|
Picked up a new 3930K setup on a P9X79 Deluxe. The whole world of Overclocking has changed again since I built a 2600K setup. I have so far gotten a stable 4.7Ghz with a +.005 offset but even with my H100 and a full Push/Pull setup, she shows 1.424vcore in HWinfo64 under full load and taps 80C with a core or two. Anything like auto (.000 = .005) or a negative offset throws a 101 BSOD. Gaming/working it hasn't risen above 68-72C for any extended period of time nor gone above 1.38vcore, but Prime95 really likes to push voltage and heat through the roof. OCCT doesn't even push it as hot but within OCCT itself it shows 1.44vcore so I don't know what to belive as far as Software voltages go. Anyone else notice how HWInfo64 seems to run like crap (pauses as it updates the sensors every 5sec or so) on a SB-E? I have a lot of tinkering left to do with it but right now I am still testing stability at 4.7 as it seems to be the sweet spot. Same voltage results from 4.5-4.7 which was kinda odd. I was hoping to push a Turbo setup of 5Ghz on 1 core and like 4.9 2 cores, 4.8 4 cores, 4.7 6 cores or something along those lines, but haven't had any luck with that method. It was so much easier to get a 5Ghz 1 core, 4.5Ghz base on the 2600K with like 1.32V max.
|
# ¿ Apr 24, 2012 19:48 |
|
|
# ¿ May 10, 2024 13:24 |
|
Factory Factory posted:More than likely a 1.38Vcore uses a VID around that 1.44V mark, hence the confusion. Now I have been meaning to ask about that. Is the VID what I belive shows above the CPU load within the sensor panel of HWInfo? If so, it usually shows a 1.357 max (or close to it) number with the Vcore being the one that goes above that. As I am not at my comp right now I cannot fully see what I am talking about. Also with the turbo method, I was able to do it great on a 2600K with a lower vcore requrirement. (I cannot fully remember, but I think it was something pretty drat low like 1.29 or 1.32vcore max) and the voltage didn't have to increase from the base. My theory was that since most apps/games at the time didn't use 4 cores, that the freq could bump up with i7's turbo for the programs that use less cores. If a app can use all 4 then hell 4.5 should be enough. But a single threaded app using 5Ghz should also give it a pretty good push and since it is only 1 core being used to max. That system has been running like that for over a year now and is 24/7 stable with the guy playing BF3 pretty much daily.
|
# ¿ Apr 24, 2012 20:12 |
|
Factory Factory posted:
Ill post a pic tonight showing what I am talking about the Vid. I see the one you are pointing out but I get it showing I believe in the sensor page above the CPU load part. I have never seen it before I got a SB-E so it may be something new. I did discover that using an old version of HWinfo instantly rebooted the machine upon launching it lol. When the entire sensor page is shown, HWinfo uses 8% cpu usage constant. I understand that about running them all at 5Ghz, but I also figured that the Vcore needed may be less to push 1core to 5 vs 4cores to 5. I know it would technically require the same vcore, but I required a lot more vcore to get them all to 5 vs just 1. Also and considering how much more Vcore is usually required to push these 6cores up to 5Ghz, it may still be the "coolest" way to get there in my case. The 2600K was partially limited due to heat. IT was built with a Antec Kuler 920 but being it was going to somewhere in Texas, it needed to be able to stay stable for this user while gaming in BF3 with SLI 480's. (The 480's were OC models I had to downclock to stock 480 speeds to keep them from causing crashes.) EdEddnEddy fucked around with this message at 21:05 on Apr 24, 2012 |
# ¿ Apr 24, 2012 21:01 |
|
Ya the H100 is a badass little kit for the money. I also have a brand new H80 I still need to sell (got the 500R free so decided to get the H100 to match). Also is it just because it is not the "Overclocker" board or something because when I do Manual Vcore in my P9X79D, I can set it to something like 1.325 and under full load, it will hit almost .15V more under full load. I am still reading up on CStates and such seeing if I can pinpoint this to some setting I am missing, but why in the world would there be this big of a difference from Set Voltage to Indicated Voltage under load? I remember the old days of Vdroop with the P5K going in the other direction about this much.
|
# ¿ Apr 24, 2012 22:47 |
|
Factory Factory posted:Vdroop still exists. If you've turned on Load-Line Calibration and cranked it up, you're overcompensating for Vdroop. Ya I am now figuring out more how the LLC works on this board. It acts almost like offset + .020v or so each setting. I pulled it back to Regular since I don't like it shooting up .030v above what I am telling it to do on its own. (The Temps were getting a bit too high for my taste). Either way I am pulling back to 4.5Ghz since at 4.7, (shooting for a Intel Burn Test stable) I could not get it to stop throwing a vCore BSOD even with 1.440v thrown threw it. Any more and the temps on 1 core hit close to 90C. I have a ton of air pushing through that H100 and repasted it again last night, but I guess I still have some airflow and other tweeks to work out. Though I think I rediscovered my old Turbo method I was talking about. I may have use for that yet.. Also at 4.5Ghz it only pulls 1.408-1.416v and so far it is a hell of alot more stable then it was. :/ (still testing, but 4.7 never got this far)
|
# ¿ Apr 25, 2012 03:42 |
|
Yes but there is always the quest for more... I think she was stable at 4.7 with my higher LLC, but the temps were not nice for an ibt. The ambient might also have something to do with it but I may just keep 4.5 for the summer. It ran for about 8 extreme runs and didn't leave the 70s too much. One thing I noticed, unlike SB boards, I cannot lower my PLLv below 1.8 where SB users are finding results with 1.7ish.
|
# ¿ Apr 25, 2012 15:19 |
|
I will say that this Offset voltage for SB(E) is sort of a royal pain in the rear end. Finding out where the CPU is unstable is sort of a crapshoot atm since it can be 100% Stable under a good constant load under a stress test, but mixing some data transfer + some random game or 3D accelerated program and the comp will BSOD quickly (like too short almost to get a glance if it is a 124 or 101 bsod) before rebooting. Little random stuff like this really makes me chuckle when someone says their comp is 100% stable yet it will get random BSOD on the OC forums. There is always that off workload that can bring out the worst. Anyway tinkering with it as I really want the ability for it to downclock under idle. Lowering LLC to High and upping the Offset to +.060 gets me a bit higher idle vcore, but a lower upper vcore of a 1.448v max which eases a bit on the temps and overall I think the bump in voltage at the lower steps might just be where my problem lies. I will still have to test the heck out of it to see if everything in the VTT/VCCP/etc is dialed in. On to IB, it is interesting how the chips main problem seems to be heat. The heat spreader is over thermal paste vs being soldered on like most past chips which is a bit odd/a bummer. I am guessing there is some hold out hope that a later revision will be soldered again. While the chip does look good, If I was to build a system outside of my 6core choice, I would have still gone SB-E and gotten a 3820 to tide me over. For some reason IB and the motherboard tech really does nothing to appeal to me over the X79 offering for the price difference. :/
|
# ¿ Apr 30, 2012 16:34 |
|
The biggest thing I saw was messing with the ASUS LLC options vs the Offset. It seems that the LLC pushes the Offset up a certain level at the higher clocks that can really mess with what you were expecting to what you are getting. With LLC on High, my 4.7Ghz OC +.060v Offset = 1.448 Full Load so far. With LLC on Very High, my 4.7Ghz OC +.020v Offset = 1.452 Full Load. So you have to tinker with it to see what you want to get. I think my upper voltage is fine with my High +.060 setting as I think I was crashing from somewhere below full load on all the cores where the +.020 offset was not giving it enough juice, and LLC doesn't work as much below full load.
|
# ¿ Apr 30, 2012 18:43 |
|
Star War Sex Parrot posted:As of right now there's no way to set strict clocks, no. At the Nvidia Launch Event I attended, it was kinda cool seeing one of the head host overclock the 680 to something like 2Ghz and still play BF3 with it. It may not be 100% set, but it does overclock like mad from what we saw.
|
# ¿ May 1, 2012 00:13 |
|
Agreed posted:5000:1 odds against that being a very carefully hand-picked card, any takers, step right up Oh No doubt, but it was still impressive to see. And I have enjoyed Zotac cards. Their support for bios updates seems to be a crapshoot, but their lifetime support is pretty solid. Just wish they stayed with more reference cooler designs for some of the past chips. Now it seems their new coolers are what the reference is going to :/ Their motherboard are also pretty sweet if you are building something using ITX.
|
# ¿ May 1, 2012 05:08 |
|
tijag posted:I did OCCT with linpack for the cpu/ram stress, and I am stable on 4.6ghz at sub 70C temperatures with offset voltage at +.075v. It can have a lot to do with other things you can tweek outside of the vCore. Make sure you read up and learn about the other settigns (or at least see what others use) since the CPU does so much more now and other parts can cause issues like the Mem controler and such. Also ASUS is good but sometimes they can give me a laugh. On one of the ROG videos where they are overclocking my exact board (The P9X79 Deluxe) they mention Offset voltage and that you can start setting it to +.200v LOL.
|
# ¿ May 4, 2012 19:09 |
|
unpronounceable posted:With SNB/IVB, the whole platform is incredibly sensitive to the bclk. I don't remember the exact details, but things like the PCIe slots are running off of it, not just the CPU. So it's best to just not gently caress with it and deal with the small difference in overclock. Using Straps you can do some good stuff with both the Unlocked and non Unlocked processors, but +- 3 to 7 Mhz can net you some good gains without that much of a stability problem depending on the motherboard. You just have to watch out about doing to much or you can run into I/O errors and such over time. I'd say though that 3mhz either way isn't going to hurt a thing though. For some reason I have yet to prove on my side, a 125Mhz BCLKC strap is supposed to require less voltage than using the multiplier to reach the same OC, though I have yet to have much success with this method. It could be partly due to the fact that I refuse to run my ram below rated specs.
|
# ¿ May 9, 2012 18:11 |
|
Love my Corsair 500R. Looks clean, reasonably quiet stock fans, proper front USB3.0 port headers (can use USB 2.0 or 3.0 headers) and was built to be used with the H100. Though no case can help my ambient temps during the summer especially with these 6 cores + SLI 560's pushing out the heat they do. So much airflow coming out of the case makes the room pretty warm lol.
|
# ¿ May 15, 2012 18:58 |
|
|
# ¿ May 10, 2024 13:24 |
|
Agreed posted:I get such an irrational boner for that thing, even though its performance is extremely disappointing compared to boring old regular aluminum fins on the same model, and you probably ought to build some scaffolding to support the installation. It's just... so... beautiful This is true, I have the non Copper one and though it isn't copper, it still does look and perform drat good for how old it is. I was tempted to get a mounting kit for it to try on my SB-E Just to see how it would perform in a push/pull.
|
# ¿ May 15, 2012 21:48 |