i hardly buy them, oh snap
Overclocking
Mac
#62
lol don’t worry cabby i havn’t missed a copy of pcpp or atomic from when they started…:blink:
nipper575
#63
Very preliminary test, took the screenshot just before completing so you can see the full load and core speed.
Clearly something is still not right, the GFlops are far too low which makes me think that the CPU isn’t being pushed as hard as it should be. After all the RAM issues, random BSODs that didn’t give any indication of what was causing them and now this, this CPU is testing me. Still, I’m making progress, couldn’t get it to cooperate at 4GHz at all yesterday. 3.3, 3.6, 3.8GHz all fine, no dramas. As soon as I’d go for 4, then it would fail before the first test would even complete, regardless of the voltage.
The most annoying thing is that I have had experience with these CPUs before and managed to get to 4Ghz with HT on within an hour or so on practically the same motherboard (UD5). Fairly confident that my RAM is having issues with this CPU, used to have some flexability with the timings but now its being incredibly finicky with it’s timings which is the difference between passing LinX or failing before the first test, it’s getting upgraded soon.
nipper575
#64
GPUs are definitely holding me back, because that CPU score is bitching, haha. If it turns out to be 10 pass stable, then this CPU is actually using less Vcore than the 940. Not a golden 920 by any means, but it’s still a nice improvement.
Azalin
#65
nice work mate.
Still a better chip than mine for sure. Temps and vcore are tops … lots of room to move 
nipper575
#67
Seems to be a good chip after all, I had my doubts after all the trouble I had getting it here though. With the right bios revision, it seems that this chip has really woken up.
As for temps, also have to remember that my ambient would be lower on average. So it gives me a bit of an unfair advantage.
Nah, she’s rock solid now. Since then dropped voltages even more and ran 10 loops of LinX.
Consistent GFlops at the usual rating for this clock speed with HT on (with this version of LinX). I still think there is more voltage to drop, but I might start going up in clock speed instead.
So Vcore : 1.264V (lower than the 940 by a good margin)
CPU PLL : 1.7V
QPI/VTT : 1.275V (Still think I can drop this down)
QPI PLL : 1.220V
DRAM : 1.580V
IOH : 1.120V
Azalin
#68
I need to go right back through mine.
Im just stumped why i can hit vtt at 1.2 when all other R2E guys are 1.275-1.35, and if the higher vcore is compensating for it at 4ghz. over 4 needs a lot of tweaking so i best tart over i reckon.
nipper575
#69
There’s no way I could get it stable at with QPI @ ~1.2V. Even still, I need to push up the QPI PLL in smaller increments with the QPI/VTT core otherwise it would still crash before one LinX test.
If you haven’t done it already, search some forums for the bios revision that gives the best results, it’s not always the latest one, definitely wasn’t in my case. Honestly, there was times over the past couple of days where I thought this CPU was never getting to 4.00GHz, lol. If my RAM wasn’t being so finicky, it would have been much easier, that’s my own fault for not noticing though.
nipper575
#70
Just point out that the guy who had this before me only ever clocked it to 3.8GHz at 1.28V. I’ve been able to undercut that by a good margin, I can run 3.8GHz on stock vcore so it seems that different set ups can cause different OCing results even with the same CPU.
Either that or he just didn’t spend the time to refine it.
nipper575
#71
Nice jump over the 940 which resulted in a score of 5.33 at the same clock speed. This is at 4.01Ghz despite what Cinebench says, it isn’t picking up the turbo multi.
Azalin
#72
Thus we start again
going to set qpi/dram ( vtt) to 1.25 as i read below can lead to droop.
vid for the cpu is 1.21 too nip … so kicking up there from stock ( id say my 4ghz clock was on the money)
The board shipped with the latest bios 1914 i think (ive read 1504 is the best so may switch back pending how this pans out today)
nice cinebench result… looks like ill only hold the edge until you upgrade gpus hehe
Edit - Also using turbov and the R2E lcd poster ( little lcd that shows running volts/temps/fan speeds while in windows - very cool littel bugger, im using it to keep an eye on all volts atm)
nipper575
#73
Maybe you should enable Load Line Calibration, that will reduce Vdroop by a good margin. I use it myself and always have since migrating to the X58 chipset (from day 1). A lot of people give it a bad name by stating that it can cause huge voltage spikes, but you’ll find that these people are talking about the LLC that their P35 or older motherboards had on them, it’s been refined since then just like every other bit of tech. I’ve never witnessed my voltage rising above where it usually sits.
BTW, using a x17 uncore multi which is probably what helped me just squeeze past your CPU score on CINEBENCH. You should give it a go if you are only using x16 (twice your ram multi). Needs bugger all QPI/VTT voltage over x16, probably not even one notch (with a 191MHz base clock anyway).
Always loved the idea of that little LCD screen, perfect for monitoring the system.
Azalin
#74
always use llc. the vcore thing seems to be a r2e thing with it giving more under load
currently @ 3.68 vcore 1.225 (1.244 load)
ive noticed my copretemp reads 98 tj max where as yours is 100 ??
nipper575
#75
I might have an older version, I’ll update it now. It’s important to know whether the TJ is 98 or 100 as it would offset temps by 2’C higher or lower depending on which is correct (as it calculates temp by subtracting from the tjmax, for those that don’t know).
Maybe it’s read off the chip? Latest coretemp still reads 100 for me. Going to get to the bottom of this later, it’s annoying me now.
Azalin
#80
yeah just gotta do it alot more methodical than this afternoons effort of quick passes up to the nitty gritty.
and see if i can hone them temps more
onwards and upwards!!!