Build New PC for Rendering Iray??

kyoto kid

...cryptomining will push a consumer card past it's normal operating envelope even more so than rendering as again a mining system is on and the card running at peak performance constantly to ensure the maximum profit gain. Consumer cards were not meant for this sort of punishment.  That is why Nvidia offered only a 3 month warranty on dedicated mining cards instead of the three years for us gamers and enthusiasts. I wouldn't purchase a used card right now no matter how good of a deal it may sound as the warranty does not usually transfer over.  

The only pre owned cards I would even consider would be refurbished Pascal Quadros.as they are tested and usually given an extended warranty.

ebergerly

kyoto kid said:

That is why Nvidia offered only a 3 month warranty on dedicated mining cards instead of the three years for us gamers and enthusiasts.

I suspect the reason why Nvidia offered a shorter warranty period becomes obvious if you look at an actual mining facility in real life. From what I've seen, they rent an empty room somewhere without the necessary airconditioning, buy some cheap racks and line up tens or hundreds of processing units and GPU's without much thought, and start running them with whatever air conditioning they can find. I saw one where the room was getting too hot so they took a 12" diameter flexible pipe from Home Depot and duct taped it to an airconditioning vent and dragged it across the room and pointed it at the racks of GPU's. 

So yeah, they're applied in ways that are FAR outside their design parameters (which is probably one or two or three in a desktop box with case cooling), so it's no wonder the manufacturer won't warranty them. 

 

kyoto kid

...exactly, the sad part is they are also treating the standard GPU cards we use like this as well all for hopes of making what usually turns out to be a very small profit.

nicstt

ebergerly said:

hphoenix said:

The 'self-protecting' parts can only respond so fast....and it's not as fast as the heat can be generated.

This is why most of us recommend that for rendering (not gaming) purposes, use a custom fan profile that starts at medium and ramps up to maximum early.  This way heat dissipation is well ahead of any dangerous potential levels.

The question is not whether heat has an effect. Of course it has an effect. But the question is whether that effect matters, and actually causes a loss of life, and whether that loss of life is relevant. Like I said, if it causes a 10 year lifespan to become an 8 year lifespan, and we change out our computers in 5 years, who cares? 

We can always find an exception, or a case in which heat can cause damage. But if that doesn't apply to 99% of users here, then does it really matter? 

I also think many in the tech community hear about something and tend to simplify it and apply it everywhere. But there are tons of people here and elsewhere who have successfully used laptops for rendering without issue. And like I showed, I even did an actual test on a certified piece of junk laptop for 24 hours continuous rendering and CPU temps stayed flat at 62C, way below any harmful temps.

And I still haven't seen any actual data and statistics which show that, for example, "instantaneous heat generation which is too fast for thermal sensors to detect can make your laptop melt, or reduce it's lifespan from 10 years to 1 year, and that heat generation happens every time you render a scene", or something like that. It's all just broad generalizations of "possible effects" without real data. IMO, "possible effects" is not supporting data unless it includes actual statistical data of laptops melting or having significant loss of life as a direct result.  

 

Does it matter?

Hindsight is the only true way to determine that little conundrum.

If, after running your kit beyond its tollerances it breaks before you're ready to replace (maybe you can't yet afford it), and it is out of warranty, then it did have an effect.

If it was inside the warranty period, then the manufacturer could very well determine that it had been used beyond its designed tollerances, and would refuse to repair/replace it.

... So yes it matters.

Might it be fine? Sure.

I don't take the chance; others will make their own decissions... But they should know the risks first.

drzap

nicstt said:

ebergerly said:

hphoenix said:

The 'self-protecting' parts can only respond so fast....and it's not as fast as the heat can be generated.

This is why most of us recommend that for rendering (not gaming) purposes, use a custom fan profile that starts at medium and ramps up to maximum early.  This way heat dissipation is well ahead of any dangerous potential levels.

The question is not whether heat has an effect. Of course it has an effect. But the question is whether that effect matters, and actually causes a loss of life, and whether that loss of life is relevant. Like I said, if it causes a 10 year lifespan to become an 8 year lifespan, and we change out our computers in 5 years, who cares? 

We can always find an exception, or a case in which heat can cause damage. But if that doesn't apply to 99% of users here, then does it really matter? 

I also think many in the tech community hear about something and tend to simplify it and apply it everywhere. But there are tons of people here and elsewhere who have successfully used laptops for rendering without issue. And like I showed, I even did an actual test on a certified piece of junk laptop for 24 hours continuous rendering and CPU temps stayed flat at 62C, way below any harmful temps.

And I still haven't seen any actual data and statistics which show that, for example, "instantaneous heat generation which is too fast for thermal sensors to detect can make your laptop melt, or reduce it's lifespan from 10 years to 1 year, and that heat generation happens every time you render a scene", or something like that. It's all just broad generalizations of "possible effects" without real data. IMO, "possible effects" is not supporting data unless it includes actual statistical data of laptops melting or having significant loss of life as a direct result.  

 

Does it matter?

Hindsight is the only true way to determine that little conundrum.

If, after running your kit beyond its tollerances it breaks before you're ready to replace (maybe you can't yet afford it), and it is out of warranty, then it did have an effect.

If it was inside the warranty period, then the manufacturer could very well determine that it had been used beyond its designed tollerances, and would refuse to repair/replace it.

... So yes it matters.

Might it be fine? Sure.

I don't take the chance; others will make their own decissions... But they should know the risks first.

People who rely on their equipment for their livelihood do not take such chances when there is insurance against these setbacks at so little cost.  Buy a well designed system, double down on cooling and don't leave these matters to chance.  As you said, it doesn't matter until it messes up your week and costs you in lost work.  There's not much difference between bad luck and unpreparedness.

ebergerly

 

drzap said:

People who rely on their equipment for their livelihood do not take such chances when there is insurance against these setbacks at so little cost.  Buy a well designed system, double down on cooling and don't leave these matters to chance.  As you said, it doesn't matter until it messes up your week and costs you in lost work.  There's not much difference between bad luck and unpreparedness.

Yeah, no question, that's perfectly reasonable. I just hate to see people running off and spending lots of money on unrealistic worries. And I see a lot of unrealistic worries in the tech community. 

 

kyoto kid

...it's not just heat either, electromechanical equipment has certain "safe" minimum/maximum temperature parameters to operate under.  For example, the order invoice printers where I last worked were located in the warehouse and during the coldest part of the winter would not function properly.  When this occurred, we had to use one of the large printers in the Digital print department, which resulted in  bumping scheduled print jobs.  This process had to be repeated four times a day

Minam

Older related post:
https://www.daz3d.com/forums/discussion/75430/buying-or-building-a-pc-for-daz-studio/p2
This right now is my favorite thread and should be updated. 

#1 PSU is important. I have a 750 W and after many diagnistoc tests, it seems to sadly not be enough for Daz 3D and not sure what it will be able to do with Blender. I mostly make games, but I can render about 5 times in iray, before I have to restart my computer, or Daz 3D will push a restart. I believe this is becuase the renders don't clear and my PSU (Power Supply) is working harder to feed my graphics card. 

I'm changing power settings to adaptive in the Nvidia control panel to try and manage power better. 

#2 The graphics card market is back to normal and thanks to Nivida introducin their new 2000's line, 1080ti are now between 400-800 USD, so now is a sweet time to pick up one of the most used cards for a lot cheaper. 
 

outrider42

The power supply has nothing to do with clearing data from the GPU. This could be a software issue as other people have mentioned Daz running up memory after repeated renders. This would be something to issue a support ticket to Daz for.

750 Watts is perfectly fine depending on your system. If you have 1 GPU, 750 is more than enough. Even with 2 GPUs 750 is likely still enough. However there are other aspects to power supplies that many people overlook, things that are in fact much more important than total wattage. Your PSU is the heart of the system, do not skimp on this. As long as your 750 is not some cheap off brand junk, it should be OK.

This video here is one of the best primers I have seen on understanding power supplies. You have an employee from Corsair and all around tech expert with great information. Gamers Nexus is a pretty respectible source of information in general. If you have any questions after the video, just ask in their forum, and they may answer.

kyoto kid

..sounds more like a GPU issue than the PSU.  I have a 750 w (Corsair) and it's been chugging along reliably for nearly 6 years handling both 3DL with UE and Iray.  Never had to reboot after rendering (and I run far more test renders just in a single session).

What GPU do you have?

. 

outrider42

I found a thread with people having similar issues. It looks like Daz has a memory leak to me. There is some way yo purge data from Daz so you don't have to close it. I can't recall how right now. But regardless your issue sounds the same as those in the thread.

https://www.daz3d.com/forums/discussion/220901/iray-memory-leak

kyoto kid

..memory leaks have plagued Daz since I first started in this over a decade ago

There is a script for purging memory as well as clearing the undo stack.

ebergerly

Minam said:

#1 PSU is important. I have a 750 W and after many diagnistoc tests, it seems to sadly not be enough for Daz 3D and not sure what it will be able to do with Blender. I mostly make games, but I can render about 5 times in iray, before I have to restart my computer, or Daz 3D will push a restart. I believe this is becuase the renders don't clear and my PSU (Power Supply) is working harder to feed my graphics card.

Not sure what you mean here. Rendering via Blender vs. DAZ Studio should be somewhat irrelevant to whether the PSU is "working harder". Generally, renders put the GPU utilization at around 100%, which means they're "working" as hard as they can. And that's measured by the power draw in watts. If you have a 1080ti, then it's rated to draw around 250 watts at 100% utilization. 

IMO, power supplies are pretty simple. In fact they're the simplest part of your PC. You size them by merely adding all the power draws of whatever you expected to be running at max power at the same time, then add a fudge factor based on your comfort level, and thats the size you need. There's really only 3 rules of PSU's:

1. Don't buy junk.
2. Don't operate them outside of their design specs by overclocking, blocking the cooling fan, etc.
3. Size them appropriately (see above). 

Other than that you should be good to go. I have a 1080ti plus a 1070, and can't get my 750watt PSU to draw more than 400watts from the wall for the entire computer system. If you're not sure, spend $30 on a simple power meter and plug your PC into it. Of course you can choose to be "better safe than sorry", but using that logic you can also justify a 1,500 watt power supply  when your computer is only drawing 200 watts. IMO, just take reasonable precautions (based on facts and ratings) and don't worry. 

Are they truly "the heart of your system", as is often repeated? Well, no more than your CPU or RAM or hard drive or motherboard or any other component. If any of those fails you could be in a world of hurt. Can a GPU cause damage to your system? Sure, anything can happen, just like failure of any other component might damage your whole motherboard, etc. But if you bought a quality PSU it's HIGHLY unlikely if you followed the 3 steps above, especially since it also has internal protection devices to limit voltages and overloads. Damage via PSU failure is probably FAR less likely for most of us than a hard drive crash that wipes out all your data, or a Windows 10 update that does the same   

davidwski_16294691f0

Minam said:

Older related post:
https://www.daz3d.com/forums/discussion/75430/buying-or-building-a-pc-for-daz-studio/p2
This right now is my favorite thread and should be updated. 

#1 PSU is important. I have a 750 W and after many diagnistoc tests, it seems to sadly not be enough for Daz 3D and not sure what it will be able to do with Blender. I mostly make games, but I can render about 5 times in iray, before I have to restart my computer, or Daz 3D will push a restart. I believe this is becuase the renders don't clear and my PSU (Power Supply) is working harder to feed my graphics card. 

I'm changing power settings to adaptive in the Nvidia control panel to try and manage power better. 

#2 The graphics card market is back to normal and thanks to Nivida introducin their new 2000's line, 1080ti are now between 400-800 USD, so now is a sweet time to pick up one of the most used cards for a lot cheaper. 
 

I don't know where the information came from, but if the PSU is overloaded, the entire computer would exhibit various problems, not just while running a single program.

ebergerly

davidwski_16294691f0 said:

I don't know where the information came from, but if the PSU is overloaded, the entire computer would exhibit various problems, not just while running a single program.

True, as I've said many times before, what often gets missed is that if you buy a good quality power supply it should come with internal protection to detect overloads and shut down for (or limit) problems before bad things happen to your components. Which is why I say Rule #1 is "Don't buy junk". 

Here's a really good explanation from Tom's on some of the details of what types of power supply protection that PSU's can have, such as overcurrent, over/under voltage, over temperature, and so on. And if you're really interested in understanding power supplies, this is part of a huge article that explains them in detail: 

https://www.tomshardware.com/reviews/power-supplies-101,4193-21.html 

 

outrider42

The power supply is the heart of the system, its not debatabe. If you back up memory then a hard drive failure is not a big deal. RAM is more expensive these days but RAM is the easiest thing to upgrade or replace. A motherboard would be a pain, true enough, but again can be replaced. And an expensive GPU would be a big set back (unless you got an extended warranty, then its just time lost waiting to get replacement.) Plus most CPUs have built in GPUs, so the computer will still run without a graphics card if needed. But the power supply is one of the only components capable of killing other parts if it dies in spectacular fashion. So instead of replacing one failed part, you might need to replaced several failed parts. And the PSU directly effects every part in the system depending on how steady and evenly it feeds everything. A less stable PSU will prevent you from getting better overclocks and effect performance across the board in less noticeable ways. While most modern PSUs have made big strides and are overall better all around, things can still happen, and it does not change the fact that power supply is the literal beating heart of a PC. You mention Tom's guide, Tom's has taken apart PSUs and found some poorly made parts and connections in name brand products over the years. Real fire hazards that you don't want in your home, much less your PC.

kyoto kid

...+1

ebergerly

outrider42 said:

But the power supply is one of the only components capable of killing other parts if it dies in spectacular fashion. 

And my only point is that power supplies are not the cruel, evil computer destroyers that people make them out to be. For the reasons I've mentioned, that if you buy a good quality one they are designed with internal protection to ensure they don't damage components. And I suspect many of us have many years of experience with computers that supports that. 

Now of course you can find cases where a PSU failure cause component damage, and you can probably even find someone on the internet claiming their power supply blew up and lit their house on fire, but I suspect there's a lot more to the story with claims like those. And yes, ANYTHING can fail in your computer, and take out other components. But if you weed out the cases where someone did something dumb and caused the failure, or overloaded their PSU and it was a piece of junk in the first place and it started smoking, I suspect you'd be hard pressed to show that power supplies are evil computer destroyers.

But of course, any component in the computer has a probability of failure, or having a bad component. That's the case with anything. So showing a case of a bad component or two doesn't prove an overall trend. You can search the web for engineering discussions of probability of failure of any electronic component (capacitors, MOV's, voltage regulators, etc.), and you'll see that everyone expects there to be failures and bad components. The only concern is if those failure rates are outside the expected rate. 

And yes, I agree that you can say that the power supply is the heart of the system, since it pumps blood (electricity) thru the veins (wires) of the computer.   But otherwise, I don't think they deserve the bad rap that some like to give them. 

By the way, for anyone interested in understanding computer power supplies and how they're designed and their different components, I'd strongly recommend you read the article I posted. Really excellent explanation of how they work. And yes, the author DOES agree they are the "heart of the system", if that matters.