High CPU usage for long periods of time
-
Has anyone had any problems with running a job for more than an hour that pushes most of the CPU cores to 100% usage? I'm planning on making some animation, and with each frame taking about 5-7 minutes each, a single second of animation is going to come in around 2 hours 20 minutes or worse. For a 10 to 20 second clip that could easily cause the CPU to run at 100% for almost 12 hours or more. I've never had anything run that long on any system I've owned and I'm not sure if it's going to overheat the box, or if it's hard on the CPU to run that long. Anyone have any experience with this on a desktop system (for the record it's an 8 core MacPro)?
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
Jim Crafton wrote:
Has anyone had any problems with running a job for more than an hour that pushes most of the CPU cores to 100% usage?
Nope, but then again I have water-cooling[^] :cool: I think the 8core macs are the same as the i7 900 series, so your CPU can handle temps up to 99 deg C (not that you want it that high, just saying). Running continuously, I would not feel happy if it was over 80 deg.
xacc.ide
IronScheme - 1.0 RC 1 - out now!
((λ (x) `(,x ',x)) '(λ (x) `(,x ',x))) The Scheme Programming Language – Fourth Edition -
Jim Crafton wrote:
Has anyone had any problems with running a job for more than an hour that pushes most of the CPU cores to 100% usage?
Nope, but then again I have water-cooling[^] :cool: I think the 8core macs are the same as the i7 900 series, so your CPU can handle temps up to 99 deg C (not that you want it that high, just saying). Running continuously, I would not feel happy if it was over 80 deg.
xacc.ide
IronScheme - 1.0 RC 1 - out now!
((λ (x) `(,x ',x)) '(λ (x) `(,x ',x))) The Scheme Programming Language – Fourth Editionleppie wrote:
Nope, but then again I have water-cooling
Pft, that's nothing compared to submerged oil cooling. :rolleyes:
-
leppie wrote:
Nope, but then again I have water-cooling
Pft, that's nothing compared to submerged oil cooling. :rolleyes:
I have seen that before, but it sure seems like a lot of hassle (and a lot of other little issues).
xacc.ide
IronScheme - 1.0 RC 1 - out now!
((λ (x) `(,x ',x)) '(λ (x) `(,x ',x))) The Scheme Programming Language – Fourth Edition -
I have seen that before, but it sure seems like a lot of hassle (and a lot of other little issues).
xacc.ide
IronScheme - 1.0 RC 1 - out now!
((λ (x) `(,x ',x)) '(λ (x) `(,x ',x))) The Scheme Programming Language – Fourth EditionYeah, and the cooling is not even that great unless you pump the oil through a tube and cool that. But it's still neat. :)
-
Develop Jim@Home and I'll gladly lend my spare cycles to your animation processing.
-
Yeah, and the cooling is not even that great unless you pump the oil through a tube and cool that. But it's still neat. :)
I had a rig like that up and running for about 6 months before the power went out (in the house) one night...it never worked again and I've never spent the time to fix it... just drained it, put the mineral oil back in the jugs and it is now boxed up in the closet. http://www.kinar.net/computer/[^] The heat stayed pretty constant in the tank even when the machine was under load.
-
Has anyone had any problems with running a job for more than an hour that pushes most of the CPU cores to 100% usage? I'm planning on making some animation, and with each frame taking about 5-7 minutes each, a single second of animation is going to come in around 2 hours 20 minutes or worse. For a 10 to 20 second clip that could easily cause the CPU to run at 100% for almost 12 hours or more. I've never had anything run that long on any system I've owned and I'm not sure if it's going to overheat the box, or if it's hard on the CPU to run that long. Anyone have any experience with this on a desktop system (for the record it's an 8 core MacPro)?
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
-
Has anyone had any problems with running a job for more than an hour that pushes most of the CPU cores to 100% usage? I'm planning on making some animation, and with each frame taking about 5-7 minutes each, a single second of animation is going to come in around 2 hours 20 minutes or worse. For a 10 to 20 second clip that could easily cause the CPU to run at 100% for almost 12 hours or more. I've never had anything run that long on any system I've owned and I'm not sure if it's going to overheat the box, or if it's hard on the CPU to run that long. Anyone have any experience with this on a desktop system (for the record it's an 8 core MacPro)?
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
Whether its going to be OK or not depends on the system and the ambient environment. Laptops are typically the worst, but not all, and there's some poorly designed and cooled desktops out there too. Given multicore chips are still fairly new and mostly underutilized, I wouldn't just assume the system cooling was sufficient for extended runs of such a CPU intensive application, not even on a Mac. About all you can do is to set up a temperature monitor to warn you if it gets too hot and hope the sensors are in the needed places. Some CPUs have internal temp sensors built in and will protect themselves from overheating, I don't know if that's Mac's CPU has that feature.
patbob
-
Has anyone had any problems with running a job for more than an hour that pushes most of the CPU cores to 100% usage? I'm planning on making some animation, and with each frame taking about 5-7 minutes each, a single second of animation is going to come in around 2 hours 20 minutes or worse. For a 10 to 20 second clip that could easily cause the CPU to run at 100% for almost 12 hours or more. I've never had anything run that long on any system I've owned and I'm not sure if it's going to overheat the box, or if it's hard on the CPU to run that long. Anyone have any experience with this on a desktop system (for the record it's an 8 core MacPro)?
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
Heating and cooling cycles will wreck a CPU over time. Too much heat will too but most systems have safeguards in place to stop that from happening (unless disabled). I have run scientific apps that hit all cores at 100% for days and haven't had any problems. I'm not sure about Macs, but they're Intel now anyways and are basically PCs with a MAC OS and proprietary drivers. Just make sure that there is adequate ventilation around the computer so that it's not sucking in hot air. The internal temp should plateau at a certain level and stay there for the duration of the run (+/- a degree or two).
-
Jim Crafton wrote:
Has anyone had any problems with running a job for more than an hour that pushes most of the CPU cores to 100% usage?
Nope, but then again I have water-cooling[^] :cool: I think the 8core macs are the same as the i7 900 series, so your CPU can handle temps up to 99 deg C (not that you want it that high, just saying). Running continuously, I would not feel happy if it was over 80 deg.
xacc.ide
IronScheme - 1.0 RC 1 - out now!
((λ (x) `(,x ',x)) '(λ (x) `(,x ',x))) The Scheme Programming Language – Fourth Editionleppie wrote:
Nope, but then again I have water-cooling[^] Cool
so do I. Sadly the gap between top end air cooling and water cooling with a 2x120 rad has been shrinking rapidly. My next upgrade is probably going to be a 3x120 rad and a bigger case; currently I'm thinking a mountain mods u2-ufo[^].
3x12=36 2x12=24 1x12=12 0x12=18
-
leppie wrote:
Nope, but then again I have water-cooling[^] Cool
so do I. Sadly the gap between top end air cooling and water cooling with a 2x120 rad has been shrinking rapidly. My next upgrade is probably going to be a 3x120 rad and a bigger case; currently I'm thinking a mountain mods u2-ufo[^].
3x12=36 2x12=24 1x12=12 0x12=18
That case is a monster :) I also feel a 2x120 rad is just a little too small for an i7. At least I can make space to add another 2x120. Anything larger, and I would need to buy something bigger too. I really like that Silverstone TJ07 though, but it is not available locally, and shipping would be a killer from anywhere else.
xacc.ide
IronScheme - 1.0 RC 1 - out now!
((λ (x) `(,x ',x)) '(λ (x) `(,x ',x))) The Scheme Programming Language – Fourth Edition -
I've been running Folding/SETI/MilkyWay@Home for years straight now. 100% all the time
// Steve McLenithan
I too have been running SETI@Home for years, so 100% all the time here too. I did have my fan turned down too low for a while and I was getting some unexpected freezes every now and again, but after turning up the fan speed it has been fine ever since.
-
That case is a monster :) I also feel a 2x120 rad is just a little too small for an i7. At least I can make space to add another 2x120. Anything larger, and I would need to buy something bigger too. I really like that Silverstone TJ07 though, but it is not available locally, and shipping would be a killer from anywhere else.
xacc.ide
IronScheme - 1.0 RC 1 - out now!
((λ (x) `(,x ',x)) '(λ (x) `(,x ',x))) The Scheme Programming Language – Fourth Editionleppie wrote:
That case is a monster
My current case[^] is a monster as well. Unfortunately its becoming a victim of the ever growing GPU length. It's only 12" front to back; after subtracting the thickness of the front panel and the fans it only just has room for a 10.5" card. ATI's 5890 card won't fit, and the more common 10.5" cards are very difficult to install. Actaully it's one of the smaller mountain mod cube cases. The extended version is bigger front to back, while the Asecendency is taller. OTOH MM sells a 12" tall expansion box to install underneath their cases if you find yourself needing more space. :omg: If I ever decide to watercool my GPUs that would give me the space to do so without needing a new case for the extra rads. OTOH the pedestal box costs as much as a normal premium case. :doh: My only disappointment is that they don't have any back panel options that would let me have a horizontal mobo (better heat dissipation with 2 aircooled cards) with exhaust fans placed above the GPUs.
leppie wrote:
I also feel a 2x120 rad is just a little too small for an i7. At least I can make space to add another 2x120.
I thought I did as well; but because my rad extends slightly beyond the edge of the fans I'd need to make new screw holes at a minimum to install it. The only rad that will fit double in the case is thermaltake's; and their watercooling products have a poor reputation both for leak resistance and performance.
3x12=36 2x12=24 1x12=12 0x12=18
-
leppie wrote:
That case is a monster
My current case[^] is a monster as well. Unfortunately its becoming a victim of the ever growing GPU length. It's only 12" front to back; after subtracting the thickness of the front panel and the fans it only just has room for a 10.5" card. ATI's 5890 card won't fit, and the more common 10.5" cards are very difficult to install. Actaully it's one of the smaller mountain mod cube cases. The extended version is bigger front to back, while the Asecendency is taller. OTOH MM sells a 12" tall expansion box to install underneath their cases if you find yourself needing more space. :omg: If I ever decide to watercool my GPUs that would give me the space to do so without needing a new case for the extra rads. OTOH the pedestal box costs as much as a normal premium case. :doh: My only disappointment is that they don't have any back panel options that would let me have a horizontal mobo (better heat dissipation with 2 aircooled cards) with exhaust fans placed above the GPUs.
leppie wrote:
I also feel a 2x120 rad is just a little too small for an i7. At least I can make space to add another 2x120.
I thought I did as well; but because my rad extends slightly beyond the edge of the fans I'd need to make new screw holes at a minimum to install it. The only rad that will fit double in the case is thermaltake's; and their watercooling products have a poor reputation both for leak resistance and performance.
3x12=36 2x12=24 1x12=12 0x12=18
Dan Neely wrote:
My current case[^] is a monster as well.
That is an interesting case :)
xacc.ide
IronScheme - 1.0 RC 1 - out now!
((λ (x) `(,x ',x)) '(λ (x) `(,x ',x))) The Scheme Programming Language – Fourth Edition -
Has anyone had any problems with running a job for more than an hour that pushes most of the CPU cores to 100% usage? I'm planning on making some animation, and with each frame taking about 5-7 minutes each, a single second of animation is going to come in around 2 hours 20 minutes or worse. For a 10 to 20 second clip that could easily cause the CPU to run at 100% for almost 12 hours or more. I've never had anything run that long on any system I've owned and I'm not sure if it's going to overheat the box, or if it's hard on the CPU to run that long. Anyone have any experience with this on a desktop system (for the record it's an 8 core MacPro)?
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
-
Sounds like a perfect fit for the Amazon EC2. If you have a couple of dollars to spend you could render all of your frames across a large farm pretty cheap. http://aws.amazon.com/ec2/#pricing[^]
Actually Houdini has a service for this (on Amazon EC2) which I'm hoping to use. However if I can't get that to work (which I can't at the moment), then I'll have to render the frames at home.
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
-
Dan Neely wrote:
My current case[^] is a monster as well.
That is an interesting case :)
xacc.ide
IronScheme - 1.0 RC 1 - out now!
((λ (x) `(,x ',x)) '(λ (x) `(,x ',x))) The Scheme Programming Language – Fourth EditionYeah, interesting if some of the ideas were a bit silly. I'd be shocked if 5% of the buyers bought any of the accesories intended for the 2nd PC. I ended up using the miniItx are to mount my pump and res. It's got a few spots where the otherwise excellent cable management (route everything behind the mobo and just pop it up at the appropriate edge hole) flopped as well. The front panel USB, etc are attached to the removable front panel, not the chassis. IF you slop the cables loosely through the HDD cage this isn't an issue; if you take advantage of the fact that there is *barely* enough space between the drive cages to route the cables it becomes a major PITA. An extra 1/4" of width would have done wonders there. That was rather frustrating considering how polished everything else was; right down to the tapped holes in the floor of the case and back of the mobo case to facilitate screwing down misc parts. :sigh:
3x12=36 2x12=24 1x12=12 0x12=18
-
Yeah, interesting if some of the ideas were a bit silly. I'd be shocked if 5% of the buyers bought any of the accesories intended for the 2nd PC. I ended up using the miniItx are to mount my pump and res. It's got a few spots where the otherwise excellent cable management (route everything behind the mobo and just pop it up at the appropriate edge hole) flopped as well. The front panel USB, etc are attached to the removable front panel, not the chassis. IF you slop the cables loosely through the HDD cage this isn't an issue; if you take advantage of the fact that there is *barely* enough space between the drive cages to route the cables it becomes a major PITA. An extra 1/4" of width would have done wonders there. That was rather frustrating considering how polished everything else was; right down to the tapped holes in the floor of the case and back of the mobo case to facilitate screwing down misc parts. :sigh:
3x12=36 2x12=24 1x12=12 0x12=18
Pretty much sums up what you should expect from hardware. Never more than 80% of what you hope for. :) Like my silly case's top fan mounting holes. Nice, but dont work on a radiator, had to drill more holes. Also, the top space is just to small. Even an extra 10mm would have made a world of difference... I had to mutilate that one fan 'touching' the CPU power connector. Lucky, I had 1mm clearance on the RAM levers.
xacc.ide
IronScheme - 1.0 RC 1 - out now!
((λ (x) `(,x ',x)) '(λ (x) `(,x ',x))) The Scheme Programming Language – Fourth Edition -
Pretty much sums up what you should expect from hardware. Never more than 80% of what you hope for. :) Like my silly case's top fan mounting holes. Nice, but dont work on a radiator, had to drill more holes. Also, the top space is just to small. Even an extra 10mm would have made a world of difference... I had to mutilate that one fan 'touching' the CPU power connector. Lucky, I had 1mm clearance on the RAM levers.
xacc.ide
IronScheme - 1.0 RC 1 - out now!
((λ (x) `(,x ',x)) '(λ (x) `(,x ',x))) The Scheme Programming Language – Fourth Edition -
Actually Houdini has a service for this (on Amazon EC2) which I'm hoping to use. However if I can't get that to work (which I can't at the moment), then I'll have to render the frames at home.
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
How many servers does it split the load over concurrently? Since amazon's using commodity servers you'd probably need at least a dozen to match your systems raw throughput. For that matter how expensive will rendering that way turn out to be; when you're potentially buying over a server-day per render the costs could add up fast.
3x12=36 2x12=24 1x12=12 0x12=18
