Abstract: The present disclosure proposes a keybed device, including: a keybed support; a keybed arranged on the keybed support; a key height limiting column arranged in one-to-one correspondence with each key in the keybed, and an end of the key height limiting column is fixed on a bottom surface of the corresponding key; a pressure detecting device arranged in one-to-one correspondence under each key height limiting column to detect a pressing strength of each key height limiting column; and a flexible support structure arranged below each key height limiting column and located on at least one side of each pressure detection device. The keybed device includes a flexible support structure, which can avoid false triggering of the Aftertouch effect. A same depth is pressed down from the flexible support structure to obtain a consistent Aftertouch effect, and the performance effect is controllable.

Abstract: An acceleration method for handshake request, includes: receiving a handshake request sent by a client towards a target domain name; feeding back a target credential bound to the target domain name to the client, where the target credential includes a specified public key so as to allow the client to utilize the specified public key to encrypt a session key of a current session; receiving an encrypted session key provided by the client, and sending a decryption request to an acceleration server, where the decryption request includes the encrypted session key so as to allow the acceleration server to decrypt the encrypted session key based on a private key bound to the target domain name; receiving and storing a decrypted session key fed back by the acceleration server, thereby completing a current handshake process.

Abstract: The present disclosure proposes a keybed device, including: a keybed support; a keybed arranged on the keybed support; a key height limiting column arranged in one-to-one correspondence with each key in the keybed, and an end of the key height limiting column is fixed on a bottom surface of the corresponding key; a pressure detecting device arranged in one-to-one correspondence under each key height limiting column to detect a pressing strength of each key height limiting column; and a flexible support structure arranged below each key height limiting column and located on at least one side of each pressure detection device. The keybed device includes a flexible support structure, which can avoid false triggering of the Aftertouch effect. A same depth is pressed down from the flexible support structure to obtain a consistent Aftertouch effect, and the performance effect is controllable.

Abstract: A hierarchical sparse tensor compression method based on artificial intelligence devices, in DRAM, not only saves the storage space of the neuron surface, but also adds a meta-surface to the mask block. When reading data, the mask is first read, then the size of the non-zero data is calculated, and only these non-zero data are read to save DRAM bandwidth. In the cache, only non-zero data is stored, so the required storage space is reduced. When processing data, only non-zero data is used. The method uses a bit mask to determine if the data is zero. There are three levels in the hierarchical compression scheme: tiles, lines, and points, reading bitmasks and non-zero data from DRAM, and saving bandwidth by not reading zero data. When processing data, if their bit mask is zero, the tile data may be easily removed.

Abstract: A hierarchical sparse tensor compression method based on artificial intelligence devices, in DRAM, not only saves the storage space of the neuron surface, but also adds a meta-surface to the mask block. When reading data, the mask is first read, then the size of the non-zero data is calculated, and only these non-zero data are read to save DRAM bandwidth. In the cache, only non-zero data is stored, so the required storage space is reduced. When processing data, only non-zero data is used. The method uses a bit mask to determine if the data is zero. There are three levels in the hierarchical compression scheme: tiles, lines, and points, reading bitmasks and non-zero data from DRAM, and saving bandwidth by not reading zero data. When processing data, if their bit mask is zero, the tile data may be easily removed.