Gigabyte has launched a suite of new PC components tailored for AI training at the Computex 2024 event. The lineup includes the TRX50 AI TOP and W790 AI TOP motherboards, the RTX 4070 Ti SUPER AI TOP and W7900/W7800 AI TOP graphics cards, the AI TOP 100E SSDs, and the UD1600PM PG5 AI TOP power supply.
These components are crucial for local AI data training that provides users with high-performance hardware that supports large AI models and intensive computational tasks. The TRX50 AI TOP motherboard, for example, offers optimal foundation efficiency for AI training, while the RTX 4070 Ti SUPER AI TOP graphics card unleashes full AI power with its advanced cooling and high VRAM capacity.
Also read: GIGABYTE GeForce RTX 4070 Ti Super Gaming OC 16GB Review
The TRX50 AI TOP motherboard supports up to 4 graphics cards with PCIe 5.0 x16 slots, up to 2TB DDR5 memory with 8 R-DIMM slots, and 4 SSDs in RAID 0 configuration. It also features dual 10GbE LAN with 25KV surge protection. With these kinds of specs, this motherboard provides an optimal foundation for AI training efficiency.
The W790 AI TOP is another new motherboard in the lineup and it supports up to 5 PCIe 5.0 x16 slots and up to 2TB DDR5 memory. It also includes support for up to 6 SSDs with 4 PCIe 5.0 x4 slots and 2 PCIe 4.0/3.0 x4 slots, along with 2 Thunderbolt 4 ports for 40Gbps bi-directional connection. Meanwhile, this motherboard maximises AI training possibilities with its extensive support for multiple high-performance GPUs and storage options.
Meanwhile, the Gigabyte RTX 4070 Ti SUPER AI TOP graphics card has a 2-slot width and turbofan, up to 64GB VRAM with 4 cards for AI model training, and an all-copper heatsink for heat dissipation. However, if the 64GB VRAM isn’t enough then you can go with the Gigabyte W7900 AI TOP 48G & W7800 AI TOP 32G Graphics Cards featuring 192GB VRAM with 4 cards for AI model training. Also with a 2-slot width and turbofan, and premium quality manufacturing claims, these graphics cards are designed for professional AI model training. They provide super high VRAM and superior cooling to handle intensive workloads reliably.
On the storage front, the AI TOP 100E SSDs offer sequential read/write speeds up to 7200/6500 MB/s and 150x TBW. For context and reference, “150x TBW” stands for “150 times Terabytes Written.” TBW is a measure of how much data can be written to an SSD over its lifetime before it is likely to fail. In this context, “150x TBW” indicates that the SSD is built to handle 150 times the typical amount of data writing compared to standard consumer SSDs. These SSDs are designed for the speed and durability needed in AI data processing and storage, as per claims.
The Gigabyte UD1600PM PG5 AI TOP PSU offers 80 PLUS Platinum efficiency and supports up to 4 GPU. Furthermore, it includes server-grade quality components for reliability and enhanced lifespan, which makes it potentially suitable for maintaining stable and powerful AI training setups.
The new Gigabyte AI TOP components seem to provide high-performance, scalable hardware for local AI training. It also seems to address privacy concerns and high cloud costs.
