(Subject to further experienced insight/input) this might be of interest to others requiring stable OC'ing for audio DSP related purposes.
I do apologize in advance: this CPU (3930k) and the term 'OC'ing', 'vCore' and 'Safe Temp' have been covered numerous times. However, I have never felt my questions and concerns directly addressed, to the point of conviction.
I've been running my 3.2GHz 3930k at 3.8GHz for the last 2 years (I was unaware that the motherboard 'decided' to run it at 3.8GHz as opposed to 3.2GHz): Seldom have I ventured into the BIOS for fear of 'royally ...', and it has served me very well, very! (the CPU/PC not the sheepish exploration!).
Before I continue any further, this is what I am working with, with respect to my PC:
64bit Windows 7 Pro computer:
#1 i7 3930k 3.20GHz (Sandybridge-E) Socket LGA2011
#2 ASUS Sabertooth X79 (Socket 2011) Motherboard
#3 Noctua NH-D14 (2011) Dual Radiator CPU Cooler
#4 2x Noctua NF-A14: above CPU cooler (breathing out) and underneath chassis (breathing in).
#5 2x Noctua NF-F12: behind CPU cooler (breathing out) and top front panel (breathing in).
#6 GIGABYTE GeForce GTX 550 Ti OC 1024MB GDDR5
#7 Intel 520 Series "Cherryville" 240GB 2.5" SATA 6Gb/s Solid State Hard Drive
#8 Coolermaster GX 650W 80PLUS PSU
#9 Corsair XMS3 32GB (4x8GB) DDR3 PC3-12800 C11 1600MHz Dual Channel Kit
The underneath chassis fan was supposed to go to the side of the case, but the Noctua CPU fan really is, too big to allow this on my R3 case. Massive shame to find this out, the hard way.
Furthermore, my projects require the following in which to 'get the job done':
Host Software: Cubase 5 Advanced 64bit
Audio Interface: TC Electronic Studio Konnekt 48 (Firewire 400 bus) with SPDIF running TC Electronic 96K Finalizer.
External Processing: Universal Audio UAD2 Solo DSP card (express card 34 PCMCIA) of which is running through a PCIe slot.
An example project runs the following (Heavy CPU load) plug-in quantity:
8* Cubase compressor (all acting on sidechain input)
1* Halion (Cubase VST Instrument)
12* Izotope Ozone 4
14* Native Instruments Massive (all running 'live') VST Instrument
3* Reverence (Cubase' convolution reverberation)
5* Slate Digital VBC compressors
2* Slate Digital FG-X
45* Slate Digital VCC Channels 4* VCC Virtual mix Bus
11* Slate Digital VTM
1* Native Instruments Guitar Rig (set to high quality)
Many other plug-in instances occur (eg. 35* Transient Designers, 40* Studio EQ, panning etc), but the afore mentioned require greater CPU (ASIO) power, and are thus note worthy.
Generally the projects will run all instances at maximum over sampling from the onset. However, as the project grows, the buffer rate (32, 64, 128, 256, 512, 1024, 2048) will go from starting at 256 to 2048 and eventually the oversampling will be reduced from 8* to *4, *2 depending on CPU handling - Every project works differently.
Subject to reading an article by Music Radar/Computer Music magazine (concerning acquiring the most potential from your work station), I disabled Hyper Threading in the BIOS from the onset, to designate as much 'strength' to the most important task at hand: The audio project(s).
The BIG query.
I haven't read enough convincing literature to persuade myself to delve further into OC'ing, until my PTTP application to Intel has been fully processed. That's not too say I haven't 'jotted down' approximate voltages for the relative GHz.
I've been doing a fair bit of reading as of late, and read that 1.4v (vCore) is the absolute maximum voltage, as dictated by Intel (physical properties), and far out/void of the 3 year Warranty if it fails subject to OC'ing.
My confusion stems from the following:
Is the 1.4v (Intel vCore maximum) a 'no questions asked' limit before you begin to decrement (throttle) the CPU' lifespan, or is there a correlation between running this voltage and the temperature increase that decrements a CPU' lifespan? Namely, if you run above 1.4v and consequently hit temperatures above +66.8 degrees Celsius that 'it's shelf life' steeps downward?
Is it fair to say that if you are running above 1.4v vCore (possibly nearing 1.5v), and your system is running heavy load audio DSP (digital signal processing), yet residing in the mid 40 degrees Celsius, that the CPU will not degrade?
Having recently upgraded the (supplied Fractal Design R3) chassis fans - a wonderful droop in temperatures all around - I opened up one of the heaviest CPU demanding projects, and began to (conservatively?) 'play' around with only the CPU multiplier (allowing the motherboard to dictate the required voltage). Incrementing from 3.8GHz to 4, 4.2, 4.3, 4.4, 4.6, 4.7, 4.8 and 4.9 - I was once more able to over sample (namely VCC VST plug-in) to x8, and decrease the interface buffer from 2048, to 512 samples. This would run comfortably and the CPU would as was mentioned be in the 46 degrees Celsius margin Running all chassis fans and CPU fan at 100% of course, - 'Because if it...' . Shocking me in awe But is it healthy to do so?
I do not plan on running this side of the OC'ing spectrum (24/7), until my audio projects require the processing power (namely 8x over sampling and lower audio buffer sample rating).
That said, when the time comes, the PC will be ran for hours at (the) higher GHz and it's relative vCore, that the project so requires. But is it healthy to do so? If the core temperature is some 3-4 degrees Celsius from breaching 50, is it healthy to run 1.4-5 vCore? Is this forbidden number exclusive to those running custom liquid coolers?
I do not dare run the 'Prime95', or it's alike (the Intel Burn sounds like murder...), having read that the CPU core temp will run into 75+ degrees Celsius and would have to do so 12-24hours. I am simply not comfortable with this level of (heat) torture.
I do understand that the software provided vCore readings are somewhat unreliable, and that one would have to physically measure the reading with a probe. None the less, these are the rough figures I have provided by Asus AI Suite II/Thermal Radar.
p.s. I appreciate that no one CPU is the same as the other, however, I feel compelled to ask of you fine folk to post relative stable vCore voltages for it's relative GHz counterpart, and any relative adjustments (PLL, VCCSA etc). I do still have a fair bit of information to get my head around, and how they 'cross-talk' per se.
So again, incase I have been too volluble per se, is running 1.4-5v vCore safe when your core temperature reads in the mid 40's degrees Celsius?
All the best, and thank you in advance!
I do apologize in advance: this CPU (3930k) and the term 'OC'ing', 'vCore' and 'Safe Temp' have been covered numerous times. However, I have never felt my questions and concerns directly addressed, to the point of conviction.
I've been running my 3.2GHz 3930k at 3.8GHz for the last 2 years (I was unaware that the motherboard 'decided' to run it at 3.8GHz as opposed to 3.2GHz): Seldom have I ventured into the BIOS for fear of 'royally ...', and it has served me very well, very! (the CPU/PC not the sheepish exploration!).
Before I continue any further, this is what I am working with, with respect to my PC:
64bit Windows 7 Pro computer:
#1 i7 3930k 3.20GHz (Sandybridge-E) Socket LGA2011
#2 ASUS Sabertooth X79 (Socket 2011) Motherboard
#3 Noctua NH-D14 (2011) Dual Radiator CPU Cooler
#4 2x Noctua NF-A14: above CPU cooler (breathing out) and underneath chassis (breathing in).
#5 2x Noctua NF-F12: behind CPU cooler (breathing out) and top front panel (breathing in).
#6 GIGABYTE GeForce GTX 550 Ti OC 1024MB GDDR5
#7 Intel 520 Series "Cherryville" 240GB 2.5" SATA 6Gb/s Solid State Hard Drive
#8 Coolermaster GX 650W 80PLUS PSU
#9 Corsair XMS3 32GB (4x8GB) DDR3 PC3-12800 C11 1600MHz Dual Channel Kit
The underneath chassis fan was supposed to go to the side of the case, but the Noctua CPU fan really is, too big to allow this on my R3 case. Massive shame to find this out, the hard way.
Furthermore, my projects require the following in which to 'get the job done':
Host Software: Cubase 5 Advanced 64bit
Audio Interface: TC Electronic Studio Konnekt 48 (Firewire 400 bus) with SPDIF running TC Electronic 96K Finalizer.
External Processing: Universal Audio UAD2 Solo DSP card (express card 34 PCMCIA) of which is running through a PCIe slot.
An example project runs the following (Heavy CPU load) plug-in quantity:
8* Cubase compressor (all acting on sidechain input)
1* Halion (Cubase VST Instrument)
12* Izotope Ozone 4
14* Native Instruments Massive (all running 'live') VST Instrument
3* Reverence (Cubase' convolution reverberation)
5* Slate Digital VBC compressors
2* Slate Digital FG-X
45* Slate Digital VCC Channels 4* VCC Virtual mix Bus
11* Slate Digital VTM
1* Native Instruments Guitar Rig (set to high quality)
Many other plug-in instances occur (eg. 35* Transient Designers, 40* Studio EQ, panning etc), but the afore mentioned require greater CPU (ASIO) power, and are thus note worthy.
Generally the projects will run all instances at maximum over sampling from the onset. However, as the project grows, the buffer rate (32, 64, 128, 256, 512, 1024, 2048) will go from starting at 256 to 2048 and eventually the oversampling will be reduced from 8* to *4, *2 depending on CPU handling - Every project works differently.
Subject to reading an article by Music Radar/Computer Music magazine (concerning acquiring the most potential from your work station), I disabled Hyper Threading in the BIOS from the onset, to designate as much 'strength' to the most important task at hand: The audio project(s).
The BIG query.
I haven't read enough convincing literature to persuade myself to delve further into OC'ing, until my PTTP application to Intel has been fully processed. That's not too say I haven't 'jotted down' approximate voltages for the relative GHz.
I've been doing a fair bit of reading as of late, and read that 1.4v (vCore) is the absolute maximum voltage, as dictated by Intel (physical properties), and far out/void of the 3 year Warranty if it fails subject to OC'ing.
My confusion stems from the following:
Is the 1.4v (Intel vCore maximum) a 'no questions asked' limit before you begin to decrement (throttle) the CPU' lifespan, or is there a correlation between running this voltage and the temperature increase that decrements a CPU' lifespan? Namely, if you run above 1.4v and consequently hit temperatures above +66.8 degrees Celsius that 'it's shelf life' steeps downward?
Is it fair to say that if you are running above 1.4v vCore (possibly nearing 1.5v), and your system is running heavy load audio DSP (digital signal processing), yet residing in the mid 40 degrees Celsius, that the CPU will not degrade?
Having recently upgraded the (supplied Fractal Design R3) chassis fans - a wonderful droop in temperatures all around - I opened up one of the heaviest CPU demanding projects, and began to (conservatively?) 'play' around with only the CPU multiplier (allowing the motherboard to dictate the required voltage). Incrementing from 3.8GHz to 4, 4.2, 4.3, 4.4, 4.6, 4.7, 4.8 and 4.9 - I was once more able to over sample (namely VCC VST plug-in) to x8, and decrease the interface buffer from 2048, to 512 samples. This would run comfortably and the CPU would as was mentioned be in the 46 degrees Celsius margin Running all chassis fans and CPU fan at 100% of course, - 'Because if it...' . Shocking me in awe But is it healthy to do so?
I do not plan on running this side of the OC'ing spectrum (24/7), until my audio projects require the processing power (namely 8x over sampling and lower audio buffer sample rating).
That said, when the time comes, the PC will be ran for hours at (the) higher GHz and it's relative vCore, that the project so requires. But is it healthy to do so? If the core temperature is some 3-4 degrees Celsius from breaching 50, is it healthy to run 1.4-5 vCore? Is this forbidden number exclusive to those running custom liquid coolers?
I do not dare run the 'Prime95', or it's alike (the Intel Burn sounds like murder...), having read that the CPU core temp will run into 75+ degrees Celsius and would have to do so 12-24hours. I am simply not comfortable with this level of (heat) torture.
I do understand that the software provided vCore readings are somewhat unreliable, and that one would have to physically measure the reading with a probe. None the less, these are the rough figures I have provided by Asus AI Suite II/Thermal Radar.
p.s. I appreciate that no one CPU is the same as the other, however, I feel compelled to ask of you fine folk to post relative stable vCore voltages for it's relative GHz counterpart, and any relative adjustments (PLL, VCCSA etc). I do still have a fair bit of information to get my head around, and how they 'cross-talk' per se.
So again, incase I have been too volluble per se, is running 1.4-5v vCore safe when your core temperature reads in the mid 40's degrees Celsius?
All the best, and thank you in advance!