*Official* NBR Desktop Overclocker's Lounge [laptop owners welcome, too]

(1.4V has been my personal max allowable voltage in all circumstances, 24/7 or otherwise)

 

IDK, is that low? What's supposed to be the "safe maximum voltage"?

Vendor recommendations take preceidence as a marker to not exceed, I don't need to burn it up all the time, just for occasional benchmarks, but if that's all you do is benchmark at highest possible OC all the time...

 

Less than 1.45V is generally considered safe, see this thread: http://www.overclock.net/forum/5-intel-cpus/1570313-skylake-overclocking-guide-statistics.html
In that thread above some users reported measurable CPU degredation after a year when operating in the region of 1.48-1.50V.

EDIT: 1.244V is my stock Vcore, 1.4V being my personal max allowable overvolt - so I've increased Vcore by 156mv or 12.5%.

 

That's higher than others suggest... more like 1.35v long term, with a maximum of 1.41v

What’s the highest core voltage for an i7-8700k?
https://www.reddit.com/r/intel/comments/78f625/whats_the_highest_core_voltage_for_an_i78700k/

Safe voltage for 8700K
https://www.reddit.com/r/overclocking/comments/7h5m26/safe_voltage_for_8700k/

8700K 24/7 Voltage & Use ???
http://www.overclock.net/forum/5-intel-cpus/1643707-8700k-24-7-voltage-use.html

"I would say you have a pretty good chip. I am running 4.8Ghz at 1.23v right now. I know the general consensus I've read is that under 1.35 and you're well within safe limits for 24/7 usage. I would not worry about your volts, and the temps are also very normal. There are some using up to 1.4v for 24/7 5.0Ghz."
8700k @ 5Ghz 1.29v | GTX 1080Ti FTW3 | 16GB 3200Mhz | Samsung Evo 500GB SSD

"I run [email protected] 24/7 on a custom loop so heat is no issue for me clocked it slightly lower for summer as things get hot here in Australia, CPU scales at around 60mV per 100 Mhz so 5.0Ghz would be around 1.255V."

"I can do 5.3 ghz at 1.392V. Haven't tested much for Realbench 2.56V, but gamed PUBG for a few hours without issues. Max temp on the hottest core was 57'C

I think I'll be running 5.1 at closer to 1.264-1,280V for 24/7. I can do 5.4 ghz at around 1.460V I think. But way too much voltage for 24/7.."

 

It depends on who you talk to, one person in the thread you quoted ran 1.44V on his 6700K for 1 year with no signs of degredation - it's all opinion in those threads. The thread I quoted earlier, 2 users in there with measurable degredation after a year at anything from 1.48-1.50V, haven't seen reports of degredation below that though.

Update to my ongoing testing to work out cause of instability: my system passed about 2 hrs of stock CPU/RAM combined with max GPU overclock on Prime95 & Firestrike run together, so instability not due to GPU overclock - so gonna test the combination of max CPU/RAM overclock combined with stock GPU - if it crashes on that then I know it's either CPU or RAM overclock instability.

 

Well I think that may be the confusion, the 6700k and 7700k aren't the same CPU, or exactly the same process, so you need to allow for variance, and can't depend on previous generation settings to promote to the next generation.

And, for the 8700K it's even more so, with 2 more cores and other features not in the previous generation, it's quite likely it needs and can allow less voltage.

Look at the low voltages those guys are getting at 5ghz+, that's why I posted them. Given those low voltages at such high OC, it would seem unnecessary to push beyond 1.35v, yes? no?

I've always searched for the lowest voltage at the highest OC to try, and then stay below that voltage for day to day use. Excursions into high voltage runs for experimentation maybe, maybe not, depends on the Cost / Pucker factor of the CPU.

I'd hate to lose a $600-$1500 CPU to an OC overvolt, heck, I'd hate to lose a $99 one too.

 

I'd run the max OC I could get for whatever I deemed my max voltage/temperature to be, but a 5Ghz 8700K is practically as fast as you could ever want for anything gaming related or almost anything really.

Oh, while I was typing this my desktop PC just froze & rebooted - this was with max CPU/RAM overclock & stock GPU. OK, so this instability I've been talking about is CPU/RAM or possibly PSU related. Possibility of PSU related as it was consuming more power as measured from the wall with CPU overclock & stock GPU in comparison to stock CPU & max GPU overclock. OK, will now test stock CPU and stock RAM combined with stock GPU while running Prime95 and Furmark - this should take PSU beyond wattage amounts already seen during this last crash - this should rule out PSU issue and therefore only leave the CPU/RAM overclock as the possible cause of instability.

EDIT: running a slight GPU overclock while running Furmark as stock GPU wasn't pushing the power of GPU up against it's TDP limits. Currently running at 380W power draw from the wall, I've got a 650W EVGA Gold rated PSU, and this is the first time I've ever heard the fans come on of the PSU - I really haven't been pushing this PSU before, so you'd figure the instability is unlikely to be the PSU right?!

 

It ran fine. I turned it off and went to bed. The next day I came into my office and the keyboard lighting was on and the machine still seemed turned off. It never booted again. It would be really silly to speculate voltage or overclocking had anything to do with it. I have never run enough voltage through any CPU to cause sudden death. I believe it had a defect of some sort. It would be foolish to speculate what the cause was, but voltage would not even be on my list of wild ass guesses. Even Intel sells a dud now and then. Bound to find on eventually. I know I am not losing any sleep over or it or adjusting my behavior because of it. I think running with scissors is a lot more dangerous, LOL.

Some people are just too eager to blame things they fear or do not understand. That kind of stupidity is why I never posted anything about it. I knew some silly person would jump to conclusions and decide what caused it with nothing to base it on. I can't tell you how many chuckles Brother @Johnksss and I have had over the years over that kind of silliness.

 

Did you try delidding it again to see if you had some LM drip onto the chip and possibly make a connection somewhere? You would think that if this was the case, it wouldn't have taken 5 months to happen though....Weird

 

Thanks for the reply. At the moment the instability I've been talking about seems to be related to my CPU/RAM overclock. Furmark & Prime 95 running right now at stock CPU/RAM to stress PSU more than it ever has before - been running for about 15mins so far with no crash, will continue for a bit longer just to rule out PSU issue.

 

It happened in December when the machine was less than a month old. I never posted anything because I knew silly people would make anecdotal comments and post wild speculation like they do with everything else. Yes, I did check and no liquid metal had escaped. I have an idea what it might have been, not related to overclocking, but I am not going to mention it because doing so might create more reasons for wild speculation with too little information to base it on. Since we do not know why, that is all we can really say about it. I did not lose any sleep over it. Since it was not a BGA piece of trash it was easy to replace, and that much is a matter if fact.

If I were going to guess I would guess the RAM overclock is flakey. I have had more issues with that than any other kind of overclocking instability. What happens if you run the same tests with default memory profile?

 

I have previously tried 4.7Ghz (my max CPU overclock) with a reduced RAM overclock, and that still failed - that was at 15-15-15-32-240-1T [email protected], which is my historical long time stable RAM tightening I did when I first did the build. My current newer RAM overclock is [email protected]. Unless my RAM has degraded I don't think the RAM is to blame, although I don't think I ever did such stringent overall system testing as I am currently doing (Prime95 all threads + Firestrike) - but I did do over a 1000% Pass on numerous instance of HCI Memtest to verify the RAM overclock stability, as well as Prime95 & Memtest86+.

Update to my testing: Passed the PSU test, 40 mins stable on stock CPU/RAM & slightly overclocked GPU while running Prime95+ Furmark. Created the highest loads I've ever seen on my PSU, 380W and first time I've ever heard the PSU fan come on. I don't think it's the PSU right, I mean it's an EVGA 650W Gold rated model.

Currently now testing a dialed back CPU overclock while keeping max RAM overclock & max GPU overclock. 4.6Ghz core & ring at a max seen Vcore of 1.344V - this was a previous long time stable overclock I used to run. Running Prime95 & Firestrike loop as test.

 

@Mr. Fox - Sorry to hear your CPU died.

We were discussing 8700K maximum safe voltage in another thread, initiated by a post about Ryzen and high core voltages.

I mentioned that I had seen you mention that your 8700K had died, and I didn't know why.

@Robbo99999 jumped over here to ask you why, and we continued the discussion on voltages while we waited

Again, sorry to hear your trouble, as you said it can happen for many reasons as a new component, and it's tough to put a specific fault to the failure - unless you see something obvious connected to the failure when it happened.

I'll let you and @Robbo99999 get back to figuring out his rig's instability.

 

Well, I think I've concluded my stability testing and worked out the cause of the instability. My final test at a dialed back CPU overclock of 4.6Ghz (on core & ring) resulted in a 2.5 hour pass of Prime95 & Firestrike loop - previously my 4.7Ghz overclock would result in freeze of 3DMark, or 3DMark App crash within a 30 minute testing window. I was able to rule out GPU overclock / RAM overclock / PSU as the cause and narrow it down to the CPU overclock.

So why has my nearly 1 yr stable 4.7Ghz overclock suddenly become unstable? Well I can't make that statement for sure, because previously I had never tested Prime95 and Firestrike together for any extended periods, so it's possible that it would have always failed that test. But, if a harsher testing regime is not the answer to that question then it could be: a) CPU degredation b) software like Firestrike or Windows changing with version number somehow affecting overclock stability. I think it's unlikely to be CPU degredation as my CPU was pretty much always under 60 degC during use, and I was 'only' using 1.4V, my dialed back overclock is using only 1.344V so at least I can now be ultra confident that degredation is gonna be a non-issue, especially as temps are even more chilly than they were. My intuition is that it's just a harsher testing regime that exposed an otherwise stable overclock.

I'll be staying on the slightly dialed back CPU overclock at 4.6Ghz @1.344V for the forseeable future.

 

Most likely never been fully stable. And Software + OS + drivers has been through version updates.

Enjoy what you have and forget about this talk about Cpu degradation. We all know the agenda on whats already said You have never run on Intels max allowed (Voltage) specs for their chips. Not even near. Same for temp!!

 

This came in yesterday along with the processor, now to wait for the RockitCool Copper IHS -

 

@bloodhawk Let me know how it is, I almost got that board but decided to go with the Taichi at the last moment because it was about $120 cheaper but still sometimes wonder if I shouldn't have gotten the Code instead.

 

3DMark 11: https://www.3dmark.com/3dm11/12706371 | http://hwbot.org/submission/3813462_

3DMark 11 Extreme: https://www.3dmark.com/3dm11/12706378 | http://hwbot.org/submission/3813464_

 

I would have gotten a Hero as a middle ground tbh. even the Prime Z370-A is great board if you wont be pushing the overclocks like me or Fox would. The VRM's are definitely better on these boards, but they are pointless for moderate overclocks. (<5Ghz if that is the case)

But from the looks of the it the Taichi might be right up there, i just haven't looked into it much.

I wanted to get the Apex, but i know in a few months i will be jumping to the EVGA X299 Dark as it is, so kinda settled for the Code, which is pretty much the Formula minus the water block on the VRM's.

Will keep you posted though, my IHS should be here tomorrow.

 

It's doing well enough so far, it looks like I'll be running 5.3Ghz at 1.33v and a -2 AVX offset for my 24/7 and that's just with a 280mm AIO and not a custom loop.

EDIT:

And the Rockit IHS is worth the money, I've got one on my chip and it's been great thus far.
I would use a floating IHS and not reseal it for optimal results.

 

This new ISP is so much better than my old one.

 

Nice! Mine came in today. Ill test out what my chip can do tomorrow night after work. And then delid.

 

Ha, and I thought my broadband was fast at 39Mbps download - but I am in the UK!

 

Your broadband is still faster than what most people in the USA have. There are only three choices where I live, and one of them cannot be taken seriously because their best offering is utter garbage (CenturyLink) with a max of 20 Mbps and horrible in other ways. The other option (the ISP I left behind - Cox) has respectable speeds, but their pricing is retarded with a mickey mouse 1TB per month data limit and punitive fees for overages. They charge more just for the add-on of unlimited data than what the crappy 20 Mbps provider (CenturyLink) charges for their crappy service and slow broadband. I was happy with Cox until they started shafting me every month for data overages. They only recently started molesting their customers' posterior orifices and I go over 1TB of data use usually within the first 8 or 10 days of each billing cycle. I'm not paying those greedy rump rangers at Cox Communications $150 per month for 200 Mbps due to their unrealistically low data limit surcharges.

 

Wow, I thought the USA had faster download speeds, we're all in the dark ages! Maybe it's mainland Europe with better broadband speeds then, I'm always hearing on the news that the UK is behind the times when it comes to broadband speeds.

 

I am with Optimum online and they are a 200 mbps provider. Fortunately no data limit here. The only other option is Verizon but there is no Fios to my address.

 

There are a few large metro areas with Gigabit service, but not that many. Some people actually opt for the slow broadband because it is so much cheaper and they pay a lot for cable TV programming and premium channels. I do not even subscribe to cable TV and only do streaming like Amazon, Netflix and Hulu. I have about 30 digital OTA stations I pick up for free, but I only watch 2 stations because it is mostly garbage programming like cable. Free garbage programming is fine. I just ignore everything i do not like. Paying out the nose to have a higher volume of similar garbage piped into my home over coaxial cable seems kind of stupid.

 

I have been wanting and waiting for FIOS for years, but they do not seem to be growing. Now that I moved to the Phoenix area they do not even offer anything at all except cellular service, so I will likely never see it. I had never even heard of data limits on residential broadband until I moved here. I thought that was limited to cellular providers. Of course when one does it, the rest of the greedy bastards do the same. It is a highly uncompetitive market area here. It almost seems like some sort of collusion or conspiracy to keep broadband prices as high and unaffordable as possible. I was paying Cox 200% more more in the Phoenix area for 150-200 Mbps with 1TB limit than what I paid Charter Communications in Washington State for the exactly the same speeds with unlimited data.

 

Some happiness from Papusan's Camp. I have struggled with awful slow internet speed in over 1.5 decade now. A few days ago I participated in the final meeting with the state municipality + broadband provider + neighbours regarding new internet. And finally I could signed an agreement on normal broadband speed where I live. We have been working on this Internet project for over 4 years. Now finally there was a solution we all could accept. Monopoly and greed are a dangerous combination. Dangerous also separately.

Today's best speed I can get is 0.39Mbps up + 5.2Mbps down, if the old Copper line work as intended. And this amazing <fast> speed is shared on +5 phones, +5 computers, +3 tv's and +++ People who haven't tried it, they will never undertand how I could survive this.

The rollout of new fiber line starts after Easter and will be finished to the summer holidays. Up to 5.2Mbps divided into 5 people has been a complicated affair task for me. Papa aka me comes last as always when all want sit on internet home. If this new 500/500 speed is not fast enough, I can upgrade to 1000/1000. But who knows, maybe they (home) manage to cripple this speed as well, LOOL

 

I dont consider Boise, Idaho a major metropolitan, but Gigabit internet is provided in my area. At $175 per month, I cant justify switching to it. Below is on my wireless card, and not ethernet.

 

Salt Lake City has Gigabit as well. Only way it would be worth $175 is if 3 families were sharing it and splitting the cost 3 ways. Otherwise, the price is about $100 more than what it is worth per month.

 

Here is Cincinnati Bell Fioptics service.
Intel(R) Dual Band Wireless-AC 8265 #2
High ping rate at 12 but I have about 4 walls between me and the router.


Here is the results not using WiFi...Ping rate still sucks though, maybe because I'm using about 40 feet worth of ethernet cable?

 

I feel like I just cannot win with this new build. Sat on Vue for a couple months until it was proven how awful it is, so I ate the $50 cost of my 2 bottles and decided I would just go with a clear coolant now. Finally ordered all my tubing and fittings and final things I needed and am about ready to build. Order my clear coolant and I take a look at it yesterday and it has this....stringy milky film like substance I noticed sticking to the bottom of the bottle. It initially disappeared when I shook the bottle yesterday but today it is forming again and now I do not trust putting this in my machine.

https://imgur.com/NV8ZQ7F

I've been working on this machine and getting parts and planning it out since October and between coolant issues and ridiculous hardware prices of computer hardware right now, I am right about at my wits end and sick of all of this ********. I just need a desktop up and running again, as great as my laptop is using it every day without a cooling pad is getting annoying, especially taking it all down and packing in my backpack for work, then coming home and setting it all up again...I need a computer. Now I think I need to go order more coolant before I can do anything. $2000 project has turned into a significantly more expensive ordeal. I wish I just went with an air cooled rig and AIO water cooler. This is not worth the headache. And all of this is for what? Maybe slightly better temperatures and better aesthetics? Which will also cost me more work and headaches in the long run dealing with maintenance on a custom loop and draining and refilling it etc etc? Honestly, what a waste of time and money. I do not think I will ever go this route on a computer again. I don't care how rewarding it is when I could have had a computer up and running 4 months ago.

 

I am going to be testing the Indigo product one last time to see if I can get it to flow correctly on the desktop. If it does, then I will try it again on the P870DM-G. Hopefully, it works well. If not, then the 3 strikes rule applies.


Big shout to the fine folks at Performance-PCs.com. When I ordered my XSPC cooling kit I had asked about promo codes and there were not any. About 24 hours later (next day) a 10% off code for St. Patrick's Day came out. Duke gave me in-store credit to take care of it and I used that to buy the Indigo XS and replenish my supply of Phobya NanoGrease Extreme.

This is what I used for coolant. Works great and no hassles of any kind. And, I see nothing unusual in my system. Just clean and clear coolant. XSPC EC6 Non Conductive Coolant - Clear

 

Nice speeds, bro. Part of the issue might be the Intel WiFi card on wireless. Killer 1535 AC is a better product and I do not see the slowdowns with them as I do with my Intel 8265. Maybe grab one on eBay and swap it out to see if it improves. Just use driver-only INF to install. Do not use the Killer Suite software. https://www.ebay.com/itm/Killer-Wir...d-AC-1535-802-11AC-Bluetooth-4-1/202033553600

 

We can use this for comparison if the Indigo XS works properly. This is with Kryosnaut on the water block. All 13 120mm chassis fans are set to "Auto" so they are pretty quiet.

 

Wow, haven't been as up to date with this thread and then went on vacation for a week, where the hell have I been. When did you make that change?

 

A week or two ago.

 

OK, results are OK... minor improvement. Hope it works well on the P870DM-G.

 

Ooops. I need to fix it. I ran the AFTER at 4.7GHz by mistake. Forgot I had set the BIOS to stock for burn-in, LOL.

 

Haha, I was just about to mention that. A 10C drop on the hottest core seemed a little too good to be true.

 

LOL, yes it was too good to be true. Fixed it. Still a minor improvement. About 3-4°C better than Kryosnaut.

 

Well that's an improvement, a small'ish improvement, but it all adds up! So, this Indigo XS ETI, I'm not sure what it is (even after I've been on their website!), is it a metal pad that sits between IHS & CPU cooler, and then it phase changes & flows into the micro pores of CPU and cooler after applying CPU heat?

 

Hmm.. think i have a decent clocker, haven't even delidded it yet. -

https://valid.x86.fr/deupqj



@Mr. Fox

Quick bench, im going to run more once i get a minimal windows 10 installed. -



8700k vs voltage is weeeird -

 

Voltage weird?

5.3Ghz is good, nothing to sniff at!

 

I believe @bloodhawk is talking about the voltage readings. With SVID disabled, the VID and CPU Package Power sensors are also disabled.

 

Very nice results. Yes, you have a good CPU sample for sure.

I will have to check and see if the ugly ASUS ROG version of CPU-Z shows my voltage properly. The regular CPU-Z never shows it correctly. It is generally like half the actual value. I have never liked the aesthetic choices of MSI or ROG skinned software, so I avoid using them.

 

Nope. Same as ordinary CPU-Z. My voltage is still shown incorrectly. This new ROG skin is not nearly as tacking looking as it used to be.

 

haha kinda decent

Mainly i was referring to what @Mr. Fox said under, and the fact that im getting 5.2Ghz @ 1.3V (i actually managed to get it stable at 1.255V), scores super close to 5.3Ghz at a very crude and much higher voltage. I still need to do a proper install of Windows for benching.

Yep!

Its pretty decent, i need to try and see if i can at least validate @ 5.4Ghz. Temps are amazing at 5.2Ghz (Ring - 4.8Ghz) . Maxes out at 60C, using the RockitItCool Copper IHS.

Yeah its kinda better, but weirdly enough with SVID disabled, the ROG skinned version reports the voltage accurately, but all other applications cant.

 

I am not using the latest BIOS because I do not want the Spectre and Meltdown fixes, so they may have addressed that to the extent you are seeing using the ROG version of CPU-Z. Security always takes a back seat to performance for me. I like how my system runs better without the patched crap, so I will likely never update the BIOS at this point. For the same reason, I do not allow Micro$lop's cancer updates to infect my OS.