So yeah as the title says, my 9700k, running on 4.6 Ghz firstname.lastname@example.org and cooled by a Corsair H50 with Arctic Silver paste (pea method) is reaching a temp that I don't think is normal. At idle I get around 50c and while playing a few hours of PUBG, my HWINFO Alert triggers a few time notifying me about temp hitting 85c.
Ok, so the fan and pump are running normally.
What's your case, how many case fans, and how do you have your airflow set up? Intake at the front/bottom and out the back/top is the ideal setup.
Are your fans positioned correctly? Air is pulled in through the fan's 'frameless' side, and goes out the framed side(usually has a little sticker in the middle with some minor details on the fan's specs.
Some people have actually managed to do this backwards.
This may just be a case of inadequate cooler, though.
I have the same airflow setup as I had before with my 8700k ( was never hitting 70c) which is push pull on the H50 (exhaust), the top 200mm is intake, the side 200mm is exhaust and the Drives bay 200mm is intake.
Also note that my case is on the floor in a basement so the ambiant temp should be pretty low at this height.
Change the top fan to exhaust. The warm air that gets exhausted out of a case rises.
When the top and rear fans are that close to one another, your top fan is pulling in and recirculating some of the warm air exhausted by your AIO.
If you have a filter for the side fan, change it to intake. The side fan becomes the closest source of cool air for the gpu. You take that away, and it'll pull air from wherever it can.
No side fan would be better than side exhaust, so if you don't have a fan filter, remove the side fan completely.
Asus motherboards have a feature called multi-core enhancement in the bios. Is this enabled?
This tends to crank up temperatures because it's sustaining the cpu's all-core max turbo clock(which I think is 4.6ghz on the 9700k), but it applies more voltage than necessary.
Intel default operation is fine.
No, MCE is disabled, I input my self found value. I put 1.175 Vcore and LLC 5. It gives a seemingly good stability with no overshoot. The highest my voltage goes is 1.163 on 24w TDP (idle basically) and 1.146 on load.
I just put my hands into the hood and look at that:
I don't know but it looked like the center was lacking some paste, I added a little bit. Idle temp at the moment is 40-45 with CPU Fans spinning at 1K RPMs.
I will test load temp and see if I do what you suggested. I'm not sure because as I said my airflow has been like that for years, it didnt allow my 4970k to get higher than 60c back then, neither did it let my 8700k get to 70c.
Load temp was still bad so I went with some changes. I noticed that my side fan was already intake, had changed it some time ago. I changed the top fan to exhaust and rotated the radiator so that the tubes are at the bottom (I dont know if it actually matters).
Even after that, my in-BIOS temp is 45-50c... could my H50 life have come to an end?
I have 2 120mm fans (dont remember the brand but really good ones) on my H50 in a push/pull setup.
Now I want to say this really weird, after un and recovering the CPU 2 times (trying to adjust paste uniformity) from the copper block thing got worse.
60c idle, 80c on pubg lobby, with a 3.8Ghz clock speed. I didnt even bother going into a game since its fairly obvious it was gonna rise over 90c.
I want to add that my H50 bracket don't have the adhessive pads that would stick on the MB. It got very glued on an old MB and I didnt want to break anything. I didn't have those for quite awhile tho so it would be strange that suddenly it causes an issue.
Ah, you don't want to add more paste. Too much will actually raise temps.
A small pea shape applied in the middle of the cpu IHS is all the paste you need to apply. The mounting pressure from securing in the copper block will do the rest.
The purpose of the paste is to fill in the tiny microscopic gaps in the IHS and the copper plate beneath the block that would otherwise trap air and hinder the cooler's performance.
But the difference between adding too much and a small dot isn't a huge difference; some paste is still better than none.
You've had the H50 for some time, right? Do you clean/dust the radiator on a regular basis? Like, every 6 months, or something?
It's looking like the 9700k is just too much for the little 120mm AIO. It worked for the 8700k, but this guy has 2 extra physical cores = higher power draw + higher heat output - need for a bigger cooler.
I'm kinda getting worried guys, I wont lie. I upgraded to a 9700k because my 8700k died few weeks ago, would only boot with 1 core enabled.
Now I noticed my temp with the 9700k (which was fine after CPU installation) was getting very high because I had a BSOD sunday evening while playing PUBG, screen froze with typical sound loop before the BSOD. The error was MACHINE_CHECK_EXCEPTION
I don't see how my H50 would suddenly not be cooling enough anymore while the pump appears to still be running at full speed and my paste application seems correct to me. What else on a water cooler can become deffective and suddenly reduce cooling effectiveness?
Could my rig be infected by some crazy rootkit or some xxxx? is it possible that even if I read a low CPU usage, that a virus is steathly juicing my CPU?
Or could it be the CPU thermal sensor readings that are erronous?
Tom's is a family friendly website. Profanity is STRICTLY forbidden and is a banning offense!
You will have to get a source for me to believe that. Win10 is totally capable of detecting a new CPU and in the end it all comes down to having the proper chipset installed which I took on Asus website.
As I said, everything is working fine, my games are running perfectly (which is a good indicator that my windows installation and drivers have no issues), until the Temp throttle triggers that is. I played some PUBG earlier and CPU temp went up to 90c (CPU Package, no cores hit 90c actually) and clock speed went down to 2.4GHz, game was still playable with fps dropping to 70 sometimes. Of course, I didn't torture the CPU for long tho.
I just have this overheating issue which I don't know the cause but I don't think it can't be software related unless an insane virus could do that without it being obvious (no process using all CPU).
The machine_check_exception could have triggered because of the fact I set 90c as max temp in the BIOS , I don't know if it would throw this error but I guess it's possible.
Hey so i just bought a new mobo Gigabyte Z370P D3, and a new CPU Intel i5-8400.In my system i have a MSI B85-G41 pc mate mobo, and Intel i3-4170 cpu.Also i have reinstalled windows 10 recently.I was wondering when i put my new mobo and cpu in my pc will i have to reinstall windows 10.I have read ...
I have a broken intel board and im on a budget so im switching from intel to amd (ryzen specifically). Ive heard that during the switch i will have to reinstall windows etc. Is it possible to do this without reinstalling windows?
I courrently have an fx8370, gigabyte 990fxa ud3 r5 and i want to buy an i7 8700k with the aorus gaming 5... do i need to do something in windows before setting off the pc to replace the parts? i just don't want to mess something :confused:
You're trusting Windows to make all the necessary driver and registry updates for the new hardware?
From your 3rd to last post, you're already running into errors...
Machine Check Exception is not being caused by your cpu throttling at 90C. The critical point for the 9700k is 100C. But you're not hitting that, due to the new bios setting.
One of the main causes of this error is incompatible, corrupt, or just plain missing drivers and other Windows files.
2 of your examples are people switching Brand and one is upgrading to a different architecture . In those cases you'd probably notice a performance hit without reinstalling Windows because it's a big move. Meanwhile, in my case, I just moved from a Coffee Lake CPU to another Coffee Lake CPU, nothing much really changed.
I dont want to argue with you because I know you're just trying to help me but this isn't the first time I do that. When I upgraded to 8700k from 4970k last year, I didn't reinstall windows for months until I decide to move to Win10 1809 to get DX12 working in BF V few months ago which is when I started having issues because as per usual, newer version of Windows mess up everything (Stuttering in games, not DPC related). Now, after fighting these issues for months, I finally fixed it and my games are running smooth as butter. You would want me to reinstall windows after that, no thanks. There is no correlation between high temp and not reinstalling windows after changing CPU.
I guess I could try with a live windows image and see if it changes something, are you taking the bet?
In one of your earlier posts, you mentioned that the old 8700k would only post with one core. Would you happen to remember what temps were like when it was completely working?
Did you also happen to reinstall the cpu at any point before? Some pins may have gotten bent/broken during this that disabled the other cores.
I finally moved my arse seriously and once again opened my case but this time I completely disassembled the H50, removed the paste, cleaned the radiator with tab water, dried it with an hairdryer. I also noticed that my H50 fans were not even correct (silly me... to my defense, it gets confusing when you do not toy with airflow that often), they were in a pull/pull configuration so the interior fan was spitting hot air back into the case.
It had been like that for years (since when I bought the H50) and I never noticed it because my temp was fine to me but I guess I could've had minus few celcius (5-10 ?) for all this time.
I applied the thermal paste following a different pea method, a single pea in the center (rather than 5 smaller dots spread)
BIOS temp after that was 30c, and on Windows (back on my usual clock speed/Vcore, 4.6GHz@1.175), it's lower than it was earlier today but it's still not very good, lowest idle temp I've seen is 41 and highest I've gotten with some load (no games, just videos and softwares launching) was 60c.
Yeah, the max LLC setting does fry cpus, especially if you left the cpu input voltage - some boards have it as VCCIN - on auto.
Although the different vendors have different levels of LLC settings, the general consensus is to use either the lowest, medium(whatever's considered medium on your particular board), or in between.
Very high and extreme are liquid and LN2 territory.
In the 2nd pic: 1.305v is perfectly fine for an 8700k. But was VCCIN left on auto? VCCIN, or input voltage is the total amount of voltage being pumped into the cpu as a whole, which then feeds the cores, internal memory controller, etc.
If VCCIN was left to auto, LLC on max, and you ran it through a stress test...
I'll use my cpu for example:
7820x currently doing 1.25 vcore and VCCIN is manually set to 1.85v. LLC 4 is used. I have an Asus board, it has levels 1 - 8.
If I run a stress test for a few hours, the cpu still reads 1.25, but the min and max values of VCCIN is now 1.840 - 1.856.
If I do the same, but with LLC 5, I get a range of 1.856 - 1.872. It's still fine, I could go all the way up to 2.0 VCCIN, but that's simply not necessary.
If we take the 1.305 on your 8700k + max LLC is 8(might be 7 on your mobo), VCCIN on auto + vendors overvolt these things more than necessary as it is =
I can only guess, but the cpu was getting fed dangerous levels of voltage at some point or other. The VCCIN - before LLC max - was likely already higher than what I had manually.
And if you apply the same for your 9700k... I would look at the cpu input voltages.
Yeah... I had left the Vcore on auto with my 8700k, I don't know how I made this stupid mistake, I had always been using high LLC in the past because I heard it would lead to the best stability (I read alot on the past weeks, I was out of computer for 2 weeks when my 8700k died, I know I was wrong now) but I forgot I'd set a manual Vcore using AI Suite, which I don't use anymore, in the past so I completely forgot to input the Vcore in the BIOS.
It still ran correctly with this setting for 1 year tho.
Note that I had a power outage in the past days before the 8700k died, if you want to read my post on the matter to have the whole story, here it is .
Now, I'm not doing the same LLC mistake with my 9700k, it is set to 5 and my vore is manually set to 1.175 with all power saving off.
You may laugh but I never heard about VCCIN, even after reading so much in the past weeks, when I search about OC, it always talk about Vcore only. Where can I see this value in windows, I found CPU VCCIO @1.344, is it that?
Voltage pictures from AIDA64
Btw, as I'm typing this, I currently have PUBG (training mode) opened in background , CPU usage is 62% and temp are :
In the first column, you can see the vcore for all 8 cores. Now, over to the 3rd column, under the mobo name(Asus Prime X299 Deluxe), the first power icon you see going down, has Vccin right next to it.
You'll see the current, min, max, and average values.
You've said that the 9700k used to run a little cooler. It's quite possible the most recent windows update may have changed something, but do take a look a the VCCIN for your chip, as LLC does affect this.