September 8, 20169 yr Author Well after overclocking to 4.6 I ran prime95 and it failed a lot so I've just gone back to stock for now. I wouldn't use Prime to be honest. If you do, use the blend test. If you let it run AVX it will generate tremendous heat. And lets face it, almost none of us run AVX in everyday use. Prime stopped being a useful stress test a few iterations back in my opinion. I would highly recommend RealBench. It's more akin to how we all use our PC's in everyday use, rather than a synthetic stress test. There's two ways you can define stability. 1) Totally and utterly stable no matter what you throw at it. Which will of course limit your max OC. 2) Totally stable for all of the applications we normally run on our PC's. The later is obviously the sensible approach. Try 4.6 @ 1.35 volts, in manual mode. Most 6700K's are comfortable with this. Then run RealBench for 30 minutes. If it passes tweak voltage down and run RealBench again, if it fails tweak up. When you get to the lowest stable voltage in RealBench, try RealBench for one hour. Then if you want to you can set an adaptive voltage. Whatever the voltage is you end up with, add .01 - .02 volts so you have a safety margin. In like to run Aida 64 as well, usually without FPU, so no AVX. Having said that, temps and stability are fine for my CPU with FPU as well. My 4790k runs at 4.5ghz at 1.15v lol I would never push it pas 1.3v. 1.45v is insane and will degrade the chip quickly. Nonsense! :smile: The mistake you are making is to compare two different CPU architectures. Intel and Asus have both confirmed that 1.45 is a safe voltage to use when overclocking SKYLAKE, the recommended upper limit. Intel make the chip, they should know. They wouldn't recommend 1.45 as a max if it was likely to be a catalyst for a thermonuclear explosion. If anyone is bothered about the voltage they are putting through their chips when overclocking, purchase Intel's Performance Tuning Plan, for a very small sum Intel will insure your CPU against overclocking related failures. Damage your chip overclocking and they'll send you a new one. We all have our own comfort level when it comes to the voltage we are prepared to put through our CPU's, for me it's around 1.42 for Skylake. But that comfort level isn't factually based, it's a "feeling" that it's high, and feelings tell us nothing about the nature of reality.
September 8, 20169 yr What's occt? It's a program similar to prime 95. Skylake run much cooler than Haswell with high vcore, no voltage regulators in the chip If you do the Aïda FPU test you have lower temps with a Skylake at 1.45 then 1.32 on a Haswell. On Extreme cooling LN2 max safe vcore on 4770k or 4790k is 1.75-1.8v over that r.i.p compared to 1.9-1.95v on a 6700k I see the same on AIO cooling you can run the Skylake with 0.1-0.15v higher vcore with similar temps. If Intel say max 1.5v with proper cooling and most of the vendors let it auto clock close to 1.45 sure Its ok if you have 80-90c on the CPU Anything above 1.5 will kill or make your 4790k very unstable in less than a year according to oc'ing experts. 1.75 is going to ruin a chip due to voltage degradation you could cool it with liquid nitrogen and you'd still get degridation,as past a certain voltage the current itself has a negative effect on the cpu ATP MEL,CFI,CFII,MEI. Type Ratings B-737, ERJ-190,ERJ-170
September 8, 20169 yr Well I think il try 4.6 at 1.30 as it seemed to be ok yesterday and to be honest I didn't exactly test it first, I whacked on fsx for an hour or so and it ran really well, temps seemed to be ok. Have you got a direct link for realbench, seems to be a lot of programs offering these test programs but never know of they are riddled with rubbish lol. One thing I did notice on the gigabyte was when I changed the memory to enhanced performance it wouldn't load. Any ideas why that would be? Not an issue as I've got 32gb but just interested why?
September 8, 20169 yr It's a program similar to prime 95. Anything above 1.5 will kill or make your 4790k very unstable in less than a year according to oc'ing experts. 1.75 is going to ruin a chip due to voltage degradation you could cool it with liquid nitrogen and you'd still get degridation,as past a certain voltage the current itself has a negative effect on the cpu Extreme OC'ing for more then 15years with LN2 liquid nitrogen both on CPU and gpu . Here is a cpuz validiation i did a couple of years ago with a 4790k , it work today in a Gaming PC http://valid.canardpc.com/hbpt3u And with a Amd http://valid.canardpc.com/tbstxh http://
September 8, 20169 yr Guys, be sure to check here: http://www.overclock.net/t/1570313/skylake-overclocking-guide-with-statistics Some good stats and figures on what to expect. They seem to agree that Prime95 is overkill and that XTU's stress test isn't enough. So try ROG and their custom x264 tests. I think I lost the silicon lottery: 4.6Ghz @ 1.376v (as measured in CPU-Z). And that's right at the beginning of the exponential curve: 4.7 requires about 1.44v and no telling what 4.8Ghz needs. Temps in the mid 80s on 4.6 so I'm about to pop the lid off. I tend to think that, even if you're below Intel's recommended max (which is probably a conservative figure to start with), the less voltage, the better for the life of the chip. Granted I could do 4.7 at 1.44v, I'd rather forgo that 100mhz and pump far less voltage through the chip. Just sayin' Garrett Frank
September 9, 20169 yr Author Well I think il try 4.6 at 1.30 as it seemed to be ok yesterday and to be honest I didn't exactly test it first, I whacked on fsx for an hour or so and it ran really well, temps seemed to be ok. Have you got a direct link for realbench, seems to be a lot of programs offering these test programs but never know of they are riddled with rubbish lol. One thing I did notice on the gigabyte was when I changed the memory to enhanced performance it wouldn't load. Any ideas why that would be? Not an issue as I've got 32gb but just interested why? Well I think il try 4.6 at 1.30 as it seemed to be ok yesterday and to be honest I didn't exactly test it first, I whacked on fsx for an hour or so and it ran really well, temps seemed to be ok. You can't be lazy with overclocking. :smile: "Didn't test it" and "seemed to be okay" is setting yourself up for trouble. Your 1.30 Volts could be borderline stable and about to fail, stable, or more voltage than you need, you have no idea. Have you got a direct link for realbench, seems to be a lot of programs offering these test programs but never know of they are riddled with rubbish lol. The following is a great guide form Asus, simple and effective. For Asus motherboards, but the principle is the same regardless of motherboard make. I would suggest reading it and then applying the simple strategy to your overclocking endeavours. I explained the steps you should be following earlier. https://rog.asus.com/19262015/overclocking/guide-overclocking-core-i7-6700k-on-the-maximus-viii-extreme/ RealBench... http://rog.asus.com/19932016/news/realbench-v2-43-new-version-available-now/ RealBench is a benchmark and stress test utility that uses open source applications and simple scripting to simulate "real world" performance of a PC system. Pay attention to "real world"! That's real world, as opposed to a synthetic stress test like Prime95 or OCCT, that bare no resemblance to how we run our PC's in real world use. I think I lost the silicon lottery: 4.6Ghz @ 1.376v (as measured in CPU-Z). And that's right at the beginning of the exponential curve: 4.7 requires about 1.44v and no telling what 4.8Ghz needs. Temps in the mid 80s on 4.6 so I'm about to pop the lid off. 1.376 volts isn't that bad to be honest, I wouldn't say you "lost the silicone lottery" exactly. I've seen examples where individuals need greater than 1.4 to be stable at 4.6 Ghz. Above average voltage but not horrendous. 1.34 volts is stable for me @ 4.6 GHz. But then it makes great sense not to set the minimal stable voltage, a margin of error is required, between .01 and .02 volts. So for me, I set 1.36 volts. 1.35 volts would also be adequate for me. So I guess I pretty much have an average chip in regard to voltage. Temps in the mid 80s on 4.6 so I'm about to pop the lid off. Mid 80's... running what, the sim, stress test? If you are running a fairly aggressive stress test, esecially with AVX, then mid 80's is a non issue. Which cooler? I'm at 75 degrees in RealBench Ambient 25C... warm in the UK lately. So 50 Delta. Would be mid 80's in Aida64 with FPU.
September 9, 20169 yr Author the less voltage, the better for the life of the chip. Absolutely... degradation clearly increases with increased voltage. I wouldn't be surprised if it was a linear relationship. A CPU even degrades slowly at stock voltage, and indeed even slower undervolted. But what we have to consider is normal expectations for a chip, expected lifespan, how long an enthusiast would expect his CPU to last before performance degrades beyond an acceptable level. Given that CPU lifespan is measured in years, and that we enthusiasts never keep our CPU's very long term, applying increased voltage through our chips up to Intel's recommended max is unlikely to result in unacceptable performance loss within the time frame we ordinarily keep or CPU's. In fact, I've seen estimated lifespans for CPU's powered up 24/7 up to 20-30 years! May be an exaggeration of course, but you get the picture. Intel give a more conservative estimate of 6 years plus for a heavily stressed CPU on 24/7. Intel and AMD determine this, not by waiting years, but by subjecting them to higher load, more voltage, higher frequencies, more heat. The variables and and failure data are then extrapolated and calculated in reverse for typical use. This is why I raise an eyebrow when enthusiasts throw their arms in the air in a panic when they hear of increased voltage... sensible voltage increases within Intel's recommended max aren't an issue. The final point of course is that Intel provide overclocking insurance for their CPU's, and overclocking necessitates higher voltage. Intel wouldn't do this if higher voltage, up to their suggested max, was destined to severely degrade your CPU prematurely.
September 14, 20169 yr Im going 1.36vcore on the 6700K @ 4.6ghz never goes above 75C with the H110i Water Cooler, P3D runs like a dream. Ron Hamilton "95% is half the truth, but most of it is lies, but if you read half of what is written, you'll be okay." __ Honey Boo Boo's Mom
September 14, 20169 yr I have a question for those using the 5WO: I have an ASUS Maximus Hero Alpha VIII (With a i6700k and Noctua 15). I cannot get my AI Suite to install properly which means I have to use the BIOS to OC. Does anyone have download link of AI suite 3 or stand alone 5WO? The official AI Suite 3 gives me constant errors and not all of the utilities work. So far I've been able to get a stable 4.6 GHz OC but I'd like to try ASUS' 5WO versus the BIOS method. Thanks. "I am the Master of the Fist!" -Akuma
September 14, 20169 yr Author Im going 1.36vcore on the 6700K @ 4.6ghz never goes above 75C with the H110i Water Cooler, P3D runs like a dream. Greetings Ron. Was that 75C in RealBench? And what was your ambient, rough guess will do? What about H110 fan speed, full blast or PWM? The reason I'm curious is that I use the same voltage, same frequency and get same CPU temp, 75C in RealBench... I'm using the NH-D15S though. Given that the H110 is supposed to be somewhat cooler, I'm surprised we get the same temp. I have a question for those using the 5WO: I have an ASUS Maximus Hero Alpha VIII (With a i6700k and Noctua 15). I cannot get my AI Suite to install properly which means I have to use the BIOS to OC. Does anyone have download link of AI suite 3 or stand alone 5WO? The official AI Suite 3 gives me constant errors and not all of the utilities work. So far I've been able to get a stable 4.6 GHz OC but I'd like to try ASUS' 5WO versus the BIOS method. Thanks. There is no stand alone 5WO, it's part of AiSuite and is specific to your motherboard. Thus, the place you download it is from the Asus support site. It's in your motherboards utility section. Didn't know which OS you are using so couldn't give you a direct link, but latest version available here... https://www.asus.com/UK/ROG-Republic-Of-Gamers/ROG-MAXIMUS-VIII-HERO-ALPHA/HelpDesk_Download/ If you are having issues, I would install the latest BIOS for your board, and also completely uninstall AiSuite and install the latest version from the support site. If you are using the version that came on the disc, and/or the bios your board shipped with, it would explain why you are having issues. AiSuite can be tricky to uninstall, so you may wish to use the Asaus AiSuite uninstall utility. It gets rid of all the left over files that can cause issues... https://rog.asus.com/forum/showthread.php?34503-AI-Suite-III-Cleaner So far I've been able to get a stable 4.6 GHz OC but I'd like to try ASUS' 5WO versus the BIOS method. As per the first post in this thread, I found it worked great, but did indeed nudge the voltage up higher than I required. Be interesting to see what it does for you.
September 14, 20169 yr Hi Martin! Thank you for your help. I have been able to install the AI Suite 3 (this time without errors). Ran the cleaner first then installed the software. Right now, I'm showing 4.6GHz with a voltage of 1.280 up to 1.4 Is it worth it to try the 5 WO? "I am the Master of the Fist!" -Akuma
September 15, 20169 yr Author Hi Martin! Thank you for your help. I have been able to install the AI Suite 3 (this time without errors). Ran the cleaner first then installed the software. Right now, I'm showing 4.6GHz with a voltage of 1.280 up to 1.4 Is it worth it to try the 5 WO? My impressions using 5WO are below... bear in mind though that my impressions of 5WO are prior to my manual overclocking endeavours which were superior in terms of voltage and temp. The settings I used in 5WO and why, are included, so hopefully that will be useful. http://www.avsim.com/topic/494147-asus-five-way-optimisation-impressions/ It's not going to hurt to give it a try, if you do though, save your present manual overclock as a profile so you can easily switch back. For me, it nudged the voltage higher than my manual overclock. Not dangerously so, 5WO is designed not to exceed a safe voltage, but higher than I needed non the less. Right now, I'm showing 4.6GHz with a voltage of 1.280 up to 1.4 In 5WO I saw about 1.41 max voltage, adaptive voltage. My manual overclocking doesn't exceed 1.36 and thus the temps are lower. I'm not sure what process you used to configure your manual overclock, but I assume that's an adaptive voltage. Some do need up to 1.4 volts of course for 4.6 GHz, but I wouldn't be surprised if that was a bit higher than you need. If I were you I'd have a look at the Asus manual overclock guide, very simple and effective. Stress testing with RealBench, which is pretty much all you need. https://rog.asus.com/19262015/overclocking/guide-overclocking-core-i7-6700k-on-the-maximus-viii-extreme/
September 29, 20169 yr My system runs stable at 4.5 with 1.3 volts (manual voltage). Max temps 73 in Realbench with NH-D15S and ambient temperature 25.
September 30, 20169 yr On my sim and allday PC i do manual overclocking yes a can do 5way as base for the manual OC. First thing a set 1.4v manual and boot in to windows start at 4.8 then up to the point it want boot. If it make 4.9ghz a good start point is 200-300mhz lower 4.6-4.7ghz Then a set adaptive in bios and dial that to the voltage a feel comfortable with. I prefer a voltage approx 0.02v less then the manual voltage , reason when you have SVID disabled in manual mode you not get the 0.025v boost when CPU feel a avx instruktion, Its sometihing you cant disable Intel thing ( manual voltage disable Svid and c-states do it) Broadwell-E have a avx clockdown , Its clockdown instead of raise the vcore the new 7700k gone a have a similar thing. The all Manual is more for Overclocking with exotisc cooling ( but some use adaptive) Its no difference in OC potential manual vs adaptive if it can do 4.8 manual it do the same with adaptive only more timeconsuming to dial in the adaptive settings . If you say the CPU do 4.8ghz with manual and only 4.6-4.7 with adaptive , you have missed something. LLC a prefer 5 or 6 on Asus boards. Memory overclocking , on a Skylake 3200mhz c14-15 is a good base in P3D each 100mhz do approx 1% and lower the cl from 16-16-16 to 15-15-15 approx 1.7-1.8% higher fps. 3200 c15 - performs as a 3400 C16 kit but if you go for a 2666 c15 kit you lose 5% or most compensate with a 200mhz OC a 4.5ghz with 3200 c15 performs as a 4.7ghz with 2666 c15. If you run a Skylake with Default 4.2 and 2133 jedec settings compared to a proper Overclocked 6700k with fast and tight mems 3600 c15 - c14 you can gain up to 30% higher fps. http://
September 30, 20169 yr Author My system runs stable at 4.5 with 1.3 volts (manual voltage). Max temps 73 in Realbench with NH-D15S and ambient temperature 25. 1.34 for me. Similar temps, also with the D15S. Great cooler isn't it. :smile: On my sim and allday PC i do manual overclocking yes a can do 5way as base for the manual OC. First thing a set 1.4v manual and boot in to windows start at 4.8 then up to the point it want boot. If it make 4.9ghz a good start point is 200-300mhz lower 4.6-4.7ghz Then a set adaptive in bios and dial that to the voltage a feel comfortable with. I prefer a voltage approx 0.02v less then the manual voltage , reason when you have SVID disabled in manual mode you not get the 0.025v boost when CPU feel a avx instruktion, Its sometihing you cant disable Intel thing ( manual voltage disable Svid and c-states do it) Broadwell-E have a avx clockdown , Its clockdown instead of raise the vcore the new 7700k gone a have a similar thing. The all Manual is more for Overclocking with exotisc cooling ( but some use adaptive) Its no difference in OC potential manual vs adaptive if it can do 4.8 manual it do the same with adaptive only more timeconsuming to dial in the adaptive settings . If you say the CPU do 4.8ghz with manual and only 4.6-4.7 with adaptive , you have missed something. LLC a prefer 5 or 6 on Asus boards. Memory overclocking , on a Skylake 3200mhz c14-15 is a good base in P3D each 100mhz do approx 1% and lower the cl from 16-16-16 to 15-15-15 approx 1.7-1.8% higher fps. 3200 c15 - performs as a 3400 C16 kit but if you go for a 2666 c15 kit you lose 5% or most compensate with a 200mhz OC a 4.5ghz with 3200 c15 performs as a 4.7ghz with 2666 c15. If you run a Skylake with Default 4.2 and 2133 jedec settings compared to a proper Overclocked 6700k with fast and tight mems 3600 c15 - c14 you can gain up to 30% higher fps. Great info as usual Hasse. :smile: Just wondered what kind of enclosures you are using for your rigs? I would imagine you are using some kind of test bench arrangement. Have to say I am loving the Lian Li X510, really glad I spent big money on it. Great case. Not perfect, the bottom vents are very restrictive, so i modded it, enabling me to turn the PSU fan up. Also the clips that attach the glass side panel tend to pop off. Not an issue they push back in but a bit annoying. An issue I'm having at the moment is with Fan Xpert 3. It's ignoring my fan profile when I reboot. Not every time, about once every three reboots. It reverts to something like the Fan Xpert Turbo profile. A bit annoying.
Archived
This topic is now archived and is closed to further replies.