1. ARM cortex V7-A architecture CPU A5, Up to 1.0GHz
2. Programmable memory management unit.
3. 32KB I-cache and 32KB D-cache.
4. CEVA X2: VLIW & SIMD Architecture CEVA DSP X2, Up to 400MHz.
5. CEVA X2: Supported four 16×16-bit, two 32×16-bit, or two 32×32 MAC operations.
6. CEVA X2: Two 32-bit multiply-accumulate.
7. CEVA X2: 32KB I-cache ,32KB D-cache,Can accesses DDR(up to 256MB)/ShareSRAM(up to
656KB)/SRAM.
8. CEVA X2: IEEE-754 standard for floating-point calculation support.
9. CEVA X2: All instructions support predication.
10. CEVA X2: Enhanced register file includes general registers for scalar operations and
address generation.
11. CEVA X2: Two Scalar Processing Units(SPUs)
12. CEVA X2: Two Load/Store Units(LSUs) for two independent accesses to the data memories
13. CEVA X2: Neural network algorithm acceleration engine: providing CNN, DNN, RNN, RCNN,
GRU.
14. CEVA X2: LSTM, Feed Forward,MFCC,TANH,ReLU etc.Neural network algorithm lib and api.
15. Baseline DCT. JPEG size from 48x48 pixels to 4800x4800 pixels.
16. Support AES-128/192/256 Encrypt & Decrypt in ECB/CTR/CBC/CBC-CTS mode.
17. 2CH on-chip SD Controller integrated.
18. Support boot from SPI NAND/SPI NOR/EMMC.
19. Build in multi channals 24 bit input sigma-delta ADCs, SNR>100dB, THD<-90dB.
20. Build in stereo 24 bit sigma-delta DAC, SNR>110dBA, THD<-90dB.
21. 4 on-chip UART Controllers inside.
22. Support RGB/CPU LCD display, solution up to 480*480.
Products Application:
1. Translation machine, pick up the voice signal, automatically translate the voice into the
desired language through the AI engine, and display it on the screen.
2. Voice central control, cooperate with smart home, you can control smart devices at
home when you speak.
3. Conference noise reduction box, using the microphone array to automatically identify
the direction of speech, through the noise reduction processing and send it to the
other party.
4. Smart speaker, you can listen to your favorite songs, listen to book programs, and
control smart home devices.
5. Noise-canceling microphone, using microphone array to process DSP, can realize
recording and restore human voice in noisy environment.
6. Audio mixer, using CEVA X2 DSP for spectrum calibration and spectrum tuning.
Product examples: