Ever Beedle on Nostr: How about consumer GPUs? I know A100s are fast, but in terms of raw throughput ...
How about consumer GPUs? I know A100s are fast, but in terms of raw throughput isn’t a 4090 like a small fraction of the cost while being much faster? Or is it using an instruction set that nvidia has gimped on the consumer models?
Published at
2023-03-09 22:49:27Event JSON
{
"id": "6fa120087c12e2c185d50cea14ca774a0b9e5e576f118042e2550ecf9feb6421",
"pubkey": "1ccff9e458848e5aa1f0e483864073e116b15f6bb3aa0a2a365f37b665df058e",
"created_at": 1678402167,
"kind": 1,
"tags": [
[
"p",
"7259c17f0879d8ab09f508b2d2c9c16b0f51fcd06ab59213402edff6ce3e83a4",
"wss://relay.mostr.pub"
],
[
"e",
"a9d7ed094148af08fdbfdac5513c6fa00b3a4275f51f7228b58e558a01b514a5",
"wss://relay.mostr.pub",
"reply"
]
],
"content": "How about consumer GPUs? I know A100s are fast, but in terms of raw throughput isn’t a 4090 like a small fraction of the cost while being much faster? Or is it using an instruction set that nvidia has gimped on the consumer models?",
"sig": "0ffbd160ca7fbf6b7c5a850a26f903fbb03f879180ce8a6dec56b9e3b894f9e35da36fb63fa4c64d28096181e5ab28bfd56b0992c7f5c308c957107518c3d6e2"
}
