Should I buy a separate PC for work if overclocking??

[KingFatty]

No no no, the clear choice here is to use this as an excuse to upgrade to an UBER gaming computer.

That way, you can use it for work and for play, because it's powerful enough to not even *need* to be overclocked, and then it's perfectly safe/stable for work too.

Best of both worlds. You could probably get a cheap motherboard and nice 4790K chip to accomplish this, that thing practically overclocks itself while being guaranteed to be stable.

 

[2is]

No no no, the clear choice here is to use this as an excuse to upgrade to an UBER gaming computer.

That way, you can use it for work and for play, because it's powerful enough to not even *need* to be overclocked, and then it's perfectly safe/stable for work too.

Best of both worlds. You could probably get a cheap motherboard and nice 4790K chip to accomplish this, that thing practically overclocks itself while being guaranteed to be stable.

4790k would be a downgrade from what he already has.

 

[Squeetard]

Same cpu and 2 - 7970's in xfire all watercooled. Used for gaming and work. For 2 years now, not a single issue. You forget that the cpu and gpu's throttle down when not taxed.
The cpu sits at 800mhz most of the time when working, the GPU's sit at 500mhz. Cool as a cucumber.

 

[DigDog]

omfg NO NO NO.

Overclocking DOES NOT affect calculations.

An unstable overclock can crash a pc. The crash can cause all the issues a crash can, but when not crashing, there is no way an overclock can affect your data.

i cant believe people would write something this stupid here on AT.


your pc is doing million of calculations, always. if overclocking had a chance of producing errors, then STATISTICALLY every overclocked pc would produce a number of errors per hour. Nearly every machine owned by every person registered to this site is overclocked, NONE of them throw out errors. Nobody here has ever posted a thread "my overclock accidentally ordered ten thousand alpaca jackets from kashmir".

Any error, ANY error would make your pc useless. It's not your calculator that would fail, it would be your registry, or cache, or flie allocation table, or any number of operations your pc is doing and you are obviously ignorant of, which would fail. Digital systems do not allow for errors.
(except, funny enough, when they do. CPU-based routines do not)

A STABLE overclock is exactly the same as a stable non-overclock. There is NO factual difference between an operation performed at 4Ghz or at 3Ghz.

You do not overclock a machine attached to medical equipment because THE OVERCLOCK MIGHT NOT BE STABLE, not because the OC might produce errors.

Errors are not something you can see. You will never, ever have an error add a zero in your tax refund, because computing does not work like that. If anything, you are going to get a BSOD, and again, thats because it is not stable.

Larry, you are talking about undervolting, not overclocking. Yes there are some procedures that can cause instability, like changing the bus, overvolting, or even extreme overclocking, none of which are "a stable overclock".

At this point, i strongly suggest you return your PC to stock clock and settings IMMEDIATELY and never, ever alter another setting because obviously you have zero understanding of computing and you could hurt somebody.

 

[myocardia]

omfg NO NO NO.

Overclocking DOES NOT affect calculations.

An unstable overclock can crash a pc. The crash can cause all the issues a crash can, but when not crashing, there is no way an overclock can affect your data.

i cant believe people would write something this stupid here on AT.


your pc is doing million of calculations, always. if overclocking had a chance of producing errors, then STATISTICALLY every overclocked pc would produce a number of errors per hour. Nearly every machine owned by every person registered to this site is overclocked, NONE of them throw out errors. Nobody here has ever posted a thread "my overclock accidentally ordered ten thousand alpaca jackets from kashmir".

Any error, ANY error would make your pc useless. It's not your calculator that would fail, it would be your registry, or cache, or flie allocation table, or any number of operations your pc is doing and you are obviously ignorant of, which would fail. Digital systems do not allow for errors.
(except, funny enough, when they do. CPU-based routines do not)

A STABLE overclock is exactly the same as a stable non-overclock. There is NO factual difference between an operation performed at 4Ghz or at 3Ghz.

You do not overclock a machine attached to medical equipment because THE OVERCLOCK MIGHT NOT BE STABLE, not because the OC might produce errors.

Errors are not something you can see. You will never, ever have an error add a zero in your tax refund, because computing does not work like that. If anything, you are going to get a BSOD, and again, thats because it is not stable.

Larry, you are talking about undervolting, not overclocking. Yes there are some procedures that can cause instability, like changing the bus, overvolting, or even extreme overclocking, none of which are "a stable overclock".

At this point, i strongly suggest you return your PC to stock clock and settings IMMEDIATELY and never, ever alter another setting because obviously you have zero understanding of computing and you could hurt somebody.

Quoted for posterity, not to mention the humor. When you are old enough to apply for your first job, you will be coming back, in an attempt to remove this post. lmfao at almost all of this post, but especially at the emboldened portions above.

 

[Maximilian]

Quoted for posterity, not to mention the humor. When you are old enough to apply for your first job, you will be coming back, in an attempt to remove this post. lmfao at almost all of this post, but especially at the emboldened portions above.

This...

Also OP don't overclock your work computer. IDC made some in depth posts a while ago about silent data corruption, overclocking can cause stuff like that.

If your work isnt terribly demanding on your hardware and you really really want to overclock your gaming rig then yeah maybe buying a cheap 2nd computer would be a decent idea. These days I just get hardware that is absolutely overkill so I dont feel a need to overclock :awe:

 

[escrow4]

No overclock is stable. Its stable what is stock out the factory. Even MCE has the potential to ruin your day (although that is highly unlikely).

 

[Dave3000]

I guess overclocking is not really free additional performance. If you do also work on your PC most people in this thread are recommending a separate PC at stock settings if you want to overclock a gaming PC. There goes the free additional performance out the window right there. So if someone that mainly played games, but also does office work on his PC even though not stressful on the CPU, only wants one PC due to lack of space, and wants to buy a new system and is deciding between one with a i5-4690k 3.5 GHz and an i7-4790k 4 GHz he would be better of with a 4790k and forget overclocking because it will be used for work as well? If all he did was just play games on his system and wanted best value for his money, he would buy an i5 4690k and overclock it?

 

[Blue_Max]

I still suggest that even the cheapest used Core2Duo or Quad running stock and a used monitor for $100 or less should be set aside for work-only purposes.

If nothing else, at least you're not mucking around with profiles and constantly changing your gaming machine around. That constant changing of clock speeds is more likely to CAUSE trouble than leaving it stably overclocked!

 

[cytg111]

sssssh .. dont listen to them, there is no empirical data to support their claims. Overclock, overclock away .. and sssssh, dont listen to them.

 

[2is]

omfg NO NO NO.

Overclocking DOES NOT affect calculations.

omfg YES YES YES.

Overclocking DOES affect calculations.

What do you think a Prime95 or IBT error is? It's a calculation error that doesn't necessarily crash the PC.

 

[DigDog]

Quoted for posterity, not to mention the humor. When you are old enough to apply for your first job, you will be coming back, in an attempt to remove this post. lmfao at almost all of this post, but especially at the emboldened portions above.

omg you. just one thing, ok?

please detail here the errors you have experienced on your system in sig.

^typed on my work pc@IBM.

What do you think a Prime95 or IBT error is? It's a calculation error that doesn't necessarily crash the PC.

so you raise the voltage and the errors disappear. im pretty certain by now that everyone in this thread is trolling, because i can't believe you fail to read basic english.

STABLE. OVERCLOCK.

yes you can have a 24/7 stable OC under prime95 or burn test. once it is STABLE the pc will NOT throw out errors.

caption: he said stable, mmk ?

 

[crashtech]

I've never overclocked a PC within livestock living quarters. Man's gotta have some pride.

 

[VirtualLarry]

What do you think a Prime95 or IBT error is? It's a calculation error that doesn't necessarily crash the PC.

This, exactly. ERRORS IN CALCULATIONS. Which CAN happen when overclocked.

 

[2is]

omg you. just one thing, ok?

please detail here the errors you have experienced on your system in sig.

^typed on my work pc@IBM.



so you raise the voltage and the errors disappear. im pretty certain by now that everyone in this thread is trolling, because i can't believe you fail to read basic english.

STABLE. OVERCLOCK.

yes you can have a 24/7 stable OC under prime95 or burn test. once it is STABLE the pc will NOT throw out errors.

caption: he said stable, mmk ?

An unstable overclock can crash a pc. The crash can cause all the issues a crash can, but when not crashing, there is no way an overclock can affect your data

Again, you're wrong. If you think you're right, you haven't got a clue what you're talking about. I already gave you examples of calculation errors that don't crash a PC. Do it yourself if you don't believe me. Run 10 passes of IBT using all your cores, customize the setting to use approx. 90% of your ram. Once those 10 passes are complete, go into your BIOS lower the voltage a notch, run another 10 passes. Keep doing it and you will eventually get to a point where IBT will generate an error but your PC is not crashing.

Bottom line, you don't know what you're talking about and you should refrain from giving advise until you do.

 

[VirtualLarry]

STABLE. OVERCLOCK.

yes you can have a 24/7 stable OC under prime95 or burn test. once it is STABLE the pc will NOT throw out errors.

But, the definition of "stable", depends on the workload. The only workload that you can "gaurantee" stable, with Prime95 testing, is, in fact, Prime95. Excel? Gaming? Nope. Prime95 cannot "prove" those other workloads stable.

Hence the anecdote regarding Rubycon's experience. Prime95 doesn't exercise the SSE 4.1 opcodes. So it's entirely possible to be "Prime95 stable", but at the same time, "SSE4.1 load UNSTABLE".

Once you learn about CPUs and computer architecture in-depth, with a bit of physics thrown in, then you will start to understand these things, and the things that everyone in this thread is trying to tell you.

Or maybe IDC can school you, I don't know.

 

[BonzaiDuck]

But, the definition of "stable", depends on the workload. The only workload that you can "gaurantee" stable, with Prime95 testing, is, in fact, Prime95. Excel? Gaming? Nope. Prime95 cannot "prove" those other workloads stable.

Hence the anecdote regarding Rubycon's experience. Prime95 doesn't exercise the SSE 4.1 opcodes. So it's entirely possible to be "Prime95 stable", but at the same time, "SSE4.1 load UNSTABLE".

Once you learn about CPUs and computer architecture in-depth, with a bit of physics thrown in, then you will start to understand these things, and the things that everyone in this thread is trying to tell you.

Or maybe IDC can school you, I don't know.

Actually, IDontCare's tutorial for running LinX "affini-tized" offered me a different perspective, which I try to use in fine-tuning my clock-settings. The GFLOPS values that were seeming outliers in a distribution, or widely fluctuating GFLOPS as iterations progress, were indications that voltage was insufficient and that error-correction was slowing things down for a particular iteration. And any lower voltage of some amount might actually cause the system to crash or BSOD during the stress-test.

So I started doing small-sample runs after achieving a "stable" overclock to see if I couldn't tweak voltage to minimize the range and variation of the GFLOPS statistic.

 

[crashtech]

Huh, I always figured variations were caused by other processes getting CPU time.

 

[BonzaiDuck]

This, anything having critical data or being important to you financially (or otherwise) should not be overclocked. There are literally billions (slight exaggeration :awe of things that can crash a modern PC mid session, you wouldn't want to facilitate any of them especially when you're in the middle of something important in fact I'd go so far as to suggest a T or S (low power) CPU variant for anything that doesn't require lots of computing power, like in your case perhaps.

I like to have it "both ways." OF COURSE we use our high-performance OC machines for mild duties that aren't so processor intensive -- I would even say "QuickBooks." But for something like that, I'd still say -- for routine, day-to-day full-time business -- don't use an overclocked system.

I don't think I'd want to use an OC'd system for programming or software development.

But it's all about risk. I'm not monitoring my event logs and watching every little irregularity because of an IMMEDIATE worry about data corruption or mistakes; I'm monitoring and maintaining the system to decrease any risk of "instability" or risk of "silent data corruption."

 

[Sunrise089]

My concern from experience is lost productivity and lost business from having a PC that stops working properly at an inconvenient time, or corrupts or loses some customer data.

I'm curious about what your estimated probability of 1) PC stops working and 2) PC corrupts/loses data is for both stock and overclocked systems. I don't know you, but I'd be quite surprised if the dollar cost of the incredibly miniscule increased risk due to overclocking was really the biggest risk you face.

The idea that there's a bright line of data corruption between stock specs and minor overclocks suggests a level of factory reliability and validation that strikes me as utterly unsupported in practice.

 

[crashtech]

I'm curious about what your estimated probability of 1) PC stops working and 2) PC corrupts/loses data is for both stock and overclocked systems. I don't know you, but I'd be quite surprised if the dollar cost of the incredibly miniscule increased risk due to overclocking was really the biggest risk you face.

The idea that there's a bright line of data corruption between stock specs and minor overclocks suggests a level of factory reliability and validation that strikes me as utterly unsupported in practice.

Basically, there are many risks that are faced every day, and my standard practice is not to add to those risks needlessly. I've had business PCs have problems, most often with HDDs and now SSDs, and even when measures are in place to protect data it's pretty taxing on a small business to have this happen. It just doesn't make sense to have an overclocked business PC unless that overclock is being used to make money commensurate with the added risk.

 

[rchunter]

I still suggest that even the cheapest used Core2Duo or Quad running stock and a used monitor for $100 or less should be set aside for work-only purposes.

If nothing else, at least you're not mucking around with profiles and constantly changing your gaming machine around. That constant changing of clock speeds is more likely to CAUSE trouble than leaving it stably overclocked!

^^^This is what I would do. That and you don't even need a separate monitor, keyboard, mouse, etc. Just get a kvm switch.

 

[2is]

Actually, IDontCare's tutorial for running LinX "affini-tized" offered me a different perspective, which I try to use in fine-tuning my clock-settings. The GFLOPS values that were seeming outliers in a distribution, or widely fluctuating GFLOPS as iterations progress, were indications that voltage was insufficient and that error-correction was slowing things down for a particular iteration. And any lower voltage of some amount might actually cause the system to crash or BSOD during the stress-test.

So I started doing small-sample runs after achieving a "stable" overclock to see if I couldn't tweak voltage to minimize the range and variation of the GFLOPS statistic.

I can attest to the whole gflops testing. When I run IBT and I see the GFLOPS fluctuate wildly, I know i'm on the cusp of instability, even if it runs through all the tests without error. It's not a condition I feel comfortable running my PC at, so I'll increase the voltage a notch. This usually stabilizes the results, i'll then go in and raise it one more notch, this often times has the benefit of slightly higher scores. I basically do this until both my GFLOPS figures stabilizes, and I'm no longer seeing a performance increase. Once I get to that voltage, I then bump it up one more notch for good measure. This is all assuming temps are well within tolerance.

 

[BonzaiDuck]

I can attest to the whole gflops testing. When I run IBT and I see the GFLOPS fluctuate wildly, I know i'm on the cusp of instability, even if it runs through all the tests without error. It's not a condition I feel comfortable running my PC at, so I'll increase the voltage a notch. This usually stabilizes the results, i'll then go in and raise it one more notch, this often times has the benefit of slightly higher scores. I basically do this until both my GFLOPS figures stabilizes, and I'm no longer seeing a performance increase. Once I get to that voltage, I then bump it up one more notch for good measure. This is all assuming temps are well within tolerance.

That sounds about right to me. It was actually IDontCare who advocated low-duration stress-testing. In the old days, we ran Prime95 all night long. It became more apparent that IBT or LinX would prove out after 25 to 50 iterations. I keep track of the long tests with GFLOPS statistics, because it's a much better sample to work with. But I'll also do 10 or 15 iteration runs. Sometimes I've logged the results to use in a spreadsheet to compute standard deviation and variance; other times, I just use the range as measure of dispersion. Sometimes, there are background processes which will throw out a low GFLOPS number, but you can actually monitor those while they happen if you have the patience to stay glued to the system throughout the stress-test.

In statistics, there's something we call an outlier. In industrial applications, it may have an "assignable cause." Here, it may be no different if something is momentarily crowding CPU usage during the stress-test. If it can be identified, you'd throw it out of the results.

Just for crude validation of stability, I run OCCT:CPU for 3.5 hours. Then I run the other tests until I've made the last voltage tweak.

I find the postures taken by this thread's participants interesting. Consider that all the Intel processors we're discussing are "already overclocking" themselves in Turbo mode, and most of us simply extend the range of clocks and call it "Turbo-Overclocking."

They made these processors to ramp up to a higher speed to begin with, but they could only have chosen the turbo-range to anticipate their own risks and costs for RMA processing, warranty returns, etc. Yet they would know from Intel's experience over that defective processor they had sold and its use in a large construction project that they have some risk of lawsuits if the processors have some defect that manifests itself in propagating errors in calculations.

So I would use that background to temper any understanding of risk in overclocking.

 

[Dave3000]

This...

Also OP don't overclock your work computer. IDC made some in depth posts a while ago about silent data corruption, overclocking can cause stuff like that.

If your work isnt terribly demanding on your hardware and you really really want to overclock your gaming rig then yeah maybe buying a cheap 2nd computer would be a decent idea. These days I just get hardware that is absolutely overkill so I dont feel a need to overclock :awe:

My i7-4930k was one of the best CPUs at the time I purchased it, even for lightly threaded software. If I wanted a better CPU the other choice was a $1000 i7-4960x which was only about 3-5% faster or giving up significant performance in heavily threaded software and going with a 4770k for only 5-7% faster lightly threaded performance (which most games are these days). Basically any CPU out there at the time that was faster than an i7-4930k would not make a noticeable difference in gaming.

If I buy a laptop, I don't think I will take it traveling with me, I just think it's too risky because it might break or get stolen and even if I can buy accidental insurance, spending an additional $130 (price may vary depending on the store) is just too much for a a laptop that cost $400-$500 in my opinion. I was also considering a NUC with an i3 and 4GB RAM, and a 128GB SSD for my "work" PC, and getting a KVM switch for my monitor and keyboard, and mouse to share with my "gaming" PC, due to space constraints. I really have no room to use a laptop in my bedroom due to my keyboard and monitor on my desk.