Futures
Access hundreds of perpetual contracts
TradFi
Gold
One platform for global traditional assets
Options
Hot
Trade European-style vanilla options
Unified Account
Maximize your capital efficiency
Demo Trading
Introduction to Futures Trading
Learn the basics of futures trading
Futures Events
Join events to earn rewards
Demo Trading
Use virtual funds to practice risk-free trading
Launch
CandyDrop
Collect candies to earn airdrops
Launchpool
Quick staking, earn potential new tokens
HODLer Airdrop
Hold GT and get massive airdrops for free
Launchpad
Be early to the next big token project
Alpha Points
Trade on-chain assets and earn airdrops
Futures Points
Earn futures points and claim airdrop rewards
Domestic optical interconnect optical switching supernode "Optical Leap Supernode 128 Card Commercial Version" deployed
At the opening ceremony of the China Home Appliances and Consumer Electronics Expo (AWE), Shanghai Yidian (Group) Co., Ltd., Shanghai Xizhi Technology Co., Ltd., Shanghai Bianren Technology Co., Ltd., and ZTE Corporation jointly released the LightSphere128 LightJump Super Node 128 commercial version (LightSphere128). This launch marks a leap from concept validation to real-world commercial deployment in just over half a year for this China-originated optical interconnect optical switching super-node solution. It is reported that, to date, the LightJump super node has been deployed at a scale of several thousand cards. According to the introduction, measured results show that, under the same scale, when training the DeepSeekV3671B model, the training performance of the LightJump Super Node 128 commercial version is significantly improved compared with a non-super-node cluster. The model switching latency is as low as the microsecond range, and the transmission latency is reduced by more than 90% compared with traditional electrical switching. At present, this super node has been adapted to the full range of the JiePiao Xingchen (Step3.5Turbo) series models, as well as mainstream large models such as DeepSeek, Minimax, Kimi, and GLM, effectively shortening the training duration and inference latency of large models. It can provide users with domestically produced computing power solutions that offer high cost-effectiveness and high energy efficiency. The LightJump Super Node 128 commercial version has achieved long-term stable training performance. (Securities Daily)