'First of One' wrote:
| Alas, it is not quite that simple. If the HSF is the same unit as on the
| Sparkle 8800GTX, then it uses a hybrid design from MACS, with a set of
| secondary heat pipes going from the base plate directly to the heat sink
| fins. The brunt of the heat output is dissipated by conventional means.
|
http://anandtech.com/casecooling/showdoc.aspx?i=2921&p=2 .
|
| The design seems to work well enough on the Sparkle 8800GTX:
|
http://anandtech.com/video/showdoc.aspx?i=2945&p=13
_____
How would anandtech have anything useful to say about the subject -
(a.) they don't even measure display adapter consumption, just total system
power consumption
(b.) the CPU Stress temperature graph in
http://anandtech.com/casecooling/showdoc.aspx?i=2921&p=2
is misleading because the power consumption levels for the CPU are not
evenly spaced; the first step is much greater than all the following steps
combined
(c.) a Peltier array consuming 50 Watts max can handle perhaps 50% of the
heat load of an 80 Watt dissipation while lowering 50% of the cold side
temperature by ~ 5 C, but then the heatsink fan must have enough capacity to
handle any extra 60% heat load, hardly a stunning efficient design
considering the same cooling performance could be obtained with LESS
heatsink/fan capacity if the Peltier array were NOT used.
(d.) for the display adapter cooling in
http://anandtech.com/video/showdoc.aspx?i=2945&p=13
the problem is even worse; the Peltier array consuming 50 Watts max and
handle less than 20% of the heat load of a 145 Watt heat source while
lowering the cold side temperature by ~ 5 C, MAYBE worth a 1 C overall
source temperature reduction.
Try
http://www.tellurex.com for information on how Peltier arrays work and
what they can actually accomplish.
Phil Weldon
| Alas, it is not quite that simple. If the HSF is the same unit as on the
| Sparkle 8800GTX, then it uses a hybrid design from MACS, with a set of
| secondary heat pipes going from the base plate directly to the heat sink
| fins. The brunt of the heat output is dissipated by conventional means.
|
http://anandtech.com/casecooling/showdoc.aspx?i=2921&p=2 .
|
| The design seems to work well enough on the Sparkle 8800GTX:
|
http://anandtech.com/video/showdoc.aspx?i=2945&p=13
|
| --
| "War is the continuation of politics by other means.
| It can therefore be said that politics is war without
| bloodshed while war is politics with bloodshed."
|
| | > 'First of One' wrote:
| >> When did this thing appear?
| >>
http://www.gecube.com/products-detail.php?prod_cat_pid=9&prod_cat_id=166&prod_id=65116
| >>
http://www.newegg.com/Product/Product.aspx?Item=N82E16814241049
| >>
| >> AGP, Peltier-equipped HSF, 675 MHz core clock. $270.
| >>
| >> The HSF unit looks suspiciously like the one on Sparkle's GF8800GTX:
| >>
http://anandtech.com/video/showdoc.aspx?i=2945&p=3
| >>
| >> Do be aware the Peltier adds another 40W to the X1950XT's already high
| >> power consumption...
| > _____
| >
| > A Peltier array that added only 40 Watts "to the X1950XT's already high
| > power consumption" would be worse than insulator, and NOT a heat pump.
| > That's what happens when the heat flow is greater than the Peltier
| > capacity. About the only way such a small capacity Peltier array could
be
| > of any benefit would be if were used for cooling just a portion of the
| > card, say, the memory alone. That could keep the memory cool at the
| > expense of adding to the overall fan cooling load. A completely useless
| > crock. Why do they make these things?
| >
| > A Peltier with a Qmax rating of 40 Watts can pump 40 Watts of heat at a
| > temperature differential of 0 degrees (between the heatsink interface
and
| > the interface with the object cooled); NO cooling. It could pump 20
Watts
| > of heat and maintain a temperature differential of perhaps 20 C. For
any
| > heat flow of more than 40 Watts there would be no cooling, but heating
| > instead.
| >
| > Phil Weldon
|
|
