running at 5 GHz on all cores.

 

[The Original Ralph]

I don't understand how removing your gpu could effect so much in prime 95. Weird.

Also, what is your vcore at while running prime 95. For example, right now with LLC on medium, I have mine set to 1.26 in bios and it stays at 1.2 under load in prime 95 with dips to 1.188. How much Vdroop are you getting and what are your LLC settings? Im wondering if something is wrong with my mobo.

My LLC is at Medium as well, when i dropped to Medium, temps came down some more (exact opposite of what i expected) - my Vcore is set to 1.275V in Bios, I'm still at 4.9 (all cores) and now that you ask, last night while running Prime95 26.6, i saw voltage on cores ranging from 1.264 to highs of 1.319 - in HWiNFO. Just now i ran Prime and HWiNFO is showing voltages of 1.286 to highs of 1.305, so something changed in BIOS

i tried using that -0.045V offset that video showed, but when i did, Vcore voltage would change to "Normal" and it'd get temps that were in the high 90c

I'm going to reflash F6 and see what it shows

I didn't get a chance to play with it at 5.0 - grandkids came over so i had to "tweak" their voltages

 

[The Original Ralph]

haven't gotten it 5.0 yet, but did drop my temps down to mid 60s with an occasional spike to 70 or 71 on one or two cores, Main drop came from pulling my GPU out and only pulled it because my cpu wouldn't run Prime's "blend" sample or HandBrake (which runs AVX instructions) unless i set the offset to 2, which would drop OC to 4.7 from 4.9.

If i left AVX at auto, it would BSOD showing conflicts with Nvidia drivers - but how the GPU being removed could drop temps over 10C ??? and it's not negative air pressure as the case cover was off since last night.

for anyone researching 9900k OCing, here's the thread on youtube i followed
https://www.youtube.com/watch?v=d8rY4TrcDXg

All you should have to is this.

Reset the BIOS to Optimized defaults.

Save and Reboot, go back into the BIOS.

Enable MCE (Multicore enhancement)
Enable XMP

Save and reboot.

It will be running at 5 GHz on all cores.

MCE has been at all cores since the beginning - the AVX offset issue doesn't respond to MCE being changed (i tried it)

 

[EatMyPizza]

I don't understand how removing your gpu could effect so much in prime 95. Weird.

Also, what is your vcore at while running prime 95. For example, right now with LLC on medium, I have mine set to 1.26 in bios and it stays at 1.2 under load in prime 95 with dips to 1.188. How much Vdroop are you getting and what are your LLC settings? Im wondering if something is wrong with my mobo.

My LLC is at Medium as well, when i dropped to Medium, temps came down some more (exact opposite of what i expected) - my Vcore is set to 1.275V in Bios, I'm still at 4.9 (all cores) and now that you ask, last night while running Prime95 26.6, i saw voltage on cores ranging from 1.264 to highs of 1.319 - in HWiNFO. Just now i ran Prime and HWiNFO is showing voltages of 1.286 to highs of 1.305, so something changed in BIOS

i tried using that -0.045V offset that video showed, but when i did, Vcore voltage would change to "Normal" and it'd get temps that were in the high 90c

I'm going to reflash F6 and see what it shows

I didn't get a chance to play with it at 5.0 - grandkids came over so i had to "tweak" their voltages

Wow, so you're getting almost no Vdroop with LLC on medium, (actually you're getting spikes) and I'm still getting plenty. Maybe it's because of the "master" vs the "pro" versions of the board. I would expect LLC to be on Turbo or extreme or whatever it is, for voltages to act like that.

 

[The Original Ralph]

something's strange with this bios - i'll make some changes and when i hit save & exit, it will not list the changes i made so i'll reboot back into bios and sure enough changes didn't take

tried to change voltage down some (1.265) and it would not take - it would display it but when i hit save & exit, and rebooted, it was back to 1.275 - something in that bios is bad or corrupted

going to try reflashing or might even go back to F5

just occurred to me, last night a second power cable i'd ordered came in (to run up to the 4 pin socet next the 4+4 cpu power socket up at the top of the board - and i installed it

I wonder if that is causing the increase in voltage range? Before i reflash, i'm going to pull that out but it'll be later after the family has left

 

[jankerson]

haven't gotten it 5.0 yet, but did drop my temps down to mid 60s with an occasional spike to 70 or 71 on one or two cores, Main drop came from pulling my GPU out and only pulled it because my cpu wouldn't run Prime's "blend" sample or HandBrake (which runs AVX instructions) unless i set the offset to 2, which would drop OC to 4.7 from 4.9.

If i left AVX at auto, it would BSOD showing conflicts with Nvidia drivers - but how the GPU being removed could drop temps over 10C ??? and it's not negative air pressure as the case cover was off since last night.

for anyone researching 9900k OCing, here's the thread on youtube i followed
https://www.youtube.com/watch?v=d8rY4TrcDXg

All you should have to is this.

Reset the BIOS to Optimized defaults.

Save and Reboot, go back into the BIOS.

Enable MCE (Multicore enhancement)
Enable XMP

Save and reboot.

It will be running at 5 GHz on all cores.

MCE has been at all cores since the beginning - the AVX offset issue doesn't respond to MCE being changed (i tried it)

If it is then it's running at 5 GHz on all cores.

 

[The Original Ralph]

nope, when i put it on "enabled" it does go straight to 5.0 but temps go to 93 to 99C right off the bat

and i'd rather figure out why this BIOS isn't stable first

reflashed F4 first, and not much better, and found it would not accept changes unless i first loaded defaults, reboot and then on the 2nd entry into BIOS it would let me make some changes. Vcore i could not change once i set it at 1.275V. One time, even though the clock ratio had been set at 49, and had booted up and ran at 4.9, the next time i booted it ran at 1.3 and sure enough, found 13 set for clock ratio in BIOS.

up to last night, i had gotten a stable 4.9 showing the temps i describe earlier mid/high 60s to low 70s, with one or two cores spiking to 73/74C with Vcore showing in HWiNFO ranging 1.268 to 1.277/1.278). Then last night, when clock ratio was set at 4.9 (i simply loaded a profile i had been using for a few days) temps went crazy up to low to high 80s, and Vcore was showing 1.288 ranging to 1.305.

I flashed F5, booted into windows a couple of times, then I shut down, turned PSU off, set BIOS switches to Dual BIOS (per Gig's manual and some of their online support, the backup BIOS will verify and repair the main BIOS supposedly). Ran it bit, rebooted a few times to let the backup "train" the main BIOS, and it seemed to return to normal, ie with Vcore set at 1.275V it would show in HWiNFO at 1.288 generally and temps were in the 72 to 80C range.

Woke up this morning at 5:00 AM, turned the computer on, checked HWiNFO and it was showing 4.9, Vcore was ranging 1.288 to 1.305V, and temps were crazy high, 88-95C.

Shut down, went into BIOS to review settings, all were as they should have been, made no changes, rebooted into windows, and this time HWiNFO is showing Vcore ranging 1.278 - 1.287 and temps were back down at 72 - 78C with a couple of cores occasionally spiking to 80C. That's not as low as it was running before last night, but for now i'll accept that

later tofay i'll try flashing F6, and seeing what that does, but i am definitely going to have a conversation with Gigabyte tech svc and see what they tell me.

 

[The Original Ralph]

well looks my BIOS chip has gone south - it got progressively worse today till it hit the point of not accepting any settings changes no matter what procedure or rain dance i did

then noticed DeBug LED showing status code of "66" which is eerily close to the code the first CPU that went south showed ("65")

both mobo and cpu are being RMA'd and going back to mothership

think i'm going to change boards to the MSI MEG Z390 Ace - same feature set as the Gig Aorus Master, hopefully better BIOS GUI

 

[EatMyPizza]

well looks my BIOS chip has gone south - it got progressively worse today till it hit the point of not accepting any settings changes no matter what procedure or rain dance i did

then noticed DeBug LED showing status code of "66" which is eerily close to the code the first CPU that went south showed ("65")

both mobo and cpu are being RMA'd and going back to mothership

think i'm going to change boards to the MSI MEG Z390 Ace - same feature set as the Gig Aorus Master, hopefully better BIOS GUI

The bios on my Aurus pro is a little weird as well. If I get a BSOD while searching for a good cvore, it will change some of the settings, and make it so you can't change them back. (and not to bios defaults) After a boot or two I sometimes get access back so I can set everything custom again.

 

[The Original Ralph]

i was going to ask you if your bios was hesitant to accept settings changes - from the very beginning selecting a value to change took anywhere from 3 or 4 cliks to 10+ and i thought it was just a sluggish chip.
Then last night i, for the hey of it, I tried booting up only on the backup BIOS, switching the main BIOS off.
initially it was fine, actually much more responsive, one clik opened whatever setting i wanted - i tried the defaults, and got crazy temps with Vcore values showing 1,205ish, it didn't make sense. Went thru a few different BIOS setting sets, and noticed that each time i went into that backup BIOS to change settings, it was becoming more and more sluggish, like i was filling up some kind of internal ram - i wish i knew more about the architecture of BIOS chips,

but mine has just gone completely south, when for whatever reason it booted up with a clock ratio setting of 13, Vcores were showing 1.19X and or but temps were high 80s to mid 90s - don't know if the board's reporting sensors were just being mis-interpreted by the CPU or something bad on the board.

but both are going back home tomorrow. Newegg has already shipped the replacements.

back to your original question, ie unlucky - i keep thinking about that, your 1.315V to get 5.0, there was another poster that posted screen captures showing, iirc, 1.272V at 5.0, that MrFinite indicating 1.250V being the expected norm -


anyways on the advice of that guy over on the geforce forum, i'm going with a MSI MEG Z390 Ace