Ivey Bridge Processor

[subair37]

Hi

I was about to get a Iveybridge processor pc , is it ok if i get an overclocked one for my video encoding , i dont do any advanced video encoding but use convertex,CCE, SP2 like that may do an MCCE like that, is it ok to go for a overclocked one as this is not going to affect my video encoding , as i heard b4 some one said overclocking is no good for video encoding , can some one please shed some light on this.

[Asmodian]

I have an i7 3770K overclocked to 4.5 GHz which I do use for encoding. The problem with overclocking the CPU during an encode is that you are running at full power for a long time. This means you need good cooling. If the CPU doesn't overheat and it is really stable at the overclock there are no issues special to video encoding.

I don't think I would do long encodes with a large overclock and a stock heatsink. Probably not a good idea to use a large overclock and a stock heatsink at all but even worse if you keep the CPU at 100% for hours.

[subair37]

I have an i7 3770K overclocked to 4.5 GHz which I do use for encoding. The problem with overclocking the CPU during an encode is that you are running at full power for a long time. This means you need good cooling. If the CPU doesn't overheat and it is really stable at the overclock there are no issues special to video encoding.

I don't think I would do long encodes with a large overclock and a stock heatsink. Probably not a good idea to use a large overclock and a stock heatsink at all but even worse if you keep the CPU at 100% for hours.

thank you very much for your info , what if i use aftermath heat sink ?

[Asmodian]

Yes that is what I have. Get a good aftermarket heat sink and you can over clock all you want. :)

I like to turn down the voltage as low as I can and still get a stable over clock. I check what the voltage is set to when at maximum turbo, using CPU-Z, and set the voltage to that in the BIOS. Then I test for maximum over clock at that voltage, maybe tweak it a little. If set to auto the BIOS will increase the voltage as the over clock goes up and using a lower voltage helps keep heat down and increases the life time of the CPU (decreases it less).

[Blue_MiSfit]

Overclocking seems like a silly idea to me for the most part. Don't get me wrong - I used to always overclock my systems, but I rarely actually noticed meaningful real-world improvement.

Weighing this negligible benefit against increased power consumption, heat dissipation / noise, potential instability, and potential shortening of equipment life, I see no reason whatsoever to overclock what is (to me) an expensive, and extremely valuable tool.

Then again, I tend to look at computing from a business perspective :)

[iSeries]

I overclock, but only to what I can get away with at or around stock voltage. Currently overclocked my 3770k to 4.2ghz at 1.11v. Rock solid stability, and a free 20% performance increase. Ivy bridge processors are not the easiest CPUs to keep cool though, if you are planning to do a lot of encoding I wouldn't use the stock cooler even at stock clocks.

[subair37]

I overclock, but only to what I can get away with at or around stock voltage. Currently overclocked my 3770k to 4.2ghz at 1.11v. Rock solid stability, and a free 20% performance increase. Ivy bridge processors are not the easiest CPUs to keep cool though, if you are planning to do a lot of encoding I wouldn't use the stock cooler even at stock clocks.

so what you saying dont overclock , really confused some say yes and some say no

[CWR03]

so what you saying dont overclock , really confused some say yes and some say no
Overclocking is good for gaming when you have bursts of high-demand processing, but for encoding you likely won't see a significant difference. As Blue_MiSfit said, it's not good for the hardware, so if you increase the load and temperature for extended periods it's not going to last, and for what? An extra few minutes of encoding time?

[iSeries]

Overclocking is good for gaming when you have bursts of high-demand processing, but for encoding you likely won't see a significant difference. As Blue_MiSfit said, it's not good for the hardware, so if you increase the load and temperature for extended periods it's not going to last, and for what? An extra few minutes of encoding time?

Greater voltage = greater heat, and this is what kills chips faster. However the Intel 'k' series are designed to be overclocked and can take a good deal of punishment. If you overclock like me though, by increasing the multi as far as you can without increasing the voltage, the chip will last just as long as at stock frequencies.

You can also buy Intel's overclocking insurance policy for something like $25, if the chip dies due to overclocking (not likely unless going for a really major overclock) they will replace the chip. Nothing to lose as far as I'm concerned.

[ramicio]

I wouldn't overclock for video encoding. You increase the risk of not getting back the correct data and having some corrupt frames. Depending on what your needs are, consumer-level hardware has zero ECC and shouldn't really be used for mission-critical calculations.

[iSeries]

I wouldn't overclock for video encoding. You increase the risk of not getting back the correct data and having some corrupt frames. Depending on what your needs are, consumer-level hardware has zero ECC and shouldn't really be used for mission-critical calculations.

If stability is properly tested then there are no problems. I have never had corrupted frames with a stable overclock.

And home encoding is not really 'mission critical' :)

[ramicio]

I never said it was mission critical. Stability testing is short term. It will not show things that will pop up in a system running 24/7, encoding for days on end. I constantly have plenty of corrupted frames in stuff with a stock clock. There's no way around cosmic ray events without some serious money.

[iSeries]

I never said it was mission critical. Stability testing is short term. It will not show things that will pop up in a system running 24/7, encoding for days on end. I constantly have plenty of corrupted frames in stuff with a stock clock. There's no way around cosmic ray events without some serious money.

Encoding for days on end is part of my stability testing routine :)

[ramicio]

Yeah, I bet :rolleyes:

[Asmodian]

I never said it was mission critical. Stability testing is short term. It will not show things that will pop up in a system running 24/7, encoding for days on end. I constantly have plenty of corrupted frames in stuff with a stock clock. There's no way around cosmic ray events without some serious money.

You have an unstable system then, I never get random corrupted frames. A system can be unstable without overclocking too.

If your stability testing is short term then you are not really stability testing. :rolleyes:

OP: Notice everyone suggested not increasing the voltage on Ivy Bridge. :)

[ramicio]

Obviously no one's ever heard of a cosmic ray event :rolleyes:

[hello_hello]

Every time I read a thread where the discussion involves overclocking, there's inevitably comments regarding shortening CPU life and hardware etc, but does anyone who advises "overclocking shortens CPU life" actually have examples of CPUs which have run overclocked for months or years, only to eventually fail at some point?

I was never into overclocking myself originally.... overclocking a Pentium 3 or 4 CPU etc seemed a lot of work for little gain, but these days, given Intel CPUs can often be overclocked around 20% without having to touch the voltages, it almost seems silly not to.

I can't imagine why stable overclocking would be likely to increase the likelihood of corrupted frames when encoding, or why there should be any at all if the system is stable. Regularly having corrupted frames at stock CPU speeds means something's definitely wrong. I can't think of a single instance of "corrupted frames" when I've encoded video.... ever. Then again, both the PCs here are protected from cosmic rays by their tin foil hats. ;)

[Bloax]

I can't imagine why stable overclocking would be likely to increase the likelihood of corrupted frames when encoding, or why there should be any at all if the system is stable.
Actually, I have one example. But notice how I'll now say that it's about GPUs.

So while this one guy on a compression forum was implementing some GPU methods in his program, some testing was going on.

It turned out that if you decompressed the output with an overclocked GPU, you'd end up with corrupted data. (No errors without OC)

But yes, I say that if you do get an Ivy Bridge, do overclock it.

[ramicio]

Cosmic ray events, radon, etc. There is radiation all around us, constantly, and it interferes with electronics. How many times do I need to spell this out for people? Consumer hardware has ZERO ECC and ZERO radiation hardening.

[Groucho2004]

There is radiation all around us, constantly, and it interferes with electronics.
I guess that's why my pocket calculator shows me these depressing numbers when I do my personal finances.

@x264 developers:
Please add code for radiation hardening so any radiation interfering with my encodes get bitch-slapped into oblivion.

[ramicio]

I doubt it. What kind of tech do they have? Stuff in the micrometer scale? The more we shrink in tech, the worse the problem gets. We have to start worrying about the chip causing its own errors. I highly doubt the chip inside of a calculator is going to be affected by background radiation.

[Groucho2004]

I highly doubt the chip inside of a calculator is going to be affected by background radiation.Sorry, I guess I should have added some sort of sarcasm icon...

[ramicio]

Haha, no, I got that it was sarcastic.

[Konrad Klar]

An overclocked system does not draw significantly more energy in idle state. It draws significantly more energy if loaded.
But hey. If for example it draws 30% more and is overclocked by 25%, then the time of computation is 80% of original.
So effectively a computing operation consumes only 4% more of energy (and is done in 20% shorter time).

[Asmodian]

Hey that also means you are less likely to have a cosmic ray event during the encode. ;)

[Phaex]

Rather than giving theoretical advice, I'll share my experience.

I have a i7 960 on clocked at 4.3 Ghz for about a year without any issues. Another e6300 overclocked to 3Ghz (stock 1.86 ver.) for 5 years without any issue. However, I am not using stock coolers. Overclocking a CPU does not shorten its life as long as it is not overheating.

On the motherboard side, you need to make sure that you are not overheating the chipest and mosfet. Normally they do not have a built in thermometer, so watch out.

If you are using the overclocked system make sure that you are treating your DRAM well too. Trust me, cooling of memory is just as important as CPU cooling.