GIGABYTE AI TOP ATOM

DGXOS-Spark-Gigabyte-GA1.2_OTA1_RC1
MD5: aed501b14a9145deb1b547ec9edabb9e
Use this package to restore the operating system and firmware to their original factory state.

How to proceed:
1. Download a USB Formatting Utility and back up your USB drive.
2. Download the DGX OS ISO and create a bootable USB installer.
3. Insert the USB, power on, and press Del to enter BIOS.
4. Set Boot Option #1 to UEFI: USB USB Hard Drive, Partition 1, then save and reboot.
5. Select DGX Spark Installation Options -> Install DGX OS 7.2.3 for DGX Spark.
6. Let the installation finish; system reboot may occur.
7. Your system will then start DGX OS and enter initial setup.

More details:
https://docs.nvidia.com/dgx/dgx-spark/system-recovery.html#recovery-process-steps

AI TOP ATOM SOC/EC/PD Firmware
Commands (case sensitive):
chmod +x AfulnxAarch64.lnx
sudo ./AfulnxAarch64.lnx ATOM_socfw_2.144.9.cap /CAPSULEONDISK
sudo ./AfulnxAarch64.lnx ATOM_ecfw_fused_2.75.3.cap /CAPSULEONDISK

BSP Version: 2.144.9
MD5: 3d5d7e0a488d5daa0511f56c6977c366
EC Version: 2.75.3
MD5: da052fd691ec16a595df90dcbce8a006

BIOS Release Notes:
(a) Merge 5.36_0ACUM_BSP4.1.20_023.
(b) NVME SSD read performance improvement.
(c) Systems could be bricked if PD FW updates first and powers cycle the system while SoC FW is still updating. (AMI UEFI Fix)
(d) AMI UEFI Security Patches.
(e) PD FW doesn't update after reflashing twice

GIGABYTE AI TOP Utility 4.2.0: AI TOP ATOM
1.Model List: Added Qwen-Image, Wan2.1 and Qwen-2.5-VL.
2.Model Download: Added more detailed model information and clearer error hints.
3.Inference: Text-to-Image now supports mounting LoRA models.
4.DashBoard: Now show CPU temperature and SSD usage.
5.ML: Reduced to one best sample per topic.

GIGABYTE AI TOP Utility 4.1.2: AI TOP ATOM
1.Model List: Added GLM-4.5-Air; removed Mixtral 8x7B and Gemma-3-1b-it.
2.RAG: Fixed known issues to improve stability and performance.
3.Inference: Fixed inference issues and Automatic cache clearing after model loading.

GIGABYTE AI TOP Utility 4.1.1: AI TOP ATOM
1. This Utility is only for AI TOP ATOM
2. Multimodal RAG: Supports multiple file types and integrates an Embedding Model with an LLM to help users organize and query information from their files through conversational interaction.
3. Model Download: Provides a user interface for downloading our verified models to enable all features of the software.
4. Inference: Allows users to perform text-to-text, text-to-image, text-to-video, and image-to-text inference entirely offline, ensuring privacy and data security.
5. Machine Learning: Prebuilt ML project templates let users build end-to-end machine learning workflows with minimal coding.