PC Pc Gaming/building thread

[dwwaino]

What frequency did you have your 8700k at?

With multi core enhance 4.7ghz across all cores. Obviously that’s going to generate butt loads of heat so I turned it off to let Intel step down the frequencies across the cores. I’m not really doing any multi threaded tasks so 4.7 and 4.6ghz on the first two cores is fine for vidya games. It didn’t affect my load temps one bit though so it had to be something else. Funnily enough I could overclock it to 5ghz and set the voltage manually, which disabled heaps of junk and I got slightly lower temps. I couldn’t quit get it stable though so it was something I was going to come back to at a later date.

Can't say I know a whole lot about it. Having a quick read, it looks like LLC is designed to prevent voltage drops when the CPU overclocks, although it's effects vary across hardware manufacturers. C-states are designed to control power consumption.

If you're after lower temps I'm guessing you should be using C-states over LLC, unless you start getting stability issues during high CPU loads.

That’s what I thought as whenever I’m playing around with overclocking and from what I’ve read I set the LLC and it keeps my voltages pretty steady. I don’t fully understand C states though. The temps have to be due to over voltage, power is voltage*current. With multi core enhancement in HW Monitor it hits 1.41v and without it was reading about 1.35v which is ok but I expected it to be lower so I’m assuming some bios setting is trying to drive the CPU harder. It’s giving me shits honestly. I’ve never had this much drama on stock settings, usually only when I try to overclock.

This is why I thought I would at least just reapply thermal paste and reseat the cooler (sigh ). Every comment mentions getting the temps on an 8700k down by 15-20c by delidding but I refuse to accept that as a solution at stock.

 

[dav3]

That’s what I thought as whenever I’m playing around with overclocking and from what I’ve read I set the LLC and it keeps my voltages pretty steady. I don’t fully understand C states though. The temps have to be due to over voltage, power is voltage*current. With multi core enhancement in HW Monitor it hits 1.41v and without it was reading about 1.35v which is ok but I expected it to be lower so I’m assuming some bios setting is trying to drive the CPU harder. It’s giving me shits honestly. I’ve never had this much drama on stock settings, usually only when I try to overclock.

This is why I thought I would at least just reapply thermal paste and reseat the cooler (sigh ). Every comment mentions getting the temps on an 8700k down by 15-20c by delidding but I refuse to accept that as a solution at stock.

Found this on the Asus ROG forums.... "Ordinarily, Intel Turbo overclocks one of the cores to 4.7 and the rest somewhat less. Asus MCE overclocks all of the cores to the max Turbo frequency of 4.7. Asus auto rules will then assign an increased voltage that works for all chips, even the bad ones. So yes, temp will increase somewhat."

If you're looking to control your temps you might need to try a manual overclock, as MCE is ramping up your clock speeds, and LLC is ramping up your voltages to keep it stable.

C-states are all about turning things down/off when not in use, so setting it to C1 will limit it to just software controlling limited power states, while C6 can theoretically let the hardware set cores to 0v if possible. There's a list of states here

 

[dwwaino]

Found this on the Asus ROG forums.... "Ordinarily, Intel Turbo overclocks one of the cores to 4.7 and the rest somewhat less. Asus MCE overclocks all of the cores to the max Turbo frequency of 4.7. Asus auto rules will then assign an increased voltage that works for all chips, even the bad ones. So yes, temp will increase somewhat."

If you're looking to control your temps you might need to try a manual overclock, as MCE is ramping up your clock speeds, and LLC is ramping up your voltages to keep it stable.

C-states are all about turning things down/off when not in use, so setting it to C1 will limit it to just software controlling limited power states, while C6 can theoretically let the hardware set cores to 0v if possible. There's a list of states here

Thanks, I’ll do some reading on the C states. I was running MCE for a long time but it was the first thing I turned off when I wanted to reduce temps. Also after I noticed 1.41v in HW Monitor. It’s ridiculous that a default setting can be more detrimental than a manual overclock. I’ll see how this new board behaves out of the box with only XMP. After that I’ll start with MCE, then try a manual 4.7ghz overclock which in theory I should have no trouble with at a sub 1.3v setting as that’s the standard Intel boost anyway.

 

[dwwaino]

This is cruel. I wonder if PCCG provide this to everybody just buying a motherboard without a CPU as they’re likely replacing it because they broke it? Lol

 

[Gold Member]

Thanks, I’ll do some reading on the C states. I was running MCE for a long time but it was the first thing I turned off when I wanted to reduce temps. Also after I noticed 1.41v in HW Monitor. It’s ridiculous that a default setting can be more detrimental than a manual overclock. I’ll see how this new board behaves out of the box with only XMP. After that I’ll start with MCE, then try a manual 4.7ghz overclock which in theory I should have no trouble with at a sub 1.3v setting as that’s the standard Intel boost anyway.

1.41v should be fine long term, as you said it's being set by Asus via MCE. If it were harmful to the chip they wouldn't be exposing themselves to the risks.

Here's some stats taken from silicone lottery about the voltages they use to get a safe and stable overclock that they can pass a 1 year warranty on. Albeit delidded with liquid metal from the 20c deg improved thermals.

https://fuse.wikichip.org/news/425/core-i7-8700k-overclockability-silicon-lottery-stats/

 

[dav3]

Thanks, I’ll do some reading on the C states. I was running MCE for a long time but it was the first thing I turned off when I wanted to reduce temps. Also after I noticed 1.41v in HW Monitor. It’s ridiculous that a default setting can be more detrimental than a manual overclock. I’ll see how this new board behaves out of the box with only XMP. After that I’ll start with MCE, then try a manual 4.7ghz overclock which in theory I should have no trouble with at a sub 1.3v setting as that’s the standard Intel boost anyway.

That's how they get good benchmarks I guess, plus it helps push people towards spending on aftermarket cooling upgrades. It sounds like temps around 80c are generally regarded as safe, as are voltages around 1.4.

 

[dwwaino]

1.41v should be fine long term, as you said it's being set by Asus via MCE. If it were harmful to the chip they wouldn't be exposing themselves to the risks.

Here's some stats taken from silicone lottery about the voltages they use to get a safe and stable overclock that they can pass a 1 year warranty on. Albeit delidded with liquid metal from the 20c deg improved thermals.

https://fuse.wikichip.org/news/425/core-i7-8700k-overclockability-silicon-lottery-stats/

I think they can get away with 1.4v+ with MCE because it’s not a 24/7 overclock and will outlast warranty. Obviously that sort of voltage is going to get hot though which is what I was experiencing. What I couldn’t understand though was why I was getting the same temps with that disabled and using Intel’s boost which is stepped down through the cores. Something else was causing the overvolt.

Not surprised with that link that they are getting a 20c improved after delidding. That’s the exact figure I keep reading. I don’t want to quite go down that path yet however.

 

[dwwaino]

My setup is almost this shiny! If I turn the rest of the lights in my house off, I basically have my own disco.

I can just about copy that picture too. I put 6 ML120 RGB fans in on the weekend and grabbed a Corsair lighting kit while I was at it. Except even for me the lighting was overkill, too bright and too distracting lol. Now that I’m putting it back together I’m pulling those strips. Too much!

 

[Gold Member]

If I can’t get a handle on these temps then I’ll just use this to get me by until I see gen3 Ryzen or what Intel have up their sleeve after 9th gen.

Intel have nothing after 9th gen on 14nm+++. They are ****ed and can't get a handle on 10nm node to literally save themselves.

Take a bucket of salt when watching the 9th gen reviews when they pop up given the extent of underhanded s**t they have stooped to recently with their pipedream 5.0ghz 28 core Xeon stunt, which turned out to be cooled with a 1 horsepower refrigerated chiller.
Not to mention the exhaustive extent of handpicked 8700k golden samples that ended up in reviewers hands for 8th gen.

 

[dwwaino]

Talk about sweating hard installing the CPU.

Checking it’s clicked into the tool
Double checking
Am I holding it right?
I better check it’s clicked
Slowly now...
Is this aligned?
What’s the slowest and most deliberate I can lower this into place???

 

[Gold Member]

You guys make me sick.
Get professional help!

 

[dwwaino]

So hypnotic.

 

[dav3]

Intel have nothing after 9th gen on 14nm+++. They are ****** and can't get a handle on 10nm node to literally save themselves.

Take a bucket of salt when watching the 9th gen reviews when they pop up given the extent of underhanded s**t they have stooped to recently with their pipedream 5.0ghz 28 core Xeon stunt, which turned out to be cooled with a 1 horsepower refrigerated chiller.
Not to mention the exhaustive extent of handpicked 8700k golden samples that ended up in reviewers hands for 8th gen.

The roadmap says Ice Lake 10nm+ in 2020, they've announced that server class CPUs should appear then, and Cannon Lake mobile CPU should be next year.

I just had a quick read and going to 10nm isn't exactly an easy process, they've basically gone past the point where Copper is a viable option and are attempting to use Cobalt instead. Samsung and TSMC (Apple) are the only ones who are currently manufacturing at that level and that class of chip is completely different to Desktop/Server class gear.

 

[dwwaino]

Was Cannon Lake always destined for mobiles? I remember when Coffee Lake was launched that many were holding off because it was meant to be short lived. Now it’s at least 2020 for Ice Lake.

 

[dav3]

Was Cannon Lake always destined for mobiles? I remember when Coffee Lake was launched that many were holding off because it was meant to be short lived. Now it’s at least 2020 for Ice Lake.

Looks like it was the code name for the 10nm die shrink of Kaby Lake. I think it was meant to be the full range of CPUs but they never got past a limited run of mobile chips. I think they're still trying to produce them while they develop Ice Lake.

This may or may not be correct, all this Lake s**t confuses the hell out of me.

Sent from my SM-G930F using Tapatalk

 

[Gold Member]

The roadmap says Ice Lake 10nm+ in 2020, they've announced that server class CPUs should appear then, and Cannon Lake mobile CPU should be next year.

I just had a quick read and going to 10nm isn't exactly an easy process, they've basically gone past the point where Copper is a viable option and are attempting to use Cobalt instead. Samsung and TSMC (Apple) are the only ones who are currently manufacturing at that level and that class of chip is completely different to Desktop/Server class gear.

Roadmaps are for the investors..

https://ocaholic.ch/modules/news/article.php?storyid=11389

And yes I know Intel's 10nm is the equivalent to TSMC's 7nm thanks to density.


Right now Intel make 1 10nm low power laptop chip which performs 25% worse than the 14nm version it's replacing AND they had to disable the integrated graphics to do it. 10nm on the desktop is miles away.

 

[dwwaino]

Computer booted no issues, back in Windows and could easily activate my OEM key because of that digital license thing. I just went into my Microsoft account and removed the PC then re-entered the key in Windows.

I was preparing to format and do a fresh install as it seems the obvious thing to do, but everything is working peachy. Given I am on the same chipset and manufacturer what's the chance I can get away without it?

Also some interesting Cinebench results. I am running the AIO fans through the motherboard CPU headers rather than the pump itself as I was getting better results originally. In BIOS when I first booted up I set XMP and actually said yes to letting the board 'enhance', then went through and sure enough multicore enhancement and all that stuff was enabled. So I left it, and running Cinebench the highest core temp I got was core 2 which maxed 83c, core 1 only hit 77c and all 6 were generally bouncing around low 70s, occasionally dipping to high 60s and the entire package temp of the CPU peaked 84c. More interestingly the VID never exceed 1.301v and averaged just under 1.3v. Of course a few Cinebench tests aren't going to tell the whole story yet and I'll need to game for a bit (especially AC: Odyssey as it's a CPU whore), but I'm starting to wonder if my Maximus X Hero board had dodgy VRMs? Either that or the Strix boards are slightly restrained in that area. If it's all good I'll put the AIO fans back through the Corsair pump connectors and compare again because obviously it gets a bit noisy when they're reacting from CPU temps and not water temps.

Something else that's bothering me too is when I was pulling the cooler off before I shagged the pins was one of the screws on the pump block wasn't very tight. The mounting bracket wasn't tight when I assembled it and Googling the issue and apparently Asus boards are a slither thinner than other manufacturers so it's quite common. When I put it together I put some small rubber washers on the bracket at the rear where the posts screw in from the front so it was firm and I know the block was tight after that. However if the case or pump has ever rattled or the washers lost their elasticity then that could cause a screw to loosen and irregular contact between the surfaces. This time I didn't require the washers and used Arctic Silver remover and surface primer and got some Grizzly Kryonaut (previously used Deep Cool Z3 paste because it was cheap but it was too runny for my liking) to be sure so I guess I'll never know.

 

[dav3]

Computer booted no issues, back in Windows and could easily activate my OEM key because of that digital license thing. I just went into my Microsoft account and removed the PC then re-entered the key in Windows.

I was preparing to format and do a fresh install as it seems the obvious thing to do, but everything is working peachy. Given I am on the same chipset and manufacturer what's the chance I can get away without it?

Should be fine, same chipset means there shouldn't be too many differences with onboard devices, and I think Windows and Windows updates have improved enough that it should recognize the need for different drivers.

Personally I probably would have reinstalled, but that's only because it's good to have a fresh start from time to time.

 

[dwwaino]

Should be fine, same chipset means there shouldn't be too many differences with onboard devices, and I think Windows and Windows updates have improved enough that it should recognize the need for different drivers.

Personally I probably would have reinstalled, but that's only because it's good to have a fresh start from time to time.

I think I will when I finish this course. I've been doing a cert iv in programming in my spare time and have a bunch of crap on here that I don't really want but I can't be bothered wiping it clean and reinstalling it all only to format it like I planned to when I was done. Android Studio and Visual Studio take forever to install with every damn package. If I run into any issues I will though I guess.

 

[Gold Member]

Let the 9th gen deception begin...

 

[dwwaino]

Let the 9th gen deception begin...

As much as I enjoy watching tubers like Jayz because they're pretty fun, guys like Steve can't be ignored no matter how much the truth hurts. Sounds like his review is going to be savage.

 

[Gold Member]

As much as I enjoy watching tubers like Jayz because they're pretty fun, guys like Steve can't be ignored no matter how much the truth hurts. Sounds like his review is going to be savage.

Steve is nothing compared to Jim.


Watch the other videos in his Con Lake series. Jim rips Steve a new one.
His videos are fascinating as they are a critical analysis of the tech industry as opposed to all the others being reviewers.

 

[dwwaino]

Steve is nothing compared to Jim.


Watch the other videos in his Con Lake series. Jim rips Steve a new one.
His videos are fascinating as they are a critical analysis of the tech industry as opposed to all the others being reviewers.

Cheers, I'll add him to my list. I usually just check out Gamers Nexus and sometimes Hardware Unboxed to catch up before going back to watching UFO and war videos.

 

[Mr_Moogle]

Let the 9th gen deception begin...

s**t like this just boggles the mind. They know they'll get caught out but they do it anyway.

Very tempted to go Ryzen in my next build.

 

[Gold Member]

s**t like this just boggles the mind. They know they'll get caught out but they do it anyway.

Very tempted to go Ryzen in my next build.

Yeah me too 100%. Next year's Zen 2 Ryzen 3000 series will be looking the goods. Intel are goneski. Everything they are doing is reactionary to Zen and even though they are still in front, they are miles behind Zen's scaleability and infinity fabric.