hi, i have magma 4 pci expander and running 3 uad1 and one scope 15dps card.
my setup is getting very hot especially now during the hot days. i have a build in fan to cool the cards, but it is still very hot inside and the real hot days are still coming.
so my question, what´s the max temperature for a dsp card? how many degrees of celsius can such card handle without getting harmed or destroyed?
thx
what´s the max temperature a dsp card can handle?
-
helldriver
- Posts: 273
- Joined: Fri Aug 31, 2001 4:00 pm
Re: what´s the max temperature a dsp card can handle?
the Sharc DSPs are designed to run pretty hot - 85 Celsius iirc
As long as you provide some airflow all should be fine.
It may be difficult, tho if there's no room above the cards and you can't leave some 'intermediate' slots open as air outlets.
cheers, Tom
As long as you provide some airflow all should be fine.
It may be difficult, tho if there's no room above the cards and you can't leave some 'intermediate' slots open as air outlets.
cheers, Tom
Re: what´s the max temperature a dsp card can handle?
This is my generalization that I use when cooling computers, explained in overly simplified form and is in no way meant to be accurate for any given part:
The way I look at it is that higher thermals reduce lifespan, but ALL use will have some effect on a part's lifespan. The effect of thermal damage could be visualized in simplified form as an exponential curve (this is useful for illustrating the idea but will actually completely differ based on application & workload). Ie, whereas 120C might fry a part and make it unusable from that point forward (and of course anything above also results in failure very shortly or immediately), the same part might operate at 110C for several months before starting to give errors or suffer issues (logic gates no longer firing reliably), and at 85C (within spec for the AD Sharc chips as noted here) the card might function fine for a few years or even longer (depending on the other components surrounding it etc). But it's possible that some other IC or a few capacitors could be aged beyond what they would have at say 75C or lower, and in the case of the lower temps the card may have lasted considerably longer.
One thing that overclockers learn when they mess around with thermal envelopes, is that all parts vary in how long they will hold up. One person might get a cpu or graphics card that will last for 6 months at speeds that noone else can attain, another person might wind up with a part that is completely worthless past stock speeds. Usually this is due to the ram or cpu/gpu (and its actually due to small variations at the atomic structure--at least by present theories) but the point to realize here is that eventually that 'overclocked' cpu will degrade enough that it will start to have errors and need to be 'tweaked'. Eventually things will just not be stable at any usable speed (or voltage) but for modern chips themselves that age is actually rather high within their operating spec (10+ years for your average intel cpu and arguably MUCH longer in most cases).
But, this only represents 1 part of the overall device, ALL electronic components on your Scope card will have an unpredictable and finite lifespan that is determined by their potential for flaws during manufacturing and use as influenced by usage (duty-cycle or work-cycle), application (workload) and your operating thermals. We can expect that everything is within specification as designed by Creamware/Soniccore on a card that doesn't fail in the first few months, but within those specs are 'tolerances' that translate into the variances seen in components.
At the same time one must recognize that putting the card on a nice cool shelf in the box and not touching it won't keep it from aging either to at least some degree, and we bought these things to use them. Personally I still have my Pulsar1 card (alongside a newer one), and it's kept in a machine with both Scope cards that is powered on all year round, but I try to keep the machine relatively cool and the Pulsar1 card is now 8-9 years old. I do hope it lasts another 5-10 years at least (and wonder if I'll be recapping it at some point or replacing it.)
Basically I'm echoing the above sentiment, I would recommend additional airflow, whether you have to do that via a slot cooler, a fan mounted in the side of the case or the front, or whatever. The cards will perform fine, but when dealing with electronics (and especially silicon & IC's) thermal effects aren't just 'will it run or wont it', they affect lifespan.
The way I look at it is that higher thermals reduce lifespan, but ALL use will have some effect on a part's lifespan. The effect of thermal damage could be visualized in simplified form as an exponential curve (this is useful for illustrating the idea but will actually completely differ based on application & workload). Ie, whereas 120C might fry a part and make it unusable from that point forward (and of course anything above also results in failure very shortly or immediately), the same part might operate at 110C for several months before starting to give errors or suffer issues (logic gates no longer firing reliably), and at 85C (within spec for the AD Sharc chips as noted here) the card might function fine for a few years or even longer (depending on the other components surrounding it etc). But it's possible that some other IC or a few capacitors could be aged beyond what they would have at say 75C or lower, and in the case of the lower temps the card may have lasted considerably longer.
One thing that overclockers learn when they mess around with thermal envelopes, is that all parts vary in how long they will hold up. One person might get a cpu or graphics card that will last for 6 months at speeds that noone else can attain, another person might wind up with a part that is completely worthless past stock speeds. Usually this is due to the ram or cpu/gpu (and its actually due to small variations at the atomic structure--at least by present theories) but the point to realize here is that eventually that 'overclocked' cpu will degrade enough that it will start to have errors and need to be 'tweaked'. Eventually things will just not be stable at any usable speed (or voltage) but for modern chips themselves that age is actually rather high within their operating spec (10+ years for your average intel cpu and arguably MUCH longer in most cases).
But, this only represents 1 part of the overall device, ALL electronic components on your Scope card will have an unpredictable and finite lifespan that is determined by their potential for flaws during manufacturing and use as influenced by usage (duty-cycle or work-cycle), application (workload) and your operating thermals. We can expect that everything is within specification as designed by Creamware/Soniccore on a card that doesn't fail in the first few months, but within those specs are 'tolerances' that translate into the variances seen in components.
At the same time one must recognize that putting the card on a nice cool shelf in the box and not touching it won't keep it from aging either to at least some degree, and we bought these things to use them. Personally I still have my Pulsar1 card (alongside a newer one), and it's kept in a machine with both Scope cards that is powered on all year round, but I try to keep the machine relatively cool and the Pulsar1 card is now 8-9 years old. I do hope it lasts another 5-10 years at least (and wonder if I'll be recapping it at some point or replacing it.)
Basically I'm echoing the above sentiment, I would recommend additional airflow, whether you have to do that via a slot cooler, a fan mounted in the side of the case or the front, or whatever. The cards will perform fine, but when dealing with electronics (and especially silicon & IC's) thermal effects aren't just 'will it run or wont it', they affect lifespan.