Overclock.net › Forums › Intel › Intel CPUs › Did we ever come a conclusion on safe SB voltages?
New Posts  All Forums:Forum Nav:

Did we ever come a conclusion on safe SB voltages? - Page 2

post #11 of 93
I did read something from ASUS saying they dont recommend going over 1.42 for 24/7 usage even on good temps. I wish I could remember how I found it but it does sound right. I have also heard 1.45 from many other sources as well. I just keep mine under 1.42v personally for 24/7.
post #12 of 93
Quote:
Originally Posted by TwoCables View Post
Besides, 0.85V - 1.3625V is just the VID Voltage Range.
This may be what I am thinking of. More confusing yet, my Cpre 2 Duo E8600's box say "1.26V max" right on it's side.
Quote:
Originally Posted by TwoCables View Post
The VID range for Sandy Bridge being 0.25V to 1.52V makes me wonder just what the hell is going on. lol I mean damn, are these CPUs tanks?
Yeah, this is also what makes it harder to believe. I mean, I' love to be wrong, but there's no way these things are good up to 1.52V or well beyond, right?

All my initial post was saying was to exercise caution. That number is just as much speculation as any other opinions that have been put forth, as while the number was printed, we know nothing of the other stuff that, the other variables, what the number really means, etc.
post #13 of 93
Quote:
Originally Posted by UsedPaperclip View Post
I did read something from ASUS saying they dont recommend going over 1.42 for 24/7 usage even on good temps. I wish I could remember how I found it but it does sound right. I have also heard 1.45 from many other sources as well. I just keep mine under 1.42v personally for 24/7.
The 1.45V I'm talking about is for 45nm Core2 CPUs (and I think 45nm Pentium Dual Cores too).

Here's the article you may be thinking of that talks about 1.425V being the safe 24/7 limit:

http://hardforum.com/showthread.php?t=1578110


There's also this reply JJ gave to another member talking about how he recommends 1.425V as a 24/7 maximum:

http://hardforum.com/showpost.php?p=...2&postcount=18
It's a computer!
(19 items)
 
  
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5GHz (1.368-1.384V fixed voltage) ASUS P8P67 EVO B3 (UEFI ver. 1850) GTX 780 ASUS DirectCU II (1228 / 6300, 1.180V) G.SKILL Ripjaws X 8GB (2 x 4GB) 1866MHz, CL9 
Hard DriveHard DriveHard DriveOptical Drive
250 GB Samsung 840 EVO (C:\) 250 GB Samsung 840 EVO (D:\) 150 GB WD VelociRaptor Samsung SH-S243N 24x DVD Burner 
Optical DriveCoolingOSMonitor
Samsung SH-S203N 20X DVD Burner Thermaltake Frio Win 7 Home Premium x64 SP1 Retail AOC G2460PG (24" 1920 x 1080 144Hz G-SYNC) 
KeyboardPowerCaseMouse
Filco Majestouch 104-key Cherry MX Blues w/NKRO Corsair HX650 (Bronze, ordered on 12-12-2009) CM 690 Intellimouse Optical (1.1A) 1000Hz polling rate 
Mouse PadAudioAudio
Basic, but premium round X-Fi Titanium HD Klipsch ProMedia 2.1 (with 16 AWG Monster Cable... 
  hide details  
Reply
It's a computer!
(19 items)
 
  
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5GHz (1.368-1.384V fixed voltage) ASUS P8P67 EVO B3 (UEFI ver. 1850) GTX 780 ASUS DirectCU II (1228 / 6300, 1.180V) G.SKILL Ripjaws X 8GB (2 x 4GB) 1866MHz, CL9 
Hard DriveHard DriveHard DriveOptical Drive
250 GB Samsung 840 EVO (C:\) 250 GB Samsung 840 EVO (D:\) 150 GB WD VelociRaptor Samsung SH-S243N 24x DVD Burner 
Optical DriveCoolingOSMonitor
Samsung SH-S203N 20X DVD Burner Thermaltake Frio Win 7 Home Premium x64 SP1 Retail AOC G2460PG (24" 1920 x 1080 144Hz G-SYNC) 
KeyboardPowerCaseMouse
Filco Majestouch 104-key Cherry MX Blues w/NKRO Corsair HX650 (Bronze, ordered on 12-12-2009) CM 690 Intellimouse Optical (1.1A) 1000Hz polling rate 
Mouse PadAudioAudio
Basic, but premium round X-Fi Titanium HD Klipsch ProMedia 2.1 (with 16 AWG Monster Cable... 
  hide details  
Reply
post #14 of 93
Quote:
Originally Posted by TwoCables View Post
The 1.45V I'm talking about is for 45nm Core2 CPUs (and I think 45nm Pentium Dual Cores too).

Here's the article you may be thinking of that talks about 1.425V being the safe 24/7 limit:

http://hardforum.com/showthread.php?t=1578110


There's also this reply JJ gave to another member talking about how he recommends 1.425V as a 24/7 maximum:

http://hardforum.com/showpost.php?p=...2&postcount=18
I dont think that was it exactly they pretty much laid it out there on the voltage recommendation but that is still a great thread to read when overclocking on sandy bridge.
post #15 of 93
Thread Starter 
Quote:
Originally Posted by Princess Garnet View Post
I know what I'm about to question doesn't give a specific answer, but I think few here will have more than speculation.

What if that 1.52V number is speculation itself. Sure, the number is printed, but that number is meaningless unless we know everything about it, not just seeing "VID" and assuming "hey, it's the maximum safe voltage regardless of other variables". Not only that, but the notes themselves say it can vary from CPU to CPU; that they can have their own range. Under what conditions is that 1.52V number okay? I'm sure the lower end listed (0.25V) wouldn't be possible across the board either, so... I'd exercise some caution. That 1.52V number might not mean it would be okay at that voltage at high frequencies/temperatures/load level for too long.

I found this from another thread...
The feedback from those using as much, or more, might be a little telling, but it's been just under ten months. How many have been running them that way the whole time with no side effects?

I mean, if someone would prove me wrong about my doubts on the 1.52V number, go ahead. I'm going to be moving to a 32nm Core i5 soon, and I'd love for that to be the limit, but I'm on a 45nm Core 2 with a limit of 1.3625V (if I remember right), so forgive me for being a bit hesitant of it.
I'm not trying to claim that 1.52V is the safe limit for SB. I agree with you, it seems high. But it's the only number that is actually official, as opposed to all the speculation. Yes, Intel could be speculating as well, but I'm more inclined to believe Intel's speculation than forum goers.

Furthermore, I think there is a huge misconception about what the voltage actually does for a chip, and how that relates to the manufacturing process. If you haven't studied it, I'd encourage anyone to go read about the CMOS manufacturing process.

In this case, for 32nm, Intel is using a combination of fancy high-k dielectric materials and metal for its gates. These materials have never before been used on a chip. These materials were first tested with the 45nm process, but are now appearing as a much-advanced second generation in the 32nm process. In this case, the high voltage could be completely safe because this new material is much more effective at limiting electron tunneling than anything used previously.
Edited by Nick2253 - 10/28/11 at 4:55pm
post #16 of 93
Quote:
Originally Posted by TwoCables View Post
The 1.45V I'm talking about is for 45nm Core2 CPUs (and I think 45nm Pentium Dual Cores too).

Here's the article you may be thinking of that talks about 1.425V being the safe 24/7 limit:

http://hardforum.com/showthread.php?t=1578110
Yes that was it, pretty much what I like to stick to.
post #17 of 93
Quote:
Originally Posted by Nick2253 View Post
I'm not trying to claim that 1.52V is the limit for SB. I agree with you, it seems high. But it's the only number that is actually official, as opposed to all the speculation. Yes, Intel could be speculating as well, but I'm more inclined to believe Intel's speculation than forum goers.

Furthermore, I think there is a huge misconception about what the voltage actually does for a chip, and how that relates to the manufacturing process. If you haven't studied it, I'd encourage anyone to go read about the CMOS manufacturing process.

In this case, for 32nm, Intel is using a combination of fancy high-k dialectric materials and metal for its gates. These materials have never before been used on a chip. In this case, the high voltage could be completely safe because this new material is much more effective at limiting electron tunneling than anything used previously.
This would explain the 0.25V side of the VID range as well as the 1.52V. I mean Intel has always listed a VID range while the actual maximum safe voltage is quite a bit higher (like 1.45V while the VID range was 0.85V to 1.3625V).
It's a computer!
(19 items)
 
  
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5GHz (1.368-1.384V fixed voltage) ASUS P8P67 EVO B3 (UEFI ver. 1850) GTX 780 ASUS DirectCU II (1228 / 6300, 1.180V) G.SKILL Ripjaws X 8GB (2 x 4GB) 1866MHz, CL9 
Hard DriveHard DriveHard DriveOptical Drive
250 GB Samsung 840 EVO (C:\) 250 GB Samsung 840 EVO (D:\) 150 GB WD VelociRaptor Samsung SH-S243N 24x DVD Burner 
Optical DriveCoolingOSMonitor
Samsung SH-S203N 20X DVD Burner Thermaltake Frio Win 7 Home Premium x64 SP1 Retail AOC G2460PG (24" 1920 x 1080 144Hz G-SYNC) 
KeyboardPowerCaseMouse
Filco Majestouch 104-key Cherry MX Blues w/NKRO Corsair HX650 (Bronze, ordered on 12-12-2009) CM 690 Intellimouse Optical (1.1A) 1000Hz polling rate 
Mouse PadAudioAudio
Basic, but premium round X-Fi Titanium HD Klipsch ProMedia 2.1 (with 16 AWG Monster Cable... 
  hide details  
Reply
It's a computer!
(19 items)
 
  
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5GHz (1.368-1.384V fixed voltage) ASUS P8P67 EVO B3 (UEFI ver. 1850) GTX 780 ASUS DirectCU II (1228 / 6300, 1.180V) G.SKILL Ripjaws X 8GB (2 x 4GB) 1866MHz, CL9 
Hard DriveHard DriveHard DriveOptical Drive
250 GB Samsung 840 EVO (C:\) 250 GB Samsung 840 EVO (D:\) 150 GB WD VelociRaptor Samsung SH-S243N 24x DVD Burner 
Optical DriveCoolingOSMonitor
Samsung SH-S203N 20X DVD Burner Thermaltake Frio Win 7 Home Premium x64 SP1 Retail AOC G2460PG (24" 1920 x 1080 144Hz G-SYNC) 
KeyboardPowerCaseMouse
Filco Majestouch 104-key Cherry MX Blues w/NKRO Corsair HX650 (Bronze, ordered on 12-12-2009) CM 690 Intellimouse Optical (1.1A) 1000Hz polling rate 
Mouse PadAudioAudio
Basic, but premium round X-Fi Titanium HD Klipsch ProMedia 2.1 (with 16 AWG Monster Cable... 
  hide details  
Reply
post #18 of 93
Id love it to be 1.52 as for me to go from 4.6 to 4.8 prime 24 hr stable takes me well into the 1.4s.

Personally i just dont feel comfortable with that as an eveyday voltage. Maybe if i could figure out the damned offset mode id be willing but not on manual mode.

Not to mention there hasnt been generally anything that i run that would significantly improve with a couple hundred mhz increase.
post #19 of 93
Quote:
Originally Posted by bce22 View Post
Id love it to be 1.52 as for me to go from 4.6 to 4.8 prime 24 hr stable takes me well into the 1.4s.

Personally i just dont feel comfortable with that as an eveyday voltage. Maybe if i could figure out the damned offset mode id be willing but not on manual mode.

Not to mention there hasnt been generally anything that i run that would significantly improve with a couple hundred mhz increase.
Oh, I am glad you asked about Offset mode. It's actually fairly easy to figure it out.

My first step is to set Load-Line Calibration to Ultra High for the smallest amount of vDroop possible while also avoiding vRise. This helps for what we will do next.

Look at this part of Real Temp (I recommend Real Temp 3.67):




Click that button if necessary.

If you get a very low number in comparison to mine here, then put some load on the CPU to see what your actual Voltage Identifier really is.

Now find the difference between the VID you get and the core voltage you want to see in CPU-Z while under full load, and then the Offset value you need to use is revealed. So what you do at this point is round up or down a little bit to see which Offset achieves the exact voltage you want (experimentation will be necessary).

For example: my VID flip-flops between 1.3561V and 1.3611V. The core voltage I wanted was a flip-flop between 1.392V and 1.400V. So the difference between these 4 numbers (the two VIDs and the two voltages) is 0.0359/0.0309 and 0.0439/0.0389. My Offset ended up needing to be +0.040V to achieve the voltage I want, but someone else with an identical system may end up needing +0.035V or +0.045V.

So yeah, that's basically it.

If you lower the LLC to High, then the Offset will need to be adjusted a bit to compensate. If the LLC is raised to Extreme, then of course the Offset will need to be adjusted the other way.
It's a computer!
(19 items)
 
  
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5GHz (1.368-1.384V fixed voltage) ASUS P8P67 EVO B3 (UEFI ver. 1850) GTX 780 ASUS DirectCU II (1228 / 6300, 1.180V) G.SKILL Ripjaws X 8GB (2 x 4GB) 1866MHz, CL9 
Hard DriveHard DriveHard DriveOptical Drive
250 GB Samsung 840 EVO (C:\) 250 GB Samsung 840 EVO (D:\) 150 GB WD VelociRaptor Samsung SH-S243N 24x DVD Burner 
Optical DriveCoolingOSMonitor
Samsung SH-S203N 20X DVD Burner Thermaltake Frio Win 7 Home Premium x64 SP1 Retail AOC G2460PG (24" 1920 x 1080 144Hz G-SYNC) 
KeyboardPowerCaseMouse
Filco Majestouch 104-key Cherry MX Blues w/NKRO Corsair HX650 (Bronze, ordered on 12-12-2009) CM 690 Intellimouse Optical (1.1A) 1000Hz polling rate 
Mouse PadAudioAudio
Basic, but premium round X-Fi Titanium HD Klipsch ProMedia 2.1 (with 16 AWG Monster Cable... 
  hide details  
Reply
It's a computer!
(19 items)
 
  
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5GHz (1.368-1.384V fixed voltage) ASUS P8P67 EVO B3 (UEFI ver. 1850) GTX 780 ASUS DirectCU II (1228 / 6300, 1.180V) G.SKILL Ripjaws X 8GB (2 x 4GB) 1866MHz, CL9 
Hard DriveHard DriveHard DriveOptical Drive
250 GB Samsung 840 EVO (C:\) 250 GB Samsung 840 EVO (D:\) 150 GB WD VelociRaptor Samsung SH-S243N 24x DVD Burner 
Optical DriveCoolingOSMonitor
Samsung SH-S203N 20X DVD Burner Thermaltake Frio Win 7 Home Premium x64 SP1 Retail AOC G2460PG (24" 1920 x 1080 144Hz G-SYNC) 
KeyboardPowerCaseMouse
Filco Majestouch 104-key Cherry MX Blues w/NKRO Corsair HX650 (Bronze, ordered on 12-12-2009) CM 690 Intellimouse Optical (1.1A) 1000Hz polling rate 
Mouse PadAudioAudio
Basic, but premium round X-Fi Titanium HD Klipsch ProMedia 2.1 (with 16 AWG Monster Cable... 
  hide details  
Reply
post #20 of 93
Two Cables,

Well if i knew how awesome your answer would be I would have asked ages ago. Thanks for the great post.

Sorry OP for semi-hijacking your thread.

Personally i think i will try to up my voltage to hit the 5ghz e-peen barrier using offset. Physcologically i feel better knowing the voltage will throttle down at idle.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › Did we ever come a conclusion on safe SB voltages?