Haswell Voltage

[Mitrailleuse]

Hello,

i was wondering,if i keep my chip at 1.42V 4.5GHZ,with speedstep that lowers freq and voltage while idling,

should i expect degradation sometime?

how can i know if it degraded?

can i expect to still use this chip the following 3-5 years?

[Lays]

I think the "recommended max voltage" is 1.3v so I'd assume over time you'll see degradation, but I'm not quite sure how much of an impact it'll make on lifespan

[AlwaysFSX]

1.42 volts is a little high, no?

[beebskadoo]

1.42@4.5......that seems like a shiz load of volts for that clock speed. 

[Giggitygoebbels]

Did you lose the lottery?

[mvitkun]

Hello,

i was wondering,if i keep my chip at 1.42V 4.5GHZ,with speedstep that lowers freq and voltage while idling,

should i expect degradation sometime?

how can i know if it degraded?

can i expect to still use this chip the following 3-5 years?

well I would do the following.

enable OSD on msi afterburner (or a similar OSD utility such as Precision X) and then download HWinfo.

http://www.hwinfo.com/download.php

 

Run HWinfo in sensors only mode.

Then click on the button at the bottom that says "configure sensors."

Click on the OSD (RTSS) tab.

then look for VCore.

when you find it click on it to highlight it in blue and click the box that says "show value in OS" below it.

 

then while keeping both the msi OSD and HWinfo open play some intensive games or do whatever you usually do on your pc and see what the vcore is usually at.

I'd say if the VCore usually hovers at or below 1.35v you should be more or less fine,and it shouldn't affect the cpus lifespan in any significant way.

If it usually hovers at 1.4+v then I'd say maybe consider downclocking to 4.2 or 4.3ghz and see if that allows you to keep a lower voltage,because 1.42v is quite high for a daily clock unless you have a very frequent upgrade cycle and can afford to replace the cpu within the next year or two.

[pummetje]

thats waaay too much voltage 

haswell runs really hot 

 

concider a 4.2ghz overclock or something less

 

if you dont know what your doing dont overclock your cpu spend a couple of hours reading and watching overclocking guides 

[ViruZ_HUN]

Hello,

i was wondering,if i keep my chip at 1.42V 4.5GHZ,with speedstep that lowers freq and voltage while idling,

should i expect degradation sometime?

how can i know if it degraded?

can i expect to still use this chip the following 3-5 years?

i think you lost the lottery

 

Go for 5 ghz on 1.7 voltage. NO PROBLEM:D

[beebskadoo]

i think you lost the lottery

Go for 5 ghz on 1.7 voltage. NO PROBLEM:D

yeah....no

[Mitrailleuse]

I did indeed lost the silicone lottery,

i used to run 4.2GHZ @ 1.27,i think ill go back to that.

[TechFan@ic]

Haswell is a pain to overclock, hopefully the Haswell refresh will be better in that regard.

[WoodenMarker]

1.42v while not recommended is perfectly safe.

[deathjester]

1.42v while not recommended is perfectly safe.

 

LOL what? Past 1.3 is considered pushing it on HASWELL. Haswell is not Sandy Bridge or AMD.

 

4.2 at 1.27 is the smart move OP. You lost the silicon lottery. My friends is just as bad. It happens. It sucks. Nothing you can do about it. That said? 4.2 on Haswell is still stupidly fast and you might gain 1.5 fps (sometimes zero fps cus you are GPU limited) going to 4.5ghz in actual games. It simply isn't worth it. Now 3 years from now, if you are considering upgrading (which is in no way guaranteed since sandy bridges are still relevant), then push the chip till it blows up, because you have the money to replace it, or replacing it might actually be worth it (probably not, I think Sandy Bridge will still be awesome and we will have more stupid die shrinks and even worse OC chips).

 

Woodenmarker you have told people 72.5C is a good IDLE or light load temp on a Haswell and now said 1.42 v is ok on a Haswell. Are you purposely trying to get people to break their @!#$? I mean seriously dude. What is your deal. 

[WoodenMarker]

@deathjester I never said 72.5c is a good temp. Don't put words in my mouth. If you want to quote something I said, do that instead.

1.4v is the max recommended safe voltage for haswell according to intel's spec. 1.42v isn't much worse.

[deathjester]

@deathjester I never said 72.5c is a good temp. Don't put words in my mouth. If you want to quote something I said, do that instead.

1.4v is the max recommended safe voltage for haswell according to intel's spec. 1.42v isn't much worse.

 

Uh huh. That is why some motherboard manufacturers BLOCKED you from going past 1.3v initially. Hell MSI made the first overclocking motherboard, and they were like 1.3v plus? HELL NO.

 

What would they know. Woodenmarker says it safe ,when he doesn't even understand that Haswell's will run at 30-40C at light load (and would even at a ridiculously high voltage even on the freakin stock cooler) because that is what they were designed to do at .1-.5 volts. Run it at light load 72.5 C and throttle temps in games, and 1.4 plus volts. 40 percent more voltage going into a chip = GOOD TO FREAKING GO. Throttle temps? Perfect.

 

Woodenmarker can you make your own blog please? You could call it "on-fire computing", or "en fuego". I really, really hope you don't work in the computer industry. 

[WoodenMarker]

Uh huh. That is why some motherboard manufacturers BLOCKED you from going past 1.3v initially. Hell MSI made the first overclocking motherboard, and they were like 1.3v plus? HELL NO.

 

What would they know. Woodenmarker says it safe ,when he doesn't even understand that Haswell's will run at 30-40C at light load (and would even at a ridiculously high voltage even on the freakin stock cooler) because that is what they were designed to do at .1-.5 volts. Run it at light load 72.5 C and throttle temps in games, and 1.4 plus volts. 40 percent more voltage going into a chip = GOOD TO FREAKING GO. Throttle temps? Perfect.

Woodenmarker can you make your own blog please? You could call it "on-fire computing", or "en fuego". I really, really hope you don't work in the computer industry. 

1.3v is simply the recommended threshold. Intel specs it's safe up to 1.4v.

 

I'd appreciate it if you would take less jabs at me and just contribute to the conversation instead.

[deathjester]

1.3v is simply the recommended threshold. Intel specs it's safe up to 1.4v.

 

I'd appreciate it if you would take less jabs at me and just contribute to the conversation instead.

 

I would appreciate you stop telling people that 72.5 C idle temps are safe and telling them that BENCHMARKING EPEEN voltage is safe for 24/7 use, that will give them slight to ZERO performance gains unless they play Guild Wars 2 or Flight Simulator X. You are giving BAD computer advice that could cost people real life money and you have a trusted advisor tag, which makes people think it is actually good. 

 

Want to know why they say 1.3v? Cus that is what won't cause HORRIBLE chip degradation . It is the same reason they give you a temp on the chip at normal load (like an encode since stress testing runs hotter).

 

You are advising people that the LOAD is idle or light load temps and that the safe voltage can safely be broken by a TON. You don't even have an understanding of how Haswell architecture works on light loads, but continue to give people advice on the parts. Your posts are pure lunacy.

 

Where are you getting this stuff. First you say 1.4, then you said 1.3. Are you just pulling these numbers out of the air? What do you do for a living. Please tell me. What gives you the right to tell people that specs past the recommended are safe. Where are your tests. Where is your evidence on HASWELL? There are a billion broken computer parts from when people passed the recommended voltage/temps. That is the mountain of evidence that anyone should need. 

 

Just because someone is running a brand new chip for less then half a year past recommended thermals and voltage, doesn't mean that is safe. What are you going to try to report me now for personal attacks? You are BREAKING PEOPLES @#$%. You recommended the stock cooler for rendering of all things on a 8 core chip and told them that 90C is perfectly fine.  

 

I mean you can report me, but I will GLADLY go if this forum allows this kind of nonsense to be thrown around at "trusted" advice. "Trusted" advice from someone who has never even worked with ONE haswell CPU and has no idea how they work. Makes PERFECT sense. 

[WoodenMarker]

@deathjester

I suggest for you not to derail the thread and talk about how you confused my explanation for Tcase temps with recommended temps. 

 

1.3v is recommended because raising the voltage usually doesn't do much to help get higher and more stable oc's. It isn't Intel's recommended voltage but the overclocking community's for oc'ing. Intel doesn't spec any specific guideline when it comes to oc'ing. 1.4v is Intel's specc'ed safe voltage threshold.

I didn't change anything. I'm talking about two different thresholds that come from two different but not always separate groups: Intel and the overclocking community.

[deathjester]

@deathjester

I suggest for you not to derail the thread and talk about how you confused my explanation for Tcase temps with recommended temps. 

 

1.3v is recommended because raising the voltage usually doesn't do much to help get higher and more stable oc's. It isn't Intel's recommended voltage but the overclocking community's for oc'ing. Intel doesn't spec any specific guideline when it comes to oc'ing. 1.4v is Intel's specc'ed safe voltage threshold.

I didn't change anything. I'm talking about two different thresholds that come from two different but not always separate groups: Intel and the overclocking community.

 

You don't even know that Haswell downclocks and undervolts on light loads. This conversation is pointless. You win. 

 

@ the op? If your chip breaks? You have only yourself to blame for listening to people that have never even worked with a Haswell, or had one in front of them and doesn't even understand how the architecture works.

 

Then again? You already figured out to do the right thing at clock it at 4.2. Why? Because you have a Haswell and don't want it to break, and you have money invested in it and you know that Intel probably puts a number there for a reason, and listening to someone who doesn't even understand what that number means? Ain't good.

[martinrox1568]

Haswell is a pain to overclock, hopefully the Haswell refresh will be better in that regard.

It won't. Lets just face the facts and accept that Intel has given up on enthusiasts.

[mgsstar]

0V-1.3V For people concerned about volts\heat. up to 1.3V is playing it SAFE. be happy with your clocks at 1.3V. using an air cooler. I have found that up to 1.3V will offer the best performance ratio in terms of GHz\Watt

1.3V-1.44V Still safe without risking CPU degradation . Requires a good cooling solution ( top notch AIO cooler) NOT recommended for people using air cooling or people concerned with heat or volts applied

1.45V and Higher. For extreme Overclockers only. from my testing going over 1.45V to achieve higher clock speeds is pointless. the performance\Watt drops heavily and massive amounts of heat is generated.

Source:http://linustechtips.com/main/topic/41234-intel-haswell-4670k-4770k-overclocking-guide/ from @ProKoN

 

Since you have an H100i it should be fine and it is best to keep the load temps under 80C.

[deathjester]

It won't. Lets just face the facts and accept that Intel has given up on enthusiasts.

^ This. Die shrinks and worse OC'rs from here on out. CEO just said he regrets Intel not spending MORE time on tablet chips and that it was a mistake to cater to enthusiasts in an article I read a few weeks back on Tom's Hardware if I remember right.

OC'd Sandybridges are the Connor MaCleod's of CPU's. They are immortal. They cannot die. There can be only one. Wish I had one. Must feel awesome to see 2 series pass and still be near the top on a 30 dollar air cooler, while getting cooler temps. I got lucky on my Haswell, but a friends is slower then many of the Sandy's on this forum on a ROG board...

Source:http://linustechtips.com/main/topic/41234-intel-haswell-4670k-4770k-overclocking-guide/ from @ProKoN

Since you have an H100i it should be fine and it is best to keep the load temps under 80C.

Funny how Prokon has a 4.9 GHZ OC on a 4770k on the cine bench thread and runs his chip at 4.6 (look at his sig) on DUAL RAD WATER huh? I wonder why? Oh yeah, because he doesn't want to lose his chip. It's almost as if...he is following Asus's recommendations on his own chip that he doesn't want to lose. How SHOCKING.

Here is Asus's guide. 90C on rendering? GOOD TO GO? See that anywhere? How about pushing past 1.3 volts?

http://rog.asus.com/242142013/labels/rog-exclusive/maximus-vi-series-uefi-guide-for-overclocking/

What is Prokon's voltage guide based off of? His personal thoughts on the matter? Extensive multi year tests? The fact that the chips didn't blow up in a whole day of testing? His one chip, when different batches of Haswell run hotter then others? Does he have an engineering sample which is why everyone thinks they should be at 4.5-4.7 due to BS initial reviews (just like the R9 290x benchmarks that were BS). Was he part of the engineering team on the architecture? He obviously knows more then they do to suggest 1.45v as a safe voltage. More then MSI also, who limited CPU voltage to 1.3 initially. More then Asus. Origin PC? Hell they guarantee a max OC of 4.4 ghz on WATER and those things cost double what it would to build one yourself. They actually have to GUARANTEE their product, and realize that gaming/streaming will give you render temps.

Anyone who is running 24/7 at 1.45 V and thinking it's safe because some random dude on the internet says so? Deserves to lose their chip. These chips came out in the summer. No one knows jack yet. They know that Intel said 1.3v. Pushing past that? Probably not the best idea if you don't plan on upgrading in 2 years. Pushing way past that? Again. You deserve to lose your chip for putting faith in a "professional overclocker". Changing numbers on a Bios doesn't make you special. It means you have the cash to lose money when you click the things on a BIOS. That guy probably will upgrade to broadwell. Will you?

A bunch of people pushed Ivy too hard. Those chips didn't last long. Why did they push them too hard? Cus of Sandy. Sandy was 2 die shrinks ago. Haswell isn't sandy. AMD chips can run a ton of voltage to. Why? Cus they are like freakin I7-920's at 4ghz.

You wan't to treat a Haswell like a Sandy or an AMD chip? Go ahead. Just don't expect it to last. Some random dude on the internet as "proof" with no test parameters and no hard evidence of "no chip degradation" wouldn't be the source I would be trusting.

Prokon also recommends stress testing for half a day at least on cache. Have fun with that. He could have clicked just cache in Aida 64 and saved himself about 11 hours. He took DAYS to find his max cache. You want to run your chips for days at high temps and volts. Go for it. Follow his guide. Break your chip. Ignore the companies specs, the companies that have to guarantee their product, and even the motherboard manufactures. Be a rebel and have your chip go out in a blaze of glory. See you on Broadwell. Have fun overclocking that...

Or you could do what Prokon does (after his guide), and run your chip at a reasonable speed and temp. Maybe he should update his guide huh?

[Analog]

Running a Haswell chip with more than 1.3V will significantly lower it's life span especially when there is no adequate cooling. By adequate cooling I mean a triple 120mm radiator. There have been Haswell processors dying in less than a month because of the use of excessive voltage. If you value your chip and indeed want to use it for 5 years, please, for your sake, back the voltage down. 

[ProKoN]

Funny how Prokon has a 4.9 GHZ OC on a 4770k on the cine bench thread and runs his chip at 4.6 (look at his sig) on DUAL RAD WATER huh? I wonder why? Oh  yeah, because he doesn't want to lose his chip. It's almost as if...he is following Asus's recommendations on his own chip that he doesn't want to lose. How SHOCKING.

 

Here is Asus's guide. 90C on rendering? GOOD TO GO? See that anywhere? How about pushing past 1.3 volts?

 

http://rog.asus.com/242142013/labels/rog-exclusive/maximus-vi-series-uefi-guide-for-overclocking/

 

What is Prokon's voltage guide based off of? His personal thoughts on the matter? Extensive multi year tests? The fact that the chips didn't blow up in a whole day of testing? His one chip, when different batches of Haswell run hotter then others? Does he have an engineering sample which is why everyone thinks they should be at 4.5-4.7 due to BS initial reviews (just like the R9 290x benchmarks that were BS). Was he part of the engineering team on the architecture? He obviously knows more then they do to suggest 1.45v as a safe voltage. More then MSI also, who limited CPU voltage to 1.3 initially. More then Asus. Origin PC? Hell they guarantee a max OC of 4.4 ghz on WATER and those things cost double what it would to build one yourself. They actually have to GUARANTEE their product, and realize that gaming/streaming will give you render temps.  

 

Anyone who is running 24/7 at 1.45 V and thinking it's safe because some random dude on the internet says so? Deserves to lose their chip. These chips came out in the summer. No one knows jack yet. They know that Intel said 1.3v. Pushing past that? Probably not the best idea if you don't plan on upgrading in 2 years.  Pushing way past that? Again. You deserve to lose your chip for putting faith in a "professional overclocker". Changing numbers on a Bios doesn't make you special. It means you have the cash to lose money when you click the things on a BIOS. That guy probably will upgrade to broadwell. Will you?

 

A bunch of people pushed Ivy too hard. Those chips didn't last long. Why did they push them too hard? Cus of Sandy. Sandy was 2 die shrinks ago. Haswell isn't sandy. AMD chips can run a ton of voltage to. Why? Cus they are like freakin I7-920's at 4ghz. 

 

You wan't to treat a Haswell like a Sandy or an AMD chip? Go ahead. Just don't expect it to last. Some random dude on the internet as "proof" with no test parameters and no hard evidence of "no chip degradation" wouldn't be the source I would be trusting.

 

Prokon also recommends stress testing for half a day at least on cache. Have fun with that. He could have clicked just cache in Aida 64 and saved himself about 11 hours. He took DAYS to find his max cache. You want to run your chips for days at high temps and volts. Go for it. Follow his guide. Break your chip. Ignore the companies specs, the companies that have to guarantee their product, and even the motherboard manufactures. Be a rebel and have your chip go out in a blaze of glory. See you on Broadwell. Have fun overclocking that...

 

Or you could do what Prokon does (after his guide), and run your chip at a reasonable speed and temp. Maybe he should update his guide huh?

 first off my 4770k never hit 4.9GHz Ever lol. where did you see that.

 

my 4770k hits 4.7 ghz as a max overclock so i scale back to 4.6 for daily usage.

 

I have had haswell since launch and i can wholeheartedly say if you keep your chip under 1.45 volts your good for 24\7 usage.

 

i have also tested 3 z87 boards and can say all z87 boards are not equal in terms of overclocking.

 

i always recommend 8 or more hours of stress testing as it simulates a " full work day" on ANY PLATFORM

 

I have received a lot of thank yous for the guide as it has helped alot of folks out there starting from scratch.

 

what is your insightful contribution to the community? how many years experience do you have?

 

how long have you had haswell in your hands?

 

if you think pushing past 1.3v will blow your chip, your wrong. pushing past 1.6V and higher will blow your chip, Ive killed one, but not to worry intel offers a three year warranty for anyone who can provide proof of purchase..

[deathjester]

 first off my 4770k never hit 4.9GHz Ever lol. where did you see that.

 

my 4770k hits 4.7 ghz as a max overclock so i scale back to 4.6 for daily usage.

 

I have had haswell since launch and i can wholeheartedly say if you keep your chip under 1.45 volts your good for 24\7 usage.

 

i have also tested 3 z87 boards and can say all z87 boards are not equal in terms of overclocking.

 

i always recommend 8 or more hours of stress testing as it simulates a " full work day" on ANY PLATFORM

 

I have received a lot of thank yous for the guide as it has helped alot of folks out there starting from scratch.

 

what is your insightful contribution to the community? how many years experience do you have?

 

how long have you had haswell in your hands?

 

if you think pushing past 1.3v will blow your chip, your wrong. pushing past 1.6V and higher will blow your chip, Ive killed one, but not to worry intel offers a three year warranty for anyone who can provide proof of purchase..

You posted it on this forum. 

 

https://docs.google.com/spreadsheet/ccc?key=0AlC81MjwelBgdEZNV3l6aHl1eUNwSUR4Rml0MXMzN1E&usp=sharing#gid=0

 

http://linustechtips.com/main/topic/62476-post-your-cinebench-r15-scores/

 

"i have also tested 3 z87 boards and can say all z87 boards are not equal in terms of overclocking".

 

That's cool man. I have the cheapest Z87 MSI makes. Chip can do 4.7 while staying within intel recommended thermal/voltage on your cooler. Instead of telling people that all z87 clock "about the same", I should be telling them to get the cheaper Z87 over your Mpower.  I can render at 4.6 with max real world temps of 75 C on a 30 dollar Evo 212. Obviously the cheaper board is better...or it doesn't matter (it doesn't).

 

"i always recommend 8 or more hours of stress testing as it simulates a " full work day" on ANY PLATFORM"

 

Or you could render videos with cooler temps on Asus realbench , or play games with cooler temps. What stress tester are you using here? Aida 64? That is one of the only ones withing 5 degrees of a real workload and none of these stress tests reflects true stability.

 

http://www.hardocp.com/article/2013/06/01/intel_haswell_i74770k_ipc_overclocking_review/7#.Urn3ofRDu30

 

These professionals say AIDA is the best and it is still not that great as a stability test. Asus realbench and clicking h.264 (simulating handbreak) or actually encoding real videos is better. Again. I like Aida for a quick 10 min run, along with 10 min runs of other things, before truly testing with encodes and games (at much lower temps then silly stress tests), and I REALLY like it for testing cache. This 8 hours stuff though? Again you are just picking a number out of your hat. Just because your method saw you get no crashes means nothing. My method and professionals method takes a fraction of the time, has lower temps and lets you do things that are fun.

 

"if you think pushing past 1.3v will blow your chip, your wrong. pushing past 1.6V and higher will blow your chip, Ive killed one, but not to worry intel offers a three year warranty for anyone who can provide proof of purchase".

 

Based on what? The dead Haswell's that another user posted about that weren't even close to 1.6v? Congratulations on killing a chip. The OP wanted to know if his voltage was safe. It isn't per Intel. I don't think he is interested in putting up E-peen benchmarks. He is interested in a 24/7 system. You know...just like yours at 4.6 on water which is a whole 100 mhz faster then my 30 dollar evo (because boards matter LOL). You killed a chip running past specs, and then immorally got a new chip from Intel, when you knew you were pushing a product too hard. People like you raise the price on our CPU's. Thanks a ton. You really are "helping the community" by putting up artificial, non 24/7 benchmarks that you have no intention of running a 24/7 system at. Did you even post temps in your screenshot on the cinebench thread? If so? That number is meaningless. 

 

The only thing you need to do for 8 hours on a Haswell, is memory testing, and you don't really need that. 1 pass of memtest and 5 passes on test 5 is pretty darn good. The reason you test for a long time with it? You don't have to worry about temps in memtest and it can be done while you sleep.

 

I mean you can disagree with me, but you disagree with the MB manufacturers, intel, the people who broke their chips, yourself who broke a chip and run at 4.6 ghz instead of 4.896 or whatever it was you posted in the cinebench thread. You also disagree with all the professionals who get their hands on much more chips and who wouldn't think of running anywhere near 1.4v let alone 1.6v in anything more then a quick benchmark and their top "overclock" or benchmark is usually close to 1.3v. Add to all that? Those were engineering chips that were better then what retailers got. They ran cooler, because they were all doing 4.7 ghz below 1.25.