G
Guest
Guest
There has been a lot of talk about isotopically pure silicon going into the Palomino. First, for anyone interested in a little education on why semiconductor companies might want to use it:
http://www.isonics.com/silicon.htm
Overall, the article is very informative. Some of their claims are outlandish (like isotopically pure silicon is going to help make a more stable Gate Oxide... That assumes the starting Silicon substrate/epi-layer is the source of Gate Oxide impurities/defects.. probably not true), however most seem reasonable.
I know people have excitedly kicked around a 60% increase in thermal conductivity. One, this doesn't directly translate into a 60% cooler processor (as touched on briefly in the article). Two, the hotter the silicon, the smaller the effects of isotopically pure silicon. As they mention at typical processor temperatures they see a 40% change in thermal conductivity from typical Silicon wafers. They don't mention several other important items. The largest of which is where do we currently remove heat from the processor? From the top of the processor, not just the substrate. So the heat still has to travel up through all of the different materials in the processor and through several interfaces. All of these things will dampen the effect of this new Silicon type. If I had to guess I might say a 10-20% increase in the overall thermal conductivity of the processor.
Without a doubt, this is a nice improvement. Something you overclockers might want to consider is the temperature/thermal conductivity trend. I imagine it holds true as you decrease the temperature too. The colder the Silicon, the better the isotopically pure wafer conducts heat (as compared to the normal Si wafer). So, the more you cool the processor, the better it conducts heat, the cooler it runs, etc. If their claims are true, you should be able to cool your processors to lower temperatures on isotopically pure silicon. This assumes you refrigerant and compressor aren't limiting factors. Temperature isn't controlled JUST by how you can conduct heat. The chiller has to be able to pull all of that heat away. Even if it can't, you may notice a degree or two difference.
Hell, call me a dork, but it would be fun to find out.
Finally, here is the article that announces the delay in their delivery along with a bunch of excuses for why they couldn't make the deadline:
http://www.isonics.com/pr67.htm
Enjoy
http://www.isonics.com/silicon.htm
Overall, the article is very informative. Some of their claims are outlandish (like isotopically pure silicon is going to help make a more stable Gate Oxide... That assumes the starting Silicon substrate/epi-layer is the source of Gate Oxide impurities/defects.. probably not true), however most seem reasonable.
I know people have excitedly kicked around a 60% increase in thermal conductivity. One, this doesn't directly translate into a 60% cooler processor (as touched on briefly in the article). Two, the hotter the silicon, the smaller the effects of isotopically pure silicon. As they mention at typical processor temperatures they see a 40% change in thermal conductivity from typical Silicon wafers. They don't mention several other important items. The largest of which is where do we currently remove heat from the processor? From the top of the processor, not just the substrate. So the heat still has to travel up through all of the different materials in the processor and through several interfaces. All of these things will dampen the effect of this new Silicon type. If I had to guess I might say a 10-20% increase in the overall thermal conductivity of the processor.
Without a doubt, this is a nice improvement. Something you overclockers might want to consider is the temperature/thermal conductivity trend. I imagine it holds true as you decrease the temperature too. The colder the Silicon, the better the isotopically pure wafer conducts heat (as compared to the normal Si wafer). So, the more you cool the processor, the better it conducts heat, the cooler it runs, etc. If their claims are true, you should be able to cool your processors to lower temperatures on isotopically pure silicon. This assumes you refrigerant and compressor aren't limiting factors. Temperature isn't controlled JUST by how you can conduct heat. The chiller has to be able to pull all of that heat away. Even if it can't, you may notice a degree or two difference.
Hell, call me a dork, but it would be fun to find out.
Finally, here is the article that announces the delay in their delivery along with a bunch of excuses for why they couldn't make the deadline:
http://www.isonics.com/pr67.htm
Enjoy