One research project I’m embarking on is one that I actually thought of accidentally. It will be looking at different thermal conductivity in thermal paste used on ASIC chips to measure performance in hashrate.
It may sound weird, I know. But while reapplying thermal paste to several hash boards, I inadvertently used two types of thermal paste with different thermal conductivity, and saw a minor difference in temperature readings when testing them.
I’ll be using different thermal pastes (based on W/mK) on a control (S19j Pro 100 TH/s) with an adequate sample size (likely 3 known fully hashing machines) and I will look at power draw, chip temperature, and hashrate over the average of 12 hours in a controlled climate.
There is, of course, a point of diminishing returns, and I hope to find that to see the peak optimization in a machine based on a tiny factor such as thermal paste. It may be minute, but it very well could be large enough to consider.
Much to be done!
Tatum / Tatum Cave
npub1w6…x0ymf
2023-11-28 03:43:47
Author Public Key
npub1w64uxt9xmuhdfq6mmeypveme46svvfphx4r69yndmjsgask6knwqjx0ymfPublished at
2023-11-28 03:43:47Event JSON
{
"id": "8580d6ddcad4fef72fd18ade23ff853aecf47ccce506515c549e2ea3e89a5795",
"pubkey": "76abc32ca6df2ed4835bde48166779aea0c624373547a2926ddca08ec2dab4dc",
"created_at": 1701143027,
"kind": 1,
"tags": [],
"content": "One research project I’m embarking on is one that I actually thought of accidentally. It will be looking at different thermal conductivity in thermal paste used on ASIC chips to measure performance in hashrate. \n\nIt may sound weird, I know. But while reapplying thermal paste to several hash boards, I inadvertently used two types of thermal paste with different thermal conductivity, and saw a minor difference in temperature readings when testing them. \n\nI’ll be using different thermal pastes (based on W/mK) on a control (S19j Pro 100 TH/s) with an adequate sample size (likely 3 known fully hashing machines) and I will look at power draw, chip temperature, and hashrate over the average of 12 hours in a controlled climate. \n\nThere is, of course, a point of diminishing returns, and I hope to find that to see the peak optimization in a machine based on a tiny factor such as thermal paste. It may be minute, but it very well could be large enough to consider. \n\nMuch to be done!",
"sig": "3b61f1e6d81ec8dd07050ee29c3bba53c04b73fe63bd9ddc83dcbd607019303f99d718531a3db79152f94485ff9c5d36a4c738a8d4764859f965e820dc85bd96"
}