Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us $3,400 per month for bandwidth bills alone, and since we don't believe in shoving popup ads to our registered users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
real_scud
Sep 5, 2002

One of these days these elbows are gonna walk all over you


I feel like very, very soon I'm going to be joining this group searching slightly higher numbers.

Finally got a new system and now that everything is relatively stable and my idle temps are around 30-31c I feel comfortable pushing it to try to get some decent OC's.

Current specs:
10900K @ Stock Speeds
MSI Z490 Godlike
64gb G.Skill 3600 / CAS 17/19/19
EK 360 AiO in full push/pull
Lian-Li OD11XL w/ a shitload of fans

Took a lot of time, and going through a bunch of lovely AiO's and a case before I settled on the EK that performs really damned well and is fairly quiet.

To make this a bit more OC related before I had to flatten my old Windows install I was able to get it up to 5.3 allcore with not much of a rise, maybe 1 or 2c in idle temps.

real_scud fucked around with this message at 08:48 on Aug 7, 2020

Adbot
ADBOT LOVES YOU

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

So I slapped an NH-U14S TR4 into my system. It has a 140mm Noctua fan (obviously). Dear God in heaven is that an annoying sounding fan when at speed. Sucks that the 140mm variant of that A12x25 is still scheduled for 2021. While those A12x25 aren't exactly dead silent when pushing air, they only sound like white noise and don't hum like that NF-A15 that's the cooler shipping with.

tehinternet
Feb 14, 2005



Combat Pretzel posted:

So I slapped an NH-U14S TR4 into my system. It has a 140mm Noctua fan (obviously). Dear God in heaven is that an annoying sounding fan when at speed. Sucks that the 140mm variant of that A12x25 is still scheduled for 2021. While those A12x25 aren't exactly dead silent when pushing air, they only sound like white noise and don't hum like that NF-A15 that's the cooler shipping with.

I went through my formative years with a computer that had fans that sounded like low powered buzzsaws— the fan sounds have always been calming white noise for me.

Kinda like a happy ASMR thing almost.

Dogen
May 5, 2002

Bury my body down by the highwayside, so that my old evil spirit can get a Greyhound bus and ride


My case is full of spectre pro 230s and a couple thermalright TY 150s push/pull on my venerable Archon heatsink. System makes a nice “whoosh” when it kicks into gear, I feel like it’s the computer equivalent of a Rolls Royce accelerating.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Kind of annoys me, too, that my mainboard's UEFI doesn't do time constants on fan control, to smooth out demands. Just because Ryzen's Precision Boost causes short intermittent temperature upswings, doesn't mean the heatpipes in the air cooler will be saturated with heat immediately. Next mainboard and CPU upgrade is way off to earliest Spring next year, so I guess I have to put up with the fans spinning up intermittently.

Vir
Dec 14, 2007

Does it tickle when your Body Thetans flap their wings, eh Beatrice?


You can run a fan control application in the operating system if the UEFI doesn't give you fine enough control over the fan curve.

spunkshui
Oct 5, 2011





Ram overclocking adventures!

What's the difference between:

https://www.corsair.com/us/en/Categ...#tab-tech-specs
$199.99
Memory Size 16GB Kit (2 x 8GB)
Tested Latency 16-18-18-36
Tested Voltage 1.35V
Tested Speed 3600MHz

and this:

https://www.corsair.com/us/en/Categ...#tab-tech-specs
$94.99
Memory Size 16GB Kit (2 x 8GB)
Tested Latency 18-22-22-42
Tested Voltage 1.35V
Tested Speed 3600MHz

After overclocking the only difference is about 0.00625V which is apparently the amount of the lowest step up.

In other news, don't open chrome when you have 32 gigs of ram humming along with a test.

spunkshui fucked around with this message at 21:15 on Sep 21, 2020

Warmachine
Jan 30, 2012

I came here to laugh at you





spunkshui posted:

Ram overclocking adventures!

What's the difference between:

https://www.corsair.com/us/en/Categ...#tab-tech-specs
$199.99
Memory Size 16GB Kit (2 x 8GB)
Tested Latency 16-18-18-36
Tested Voltage 1.35V
Tested Speed 3600MHz

and this:

https://www.corsair.com/us/en/Categ...#tab-tech-specs
$94.99
Memory Size 16GB Kit (2 x 8GB)
Tested Latency 18-22-22-42
Tested Voltage 1.35V
Tested Speed 3600MHz

Looks like this difference is about 0.00625V which is apparently the amount of the lowest step up.

In other news, don't open chrome when you have 32 gigs of ram humming along with a test.

If I want to crack-ping my computer, I try and do anything other than read sensors while running Prime 95 Small AVX.

The Electronaut
May 10, 2009


spunkshui posted:

Ram overclocking adventures!

What's the difference between:

https://www.corsair.com/us/en/Categ...#tab-tech-specs
$199.99
Memory Size 16GB Kit (2 x 8GB)
Tested Latency 16-18-18-36
Tested Voltage 1.35V
Tested Speed 3600MHz

and this:

https://www.corsair.com/us/en/Categ...#tab-tech-specs
$94.99
Memory Size 16GB Kit (2 x 8GB)
Tested Latency 18-22-22-42
Tested Voltage 1.35V
Tested Speed 3600MHz

Looks like this difference is about 0.00625V which is apparently the amount of the lowest step up.

In other news, don't open chrome when you have 32 gigs of ram humming along with a test.

Timings are better on the first kit.

spunkshui
Oct 5, 2011





The Electronaut posted:

Timings are better on the first kit.

Oh I know.

I paid for the second kit and I'm running at the speeds of the first kit with 1.35625 volts to make the error go away.

No errors at all but also no noticeable changes in performance which is why I never really played with ram before but ASUS is taking their sweet loving time with the ROG 3080 and I want to make lines go up.

hobbesmaster
Jan 28, 2008



spunkshui posted:

Oh I know.

I paid for the second kit and I'm running at the speeds of the first kit with 1.35625 volts to make the error go away.

No errors at all but also no noticeable changes in performance which is why I never really played with ram before but ASUS is taking their sweet loving time with the ROG 3080 and I want to make lines go up.

If it isn't a different chip (ie Samsung-B vs Micron-E) then you're paying for the time they spent sorting through the chips to find the best ones that'll pass whatever standards there are for the listed QVLs. You just got lucky buying the second one and just happened to get chips that might've qualified for the first SKU.

spunkshui
Oct 5, 2011





hobbesmaster posted:

If it isn't a different chip (ie Samsung-B vs Micron-E) then you're paying for the time they spent sorting through the chips to find the best ones that'll pass whatever standards there are for the listed QVLs. You just got lucky buying the second one and just happened to get chips that might've qualified for the first SKU.

Yeah, and the better ones can probably be pushed even more.

Its not fully luck though because it crashed the test after 2 minutes. It cant handle C16 at 3600 with 1.35V like the more expensive ram does.

But all I had to do was click up once on the Dram voltage and it passed a 2 hour test.

Badger of Basra
Jul 25, 2007



Badger of Basra posted:

Hopefully the last new system question: everything gets very unstable (like, crashing 1-2 minutes after Windows loads) when I turn on my memory's XMP profile (or DOCP in this case).

CPU: AMD Ryzen 7 3700X 3.6 GHz 8-Core Processor ($294.99 @ Amazon)
Motherboard: Asus TUF GAMING X570-PLUS ATX AM4 Motherboard ($164.99 @ Amazon)
Memory: G.Skill Trident Z Neo 16 GB (2 x 8 GB) DDR4-3600 CL16 Memory ($114.99 @ Amazon)

This memory is on the QVL for my motherboard. From googling around I can't tell if the best step would be to raise the voltage from 1.35 to 1.36, enter all the settings manually, or something else. I ran the windows memory test and no errors there, and everything seems to run fine if I leave it at the default speed. Any suggestions?

Putting this question here so I stop making GBS threads up the pc building thread. Also I did try raising the DRAM voltage from 1.35 to 1.36 and that seemed to be stable, but then 1) I was getting popping/crackling sounds in audio 2) the BIOS froze when I went back to turn off the XMP. Anyone else have ideas? I am hesitant to use the DRAM Calculator and enter everything myself since I'm afraid I'll gently caress something up, not having done this before. Could I turn on the XMP but then just lower the speed to 3400 instead?

hobbesmaster
Jan 28, 2008



The ryzen DRAM calculator is only going to be more aggressive with timings than XMP. The worst that can happen when messing with timing and speed is that your computer won’t post and you’ll need to reset your bios’s config to get back to stock timings. Don’t worry about miskeying anything there. All you have to lose is time.

Voltage can actually kill components but it is very unlikely until you’re doing truly ridiculous things like 1.6V+. General recommendation for DDR4 is up to 1.45V is safe for normal 24/7 stuff.

All that said, if it’s on the QVL and the XMP profile doesn’t work you should return the RAM. It doesn’t work as advertised.

spunkshui
Oct 5, 2011





Badger of Basra posted:

Putting this question here so I stop making GBS threads up the pc building thread. Also I did try raising the DRAM voltage from 1.35 to 1.36 and that seemed to be stable, but then 1) I was getting popping/crackling sounds in audio 2) the BIOS froze when I went back to turn off the XMP. Anyone else have ideas? I am hesitant to use the DRAM Calculator and enter everything myself since I'm afraid I'll gently caress something up, not having done this before. Could I turn on the XMP but then just lower the speed to 3400 instead?

The frozen bios sounds crazy bad.

Id try to get some ram at a local store or swap ram with a friend to to see if you can confirm you got a lemon.

Quaint Quail Quilt
Jun 19, 2006



The Ryzen (Zen 2) auto overclocker from 1usmus of the dram calculator drops 7pm Eastern
https://mobile.twitter.com/1usmus/s...526652758274051

spunkshui
Oct 5, 2011





Weirdest thing. Z390 Prime A, 9600k @5.0ghz

A game crashed out of nowhere so I decide to run a stress test. The OCCT stress test shuts down due to heat (85C). I run it again and my voltage goes from 1.385 to 1.39... 1.40 1.41 1.42 1.43 all in like 30 seconds. I shut it off in a confused panic.

I have been screwing with this CPU like 2 years and normally its voltages are very stable at like 1.385 +/- 0.01

I recently changed all my ram and temporarily overclocked it but I definitely tested my cpu after that, though I might have only been looking at thermals. Still the thermals went nuts as voltage climbed so I think I would have noticed. I actually was on zoom all day with this computer with 30 kids faces on 1 screen and a bunch of chrome windows on the other. It was only after class when I was playing Overwatch that the crash happened and I ran the stress test.

For the record I teach off 2 computers and the only one that's bluescreened during class is the Microsoft surface pro 7. Poor thing has to run OBS to record itself and zoom all without an internal fan.

So uhhh.... anyone else ever had their motherboard forget how to voltage before?

Does changing your ram sticks effect cpu voltage? Or the number of ram sticks?

How about ram voltages? My ram is back on stock for now.

Edit: forgot to mention load line calibration was on auto all 2 years. Im now on “5” on the 1-7 scale. 4 was too droopy. This fixes the problem so far. Going to watch vcore.

spunkshui fucked around with this message at 06:17 on Oct 2, 2020

VelociBacon
Dec 8, 2009



spunkshui posted:

Weirdest thing. Z390 Prime A, 9600k @5.0ghz

A game crashed out of nowhere so I decide to run a stress test. The OCCT stress test shuts down due to heat (85C). I run it again and my voltage goes from 1.385 to 1.39... 1.40 1.41 1.42 1.43 all in like 30 seconds. I shut it off in a confused panic.

I have been screwing with this CPU like 2 years and normally its voltages are very stable at like 1.385 +/- 0.01

I recently changed all my ram and temporarily overclocked it but I definitely tested my cpu after that, though I might have only been looking at thermals. Still the thermals went nuts as voltage climbed so I think I would have noticed. I actually was on zoom all day with this computer with 30 kids faces on 1 screen and a bunch of chrome windows on the other. It was only after class when I was playing Overwatch that the crash happened and I ran the stress test.

For the record I teach off 2 computers and the only one that's bluescreened during class is the Microsoft surface pro 7. Poor thing has to run OBS to record itself and zoom all without an internal fan.

So uhhh.... anyone else ever had their motherboard forget how to voltage before?

Does changing your ram sticks effect cpu voltage? Or the number of ram sticks?

How about ram voltages? My ram is back on stock for now.

Edit: forgot to mention load line calibration was on auto all 2 years. Im now on “5” on the 1-7 scale. 4 was too droopy. This fixes the problem so far. Going to watch vcore.

I was about to say this feels like a LLC thing. If anyone else is reading this, never set auto for LLC, you need to be able to see your max load voltage to be able to know exactly what it will be when you increase or decrease it. If your mobo is doing it's own thing for LLC you might make a change to vcore in your BIOS of +0.03mv and not see the same +0.03mv change at the CPU under load.

Lockback
Sep 2, 2006

All days are nights to see till I see thee; and nights bright days when dreams do show me thee.


Also looks like there was a BIOS update 6 weeks ago or so that addressed "stability" which could include a LLC bug. May be a good idea to run through and update bios/chipset drivers.

spunkshui
Oct 5, 2011





Lockback posted:

Also looks like there was a BIOS update 6 weeks ago or so that addressed "stability" which could include a LLC bug. May be a good idea to run through and update bios/chipset drivers.


Thanks, I think ill update, but im staying on 5 LLC anyway which I probably should have done from the start.

VelociBacon posted:

I was about to say this feels like a LLC thing. If anyone else is reading this, never set auto for LLC, you need to be able to see your max load voltage to be able to know exactly what it will be when you increase or decrease it. If your mobo is doing it's own thing for LLC you might make a change to vcore in your BIOS of +0.03mv and not see the same +0.03mv change at the CPU under load.

On my wife’s computer I always use load line calibration but on my computer I started with it off and the voltage readings, temps and stability has always been rock solid.

Kind of figured if it ain’t broke don’t fix it. Most people online say LLC 6 and 5 are good and 5 is better for cpu health if its stable.

VelociBacon
Dec 8, 2009



spunkshui posted:

On my wife’s computer I always use load line calibration but on my computer I started with it off and the voltage readings, temps and stability has always been rock solid.

Kind of figured if it ain’t broke don’t fix it. Most people online say LLC 6 and 5 are good and 5 is better for cpu health if its stable.

My limited undertanding is that realistically if you're monitoring what your CPU is getting and you have an idea of what the limits are for your chip, which of the LLC settings you choose shouldn't really matter too much because when setting up the overclock you'll see x number as being the under-load voltage and you'll adjust the BIOS up or down to move this to whatever your target is, or within the bounds you understand to be the reasonable limits. The problem with "AUTO" is some boards might decide for some reason to change the LLC from 3 to 5 or something like that, and then the CPU sees a different voltage that you may not have intended to change.

spunkshui
Oct 5, 2011





After my bios was updated and a quick game of "why wont my OS load" I'm going to settle for 4.8ghz @ 1.376V while loaded. 10 mins in OCCT makes my rad pretty hot though since my cooler came with 1200rpm fans. Amazon decided I needed 2000rpm Noctua's.

I cant get the 5.0ghz to last 10 minutes on OCCT now at voltages I want. Either the crazy ramping voltages hurt it or the fact I'm use only LLC5 instead of what once was a surprisingly competent auto.

The other rig in our house has a Z370 board a 4.8ghz is all I have managed with the same cpu anyway.

Ill take the low voltages for now, this cpu needs to last through the pandemic I don't want to have to knife fight someone outside a microcenter for the new AMD cpu.

spunkshui
Oct 5, 2011





And now the ram OC is also back after an hour of testing this morning.

under load near the end of the ram test (excited for new radiator fans today and a deep cleaning)


Idle vcore and ram info, box says CL18 ram

VelociBacon
Dec 8, 2009



spunkshui posted:

And now the ram OC is also back after an hour of testing this morning.

under load near the end of the ram test (excited for new radiator fans today and a deep cleaning)


Idle vcore and ram info, box says CL18 ram


I know those are max temps during ram testing but what do you get for max cpu temps during cpu testing? If they're the same I think you have a lot more thermal headroom. Not sure if AVX workloads are tested with that ram tester though.

I thought I was stable with my 9900k at 5ghz all core and I think 1.31v but rendered a video recently that got me to 100C. Never even comes close to that with any other actual workload so it's funny. Just ended up turning down the render rate instead of detuning.

spunkshui
Oct 5, 2011





VelociBacon posted:

I know those are max temps during ram testing but what do you get for max cpu temps during cpu testing? If they're the same I think you have a lot more thermal headroom. Not sure if AVX workloads are tested with that ram tester though.

I thought I was stable with my 9900k at 5ghz all core and I think 1.31v but rendered a video recently that got me to 100C. Never even comes close to that with any other actual workload so it's funny. Just ended up turning down the render rate instead of detuning.







Water idles at 30C but its fans max at 1200 so I'm finally getting some nice Noctua 2000 rpm 140mm fans for it.

The problem with 5.0 ghz is my idle voltage is 1.394 and its not stable at 5ghz under load because of the droop. Stepping up the voltage by 0.1 still throws and error after 3 minutes in OCCT.

Im going to try LL6 for giggles again. but if it doesn't work I'm happy with the above.

VelociBacon
Dec 8, 2009



spunkshui posted:







Water idles at 30C but its fans max at 1200 so I'm finally getting some nice Noctua 2000 rpm 140mm fans for it.

The problem with 5.0 ghz is my idle voltage is 1.394 and its not stable at 5ghz under load because of the droop. Stepping up the voltage by 0.1 still throws and error after 3 minutes in OCCT.

Im going to try LL6 for giggles again. but if it doesn't work I'm happy with the above.

Idle temps and voltage don't really mean too much! What are your temps under max load? You should just pick an LLC and stick with it, look at what the voltage your CPU is getting is at whatever LLC setting and just use offset or manually adjust the voltage up or down in bios to hit the at-CPU Vcore you're targeting. Quickly looking around online it looks like you should be keeping your vcore 1.35-1.40v depending on if you can get the temperature under control. Too aggressive an LLC and you'll be seeing over 1.40v if you're idling at no load at 1.394v so I'd be careful. Maybe set it to 5 and see what your vcore is during your stress test to make sure it's under 1.40v?

You have c-states enabled right? It shouldn't be idling at max vcore unless we're not talking about the same voltage reading.

VelociBacon fucked around with this message at 17:52 on Oct 3, 2020

spunkshui
Oct 5, 2011





VelociBacon posted:

Idle temps and voltage don't really mean too much! What are your temps under max load? You should just pick an LLC and stick with it, look at what the voltage your CPU is getting is at whatever LLC setting and just use offset or manually adjust the voltage up or down in bios to hit the at-CPU Vcore you're targeting. Quickly looking around online it looks like you should be keeping your vcore 1.35-1.40v depending on if you can get the temperature under control. Too aggressive an LLC and you'll be seeing over 1.40v if you're idling at no load at 1.394v so I'd be careful. Maybe set it to 5 and see what your vcore is during your stress test to make sure it's under 1.40v?

You have c-states enabled right? It shouldn't be idling at max vcore unless we're not talking about the same voltage reading.

VelociBacon posted:

You have c-states enabled right? It shouldn't be idling at max vcore unless we're not talking about the same voltage reading.

I always have c-sates off and downclocking for stability. Sure my chip wont live as long but its how a treated my i7-920 @ 4.0ghz and E-6600 @ 3.6ghz and I have never even had to back off the overclocks before they became completely obsolete chips. The i7-920 is still living out its golden years at my highschools robotics club.

Thanks to overclocking those are the only CPUs I have needed since the C2duo hit shelves.

Everything is back to completely normal now stability and voltage wise @5.0ghz. I'm 100% sure LLC6 is what auto was using, my only problem is that due to age my 280mm rad isn't working as well as it used too and a core flicks past 85 after 5 minutes of letting the water heat up.





The rad is older then the 9600k, it was bought to replace a120mm AIO that was cooing the i7-920 since its pump died. I couldn't hit 4ghz without this rad so it was a fun upgrade to have to buy. Its fans top out at 1200 rpms though so I'm just keeping them on max right now.

If the water cools down Cinebench stays at 80 and under no problem.



Rad fans going from basic 1200 rpm corsairs to 2000 rpm Noctuas tonight is going to be sick. Also swapping rear case exhaust for an a12x25 too since I'm embarrassed to admit my case exhaust fan is the old cooling fan for my 120mm radiator and its clearly optimized for static pressure not airflow.

TLDR: AUTO LLC BAD EVEN IF IT WORKS FOR 2 YEARS, COMPUTER NOW GOOD BACK AT 5.0ghz

spunkshui fucked around with this message at 00:12 on Oct 4, 2020

spunkshui
Oct 5, 2011





Today was a good day.





VelociBacon
Dec 8, 2009



How much did your temps drop under full load compared to before? I have the same AIO so if it's much better I'll upgrade.

spunkshui
Oct 5, 2011





VelociBacon posted:

How much did your temps drop under full load compared to before? I have the same AIO so if it's much better I'll upgrade.

Make sure you have the AIO pump on “extreme”, helps a ton with little noise impact.

Initial load is the same from idle, whats improved is sustained temps since I can keep the water from heating as much.

You can see my cores are all just under 80 even after 10 minutes of torture.

Water used to hit 36-37C in 5 mins before OCCT shut off, now it stays around 32-33C and passes the 10 min mark fine.

I like the upgrade because my fans are Idle at 1000 now and I have 1800 rpm cooling headroom If I need it. Vs the stock fans really only having like 800 to 1100 RPM in total to play with. The old fans also feel very cheap in comparison and have no sound dampening rubber.

Curve: 29C water 60% fan, increasing by 10% every 1C

spunkshui fucked around with this message at 16:07 on Oct 4, 2020

VelociBacon
Dec 8, 2009



spunkshui posted:

Make sure you have the AIO pump on “extreme”, helps a ton with little noise impact.

Initial load is the same from idle, whats improved is sustained temps since I can keep the water from heating as much.

You can see my cores are all just under 80 even after 10 minutes of torture.

Water used to hit 36-37C in 5 mins before OCCT shut off, now it stays around 32-33C and passes the 10 min mark fine.

I like the upgrade because my fans are Idle at 1000 now and I have 1800 rpm cooling headroom If I need it. Versus the stock fans really only having like 800 to 1100 RPM in total to play with. The old fans also feel very cheap in comparison and have no sound dampening rubber.

The one thing I don’t like about the new fans is 75% power is 1000 RPMs so my fan curve starts at 70% fan speed at 27° and it reaches 100% at 33°.

50% power should be 1000 RPMs, not really sure why it’s so far off but I’m still able to make a curve that does what I want.

Thanks for taking the time to reply with all the info. I could swear my AIO only has 'performance' or 'quiet' as options. It's a corsair h115i pro right? Maybe there was a software update? The whole system is essentially silent, I never hear it on my floor in my closed case.

Was it a pain to swap the fans? I can't remember where the fan connectors actually attach, I know on my EVGA Hybrid kit the fan PWM connector is buried somewhere under the shroud on the card.

Anyways, thanks for this. Seems like I have some stuff to play with which is great. I don't know if your 9600k has a solder TIM but I feel like that's the bottleneck on my 9900k anyways but it doesn't hurt to have better cooling on the other side of the spreader.

spunkshui
Oct 5, 2011





VelociBacon posted:

Thanks for taking the time to reply with all the info. I could swear my AIO only has 'performance' or 'quiet' as options. It's a corsair h115i pro right? Maybe there was a software update? The whole system is essentially silent, I never hear it on my floor in my closed case.

Was it a pain to swap the fans? I can't remember where the fan connectors actually attach, I know on my EVGA Hybrid kit the fan PWM connector is buried somewhere under the shroud on the card.

Anyways, thanks for this. Seems like I have some stuff to play with which is great. I don't know if your 9600k has a solder TIM but I feel like that's the bottleneck on my 9900k anyways but it doesn't hurt to have better cooling on the other side of the spreader.
yup h115i pro


Once I realized it was less work to unscrew the RAD from the top of the case it wasn't too bad. Fan connectors come off the Pump/CPU block combo.

TheFluff
Dec 13, 2006

FRIENDS, LISTEN TO ME
I AM A SEAGULL
OF WEALTH AND TASTE


Slippery Tilde

spunkshui posted:

I always have c-sates off and downclocking for stability.

I'm pretty sure this doesn't actually improve stability by any meaningful margin on modern CPU's. For a daily overclock nobody actually recommends disabling C-states and downclocking anymore.

spunkshui
Oct 5, 2011





TheFluff posted:

I'm pretty sure this doesn't actually improve stability by any meaningful margin on modern CPU's. For a daily overclock nobody actually recommends disabling C-states and downclocking anymore.

I enabled C states and speed step but my multiplier wont leave 50.

Google just tells me to change the settings I think I already changed.

VelociBacon
Dec 8, 2009



spunkshui posted:

I enabled C states and speed step but my multiplier wont leave 50.

Google just tells me to change the settings I think I already changed.

You may already be doing this but make sure you're letting the PC idle for 10 min and seeing if it drops then, as it'll keep the x50 multiplier up for background processes until it catches up with them. I'd also make sure your PC power settings are set to "Balanced".

Something else to look at, I had an ASUS Maxiumus Big Dick Hero lordkING VIIIXXCCMM a few years ago and it straight up would not let the clocks and voltages idle in 'manual' CPU voltage setting. I had to use the 'offset' voltage setting, basically looking at the voltage under load, comparing it to what it was when I set it manually, and then + or - the difference with the offset to get the same voltage I was getting in manual mode. It then would idle properly.

spunkshui
Oct 5, 2011





New case! Corsair 500D non RGB, no more loving fan knobs for me!



Goodbye useless 5.25 bays, hello triple Noctua a12x25 intake fans. Washed the AIO radiator and repasted with thermal grizzly non conductive grease.

Average load cpu temps are down by nearly 10C and its a quieter case with much more effective looking filters.

Ran it for 10 minutes and 2 cores didn't even go above 75 the entire time.

I mean just look at that loving cinebench





Low 70s

VelociBacon
Dec 8, 2009



spunkshui posted:

New case! Corsair 500D non RGB, no more loving fan knobs for me!



Goodbye useless 5.25 bays, hello triple Noctua a12x25 intake fans. Washed the AIO radiator and repasted with thermal grizzly non conductive grease.

Average load cpu temps are down by nearly 10C and its a quieter case with much more effective looking filters.

Ran it for 10 minutes and 2 cores didn't even go above 75 the entire time.

I mean just look at that loving cinebench





Low 70s

Cool, let's see OCCT results? I looked at getting noctua fans for the rad but they were like... $33 bucks each for me. Seems like a lot. I'd love to see your custom fan curves btw.

My temps are way hotter than yours with the 9900k at 5GHz on all cores. I guess that's just the way it goes with the extra cores but it's too bad.

Adbot
ADBOT LOVES YOU

spunkshui
Oct 5, 2011





VelociBacon posted:

Cool, let's see OCCT results? I looked at getting noctua fans for the rad but they were like... $33 bucks each for me. Seems like a lot. I'd love to see your custom fan curves btw.

My temps are way hotter than yours with the 9900k at 5GHz on all cores. I guess that's just the way it goes with the extra cores but it's too bad.

Yeah you have way more going on in there I have 6 cores. Do get 2000rpm fans for the RAD, mine aren't impressively quiet so you can def get something else.

Oh those are long duration OCCT results. That was 7ish mins, heres a solid 10minutes.



Fan1 is CPU RAD https://www.amazon.com/gp/product/B...0?ie=UTF8&psc=1 Not going to recommend these, kinda loud 1500-2000rpm. I'm going to replace them with 140mm "a12x25s" when Noctua finishes designing it.

Fan2 is CASE FANS https://www.amazon.com/Noctua-NF-A1...t/dp/B07C5VG64V These are loving good.

Ok there was some spikes in there clearly, (87C), but it really does hang out around 72-73 average even after 10 minutes.

I'm thinking washing dust off the RAD and the repast did most of the work here, but new case could definitely be a factor too now that I have 4 pwm fan responding to water temp. A12x25 fans are a sound investment. heh

Ok, fan curves!







So far that simple line has worked out great.

edit: link fixed https://www.amazon.com/gp/product/B...0?ie=UTF8&psc=1

spunkshui fucked around with this message at 14:03 on Oct 16, 2020

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply