What's the likelihood that something other than the PSU is fried? In other words, would it be prudent to buy a new PSU to try and revive the host?
Hi! It would be interesting to look at the electrical specs of those products. What are the exact model of the PSU and motherboard and GPU's that were involved in this cooking?
If you don't see obvious overheating damage to the slots the cards were plugged into or damage to the 24pin and cpuaux power connectors, or damage to the PCIe power cables, you likely just had a failed PSU.
I unfortunately don't have any experience at all with AMD cards, only Nvidia. I use the Linux Nvidia application to power limit my RTX 2080 and GTX 1080Ti to 200 watts.
nvidia-smi -i 1 -pl 200
nvidia-smi -i 3 -pl 200
is the command I use in my overclocking script for those cards. There might be something similar for AMD. I would bet even money that there is over in the crypto mining forums as AMD cards are popular there and I know they limit the power levels for Nvidia cards. So expect they did the same for AMD.
Prudence be damned - I just bought a new PSU from a local shop. Will know in a day or two whether anything else was fried.
Slow down a bit before you fry a second one :-). A few golden rules :-). Make sure your replacement PSU is able to supply virtually all its rated output at 12V. For example, you implied that your fried PSU had a 400W rating. Did the label say something like 12 volts 34 amps (=408 watts @ 12 volts) or was the 400W shared between all rails with the 12V current somewhat less than 34 amps (perhaps even less than 30 amps)?
If your machine draws 400W when loaded, you need at least 55 amps at 12V (=660W), preferably a little higher. That calculation is based on 400W being 60% of 660W. PSUs are most efficient when you limit the draw to around 40-60% of their rated output. You can get away with a bit more than 60% but you may shorten its life considerably. It's very important NOT to buy the cheapest unit available as generic PSU makers tend to lie in their claimed output specs. You want an 80+% efficient unit (preferably at least 80+% Gold rating) where the manufacturer has enough confidence to give you something like a 5 year warranty (or more). These can be a little expensive so it pays to do some research and to shop around. There is *lots* of advice available on line.
It's quite likely that the rest of your hardware may be OK. When PSUs self destruct spectacularly through overload, they don't usually take other stuff with them. PSUs are supposed to have overload protection. Budget PSUs may very well skip on protections though. If you want to avoid the same fate happening to the next one, make sure you list the details of what you propose to buy. I'm sure there will be several willing to give advice :-).
That calculation is based on 400W being 60% of 660W. PSUs are most efficient when you limit the draw to around 40-60% of their rated output. You can get away with a bit more than 60% but you may shorten its life considerably. It's very important NOT to buy the cheapest unit available as generic PSU makers tend to lie in their claimed output specs. You want an 80+% efficient unit (preferably at least 80+% Gold rating) where the manufacturer has enough confidence to give you something like a 5 year warranty (or more).
I think that advice for heading towards 40-60% area sounds good, but at the same time I wouldn't be too worried about stressing a quality power even more. Because if we're talking about a quality power that's been given a long warranty (5+ years) it also means those PSU's are built to take the torture quite well. Also these days the 'good efficiency area' is very wide on a quality PSU... or at least the relative percentage drop is very small even when the PSU is driven way outside that 40-60% area (excluding extremes). I think that is a thing where manufacturers have been able to better the products quite much from the past. Here's efficiency graphs of three quality PSU's that are "Gold":
You could basically let the computer sit idle, drive it somewhat or drive it freakin' hard and the efficiency would change only a little.
For me personally the acoustic performance would be the most important factor if the PSU is otherwise reasonably 'good quality'. That's where I would put some thought... to figure out how much the total load of the computer under crunching would be... and then I'd take a moment to look at the acoustic specs of a PSU. Under what circumstances will it turn on the fan?
I have a few years old Corsair RM850 in a system with RX580 now.. It doesn't spin the fan at all and there's no option to make it spin (load is too low and there's no switch). That's acoustically optimal, I don't complain about that. I believe the PSU doesn't currently run too warm either, because it's at the floor level, fan-side up... and gets cool air. So the warm air can freely exit the box. But if that PSU would be installed upside-down, then that zero rpm might not be a good thing in a long run (temp could be higher and fan still not spinning... which could shorten life of the PSU). So I would say the PSU orientation could be a thing to take into account sometimes... with wattage capacity and how the fan is set to work).
edit: I think that the a quality PSU even with medium capacity can stand the electrical stress quite well, but if the fan curve on a smaller capacity PSU is such that the fan will be spinning a lot of the time... then the fan tends to wear out sooner. Generally a heavier duty PSU might operate the fan in a more relaxed manner which in turn will help it wear less and the bearing will remain silent longer.
Just some thoughts in the middle of night... already outside the topic, sorry.
It's very important NOT to buy the cheapest unit available as generic PSU makers tend to lie in their claimed output specs. You want an 80+% efficient unit (preferably at least 80+% Gold rating) where the manufacturer has enough confidence to give you something like a 5 year warranty (or more). These can be a little expensive so it pays to do some research and to shop around. There is *lots* of advice available on line.
One other thing to consider is the heat the lower quality psu's put out as opposed to the ones that are 'rated' by color codes, I've been replacing my non color rated ones and my a/c units don't run as hard any more. There is also a big difference between Gold and Bronze too, with Gold being lots cooler. I think Platinum is the highest , the highest I've seen anyway, and it puts out even less heat.
It's very important NOT to buy the cheapest unit available as generic PSU makers tend to lie in their claimed output specs. You want an 80+% efficient unit (preferably at least 80+% Gold rating) where the manufacturer has enough confidence to give you something like a 5 year warranty (or more). These can be a little expensive so it pays to do some research and to shop around. There is *lots* of advice available on line.
One other thing to consider is the heat the lower quality psu's put out as opposed to the ones that are 'rated' by color codes, I've been replacing my non color rated ones and my a/c units don't run as hard any more. There is also a big difference between Gold and Bronze too, with Gold being lots cooler. I think Platinum is the highest , the highest I've seen anyway, and it puts out even less heat.
They're still fairly rare but 80+ Titanium has been out for some years. Although not of much interest to full time Boinc crunchers users, it added a 90% efficiency at 10% load requirement (vs 20% being the minimum in prior standards) in addition to the usual ~2% higher efficiency at higher load levels. With newer PCs drawing much less power when the CPU is idle efficiency at 10% load is relevant a lot more often than at much higher levels.
As far as what will pay for itself in reduced power costs over the lifetime of a system, at US retail prices at ~13 cents/kWh for power (about the US average) last summer was the first time a Titanium PSU for one of my crunchers would pay for itself in reduced power costs (ignoring AC savings) over about 6 years of 24/7 use. While that appears to've been a sale price (it's $18 more now) previously the quality of sales only toggled cheapest lifetime costs between Gold and Platinum. With many of my previous PSUs dieing around 7-9 years (generally 1 or 2 years outside of warranty) Seasonic offering a 12 year warranty was also a factor since it means I should get a few more years out of the purchase; which should pay for the premium even ignoring the power savings.
I use Linux Kubuntu and have no problems so I suppose you will have no problems with Lubuntu too.
As for adjusting AMD parameters, as far as I know so far there are only ways via a text command window, and if I remember right, something with a graphic is being worked on. Having just a 580 I never messed with that.
I suppose you successfully installed AMD OpenCL on it!
Hi! It would be interesting to look at the electrical specs of those products. What are the exact model of the PSU and motherboard and GPU's that were involved in this cooking?
The PSU is/was a CoolerMaster RS-460-PSAR-J3. The motherboard is a GIGABYTE, circa 2009, with an AMD Phenom II X4 CPU, but at the moment the MB model number is covered by the GPU's, which I don't want to take out for the umpteenth time. (I thought I had that model number written down somewhere....). The GPUs are a RX 460 1024SP (OEM pulled new from a HP system) and a RX 570 (used, XFX RS, Black Edition).
Under Win7 that same system configuration was pulling 370W (~340W with both GPUs underclocked), so seeing it pull 400W just before it blew makes me think that the extra draw was a symptom of the PSU giving itself last rites. The resting state draw of the Lubuntu system (no E@H running) was 110W.
Ideas are not fixed, nor should they be; we live in model-dependent reality.
I suppose you successfully installed AMD OpenCL on it!
Yes! I installed the AMD Ubuntu drivers easy-peasy by clicking on the install script, then, not sure whether OpenCL had installed (the AMD install notes made me think they hadn't), I used AMD's terminal codes and got AMD OpenCL installed no problem.
Ideas are not fixed, nor should they be; we live in model-dependent reality.
If you don't see obvious overheating damage to the slots the cards were plugged into or damage to the 24pin and cpuaux power connectors, or damage to the PCIe power cables, you likely just had a failed PSU.
I unfortunately don't have any experience at all with AMD cards, only Nvidia. I use the Linux Nvidia application to power limit my RTX 2080 and GTX 1080Ti to 200 watts.
nvidia-smi -i 1 -pl 200
nvidia-smi -i 3 -pl 200
is the command I use in my overclocking script for those cards. There might be something similar for AMD. I would bet even money that there is over in the crypto mining forums as AMD cards are popular there and I know they limit the power levels for Nvidia cards. So expect they did the same for AMD.
Excellent. Good to know. I'll check it out for sure. TY!
Ideas are not fixed, nor should they be; we live in model-dependent reality.
cecht wrote:What's the
)
Hi! It would be interesting to look at the electrical specs of those products. What are the exact model of the PSU and motherboard and GPU's that were involved in this cooking?
If you don't see obvious
)
If you don't see obvious overheating damage to the slots the cards were plugged into or damage to the 24pin and cpuaux power connectors, or damage to the PCIe power cables, you likely just had a failed PSU.
I unfortunately don't have any experience at all with AMD cards, only Nvidia. I use the Linux Nvidia application to power limit my RTX 2080 and GTX 1080Ti to 200 watts.
nvidia-smi -i 1 -pl 200
nvidia-smi -i 3 -pl 200
is the command I use in my overclocking script for those cards. There might be something similar for AMD. I would bet even money that there is over in the crypto mining forums as AMD cards are popular there and I know they limit the power levels for Nvidia cards. So expect they did the same for AMD.
cecht wrote:Prudence be
)
Slow down a bit before you fry a second one :-). A few golden rules :-). Make sure your replacement PSU is able to supply virtually all its rated output at 12V. For example, you implied that your fried PSU had a 400W rating. Did the label say something like 12 volts 34 amps (=408 watts @ 12 volts) or was the 400W shared between all rails with the 12V current somewhat less than 34 amps (perhaps even less than 30 amps)?
If your machine draws 400W when loaded, you need at least 55 amps at 12V (=660W), preferably a little higher. That calculation is based on 400W being 60% of 660W. PSUs are most efficient when you limit the draw to around 40-60% of their rated output. You can get away with a bit more than 60% but you may shorten its life considerably. It's very important NOT to buy the cheapest unit available as generic PSU makers tend to lie in their claimed output specs. You want an 80+% efficient unit (preferably at least 80+% Gold rating) where the manufacturer has enough confidence to give you something like a 5 year warranty (or more). These can be a little expensive so it pays to do some research and to shop around. There is *lots* of advice available on line.
It's quite likely that the rest of your hardware may be OK. When PSUs self destruct spectacularly through overload, they don't usually take other stuff with them. PSUs are supposed to have overload protection. Budget PSUs may very well skip on protections though. If you want to avoid the same fate happening to the next one, make sure you list the details of what you propose to buy. I'm sure there will be several willing to give advice :-).
Cheers,
Gary.
Gary Roberts wrote:That
)
I think that advice for heading towards 40-60% area sounds good, but at the same time I wouldn't be too worried about stressing a quality power even more. Because if we're talking about a quality power that's been given a long warranty (5+ years) it also means those PSU's are built to take the torture quite well. Also these days the 'good efficiency area' is very wide on a quality PSU... or at least the relative percentage drop is very small even when the PSU is driven way outside that 40-60% area (excluding extremes). I think that is a thing where manufacturers have been able to better the products quite much from the past. Here's efficiency graphs of three quality PSU's that are "Gold":
https://www.tomshardware.com/reviews/super-flower-leadex-gold-550w-power-supply,4416-5.html
https://www.tomshardware.com/reviews/evga-supernova-650-g3-psu,5533-5.html
https://www.tomshardware.com/reviews/corsair-rm750x-v2-psu,5585-5.html
You could basically let the computer sit idle, drive it somewhat or drive it freakin' hard and the efficiency would change only a little.
For me personally the acoustic performance would be the most important factor if the PSU is otherwise reasonably 'good quality'. That's where I would put some thought... to figure out how much the total load of the computer under crunching would be... and then I'd take a moment to look at the acoustic specs of a PSU. Under what circumstances will it turn on the fan?
I have a few years old Corsair RM850 in a system with RX580 now.. It doesn't spin the fan at all and there's no option to make it spin (load is too low and there's no switch). That's acoustically optimal, I don't complain about that. I believe the PSU doesn't currently run too warm either, because it's at the floor level, fan-side up... and gets cool air. So the warm air can freely exit the box. But if that PSU would be installed upside-down, then that zero rpm might not be a good thing in a long run (temp could be higher and fan still not spinning... which could shorten life of the PSU). So I would say the PSU orientation could be a thing to take into account sometimes... with wattage capacity and how the fan is set to work).
edit: I think that the a quality PSU even with medium capacity can stand the electrical stress quite well, but if the fan curve on a smaller capacity PSU is such that the fan will be spinning a lot of the time... then the fan tends to wear out sooner. Generally a heavier duty PSU might operate the fan in a more relaxed manner which in turn will help it wear less and the bearing will remain silent longer.
Just some thoughts in the middle of night... already outside the topic, sorry.
Gary Roberts wrote:It's very
)
One other thing to consider is the heat the lower quality psu's put out as opposed to the ones that are 'rated' by color codes, I've been replacing my non color rated ones and my a/c units don't run as hard any more. There is also a big difference between Gold and Bronze too, with Gold being lots cooler. I think Platinum is the highest , the highest I've seen anyway, and it puts out even less heat.
mikey wrote:Gary Roberts
)
They're still fairly rare but 80+ Titanium has been out for some years. Although not of much interest to full time Boinc crunchers users, it added a 90% efficiency at 10% load requirement (vs 20% being the minimum in prior standards) in addition to the usual ~2% higher efficiency at higher load levels. With newer PCs drawing much less power when the CPU is idle efficiency at 10% load is relevant a lot more often than at much higher levels.
As far as what will pay for itself in reduced power costs over the lifetime of a system, at US retail prices at ~13 cents/kWh for power (about the US average) last summer was the first time a Titanium PSU for one of my crunchers would pay for itself in reduced power costs (ignoring AC savings) over about 6 years of 24/7 use. While that appears to've been a sale price (it's $18 more now) previously the quality of sales only toggled cheapest lifetime costs between Gold and Platinum. With many of my previous PSUs dieing around 7-9 years (generally 1 or 2 years outside of warranty) Seasonic offering a 12 year warranty was also a factor since it means I should get a few more years out of the purchase; which should pay for the premium even ignoring the power savings.
https://en.wikipedia.org/wiki/80_Plus
I use Linux Kubuntu and have
)
I use Linux Kubuntu and have no problems so I suppose you will have no problems with Lubuntu too.
As for adjusting AMD parameters, as far as I know so far there are only ways via a text command window, and if I remember right, something with a graphic is being worked on. Having just a 580 I never messed with that.
I suppose you successfully installed AMD OpenCL on it!
Richie wrote:Hi! It would be
)
The PSU is/was a CoolerMaster RS-460-PSAR-J3. The motherboard is a GIGABYTE, circa 2009, with an AMD Phenom II X4 CPU, but at the moment the MB model number is covered by the GPU's, which I don't want to take out for the umpteenth time. (I thought I had that model number written down somewhere....). The GPUs are a RX 460 1024SP (OEM pulled new from a HP system) and a RX 570 (used, XFX RS, Black Edition).
Under Win7 that same system configuration was pulling 370W (~340W with both GPUs underclocked), so seeing it pull 400W just before it blew makes me think that the extra draw was a symptom of the PSU giving itself last rites. The resting state draw of the Lubuntu system (no E@H running) was 110W.
Ideas are not fixed, nor should they be; we live in model-dependent reality.
Millenium wrote:I suppose you
)
Yes! I installed the AMD Ubuntu drivers easy-peasy by clicking on the install script, then, not sure whether OpenCL had installed (the AMD install notes made me think they hadn't), I used AMD's terminal codes and got AMD OpenCL installed no problem.
Ideas are not fixed, nor should they be; we live in model-dependent reality.
Keith Myers wrote:If you
)
Excellent. Good to know. I'll check it out for sure. TY!
Ideas are not fixed, nor should they be; we live in model-dependent reality.