Post your New Ivy Bridge Processor Thermals, Type of Cooling, and First Impressions:

RamonGTP said:
Neither of the two Intel engineers would give a solid answer as to what was "safe" for this reason, can't really blame them. If I had a nice job at Intel I wouldn't want to jeaprodize it either. But both essentially said the same thing, which is that the voltages people were discussing (which was around the 1.35 mark) is most certainly not safe. He did say that when they test at Intel, they have trouble maintaining thermal accuracy above 1.2v
Click to expand...

They probably also don't know what is "safe" since I'm sure it varies by chip. And it's not like once you hit some unsafe voltage then, boom, the chip is dead - it's still a slow process.
 
Forceman said:
They probably also don't know what is "safe" since I'm sure it varies by chip. And it's not like once you hit some unsafe voltage then, boom, the chip is dead - it's still a slow process.
Click to expand...

I'm sure there isn't a single person here or there that thought there was a terminal voltage where the chip goes boom. No need to be so rigid in your interpretation. Everyone was understanding what was being implied by "safe"
 
RamonGTP said:
I'm sure there isn't a single person here or there that thought there was a terminal voltage where the chip goes boom. No need to be so rigid in your interpretation. Everyone was understanding what was being implied by "safe"
Click to expand...

My point is that there is no definition of "safe". Does safe mean you can run it at that voltage for 3 years with a 10% chance of failure, or a 50% chance of failure? What about 1 year with a 1% chance of failure? It isn't as simple as "1.45 volts is safe, but 1.55 is not". I'm not being rigid, I'm just saying that Intel themselves probably hasn't quantified a "safe" voltage for the chips - as far as they are concerned, the safe voltage is the one they validated the chip at, and that's all.
 
swaggerbot said:
Well I'm not sure what to make of my temps. Either I have an incredibly hot chip or I'm doing something wrong.

i7 3770k
Asrock z77 extreme 4
Zalman CNPS 10x performa w/MX-4 thermal paste
Corsair Obsidian 650D- ambient temp 72 deg

Running 45 x 100.

I've re-seat and re-applied paste literally like 4-5 times to make sure I've got a correct fit, and any voltage in 1.18-1.19V range takes me up to 92-94 deg C on intel burn test and prime after 20 minutes approaches 90 as well. I've tried all sorts of vcore fixed vs offset, LLC tweaking, PLL tweaking and as as soon as the vcore settles into 1.18 or 1.19, those are the temps I get.

Do I have a garbage cooler? A dud chip? I'm almost inclined to head down to microcenter and pick up an H80/100 or something because these temps seem absurd.

PS - I had my "old" 2600k in the system as a placeholder and didn't do a clean install with the cpu switch, altho Win7 did re-install a bunch of stuff. Silly quesiton, but could that somehow be throwing off the temps or anything? Sorry, I'm just scratching my head at these temps.
Click to expand...

I tried out your settings and with push-pull Megahalems the hottest core was 73. It failed after about 30 minutes on my machine though, need more voltage.
 
Forceman said:
My point is that there is no definition of "safe". Does safe mean you can run it at that voltage for 3 years with a 10% chance of failure, or a 50% chance of failure? What about 1 year with a 1% chance of failure? It isn't as simple as "1.45 volts is safe, but 1.55 is not". I'm not being rigid, I'm just saying that Intel themselves probably hasn't quantified a "safe" voltage for the chips - as far as they are concerned, the safe voltage is the one they validated the chip at, and that's all.
Click to expand...

Agreed.
 
swaggerbot said:
Well I'm not sure what to make of my temps. Either I have an incredibly hot chip or I'm doing something wrong.

i7 3770k
Asrock z77 extreme 4
Zalman CNPS 10x performa w/MX-4 thermal paste
Corsair Obsidian 650D- ambient temp 72 deg

Running 45 x 100.

I've re-seat and re-applied paste literally like 4-5 times to make sure I've got a correct fit, and any voltage in 1.18-1.19V range takes me up to 92-94 deg C on intel burn test and prime after 20 minutes approaches 90 as well. I've tried all sorts of vcore fixed vs offset, LLC tweaking, PLL tweaking and as as soon as the vcore settles into 1.18 or 1.19, those are the temps I get.

Do I have a garbage cooler? A dud chip? I'm almost inclined to head down to microcenter and pick up an H80/100 or something because these temps seem absurd.

PS - I had my "old" 2600k in the system as a placeholder and didn't do a clean install with the cpu switch, altho Win7 did re-install a bunch of stuff. Silly quesiton, but could that somehow be throwing off the temps or anything? Sorry, I'm just scratching my head at these temps.
Click to expand...
if you are running at 4500 and have temps that high, doing what you say you are doing then ... i'd
1) check to make sure cooler is really touching chip well.
2) if 1 ok, take chip back within 15 days, and get another,
say it was the wrong chip.
3) get a better cooler ... nhd14... h100 is noisy
but i'd bet the chip is bad.

i loaded win7 on a 2600k, i then replaced with 3770k, win7 x64 booted fine
added a whole bunch of drivers inc 4 3770k cores etc.
i have no problems.
that is probably not it. ... you could image what you got reinstall win7 clean
just add prime95, realtemp or cooltemp, and cpu-z quickly, run 20 min and see
if anything has changed.

i'm running 4700, vcore 1.33, load line med, cpu pll 1.625 i get
83cmax ttp 73c. my 3770k runs about 8-9c hotter than my
2600k running in my rog setup at 4700. just about same everything else.
so that's what i'd do.
PS. i'd also make sure you have new good paste you are applying. it should be jell like not hard and old.
 
Last edited:
Forceman said:
My point is that there is no definition of "safe". Does safe mean you can run it at that voltage for 3 years with a 10% chance of failure, or a 50% chance of failure? What about 1 year with a 1% chance of failure? It isn't as simple as "1.45 volts is safe, but 1.55 is not". I'm not being rigid, I'm just saying that Intel themselves probably hasn't quantified a "safe" voltage for the chips - as far as they are concerned, the safe voltage is the one they validated the chip at, and that's all.
Click to expand...
i worked with intel all my life. they are great chip designers.
their processes are 2nd to none. that said ...
they usually build in a whole of margin into their chip designs.
i doubt seriously that 1.35 volts external, before droop, would cause
any serious problems to the chip especially at slow speeds with speedstep on.

it is the combo of volt, freq, and heat, over time, that may kill a chip.

i don't stress things as much as some of you may. yet i leave my system on 24/7. it's happy! how many of you have ever left you power amps on,
or your systems on 24/7 and found they failed overnight????
i have had 0 such problems. it is turning them on that causes issues!

i also don't buy AMD ... it is just intel is worth $50 more as insurance.

jet
 
Yes, if you run 1.35, don't mess with vdroop, leave speestep on and don't OC it probably won't cause a problem. Here's the problem with all those selective variables.

Who's going to push 1.35 for low speeds?
vdroop and speedstep effectively means even though you have 1.35 set, you'll rarely if ever actually supply that voltage. Anyone supplying 1.35v is doing it because they need that much juice for HIGH speeds and quite possibly would need to enable some form of LLC so there goes the vdroop failsafe. No one is going to raise voltage just for the sake of raising voltage. (unless they simply have no idea what they're doing)

So... What happens when you're supplying 1.35 at high loads and speed over a period of time? That's the question. I've experienced CPU degradation on several processors with high levels of voltage and overclocks. It's something that happens and at a certain point, the process gets accelerated far quicker than some may realize.

I've seen it on several AMD processors and thought I was startign to experience it on my old Q6600 as I needed to raise voltage a few times through the course of it's life to maintain overclocked stability. That one ended up being the board starting to degrade.
 
in the end... intel is NEVER gonna say what voltage is safe. there are too many variables, and safe would limit overclocking severely and that's not what they want.
they want you to pick what you think you need. they expect you to have some reasonable
understanding of what the chip will do, and how to keep things in balance. you need to know what you are doing!!! read. it really isn't that hard.

and i will repeat one more time ... intel is NEVER gonna say what voltage is safe!!!
so only question is .. what are you gonna do?

jet
 
one last post ... slight off topic ... i have an asus g2s laptop.
it has dual core 2.4g gt8600m graphics ... it has room for only one fan.
where did they put it??? guess?? of course the cpu!
guess what get really hot? yeh that graphics card ...
how hot do you think it gets????
google asus g2s graphics hot

can these 3770k chips take 110c? i'd bet yes, but i'm not going try.
does my g2s temp monitor get there and fail? never fails!!!! 4 years old.
fan noise is insane at 95c when it just starts getting hot, but does it ever fail??? NO!!!

jet
 
jetli said:
in the end... intel is NEVER gonna say what voltage is safe. there are too many variables, and safe would limit overclocking severely and that's not what they want.
they want you to pick what you think you need. they expect you to have some reasonable
understanding of what the chip will do, and how to keep things in balance. you need to know what you are doing!!! read. it really isn't that hard.

and i will repeat one more time ... intel is NEVER gonna say what voltage is safe!!!
so only question is .. what are you gonna do?

jet
Click to expand...

I agree
 
I'd down to 1.14v stable 4.5GHZ with 1.55v Pll on my 3570k.

With this voltage temps are almost the same as SB for same speed with voltage required for stability of 2500k.


I think the big problem with IB right now is figuring out the proper voltage for stability. I think most people are putting too much voltage through them. Even at default voltage setting.
 
SonDa5 said:
I'd down to 1.14v stable 4.5GHZ with 1.55v Pll on my 3570k.

With this voltage temps are almost the same as SB for same speed with voltage required for stability of 2500k.


I think the big problem with IB right now is figuring out the proper voltage for stability. I think most people are putting too much voltage through them. Even at default voltage setting.
Click to expand...

That may be true of some people, but that isn't the problem. The problem is that it runs hot. I doubt you're as stable as you think at those settings. I can pass prime and IBT all day with lower voltage than what I'm running, but I also notice it does some odd stuff every once in a while, even when it's not under high load. The odd behavior is completely eliminated with a slight voltage bump. You'll also likely notice that your IBT scores will go up a few gflops if you raise the voltage even while keeping the speed the same and thats because the CPU is having to do less error correcting and recalculating.
 
jetli said:
i think your system is incredibly good. don't complain.

i still think prime95 for 12 hrs stresses chip pretty hard.
i'd run that if you think it is stable. start in morn. don'rt run overnight.
keep an eye on it.

i'd be surprised if your system still is stable. just call me skeptical.
jet
im at 4700, load line med, vcore 1.33, cpu pll 1.625, nh-d14, temps 83c max

what are your parameters?
Click to expand...

Link said:
Can you share other settings in the bios?
Click to expand...

The only noteworthy stuff I did was disabling HT and spread spectrum
 
RamonGTP said:
That may be true of some people, but that isn't the problem. The problem is that it runs hot. I doubt you're as stable as you think at those settings. I can pass prime and IBT all day with lower voltage than what I'm running, but I also notice it does some odd stuff every once in a while, even when it's not under high load. The odd behavior is completely eliminated with a slight voltage bump. You'll also likely notice that your IBT scores will go up a few gflops if you raise the voltage even while keeping the speed the same and thats because the CPU is having to do less error correcting and recalculating.
Click to expand...

Yeah I've seen that, always wondered why.
 
RamonGTP said:
That may be true of some people, but that isn't the problem. The problem is that it runs hot. I doubt you're as stable as you think at those settings. I can pass prime and IBT all day with lower voltage than what I'm running, but I also notice it does some odd stuff every once in a while, even when it's not under high load. The odd behavior is completely eliminated with a slight voltage bump. You'll also likely notice that your IBT scores will go up a few gflops if you raise the voltage even while keeping the speed the same and thats because the CPU is having to do less error correcting and recalculating.
Click to expand...

That's interesting to know, thanks for the tip.
 
i ran intel burn test for 1st time ... passed ... about 108-109 gflops ...
is that good or not?
transcoding seemed real good.
pcmark07 was excellent ~ p5900
3dmark seemed way weak ... ~p2600 as i recall ... gtx550ti

need to check virtu

what aida64 tests are good? and what results? anyone

jet
 
So far best I've managed that's semi-stable without dumb voltage is 4600. Temps got into the upper 60s during 10 minutes of Prime 95. Required voltage to be set to 1.275 in Asus AI, which translated into 1.352 under load according go CPU-Z. 4700 required voltage of almost 1.4 according to CPU-Z. Testing what I think might be a daily OC in IBT (Maximum stress) now. 4400, which requires 1.145 in Asus AI to be stable, translating to 1.23-1.25 under load. So far its running between 62 and 70C, pulling about 120 gflop.

Using a 3570k in an Asus P8Z77-V Pro. I'm still not quite sure how the voltage I set in Asus' OC tool translates to actual voltage. Up to 4300, I don't have to touch voltage controls at all.
 
I have the same board. The only setting I mess with is the voltage offset. I actually don't "need" to mess with it but if I leave it at auto, the voltage going to the CPU is quite a bit higher than it actually needs to be.
 
RamonGTP said:
but I also notice it does some odd stuff every once in a while, even when it's not under high load. The odd behavior is completely eliminated with a slight voltage bump. You'll also likely notice that your IBT scores will go up a few gflops if you raise the voltage even while keeping the speed the same and thats because the CPU is having to do less error correcting and recalculating.
Click to expand...

I haven't noticed any odd behavior. I have noticed that high speed ram with proper voltage settings for stability has a nice performance boost for gflops.

At [email protected] with ram at 1600mhz 7-9-8-21-1T 1.5v glflops are around 116.
Ambient Temps 26c

http://valid.canardpc.com/show_oc.php?id=2361977
LinX4point53570kmax4gb.jpg






At 4,[email protected] with ram at 2133mhz 9-11-10-1T 1.63v glfops are around 121.
(results below in post#624)

Seems to me like the heat is beat by lowering voltage and balancing out pll and core voltage for stability.
 
Last edited:
RamonGTP said:
I have the same board. The only setting I mess with is the voltage offset. I actually don't "need" to mess with it but if I leave it at auto, the voltage going to the CPU is quite a bit higher than it actually needs to be.
Click to expand...

My MSI MB doesn't have LLC offset voltage settings but it doesn't seem to hurt performance or temps.
 
SonDa5 said:
I haven't noticed any odd behavior. I have noticed that high speed ram with proper voltage settings for stability has a nice performance boost for gflops.

At [email protected] with ram at 1600mhz 7-9-8-21-1T 1.5v glflops are around 116.
Ambient Temps 26c


At 4,[email protected] with ram at 2133mhz 9-11-10-1T 1.63v glfops are around 121.
(results below in post#624)

Seems to me like the heat is beat by lowering voltage and balancing out pll and core voltage for stability.
Click to expand...

You're beating heat by lowering voltage to the point that the CPU is wasting CPU cycles due to calculation errors. If it were that easy I assure you Inte, engineers would have thought of it. You didn't outsmart them. Setting voltage to just above the point you don't crash doesn't equal stable.
 
I'm going to have to read up on PLL voltage...

Mine was 1.8v default and I bumped it up to 1.82v for 4.5ghz (for no real reason.. but I tought I could help with stability). But you guys are running lower PLL voltages.

What's a good way to figure/test to find optimum PLL voltage required?
 
RamonGTP said:
You're beating heat by lowering voltage to the point that the CPU is wasting CPU cycles due to calculation errors. If it were that easy I assure you Inte, engineers would have thought of it. You didn't outsmart them. Setting voltage to just above the point you don't crash doesn't equal stable.
Click to expand...
i'm at 4700, 1.33 vcore & i'm getting 109 gflops
is having a higher freq clock @4700,
and thus higher vcore @1.33 to start, causing my gflops @109
to be less than others are reporting here running their systems @4500 with
lower voltages?

does their ram speed have any effect on gflops?

others are reporting gflops around 120, @4500 w/vcore 1.2 or less!

jet
 
jetli said:
i'm at 4700, 1.33 vcore & i'm getting 109 gflops
is having a higher freq clock @4700,
and thus higher vcore @1.33 to start, causing my gflops @109
to be less than others are reporting here running their systems @4500 with
lower voltages?

does their ram speed have any effect on gflops?

others are reporting gflops around 120, @4500 w/vcore 1.2 or less!

jet
Click to expand...

Yes IBT tests ram so more bandwidth = more performance in IBT
 
passed Linx test 20times easy with 1.150v

But not like Sandy Bridge it failed on Prime Test

Weird, Sandy was Easier on Prime and Harder on Linx.

anyways i managed to pass both of them @ 1.180

but i decided to use it @ 1.190 for safe.

my room Temp was around 68 here in NY

and this Temp is pretty nice i guess.

here is pic


22.jpg



3770k @ 4.5 1.190v
Dorsair Dominator GT 2133mhz 1.5v
Maximus V Gene
H100
Gtx 680 SLI
Thermaltake Level 10 GT snow
Samsung 830 256gb
Seasonic x-1050
 
maxkilla said:
passed Linx test 20times easy with 1.150v

But not like Sandy Bridge it failed on Prime Test

Weird, Sandy was Easier on Prime and Harder on Linx.

anyways i managed to pass both of them @ 1.180

but i decided to use it @ 1.190 for safe.

my room Temp was around 68 here in NY

and this Temp is pretty nice i guess.

here is pic

3770k @ 4.5 1.190v
Dorsair Dominator GT 2133mhz 1.5v
Maximus V Gene
H100
Gtx 680 SLI
Thermaltake Level 10 GT snow
Samsung 830 256gb
Seasonic x-1050
Click to expand...

Your gigaflops is the same as mine. I'm running 4.5ghz at 1.250 volts. It was a .5 volt lower but firefoxed crashed while I was stress testing. I wanted to make sure my system was stable under load while using it. One of my cores hits 82c with hyperthreading on using linx with avx. I'm using the same speed ram and voltage too.
 
Galvin said:
Your gigaflops is the same as mine. I'm running 4.5ghz at 1.250 volts. It was a .5 volt lower but firefoxed crashed while I was stress testing. I wanted to make sure my system was stable under load while using it. One of my cores hits 82c with hyperthreading on using linx with avx. I'm using the same speed ram and voltage too.
Click to expand...


To Be Honest.

Gigaflops.... No Human Beings can feel the Difference between 90 g flops and 130g flops.
it's just a number.
i will say anything over 100gflops with OCed sandy or Ivy is good.
i can push my Gflops around 130 by giving it more voltages around 1.60v and squeeze timings.

and as i described above
i was able to pass linx at 1.150v 20times with no error and i gave it 8GIG test
not short 3GIGs as other guys above.

if you dont wanna see BSOD while surfing the web you should pass the Prime for atleast 2hours as well.

and i recommend you to give it 0.005 volt above the Settings you have passed on both prime and linx good luck
 
After my initial Asus auto OC I was at 4.5 I did some more tinkering and now I'm at 4604.4. cpu v 1.365 DDR3@1600 I have run intel burn, prime 95, and cinebench with no problems with max temp 78. I did have it up to 4704.4 but cinebench would crash with cpuv at 1.4 Looks like this is max for me.

Core i7-3770k
Corsair H100 in 700D
GSkill DDR3 1600
OCZ Agility 3 240gig
 
maxkilla said:
To Be Honest.

Gigaflops.... No Human Beings can feel the Difference between 90 g flops and 130g flops.
it's just a number.
i will say anything over 100gflops with OCed sandy or Ivy is good.
i can push my Gflops around 130 by giving it more voltages around 1.60v and squeeze timings.

and as i described above
i was able to pass linx at 1.150v 20times with no error and i gave it 8GIG test
not short 3GIGs as other guys above.

if you dont wanna see BSOD while surfing the web you should pass the Prime for atleast 2hours as well.

and i recommend you to give it 0.005 volt above the Settings you have passed on both prime and linx good luck
Click to expand...
the reason i ask is, i plan on alot of video encoding in my system.
it is my understanding that both clock freq and gflops do matter, but
i guess i will likely have to live with 108gflops and 4700 since so far in my testing i cant get those numbers up by lowering volt and freq like these others here
are touting. i am going to try to overclock mem next to see if i cant get
numbers higher. this board and chip are supoose to be the most forgiving
mem system built yet so lets me what it can do.

jet

i guess some chips yield better gflops than others????
 
maxkilla said:
To Be Honest.

Gigaflops.... No Human Beings can feel the Difference between 90 g flops and 130g flops.
it's just a number.
i will say anything over 100gflops with OCed sandy or Ivy is good.
i can push my Gflops around 130 by giving it more voltages around 1.60v and squeeze timings.

and as i described above
i was able to pass linx at 1.150v 20times with no error and i gave it 8GIG test
not short 3GIGs as other guys above.

if you dont wanna see BSOD while surfing the web you should pass the Prime for atleast 2hours as well.

and i recommend you to give it 0.005 volt above the Settings you have passed on both prime and linx good luck
Click to expand...

I've done over 30 hours of stress testing no issues. I started at 1.1v and went thru around 30+ reboots before BSOD's stopped. Then started stress testing from there. I went in 0.005 increments.
 
jetli said:
the reason i ask is, i plan on alot of video encoding in my system.
it is my understanding that both clock freq and gflops do matter, but
i guess i will likely have to live with 108gflops and 4700 since so far in my testing i cant get those numbers up by lowering volt and freq like these others here
are touting. i am going to try to overclock mem next to see if i cant get
numbers higher. this board and chip are supoose to be the most forgiving
mem system built yet so lets me what it can do.

jet

i guess some chips yield better gflops than others????
Click to expand...


Asrock Motherboards give you highest Gflops scores.
and 3570k also gives more Gflops Due to no Hyper Thread

but overall in game,stable,video encode etc,
Asrock can't beat Asus speacially ROG series Many Reviews Approves it.

remember this,
End Of Overclocking = Stock speed.
 
RamonGTP said:
You're beating heat by lowering voltage to the point that the CPU is wasting CPU cycles due to calculation errors. If it were that easy I assure you Inte, engineers would have thought of it. You didn't outsmart them. Setting voltage to just above the point you don't crash doesn't equal stable.
Click to expand...


You are mistaken about my cpu.

I lowered my vocore voltage and tweaked my memory and my gflop perfromance increased from the higher voltage settings.

I see lower heat helping CPU perform better. I understand what you are saying about the CPU possibly not performing as well with lower voltage but that isn't happening with my cpu as the linX performance shows above.


My IB runs fine at low voltages.
 
You must log in or register to reply here.
Back
Top