- Kioxia details its plans for a new SSD for AI work loads that reach 10 million IOPS
- Iops measures how quickly a storage device can handle small random requests
- The new Use XL-Flash unit, a type of SLC NAND and a new internal controller
Kioxia has presented plans for a new SSD that says that the impressive 10 million IAP could reach, a level of performance aimed directly to the demands of the systems promoted by AI.
The SSD will use XL-Flash, a single-level cell NAND type (SLC), combined with a new internal controller.
A kioxia spokesman said Techpowerup“We are taking our Flash XL Ultra Fast Memory Chips, which use single -level cells, and combining them with a completely new controller … We are pointing to more than 10 million IOP, and we plan to have samples ready for the second half of 2026”.
Difference between IOPS and GBPS
IOP, or input/output operations per second, measure the speed with which a storage device can handle small, particularly important applications in AI and server applications where quick access to small files is key.
This is different from GBPS, which refers to the real data transfer speed and is used to measure how fast you can read or write large files.
An impulse with high GBP can excel in the video edition or transfers of large files, but for automatic learning tasks where thousands of small data packages are constantly read or written, high IAPs matter more.
The Kioxia approach for next -generation storage includes not only unique projects, but also a broader effort to comply with various cases of use. Its CM9 series, which is now shown to customers, focuses on speed and reliability to match the high -end GPUs used in AI, while the LC9 series offers massive 122 TB capabilities for large databases.
Behind these products is the eighth BICS Flash generation, which presents CBA technology to increase performance and efficiency.
Kioxia is also preparing future flash memory generations using two methods. The first will add more layers for capacity, while the second combines new CMOS designs with older cellular structures to keep investment costs under control.
