How does the architecture of Kirin processors compare to the common solutions from other manufacturers? I'm particularly interested in efficiency in multicore performance and optimization for mobile applications. Are there general differences in cache structure or instruction set optimization that make Kirin chips stand out? How do you evaluate such approaches in the context of current mobile generations? I'd love to hear your opinions on this—maybe even include experiences with other manufacturers to get a comprehensive picture.
How do Kirin chips compare to others?
👁️ 4 views💬 2 replies❤️ 0 likes
2 Replies
Kirin-Chips from Huawei are doing their own thing when it comes to multicore efficiency – mainly thanks to the Da Vinci architecture with hybrid cores (big, medium, small cores like the Snapdragon 8 Gen 3) and its own NPU for AI tasks. The cache structure is similar to Qualcomm's (e.g., 6 MB L3 cache in the Kirin 9000), but Huawei heavily relies on branch prediction and memory bandwidth, which is noticeable in gaming and AI apps.
A major advantage is the strong AI integration: the Kirin 9000E, for example, delivers 26 TOPS of NPU performance – easily outpacing even the Snapdragon 8 Gen 2. However, Kirin faces drawbacks in 5G modernity due to US sanctions (though Huawei is working on its own solutions like the Balong 5000 chip). So, if you want maximum AI efficiency and power savings, Kirin is the way to go – otherwise, Snapdragon might be the better choice.
Hey man, I've always been curious about Kirin's architecture. Especially how it differs from Qualcomm's Snapdragon in terms of multi-core efficiency—have you ever looked into that? The cache structures and all that can be completely different, right?