Originally Posted by slothfish
I've had my trusty D0 920 for about 10 years now (wow time flies!), and recently I've been having trouble keeping my OC at 4ghz with HT on like I have been doing for years on a Rampage II Extreme. With this in mind, I have several questions:
- My stock voltage is a super-low 1.18v. I was able to get to 3.8ghz HT-on at 1.20v and 4ghz with HT-on at 1.22v for years, but now my 4ghz OC is failing even at 1.26v with HT-on. Is this just obvious chip degredation, or do you think my IHS has possibly just become separated a bit and I would be able to go back to my old voltages with a delidding?
-I was running a Prolimatech Megahalems for years, but it doesn't fit in my case so I got a Corsair H150i. My temps on the Megahalems were rising for years and I thought maybe I hit the limit of the heatsink, but I am getting the same temps on my H150i. Again, I'm thinking a delidding might help this (P95 small FFT temps are up around 80c at load now, 4ghz HT-on). This also seems WAY too high for my voltage, unless all the temps I'm seeing reported online are with HT off, in which case this would make sense.
- If I turned HT off in order to lower my voltage and heat output, would I really see much of a hit in performance? I do a gazillion different things on this computer from gaming to encoding video, so I definitely use all the threads at least part of the time. Would 3.6-3.8ghz HT-on perform better in general than 4ghz HT-off?
Thanks in advance everyone! I'm sad my little trooper is slowing down in his old age, but I'm hopeful he can be revived!
I unfortunately just had a very similar situation happen with an i7-6850k that I recently Purchased. The Guy had been running his OC Stable for years at 4.5Ghz with 1.3v and 3000Mhz DDR4. And as of a few weeks before I ended up with the platform for incredibly cheap, it started blue screening like crazy. He had to bring his OC down to 4.2Ghz to keep it stable at 1.35v, this was on the same h150i that you are using, and he had to drop the Quad Channel Ram down to the JDECC Speed of 2133Mhz or it wouldn't even post. Furthermore even at 2133Mhz it would only read 4 of the 8 4GB Sticks of Ram, no matter what he did he could only get Dual Channel to boot.
So he ended up trying all different manners of things with his limited knowledge, and long story short ended up both bending the pins in his socket, AND Corrupting the Boards CMOS, making it completely unusable for him. Because of this, he was just trying to sell it to get something towards a new Ryzen Build. I offered $200 thinking there was no way he would say yes, and he countered with $250 and I bought it. I have years of experience fixing Bent Pins in LGA Sockets, so after a few hours of very Intricate work I was able to fix that issue. Next I pulled the CMOS Chip out and stuck it in my EPROM Reader and succesfully flashed a fresh new UEFI on it. Stuck in just 4 Ram Modules and it booted right up in Quad Channel Mode.
But to make a long story short, after extensive testing, I found that the Memory Controller had degraded, making 8 DIMM Quad Channel impossible at any Speed, 4 DIMM Quad Channel was still possible at 3000Mhz but required 1.2v SA Voltage which is WAY HIGHER than I would ever normally run. And to top it off the CPU could still run a 4.5Ghz All Core OC, but needed 1.45v to do it, which would end up with TEMPs Close to throttle Territory, even when just doing a CB15 run.
I have seen it many times before, and in both your case and mine this is what CPU Degradation looks like and will only continue to happen if the CPU's Continue to be pushed to its limit. In these cases if you intend to continue to use your CPU you can either Raise your Voltage to get back to where you were and hopefully get another Year out of your platform (while continuing to raise Voltage every few months to compensate for the loss of stability). Or if you want it to last another few years or more you should Lower the Voltage by about .1V and then find your Highest all Core OC at that point. This should reduce the heatload on your CPU and therefore slow down the rate of degradation. When doing this in the past I have gotten a CPU that was on its last Leg to last in its Server home for almost 4 More Years before I finally replaced it. This is by no means a guarantee, each architecture is different, and even each Individual CPU reacts differently, but you definitely will have the most luck if you reduce the heat load, and reduce any high LLC settings you may have had.
Also, others in this thread have recommended X58 Xeons, that is another excellent way to extend the life of your platform, just remember these CPUs have often been running for just as long as yours has, in fact its likely that they have been running non stop for their entire life. But because they were most likely in actual servers, they likely were not pushed to their maximum, so you can slot it in an x58 mobo and really push it, in some cases past your original i7 part. Just also remember once you find your Maximum OC of that part, if you want it to last for another 3 to 5 years, its best to back off your Core Voltage to around 1.25v or so and then find your highest All Core OC there, then you will have much more manageable Heat, which will allow your build to purr along for years to come. If you don't really care and just want to push it to the max, then you can really get some high performance out of the part, just beware of the degradation and have another Chip on Stand by in the event that this one finally gives out.
Anyways, hope that helps.