New Loop - Good Or Bad??

[bschmidt71]

i was hoping someone could help me with this:

 

my GPU consistantly runs at water temp (WT)+20C load -- WT+5C idle (max stable OC with no voltage mods/increase)

my CPU consistantly runs at WT+40C load -- WT+10C idle (3.8 GHz w/ HT+turbo enabled -- all voltages set to auto)

 

questions:

 

note: WT stays at 30C due to my Boreas unit (have i mentioned this thing kicks ARSE!!)

 

1. i thought GPU's were supposed to be HOT?? at 30C WT my GPU stays very constant at 50C - even OC'ed to its max stable!! (no voltage mods)

 

2. could something be wrong with my CPU block/mounting/TMI ?? or, is this normal?? comparing it to my GPU block, something looks very....off (assuming similar heat outputs for both GPU and CPU). for reference my CPU temps are 70C load (WT+40C) and 40C idle (WT+10C)....as soon as i put 100% load on all "8" threads the CPU immediately jumps to 70 and stays there - they never rise since my boreas keeps the water at a constant 30; however, i have raised my set point to 40 to see what would happen, and the CPU went up to 80 in lock-step with the WT rising to 40 within about 5-10 minutes using handbrake as my stress medium (since that it what i use on a regular basis)

 

FEEDBACK WELCOME!!

[Hyper]

With voltages on auto on the cpu, it might be overvolting too much causing higher heat.

 

Check to see what your voltages on the CPU are. you can probibly lower then and drop the heat a lot.

 

 

[TheHippi]

You don't have the gpu hooked up first, do you?

[bschmidt71]

You don't have the gpu hooked up first, do you?

 

nope. my loop goes as follows:

 

pump1 -> CPU -> rad1 -> pump2 -> GPU1 -> GPU2 -> rad2 -> Boreas -> pump1

 

 

Check to see what your voltages on the CPU are.

 

i'll check when i get home, but i think they do sit kind of high (i wanna say 1.4ish??) since i'm pretty inexperienced with OCing, i figured auto was the safe bet. would .05-.1V make that much difference on an extreme binned part such as the 965 is? certainly not 20C?? note: i actually down clocked it from what i have had it running before on air - which was 4G. i've booted to 4.1 but could not run stable. now, i can't even run stable at 4.0 (very likely due to high temps running on air - i was frequently close to 90C even with my Noctua NH-U12P SE 1366)

 

so my temps do seem a bit high then for the CPU? what *should* they be?? (is anyone else using the heatkiller 3 block that can monitor water temps/diff??) i doubt i will be able to get them close to the GPUs temps - around 50 at load (WT+20) - or should i be able to do exactly that? if so, i'm thinking the block isn't on good enough or the TIM isn't placed well enough/too little. i still need to take my busted GPU out of the loop....so when i do that i will also check my TIM on the CPU to see what it looks like.(probably this weekend) that will tell me if my pressure is good or if i need more/less TIM...it will also increase my water flow a slight bit...

[slick2500]

Wth is a Boreas?

[Hyper]

nope. my loop goes as follows:

 

pump1 -> CPU -> rad1 -> pump2 -> GPU1 -> GPU2 -> rad2 -> Boreas -> pump1

 

 

 

 

i'll check when i get home, but i think they do sit kind of high (i wanna say 1.4ish??) since i'm pretty inexperienced with OCing, i figured auto was the safe bet. would .05-.1V make that much difference on an extreme binned part such as the 965 is? certainly not 20C?? note: i actually down clocked it from what i have had it running before on air - which was 4G. i've booted to 4.1 but could not run stable. now, i can't even run stable at 4.0 (very likely due to high temps running on air - i was frequently close to 90C even with my Noctua NH-U12P SE 1366)

 

so my temps do seem a bit high then for the CPU? what *should* they be?? (is anyone else using the heatkiller 3 block that can monitor water temps/diff??) i doubt i will be able to get them close to the GPUs temps - around 50 at load (WT+20) - or should i be able to do exactly that? if so, i'm thinking the block isn't on good enough or the TIM isn't placed well enough/too little. i still need to take my busted GPU out of the loop....so when i do that i will also check my TIM on the CPU to see what it looks like.(probably this weekend) that will tell me if my pressure is good or if i need more/less TIM...it will also increase my water flow a slight bit...

 

 

My 920 C0 stepping i7 doesnt require that much voltage to hit 3.8... the 965 should be able to do that at almost stock v... also check the qpi voltage. Auto settings tend to over volt way too much and you will get a lot higher temps that way.

 

BTW 40C at load is actually pretty good for an oc'ed i7.

[bschmidt71]

Wth is a Boreas?

 

this is a Boreas:

 

(space)(space)

 

 

the blue block you see there is one of 4 water blocks - which houses 3 TEC's (thermo electric chiller) each for a total of 12. the water is cooled to subambient by the "cold" side while the "hot" side is attached to the fins the blocks are attached to. the fins create a wind tunnel of sorts, and the fans at each end suck air between the fins (into this "tunnel") and out the sides. allthough the unit is capable of subambient temperatures, my unit will not - by design. condensation is a very real problem with these devices, so i have negated this capability by adding my water loop in series with the Boreas; thereby limiting its cooling ability to "merely" ambient. but, since it still has the same ability to lower [water] temperatures, it means my water loop stays at a constant temp. the Boreas comes with software that controls "how much" it cools -- i have mine set for 30C.

 

here are some specs:

 

(space)

 

 

for those who are interested, here is a review frmo Hardware Canucks. you can also google "Boreas MTEC review" to find other reviews from different sites

 

BTW 40C at load is actually pretty good for an oc'ed i7.

 

acually, it's 70C: water temp + 40 (as explained above, my water temp stays pretty constant at 30C)

[bschmidt71]

i played with my voltages a bit this morning before work (well, OK...i played with the CPU voltage)

 

i set the CPU to 1.3V, then turned the mult down to 12 and the RAM to the lowest allowable setting - i was able to raise the bclk to 218, but raising the voltage as high as 1.41V would not let me get anymore out of the chip (note: i kept RAM speeds below 1300MHz - its rated for 1600) (note2: all speeds boot only, no stressing was done).

 

i then started playing with my multiplier (get your mind out of the gutter you pervs): lowering bclk to 133 (default) and keeping RAM at the lowest setting i was able to boot at a 31 mult, but crashed during the physics test on 3DMark. i left off (had to get ready for work) at 160x25 with RAM at1604 -- still at 1.3V. i'll try running 3DMark/PCMark when i get home in a few hours, and if those go well i'll try some encoding (season 1 of Star Trek: Enterprise + Handbrake should be sufficient stress) note: i turned off Turbo, but left HT on.

 

since i was just trying to get a good boot and wasn't really testing stability, i never checked temps with the 1.3V. i'll be checking that as well during the heavier testing and will post later on.

[bschmidt71]

i played with the OC a little this weekend. after setting the voltage to 1.3 the temp did seem to come down quite a bit (more than i expected it to). i'm now running at a much more reasonable 60C (dropped about 10C just going from 1.41 to 1.3!!) i also tried getting it back up to 4G, but alas, i think that will no longer be possible. i DLed and ran OCCT, Prime95, and even memtest86+ to check stability and everything ran fine for several hours (more than 1 hour - less than 8). overall, i think ill stay with how it is now. it idles at WT+6 and loads at WT+30 - give or take a few C's. i also checked the TIM when i took my busted GPU out of the loop. it looked a little wierd - the TIM was "squished" out to the ends of the CPU and the CPU did not look too well covered in all areas. i think i might have tightened the block too much? i put a little extra TIM on and re-tightened; then i ran F@H overnight to "cure" the TIM before i got my temps, as stated above, the next day.