Temperatures and voltages OC 965 C3

Will something happen if my cpu is at 62 c while OC to 4ghz? I don't care much for its durability; I'm switching to fusion as soon as it comes out.

Also, could you propose all voltages under every single acronym. I know that every chip is different, but I can't bother with going from down to up. I'd rather go for the highest settings and go down to stability.

I plan to repasting thermal paste, because I think i screwed up with that. I've yellow 14 hf, pump with thats 1000l/h, so I suppose that at 1.55 volts and 4 ghz going all the way up to 65 is awful.

P.S. 65C before crashing. It seems like 1.55 volts is best, as lower volts crash.
Currently, most successful OC run for 6-9 minutes, at this settings:
258FSB (didn't check max possible)
15.5X ratio
DRAM 1720mhz at CL9 (supports 1866 CL9)
CPU/NB 2580
HT Link 2322
CPU Voltage 1.55
CPU/NB 1.4
CPU VDDA 2.9
DRAM 1.722 (supports 1.65)
HT auto
NB auto
NB 1.8V auto
SB auto
S5 1.2V auto

I listed all voltages available at BIOS. Please, point out if there is any disproportion between voltages. Looks like 250-6 FSB is optimal, and 1.55V. I'll reapply thermal paste.

965 C3
RAM 1866mhz, CL9, 1.65V
Asus Crosshair Formula 3
HAF 932
3 fan radiator
Egheim pump... 1000 Station, Rev 2
900Wat; 80+ Xigmatek supply

There is also GPU- HD 5850- in loop. Which smart universal waterblock.

62C is the designed Max operating temp. for the C3 965 BE. It's not like it blows up at 62C but if run continuously at this temp it may experience BSODs, reboots, display issues, etc. Higher temps may result in system shut-down?

http://products.amd.com/en-us/Deskt...17&f1=&f2=&f3=&f4=&f5=&f6=&f7=&f8=&f9=&f10=No

Your voltages look pretty maxed right now. There are diminishing returns as you exceed the designed operating voltages.

I found with my 965BE, that after 1.525v I didn't get stable at 4Ghz. So I've settled at 3.9Ghz. I tried voltages up to 1.6v and still no difference. I think I still have some tweaking to do. My load temp is somewhere between 54 - 56c depending on ambient I think. coretemp recorded a max of 56c but when I was monitoring it was flicking between 54 and 55c. This is with 5 hours + of prime 95. Will still do a longer test when I can. (current draw is over 1 Amp when loaded ). My cooler is a corsair H70. I would expect you to have cooler temps than I as mine is an off the shelve setup

beenthere said:

62C is the designed Max operating temp. for the C3 965 BE. It's not like it blows up at 62C but if run continuously at this temp it may experience BSODs, reboots, display issues, etc. Higher temps may result in system shut-down?

Click to expand...

It seemed like 1.55V was optimal, or was it just other volts weren't set right. I wouldn't know, because I haven't checked it out at lower temperatures, and I'm determined to get 4Ghz.

beenthere said:

Your voltages look pretty maxed right now. There are diminishing returns as you exceed the designed operating voltages.

Click to expand...

I'm don't think so, due to the fact that again 1.55V seemed optimal for me, although that did raise temperature. I'm willing to reduce that if there're other ways of getting there, to 4Ghz.

Temperatures seem too high. Just by switching my setup from Air cooled standard HS that came with the CPU to a Corsair H70 I was able to get the CPU stable from 3.6Ghz up higher with the same volts.
I've taken the approach "put the volts at max for the temperature range and then see where the OC gets to", however this didn't give good results (on a previous OC).

to OCF!


I'd lower the cpuNB voltage if I were you. 1.30v should be as high as you want to go with water and I doubt you'll even need that much.

I am curious why you would use the clock instead of the CPU/cpuNB multipliers??? You might use a small clock OC to ramp up the RAM speed a little from 800 MHz but I can't imagine any other use for a clock OC.

Usually the X4's like the HT Link at stock speeds or lower ...

janat08 said:

It seemed like 1.55V was optimal, or was it just other volts weren't set right. I wouldn't know, because I haven't checked it out at lower temperatures, and I'm determined to get 4Ghz.

I'm don't think so, due to the fact that again 1.55V seemed optimal for me, although that did raise temperature. I'm willing to reduce that if there're other ways of getting there, to 4Ghz.

Click to expand...

Seems like a very over-volted set-up to try and reach 4 GHz.? See my sig. for comparison. More voltage is not always better.

QuietIce said:

to OCF!


I'd lower the cpuNB voltage if I were you. 1.30v should be as high as you want to go with water and I doubt you'll even need that much.

I am curious why you would use the clock instead of the CPU/cpuNB multipliers??? You might use a small clock OC to ramp up the RAM speed a little from 800 MHz but I can't imagine any other use for a clock OC.

Usually the X4's like the HT Link at stock speeds or lower ...

Click to expand...

I read somewhere that higher HT along with CPU/NB frequency makes processor and system overall more productive. I thought that clock- FSB is what you're referring to right?- increases overall system performance, so I'd imagine it comes first in priority. I'll reduce it to stock along with CPU/NB to see what happens than I'll increase it and end.

Please, comment on other voltages. It would be good if some one provided link for recommended proportions of voltages from one to other, encompassing all CPU voltages.

P.S. I've got Egheim 1000 Station Rev. 2, and I'm thinking if it's working constantly at maximum capacity, or if it regulates workflow. It would be nice for it to reach maximum water flow before 50-55C.

P.S.S. I'm wandering because temperatures suddenly increase from 40 to 50, from there it gradually increases to 55-8. Although all fans are working at max.

Dolk's Phenom OC'ing thread goes into the voltages, frequency, etc. settings/options/challenges.

http://www.overclockers.com/forums/showthread.php?t=596023

Yea, whatever the voltages it tends to stay around 55-60. Also when I start stress testing in may increase to 57-8 and then decrease to 55-6.

P.S. The temperature anyways gradually increases from 55 to 58-9 gradually, so I decreased volts a lot. Still just under 45. I suppose that it's working at it's max capacity.

What if I connect Laing DDDC Pump 12v Pro, into the loop. Connect output of two pumps through T ... connector (?), and then connect them back near input with T thing? Any worries that one of two will break, or that their added flow and pressure will begin leak somewhere? If you agree, than advise on how to set them up; should pumps press against each other in upper of T (facing each other), or should the one with least flow rate or serving height-pressure- be in lower of T.

I read some of it. I don't think that he's covering all voltage mentioned in my list. I set them to half way through between max and mins, if their frequencies were at half of "medium" and stock.

janat08 said:

I read somewhere that higher HT along with CPU/NB frequency makes processor and system overall more productive. I thought that clock- FSB is what you're referring to right?- increases overall system performance, so I'd imagine it comes first in priority. I'll reduce it to stock along with CPU/NB to see what happens than I'll increase it and end.

Click to expand...

The cpuNB speed is very important for two reasons. One, it often stabilizes the CPU overclock usually leading to higher possible clocks. Second, it increases memory performance, both L3 and RAM. You definitely cannot go wrong running the cpuNB up as high as it will go as long as you don't have to decrease the CPU overclock.

Unless you're running a pair of high-performance video cards you'll never see any difference between an HT link of 2400 and an HT Link of 1800. At 1800 MHz the HT Link can handle all the video and other system data you can throw at it.

The "clock" (short for reference clock) is technically referred to as HTRef and is sometimes called the FSB, though that's very inaccurate and may be the cause of your confusion. On outdated systems with a FSB all the RAM and system data must go through the FSB to get to/from the CPU. AMD did away with that in 2005 (04?) with the K8 CPUs, which replaced the outdated FSB architecture with the HyperTransport Link (HT Link) and Integrated Memory Controller (IMC). System information goes through the HT Link and RAM data goes through the IMC, which is part of the cpuNB on the K10 (Phenom and Athlon II) CPUs. The clock is only a reference clock used by other components, along with their respective multipliers, to get their operating speed. It carries no data. A CPU running at 240x15 is just as fast as a CPU running 200x18 - and the other components (cpuNB, RAM, and HT Link) all work the same. Its only the final component speeds that matter.


PS
As far as your other voltages go (not sure what "S5" is, though) I think they're fine except that cpuNB VID. If I didn't think that I would have said so earlier ...

Please, provide more examples of your over-clock, and comment on my WC suggestion.

I think mixing pumps is a bad idea - especially those two pumps, which are at opposite ends of the spectrum. The DDC is a low-flow, high-pressure pump while your Eheim is a high-flow, low-pressure pump (both relatively speaking). A better match would be a Laing D5 or Swiftech MCP-655 (virtually the same pump) because it's also a high-flow, low-pressure pump. In my opinion, though, it's still not a good idea to mix pumps. If you want to add another pump get another Eheim 1000.

As for routing them in parallel (using the T's as you described) that's normally not recommended. Putting them in parallel increases flow but has the same pressure (actually, an average of the pressures - I think). Better to run them in series where the pressure is increased and the flow remains (more or less) the same.


It's normal with a water cooling loop for the core temps to increase quickly then slowly increase until the system reaches equilibrium. Often the loop will go above equilibrium for a short time them come back down. What you're seeing is normal.

What CPU block are you running (or did I miss that information)? Which radiator model are you using (not all 3x120's are alike)? What about rad fans? Anything else in the loop besides the video card?

QuietIce said:

I think mixing pumps is a bad idea - especially those two pumps, which are at opposite ends of the spectrum. The DDC is a low-flow, high-pressure pump while your Eheim is a high-flow, low-pressure pump (both relatively speaking). A better match would be a Laing D5 or Swiftech MCP-655 (virtually the same pump) because it's also a high-flow, low-pressure pump. In my opinion, though, it's still not a good idea to mix pumps. If you want to add another pump get another Eheim 1000.

As for routing them in parallel (using the T's as you described) that's normally not recommended. Putting them in parallel increases flow but has the same pressure (actually, an average of the pressures - I think). Better to run them in series where the pressure is increased and the flow remains (more or less) the same.


It's normal with a water cooling loop for the core temps to increase quickly then slowly increase until the system reaches equilibrium. Often the loop will go above equilibrium for a short time them come back down. What you're seeing is normal.

What CPU block are you running (or did I miss that information)? Which radiator model are you using (not all 3x120's are alike)? What about rad fans? Anything else in the loop besides the video card?

Click to expand...

My radiator is NexXxoS Alphacool Xtreme III. My CPU block is Yellow Stone 14 HF. 3 Coolmove 120x120x25 Alphacool ventilators. What do you mean by having them run in series, don't they have to share radiator? The whole point is to make CPU cooler. I might have just applies thermal paste badly. Do you think I should repaste it? I remember that I was kinda wonky when I was installing cpu water block.

Running them in series means first one pump then the other - not side-by-side.


If you have any doubts about the block mount you should redo it.


PS
It's possible you just don't have enough radiator for the CPU and video card - but I'd reseat the block anyway.

120x38mm fans might also help if it's not the block ...

Could you elaborate? I understand side-by-side, but how will the liquid get cooled, they will have to merge at radiator anyways.

A simple example of two pumps in series: res > pump > pump > block > rad > res