Huawei Ascend 310 vs Huawei Atlas 300V Pro
The Atlas 300V Pro delivers 8.8x the FP16 throughput of the Ascend 310 (70 vs 8 TFLOPS dense).
Huawei Ascend 310
Full specs →Huawei Atlas 300V Pro
Full specs →Specifications
Performance (TFLOPS)
FLOPS by Precision
What actually differs
The Atlas 300V Pro is the newer part: Da Vinci, launched in 2022, against the Ascend 310's Da Vinci from 2018. Newer architectures typically add lower-precision formats and better throughput per watt, so check the precision rows your workload actually uses.
Dense throughput for the Ascend 310 against the Atlas 300V Pro: FP16 8 vs 70 TFLOPS. Sparse figures, where the vendor publishes them, appear under each dense number in the performance table.
For LLM training and inference, weight the FP8 and FP16 rows and memory capacity most heavily. For scientific and HPC workloads, the FP64 row is the one to read.
Frequently asked questions
Is the Ascend 310 faster than the Atlas 300V Pro?
At FP16 precision the Atlas 300V Pro reaches 70 TFLOPS dense against 8 TFLOPS for the Ascend 310. The performance table on this page lists every published precision for both GPUs.
Get Comparison Updates
New GPUs added weekly. Be the first to see how they compare.