Comment by karmakaze 13 hours ago It seems we could use a new kind of memory that streams the weight data in, like GDDR in reverse. 5 comments karmakaze Reply 0x457 13 hours ago High Bandwidth Flash? https://www.sandisk.com/company/newsroom/blogs/2025/scaling-... rzzzt 9 hours ago Optane? (Too soon?) 0x457 8 hours ago Optane was targeting the latency gap, while HBF targets bandwidth. Given how LLMs work, HBF is perfect for offloading. carloslfu 13 hours ago interesting! carloslfu 13 hours ago yes! I guess future hardware designs will have something like that!
0x457 13 hours ago High Bandwidth Flash? https://www.sandisk.com/company/newsroom/blogs/2025/scaling-... rzzzt 9 hours ago Optane? (Too soon?) 0x457 8 hours ago Optane was targeting the latency gap, while HBF targets bandwidth. Given how LLMs work, HBF is perfect for offloading. carloslfu 13 hours ago interesting!
rzzzt 9 hours ago Optane? (Too soon?) 0x457 8 hours ago Optane was targeting the latency gap, while HBF targets bandwidth. Given how LLMs work, HBF is perfect for offloading.
0x457 8 hours ago Optane was targeting the latency gap, while HBF targets bandwidth. Given how LLMs work, HBF is perfect for offloading.
High Bandwidth Flash? https://www.sandisk.com/company/newsroom/blogs/2025/scaling-...
Optane? (Too soon?)
Optane was targeting the latency gap, while HBF targets bandwidth. Given how LLMs work, HBF is perfect for offloading.
interesting!
yes! I guess future hardware designs will have something like that!