X79 and/or rampage IV OC'ing thread for those of us still left....

[mouacyk]

That was really hard to comprehend. I would recommend buying a new CPU EPS cable and not re-use one that's partially damaged, if your PSU is modular. Some thing in the connection caused resistance or the cable gauge wasn't sufficient for the load and caused it to burn in the first place.

 

[dalekdukesboy]

Yes I tend to write run on sentences and stream of consciousness. I know what you're saying and I will get another PSU but load wasn't that high and PSU is rated for far higher than this setup needs so yeah or I can buy new cable for this thing but it was two different cables same result...anyway I have a laptop as a backup if things go awry again and I'm keeping an eye on it. I'm wondering if I kinked cables too much or did something to cause the issue.

 

[PolRoger]

And I was getting random shutdowns etc on last build near end before it wouldn't boot and now I've had increasing issues with this one but even at first every so often under stress it would just stop and restart. I just assumed it was reaction to a bad overclock though strange it has happened before on other builds just not so frequently.

Your PSU could be developing reliability issues.

I have multiple psu for different setups and two different ones... a 650w and a 750w (both out of warranty) still turn on and function but have become unreliable and cause running/overclocking issues. Not too long ago I went out and bought new psu to replace those two. Actually, just earlier this week I was clocking/testing on my X58 setup and was using my "suspect" 650w psu and it started acting up (again) with some hard shut downs... I went and swapped it out with an exact/identical 650W PSU and everything cleared right up. I'm probably going to go ahead and send those two faulty psu out to the local county e-recycler.

 

[mouacyk]

Depending on who the maker is, some PSU's have fairly long warranty periods - 7 to 10 years.

 

[dalekdukesboy]

Your PSU could be developing reliability issues.

I have multiple psu for different setups and two different ones... a 650w and a 750w (both out of warranty) still turn on and function but have become unreliable and cause running/overclocking issues. Not too long ago I went out and bought new psu to replace those two. Actually, just earlier this week I was clocking/testing on my X58 setup and was using my "suspect" 650w psu and it started acting up (again) with some hard shut downs... I went and swapped it out with an exact/identical 650W PSU and everything cleared right up. I'm probably going to go ahead and send those two faulty psu out to the local county e-recycler.

Thanks this is very helpful, I've never had an issue with a PSU except if they literally didn't turn on and just were DOA. Your situation sounds similar to mine, I'm just disappointed considering Rosewill is great company my case is a Rosewill as well as many fans and the PSU obviously. So I turned on overvoltage protection and surge protection both of which are on this motherboard so I feel fairly confident it won't break the board, hell the other board minus the melted connectors may work still for all I know.

However, like you as of today/last night I figured out the the PSU was the culprit one way or another or at least is part of problem. So now do I want a replacement from this company or do I just get totally new ones from another manufacturer and call it a day.

Depending on who the maker is, some PSU's have fairly long warranty periods - 7 to 10 years.

Yeah I believe this one has a warranty of 5 years, I just am not really anxious to replace it with another of same type that could also have issues....I'll look into how warranty works and take it from there I guess.

 

[dalekdukesboy]

That was really hard to comprehend. I would recommend buying a new CPU EPS cable and not re-use one that's partially damaged, if your PSU is modular. Some thing in the connection caused resistance or the cable gauge wasn't sufficient for the load and caused it to burn in the first place.

I did better than that and my truck was in the shop so I had a relatively nearby Best Buy ship me a Corsair PSU 1000 watt and I got it all put in today. However even after resetting bios etc system is doing what it did last night after that good stress run, which is it takes multiple tries to start and sometimes doesn't and I'm getting the familiar windows is broken messages often and sometimes it makes it to windows.

This is beyond frustrating. I'm sure the new PSU by Corsair is good. All I can surmise is the damage was largely done from whatever that Silverstone PSU POS did. I had it barely two years and now it appears it may have killed two boards. Regular rampage board was flakey but I assumed it was everything but the PSU.

I didn't think one big name brand psu rated for 1200 watts could be such a POS and it literally is first PSU I ever had to do anything wrong, except one that was DOA when I got it and it never started....which I wish this one never did.

So yeah, as of last night it resets and takes several times and eventually starts. Doing exact same thing as I said with new Corsair PSU...lol ironic now this is the most troubled platform I've ever had and only one I started my own thread on...I had x58 EVGA and sold it to someone and it was working long after I sold it, most of my core 2 duo systems worked till I stopped using them although my biostar board did die but it was after I gave it to my sons so I didn't witness it stop working. Just for giggles I'll throw the 6 core back in it and see what it does when board resets with new CPU.
Only other thing I'm thinking of trying is flashing bios and see if bios was corrupted just not feeling confident any of that stuff will make huge difference. One thing I can say is that PSU will be sledge hammered soon by me just for satisfaction.

 

[mouacyk]

Test CMOS battery for 3.3v and replace if < 3.2v. Mine did exactly what yours was doing and the battery tested at 2.98v. After swapping a 3.05v in, it booted right away. Also clear CMOS completely for good measure.

 

[dalekdukesboy]

Test CMOS battery for 3.3v and replace if < 3.2v. Mine did exactly what yours was doing and the battery tested at 2.98v. After swapping a 3.05v in, it booted right away. Also clear CMOS completely for good measure.

Interesting, it was doing same thing? Yeah it's awful now, I did that one good run with old PSU yesterday, got the new PSU overnighted and put in today after I was frustrated at it starting/restarting with 00 codes and rebooting several times and then eventually it booted. That funky sequence was what convinced me to just get the new PSU but literally it is acting exactly the same. ONLY thing that is different is once it is in windows it has not shut down but I also haven't had it on a long time just long enough to try to test it and troubleshoot it.

Yeah what mine does is it now starts boot with 00 code and starts/restarts multiple times each time getting further down boot sequence and eventually gets to Bios/windows then it boots up fully through Windows, but so far I get a single run stable on linx and that's it so it still seems shaky.

I have hit the clear CMOS and put in all my OC settings if that is what you mean by clearing CMOS completely. Also how do you "test" that battery? That is the round battery in the board yes? Now that I think of it, I should have a few of those batteries around from multiple boards, I've literally never replaced one that I can remember anyway. Usually something on system dies or gets outdated and/or replaced first.

Test CMOS battery for 3.3v and replace if < 3.2v. Mine did exactly what yours was doing and the battery tested at 2.98v. After swapping a 3.05v in, it booted right away. Also clear CMOS completely for good measure.

I admit I did consider that little round battery as a possible culprit but admittedly didn't give it too much credence as possibility. However I just took battery from the last Rampage I had and popped it in and it isn't perfect but it basically boots on one or two tries at most so appears battery is at least part of problem if not all of it.

Is only way to test that battery OUT of the system with a battery volt reader? I mean board has system measurements for 3.3 and 12 volt etc but I have little faith in their accuracy especially down to a few tenths of a volt which it sounds like in your experience makes difference of a system that barely starts and limps into Windows and one that boots perfectly...which obviously is a huge difference, especially if stability once in Windows was affected by that battery at >3 volts?

 

[mouacyk]

I used a digital multimeter. With hidden firmware in today's motherboards and cpus, no one knows what else uses that battery besides the rtc when computer is off.

 

[dalekdukesboy]

I used a digital multimeter. With hidden firmware in today's motherboards and cpus, no one knows what else uses that battery besides the rtc when computer is off.

Yes multimeter was term I was looking for just couldn't think of it but I've used them many times on just household batteries AA/C/D cells etc just to test back in the day to see if they were good enough to use in landline message machines and various household appliances, been a while since I've used one not sure I even have one now.

I tried an interesting experiment to totally bypass that battery...I took it out, lol. So I'm running the board with no battery, so as I understand it on total shut-off it won't store any settings but when you restart PC or go into bios and makes changes it still should work correct? I'm trying it that way to rule it out as the issue. Currently I'm on laptop because I'm doing stress tests and board hums along fine at stock settings even with all memory used.

However, any attempt to overclock in bios gets me those weird starts and restarts and it almost boots but either hangs on a couple codes in bios if I try hitting reset as it's booting OR if I do nothing it just gets close to booting but shuts off and restarts with 00 code just as the last couple codes before entering bios screen show up.

So kinda weird that at stock settings it's perfectly stable, it wasn't with old power supply last day I tried it. It would boot with OC and stock but begrudgingly and neither was really stable and I didn't try many times because it was obvious it just wasn't going to work and PSU was at least one issue I had.

So I'm typing this on laptop but it is on third run of all of 16 gigs of memory linx testing with cpu/memory at stock and every run stable. So at the least it appears board will work at stock settings if nothing else. I have at least 3 of those batteries here but all are from boards minimum 8 years old...so I'm just going to buy a bunch locally they also are used in a lot of other household things you'd not expect when you pry off the back of battery cover so they are cheap enough and I will find use for them.

Would NOT having battery in effect overclock? I'm just wondering if when you reset from stock to OC if battery is needed to store info even long enough to restart, either way was interesting to see what it'd do with no battery...oh and of course my time/clock is wrong now lol but that's easy enough to get reset.

Your PSU could be developing reliability issues.

I have multiple psu for different setups and two different ones... a 650w and a 750w (both out of warranty) still turn on and function but have become unreliable and cause running/overclocking issues. Not too long ago I went out and bought new psu to replace those two. Actually, just earlier this week I was clocking/testing on my X58 setup and was using my "suspect" 650w psu and it started acting up (again) with some hard shut downs... I went and swapped it out with an exact/identical 650W PSU and everything cleared right up. I'm probably going to go ahead and send those two faulty psu out to the local county e-recycler.

New PSU, new life. You called it. I also found my micron memory is either junk/and/or this board despises it. I was having same problems as I said earlier but at stock mode and micron modules at stock speed everything passed fine but any overclock like I used in past of memory/cpu wouldn't boot at all. I noticed this board gave fairly crappy results with micron modules even before and seemed finicky about them so swapped out the 2800 mhz Klevv's and this was result....

That one run with shit PSU using only 1.328 volts under load for 4.5 ghz wasn't a fluke, it's what this CPU is really capable of but that PSU and Micron modules were holding it back. This 4.6 ghz run required 1.375 volts under load, it passed a run or two with 1.36 volts so it may even work at 1.368 or something like that. Either way, this is much less voltage than I was using with that crappy PSU to get 4.5 ghz stable except for that last successful run of it's life. I thought this board was toast, but far from it....I also put in a new battery just to rule that out and did that before I swapped memory out so that could have been another problem I eliminated but I don't know didn't test old batteries.

Ran Cinebench again and almost broke 1700, I did once but I believe I had it at 4.7 or more and wasn't stable but this was 4.62 ghz linx stable and I got 1690 points so pretty happy with that result.

 

[dalekdukesboy]

A few more mhz 8 to be precise, simply to say it's 4.6 ghz and not 4.592

 

[johnspack]

Dammit.. I want the 1680 v1... I know it will do 4.7 stable....

 

[dalekdukesboy]

Dammit.. I want the 1680 v1... I know it will do 4.7 stable....

Just FYI I tried like hell to get it stable at exactly 4.7 ghz and with 125 multiplier pushing bclk or using 100 multiplier I couldn't get it to do 4.7 I tried that for a while before playing around with 4.6 or so. Problem is it appears to need exponentially more voltage over 4.6 and I could get somewhat close and get run or two stable with only 2 or 4 gigs of ram but no more and temp of 2 hottest cores were 90 quickly. So yes I guarantee it will do 4.7 or possibly even more if I got a better pump/radiator.

I already got it stable 4.7 ghz if you're willing to run memory at 1333 mhz or so lol. Hardly even ran hot but it's because rather than almost 1.3 volts for VTT and VCCSA voltage it can use around a volt or so and memory controller stays cool etc, one problem of having memory controller on the CPU is it really limits overclocking if you don't have below ambient cooling to deal with extra voltage and heat that comes with it. My ultimate goal is to down the road look into what the best radiator I can get is and I've already got a good pump which pumps like 1500 ml an hour so very high performance way more water moving than in these AIO's I"m using but I have no experience setting up a cooling loop and will take my time to figure that out and set it up correctly with aim of best cooling performance possible.

 

[dalekdukesboy]

So I've been messing around and I didn't save it but 4.655 ghz was stable with 2400 mhz memory with meh timings. I tried 4.675 got auto reset, but I had couple voltages intentionally low hoping it'd work and that was after 1 successful 4 gig run. I upped the VCCSA and VTT ( I think) voltages slightly and it got to second run without shutdown but failed stability test so bit better but no go. Long story short, 4.7 ghz on 8 cores/ 22nm process with a 280 mm Corsair AIO water cooler even with (4) 175 CFM 140 mm silverstone fans is a bitch to cool and keep stable. I got within 50 mhz but again not sure 4.7 will happen as it is.

 

[mouacyk]

Got Extreme 9 board installed and 1680 v2 is up running again. However, overclocking is back to basics for now -- 4.4GHz at 1.325v seems to be running without issue, any less voltage will BOSD within minutes in Prime 95. Gonna have to try to get this CPU back to 4.6GHz...

 

[dalekdukesboy]

Got Extreme 9 board installed and 1680 v2 is up running again. However, overclocking is back to basics for now -- 4.4GHz at 1.325v seems to be running without issue, any less voltage will BOSD within minutes in Prime 95. Gonna have to try to get this CPU back to 4.6GHz...

I considered getting an Asrock when I ended up with the Asus RIVE so definitely be interested in seeing how it goes.

 

[mouacyk]

My primary interest in the Asrock board was the 2x 8pin EPS. Fairly certain the single 8pin on my ASUS P9X79 is what caused the fry -- 1.4v at ~4.6GHz might have been maxing out the cable (guessing near 300W?) Not really sure if having 2 cables will affect stability, but I'll find out.

 

[dalekdukesboy]

My primary interest in the Asrock board was the 2x 8pin EPS. Fairly certain the single 8pin on my ASUS P9X79 is what caused the fry -- 1.4v at ~4.6GHz might have been maxing out the cable (guessing near 300W?) Not really sure if having 2 cables will affect stability, but I'll find out.

I'm using one 8 pin on this board it has a 4 and an 8 pin, I was only using the 4 pin with old power supply and when that was giving me issues and I smelled it and the plastic from the cable melted in the 4 pin connection is when I realized my power supply was crap. Thankfully I didn't use the 8 pin because I didn't have 8 pins that weren't broken I only had the 4 lol. First time I ever seen a power supply just melt where its' connectors....not a surprise I checked online and that power supply available not even 2 years ago is already discontinued. I'm assuming they had a lot of problems like this with it and it got pulled fast.

Thankfully I've found my cpu does 4.6 ghz quite happily with 1.376 volts under load so it uses roughly 200 watts on its' own not sure total wattage at the wall that is ONLY the cpu.

 

[Aquinus]

My primary interest in the Asrock board was the 2x 8pin EPS. Fairly certain the single 8pin on my ASUS P9X79 is what caused the fry -- 1.4v at ~4.6GHz might have been maxing out the cable (guessing near 300W?) Not really sure if having 2 cables will affect stability, but I'll find out.

Power isn't what would melt the connector, it would be current and resistance. For example, if it's not making good contact, that could make it run hot because there's more resistance which is what emits heat. It's 12v so that's about 25A total for the entire EPS connector (at 300w,) if that's split across 4 pairs of +12v and ground pins, that's roughly 6.25A per pin. You would probably need at least 4-6 more amps per pair to get 18 AWG hot enough to melt the plastic. My bet is that your +12v EPS connector was just loose fitting or not making great contact with one or more pins and that is what made the connector get hot and melt.

 

[mouacyk]

Thankfully I've found my cpu does 4.6 ghz quite happily with 1.376 volts under load so it uses roughly 200 watts on its' own not sure total wattage at the wall that is ONLY the cpu.

Think I'm settling down at 46x with 1.384v as well. Linux doesn't like BCLK at all, have to leave it at 100. Gonna have to investigate this + RAM next. Doing my standard test by compiling 10 loops of GCC now.

4.7GHz seemed stable with Linpack Extreme at 37x127 at 1.405v but temps were 95C's.

 

[dalekdukesboy]

Think I'm settling down at 46x with 1.384v as well. Linux doesn't like BCLK at all, have to leave it at 100. Gonna have to investigate this + RAM next. Doing my standard test by compiling 10 loops of GCC now.

4.7GHz seemed stable with Linpack Extreme at 37x127 at 1.405v but temps were 95C's.

Yeah 95 C is literally thermal shutdown # for me you're just lucky that your CPU is rated for over a 100 so you can technically do that and it won't throttle or shut off but regardless that's just very hot and shows at that level the CPU needs better cooling but I'd like to see posts of your various runs. Not sure why you'd have issue at over 100 bclk for me that hardly matters other than that 125 strap is best for 2400 mhz and above memory by far and 100 strap it runs about same temps give or take but uses 30-40 more watts at same 4.5-4.6 ghz.

This is a successful 7 gig run at 4.62 with load voltage of 1.408. I had it so it drooped to 1.392 or less and it passed 7 runs with 10 gigs being tested. Weird bit is I tried 10 gigs again with this voltage of 1.408 and it passed one run and shut down/reset/restarted. I can run well over 1.4 volts @ 4.6 ghz or less. But even at 4.62 ghz if I test all or most of my memory it flakes out. But doing 3 gigs less and doing 7 gigs it passes just fine. So seems an auto shutoff kicks in whether it's vrm temps are too high or something because it only happens when I clock it high enough AND give it over 1.4 volts. I'll look through bios and see if I have any settings in there that may kick in and tell it to shut off as well it could be something in there I didn't notice.

I have a fan right over vrms on top of board but nothing really on the back/output area of board where there is a very large heatsink on the black edition, whereas reg RIVE has a smaller stack heatsink. At any rate an interesting problem to solve to at least see what causes that. Because otherwise OC is rock stable I don't even get BSOD's anymore unless I intentionally set memory settings so tight I know it'll happen.

 

[dalekdukesboy]

Think I'm settling down at 46x with 1.384v as well. Linux doesn't like BCLK at all, have to leave it at 100. Gonna have to investigate this + RAM next. Doing my standard test by compiling 10 loops of GCC now.

4.7GHz seemed stable with Linpack Extreme at 37x127 at 1.405v but temps were 95C's.

I used Linux successfully up to 131.5 bclk or so, I found out my "wall" at 4.6 ghz/1.4 volts was BCLK not anything else I suspected. I did a run with 1.42 volts or so and up to 1 gig of memory used I could get stable for 10 runs with Linx with 4.7 ghz. May need more/less voltage but yeah temps quickly got to 92 even for the short seconds it did the runs so has no prayer using 2 gigs- 16 gigs of memory I'm using. Anyway platform is very BCLK sensitive in general so not surprised we both have hard wall limits. However I'd be surprised if you couldn't do up to about 107 bclk stable or using 125 strap the 131.5 or so I'm using. I think CPU is very willing it's the individual board and PCI-E bus and other things that get overclocked with the BCLK that go wonky and cause problems.

 

[mouacyk]

I didn't realize that anything past the straps (100, 125, 166) are fsb overclocks, that were contributing to my instability in Linux. For example:

36 * 125.0 = 4.5GHz, 1.866 RAM strap = 2333MHz on RAM (No FSB overclock)
35 * 128.6 = 4.5GHz, 1.866 RAM strap = 2400MHz on RAM (3.6MHz FSB overclock)

So, to maximize the RAM on this motherboard, I oc'ed the BCLK to get the RAM to 2400MHz and maintain the same clock speed. However, the 3.6MHz bus overclock must be causing problems for Linux, because it's much more sensitive.

I've now settled for 36 x 125 at 1.315v and tuned the memory at 2333MHz (72GB/s read, 67GB/s write, 65GB/s copy, 54.4ns latency) for 24/7. It's great and fast for distributed compiling things in Gentoo now.

My RAM is 4x8GB that can run 2666MHz at 11-13-13-35-2T but this motherboard just will not boot above 2400MHz. I've tried 1.7v on DIMMs and 1.25v SA/VTT. I think it doesn't know how to train tertiary timings above 2400MHz. May have to experiment with setting all the timings manually -- I can't believe that there's a hard limit by the board to just 2400MHz.

Dug up an old post on OCN, that might be helpful when trying for 2666MHz (especially the need to increase VCCSA upwards of 1.4v):
https://www.overclock.net/forum/6-i...-discussion-owners-club-159.html#post25955435

 

[dalekdukesboy]

I didn't realize that anything past the straps (100, 125, 166) are fsb overclocks, that were contributing to my instability in Linux. For example:

36 * 125.0 = 4.5GHz, 1.866 RAM strap = 2333MHz on RAM (No FSB overclock)
35 * 128.6 = 4.5GHz, 1.866 RAM strap = 2400MHz on RAM (3.6MHz FSB overclock)

So, to maximize the RAM on this motherboard, I oc'ed the BCLK to get the RAM to 2400MHz and maintain the same clock speed. However, the 3.6MHz bus overclock must be causing problems for Linux, because it's much more sensitive.

I've now settled for 36 x 125 at 1.315v and tuned the memory at 2333MHz (72GB/s read, 67GB/s write, 65GB/s copy, 54.4ns latency) for 24/7. It's great and fast for distributed compiling things in Gentoo now.

My RAM is 4x8GB that can run 2666MHz at 11-13-13-35-2T but this motherboard just will not boot above 2400MHz. I've tried 1.7v on DIMMs and 1.25v SA/VTT. I think it doesn't know how to train tertiary timings above 2400MHz. May have to experiment with setting all the timings manually -- I can't believe that there's a hard limit by the board to just 2400MHz.

Dug up an old post on OCN, that might be helpful when trying for 2666MHz (especially the need to increase VCCSA upwards of 1.4v):
https://www.overclock.net/forum/6-i...-discussion-owners-club-159.html#post25955435

I also have 16 gigs of memory x 4, or 4x4 like 4x4 truck lol, whereas you have 4x8 for 32 gigs which is twice as much memory for the board to handle THAT by itself even though you are only using 4 memory slots like me is certainly enough to lower the limits of your memory overclocking and BCLK overclocking. I've never tried 32 gigs in this board but if I had a set I would it'd be interesting but nothing I do uses anywhere near 16 gigs so seems pointless but it'd be fun to try

....like just now I got it to boot into windows and into cpuz- @ 4.962 ghz with 131.5 FSB....however only long enough to open cpu-z and it shut itself down lol. So for suicide screen shots and some benchmark fun I think if I lowered the FSB I may be able to get this to boot near 5 ghz I'll have to try just to see what she's willing to do. I was surprised I just upped multiplier and it booted very readily and made it into windows no problem. However, after a minute or so and me starting to open couple programs it just hard shut off. Again, I'll have to play around with it see if I can get it to stay on long enough for a run of something or at least a screenshot of CPUZ just to show I did it.

 

[dalekdukesboy]

My current stable Overclock. This is with the "new" memory and is about the best results with it I've gotten. Ironically running it at the rated 2800 mhz speed seemed to give far worse memory benchmarks and I didn't compare to all other cpu related stuff but at any rate 2455 mhz cr1 timings thus far are significantly better than 2800 mhz.