TR Forums

I never spent a thousand dollars on a processor before so forgive me if this question is totally lame.
Is it safe to run this processor at 100% load for 20, 30 hours straight - or even longer?

I'm rendering in C4D. Using GI and Multi-Pass makes for a long process - total will be about 30 hours.

RealTemp reports all 6 cores at 52C or under. I running it at stock voltage, no OC.

In general would you say that modern processors - desktop procs - are designed to handle this type of load? Or should you really be using server parts for 24/7 work?

Thanks!

Assumign the temperature is kept within check (as it seems to be in your case), you're perfectly safe.

thanks morphine, I appreciate the quick response.

I was curious because I haven't seen any 100%-load temps like mine - all the sites I've checked out everyone is in the 70-85 range, or even higher - on Prime, etc. I'm not complaining - just wondering if I got lucky or what...

Thanks again for your reply!

One thing you might want to consider is using ECC RAM, provided the i7 and the motherboard support it (I don't know at this time).

Thanks, I considered going that route on this build - decided against it because it was my first one, trying to keep things simple. I built a quasi-workstation I guess you'd call it - kind of goofy but it rocks the few benchmarking apps I've thrown at it - Everest, ATTO, Cinebench - so it does what I need for now.

Def will go the ECC route next time, when I do a "real" workstation - dual Xeons, Quadro GPU, etc. But for now - as long as the Gulftown can handle 20-30 hours of 100% workload every few weeks I'm okay with it.

Thanks again for your help morphine - it is much appreciated!

May be you shouldn't spent a thousand dollars on the CPU to begin with? I know this may not sound good, but if you are so scared why do it? Computer equipment loses value constantly, don't treat it as an "investment". It is just a means to achieve your goals. In this case, it is something to help with your rendering and stuff. It saves time and time is money in your business. Then this $1K becomes just an expense. If you are doing rendering as a hobby, then you need to spend according to what you can spend with your disposable income.

I figured I'd get at least one response like yours - because of the way I worded my original question. No fault to you - it does sound like I'm chicken to run the thing! But I'm not, not at all. I just needed a gentle reminder that it's perfectly fine to work these processors at 100%. And when that proc in question is a thousand-dollar unit...maybe I can be forgiven for having a moment of doubt.

I am super happy with my "investment." or whatever you want to call it....

Hrm. 52c on the individual core sensors, on the stock heatsink? Doesn't sound like the rendering is pushing the CPU very hard. Just wondering if there might be a bottleneck somewhere else holding it back.

I don't mean this to sound douchey, but why did you spend $1000 on a cpu? You can buy a similar one for $200-300 and just overclock it. Also, AMD's 6 core is pretty much just as good as Intel's, and its like $250.

I never spent a thousand dollars on a processor before so forgive me if this question is totally lame.
Is it safe to run this processor at 100% load for 20, 30 hours straight - or even longer?

As long as you're keeping it adequately cool, no problem.

Def will go the ECC route next time, when I do a "real" workstation - dual Xeons, Quadro GPU, etc. But for now - as long as the Gulftown can handle 20-30 hours of 100% workload every few weeks I'm okay with it.

Before spending a gazillion dollars on a "real" workstation, you might consider looking at what you already have, identifying the bottlenecks, and upgrading accordingly. As others have mentioned, it doesn't sound like your CPU is being pushed to the limit as it is now; and if that's the case, buying a faster one isn't going to improve your performance.

I don't mean this to sound douchey, but why did you spend $1000 on a cpu? You can buy a similar one for $200-300 and just overclock it. Also, AMD's 6 core is pretty much just as good as Intel's, and its like $250.

Why does this question keep occuring? Did we stop teaching the answer to this question back in overclocking 101 classes?

Mentawl wrote:

Hrm. 52c on the individual core sensors, on the stock heatsink? Doesn't sound like the rendering is pushing the CPU very hard. Just wondering if there might be a bottleneck somewhere else holding it back.

TskMngr shows 12 threads all running at 100%...am I missing something obvious here (or maybe not-so-obvious)?

CPU-z reports 3462.36 MHz (133.17x26)

Why does it matter AT ALL that he bough this processor? He has it, shut up and just answer his questions.

Do you have the stock heatsink? If so, the temps sounds a little too low.

Do you have a case with amazingly high airflow? If so, than this might give us our answer.

Ensure that you dust out the computer frequently, and I would say every month monitor the temps to ensure no drastic changes.

In regards to the 12 threads showing 100%, what program are you running? Get Intel burn test to ensure that you are REALLY stressing the processor.

StuG wrote:

Why does it matter AT ALL that he bough this processor? He has it, shut up and just answer his questions.

Do you have the stock heatsink? If so, the temps sounds a little too low.

Do you have a case with amazingly high airflow? If so, than this might give us our answer.

Ensure that you dust out the computer frequently, and I would say every month monitor the temps to ensure no drastic changes.

In regards to the 12 threads showing 100%, what program are you running? Get Intel burn test to ensure that you are REALLY stressing the processor.

Thanks Stu - I understand people's "concern" whenever the 980x is mentioned - and not just here, it's the same story everywhere. It's a thousand-dollar processor, it's bound to provoke...!

This is a Cinema 4D render pushing the cores to 100% - C4D is multi-threaded, 64-bit app and is very reliable - I'm sure you're familiar with Cinebench - C4D is the modeling app it's based on.

I added a 140mm to blow fresh air onto the GPU and did a meticulous job of routing cables - pretty proud of it actually, for my first build. I am using the stock HSF but apparently it's a good one - gets good reviews around the web. The consensus seems to be that Intel finally released a decent stock HSF.

I'm pretty diligent about dust build-up. Living in NYC there's dust and everything else in the air at all times, so I clean my screens every 2 weeks or so. My idle temps (using RealTemp) are from 17-22 and I've yet to hit 65 at full load. Either something is goofy with RealTemp (I'll prob install Speedfan at some point) or else ... or else I don't know, maybe I just did a good job!

[edit: I also have 2 USB drives with a clip-on fan blowing room air onto them, and the fan is pointed directly inline with the 200mm intake fan on the front of the HAF-X. So that's probably contributing some cooling as well inside the case.]

Rather than Speedfan, you might want to try OpenHardwareMonitor. I'm a fan because it doesn't need installed as such, just runs from the folder. Seems to report actual sensor temps on every machine I've tried it on so far.

http://openhardwaremonitor.org/

I'd forgotten that the 980X heatsink was a chunky tower model. 52c does still sound fairly cool, though! I could just be comparing a new 980X to my old 920 though, and I know the 980X is a fair bit more efficient clock for clock.

Thanks for the link Mentawl - I definitely prefer an app that doesn't need to be installed - that's why I haven't bothered with Speedfan yet, I'm trying to keep the installed apps (esp small utility apps) to a minimum.

I keep RealTemp running all the time but I'll try OpenHWMonitor and see what it reports. Thanks again!

As for the processor, I'm quite sure it is safe to run it at stock speeds during longer periods. Processors are quite hard to kill at stock voltage, stock speeds and good enough cooling (and I'd guess that if that fails, the processor will throttle back to save itself). As for the value of a 1000$ processor - If I remember well, it used to be right on spot for performance/watt and performance/dollar (check the TR reviews). Ofcourse, since you're not overclocking, you could have gone for the 900$ 970, and be saved from all this discussion. But then, I think it would make a few hours longer rendering.

The question is: are all your other components up to that task? Good quality PSU, motherboard, storage system? Personally, I'd guess the psu (and do you have an ups?) are the most important. If they get flaky after some hours, then everything else could suffer from that.

Running all cores at 100% does not necessarily mean it is generating the most heat. The mix of instructions being executed by the threads matter. That is why apps like Prime95 and IntelBurnTest can generate the most heat as they call the set of instructions that will make the circuit sweat the most. They are mostly floating point / SSE instructions AFAIK but you need to call specific ones often enough to generate the max heat.

Thanks Ozy - good points to consider. I've always run a high-quality UPS (Tripplite 1000W) because where I live, the apartment house is pretty old and has ... "interesting" ... electrical service. Around here a UPS isn't something you "might want to have" it's a necessity.

I went with Corsair's HX850 for a PSU. It's a modular unit with an 80 Plus Silver rating and wasn't crazy-expensive - came with a ton of cables too. I highly recommend it if 850W and 12V@70A suits your needs.

Thanks FFox - I'll give IntelBurn a shot at some point.

canoli wrote:

Thanks for the link Mentawl

I keep RealTemp running all the time but I'll try OpenHWMonitor and see what it reports. Thanks again!

OpenHardwareMonitor confirms the temps I'm getting in RealTemp - thanks! OpenHWMon has quite a bit more info too - reads GPU temps, voltages, etc. Nice! and all for free....can't beat it.

Thanks again Mentawl!

Flying Fox wrote:

That is why apps like Prime95 and IntelBurnTest can generate the most heat as they call the set of instructions that will make the circuit sweat the most. They are mostly floating point / SSE instructions AFAIK

But, but... nobody uses those instructions anyway!!!!!111oneoneone

canoli wrote:

Thanks FFox - I'll give IntelBurn a shot at some point.

Now don't run that 24x7!

NO - I'll follow the directions when I try that.

Not entirely sure why I'd want to run it though - if C4D is pushing 12 threads to 100% (in my 30th hour) and my temps are hovering around 50C, I think I'm happy with that. I understand what you said about the different instructions - my question is "should I bother?"

As was explained so eloquently in another thread, nobody needs to use floating-point numbers anyway - oh well except in video editing, image editing, HDR processing ... and in the very small, insignificant industries like mining, geologic, archeological, architectural and medical applications - not to mention computer science - basically - the entire scientific community...but I mean besides that there's really no call for FP ( !

No but seriously - What I mean is, if C4D rendering - for 30+ hours straight at 100% - hasn't pushed temps past 54...and that is by far the most stressful workflow I'll ever need... shouldn't I maybe...um, kinda sorta...leave well-enough alone?

canoli wrote:

NO - I'll follow the directions when I try that.

Not entirely sure why I'd want to run it though - if C4D is pushing 12 threads to 100% (in my 30th hour) and my temps are hovering around 50C, I think I'm happy with that. I understand what you said about the different instructions - my question is "should I bother?"

As was explained so eloquently in another thread, nobody needs to use floating-point numbers anyway - oh well except in video editing, image editing, HDR processing ... and in the very small, insignificant industries like mining, geologic, archeological, architectural and medical applications - not to mention computer science - basically - the entire scientific community...but I mean besides that there's really no call for FP ( !

No but seriously - What I mean is, if C4D rendering - for 30+ hours straight at 100% - hasn't pushed temps past 54...and that is by far the most stressful workflow I'll ever need... shouldn't I maybe...um, kinda sorta...leave well-enough alone?

At that point running IBT becomes a curious exercise to check out the cooling potential of your setup. If it runs stable with your workloads then it should be fine. Now enjoy the new system instead of freaking out of minuscule details!

canoli wrote:

Now enjoy the new system instead of freaking out of minuscule details!

What, I can't do both?

canoli wrote:

NO - I'll follow the directions when I try that.
"should I bother?"

Double checking? Getting a second opinion? Better safe then sorry?

To me if your going to spend money on a computer then spending a few hours to ensure its working properly? If I was you I would rather test it now even if it is just extra peace of mind instead.