BTC $64,457.39 -0.74%
ETH $1,805.27 -1.34%
BNB $591.62 -0.35%
XRP $0.9778 +1.27%
SOL $74.89 +3.21%
TRX $0.3078 +0.72%
DOGE $0.0593 +0.24%
ADA $0.1535 -0.36%
BCH $208.84 -0.19%
LINK $9.23 +0.47%
HYPE $68.73 -2.14%
AAVE $111.37 -0.68%
SUI $0.7297 +0.42%
XLM $0.1405 +0.98%
ZEC $953.79 +3.85%
AAPL $350.73 -0.92%
AMZN $259.23 +0.32%
GOOGL $282.25 -0.91%
MSFT $507.43 -2.17%
META $620.11 -0.89%
NVDA $157.12 -0.39%
TSLA $382.68 +0.59%
SNDK $815.36 -0.85%
INTC $55.67 +1.27%
SPCX $142.03 +2.79%
MU $726.51 -0.48%
AMD $510.49 -0.89%
BTC $64,457.39 -0.74%
ETH $1,805.27 -1.34%
BNB $591.62 -0.35%
XRP $0.9778 +1.27%
SOL $74.89 +3.21%
TRX $0.3078 +0.72%
DOGE $0.0593 +0.24%
ADA $0.1535 -0.36%
BCH $208.84 -0.19%
LINK $9.23 +0.47%
HYPE $68.73 -2.14%
AAVE $111.37 -0.68%
SUI $0.7297 +0.42%
XLM $0.1405 +0.98%
ZEC $953.79 +3.85%
AAPL $350.73 -0.92%
AMZN $259.23 +0.32%
GOOGL $282.25 -0.91%
MSFT $507.43 -2.17%
META $620.11 -0.89%
NVDA $157.12 -0.39%
TSLA $382.68 +0.59%
SNDK $815.36 -0.85%
INTC $55.67 +1.27%
SPCX $142.03 +2.79%
MU $726.51 -0.48%
AMD $510.49 -0.89%

self-developed

All
Article
Flash

OpenAI's self-developed Git backend Pando has launched and has taken over the main code repository

After Cursor and Anthropic, OpenAI announced its self-developed Git backend Pando, which has now taken over the company's entire main code repository. Over the past year, Codex has enabled engineers to deploy a large number of Agents to write code simultaneously, resulting in an increase in code branches, pushes, merges, and pull requests. OpenAI's original system experienced frequent stalls and merge failures, with reliability even dropping below 90%.One engineer is responsible for the core architecture design, and multiple Codex Agent teams participated in the implementation, with Pando going into production in 10 weeks. This backend allows code pushes from different branches to be processed in parallel without queuing each other. When pulling code, data packets are no longer generated separately for each client but instead reuse existing Git data packets, distributing them directly through caching. After the transformation, the median latency of code pushes was reduced by nearly 80%, distributing nearly 1 PB of data daily, with a cache hit rate exceeding 99%, but different branches no longer guarantee to present the same state at the same moment.Cursor's Origin is still based on Git, primarily aimed at replacing GitHub as a code hosting service. Anthropic's Scion, on the other hand, has abandoned Git, placing commits, merges, and historical queries on the server side. OpenAI retains the Git client and protocol, only reconstructing the backend, so engineers and Agents do not need to change their original development processes significantly.

first_img Pump.fun Co-founder Alon: Self-developed cross-chain bridge has been launched, purchase error rate has decreased by 92%

The co-founder and Chief Operating Officer Alon of the meme coin issuance platform Pump.fun stated that the team had previously focused on launch pads for a long time and only began supporting cross-chain exchanges about 5 months ago. After investigating the cross-chain service provider Relay used at that time, he noted that over one-ninth of cross-chain purchases on Relay encountered errors, with typical purchase landing times close to 4 seconds, and one-tenth exceeding 7 seconds. The number of tokens users received was usually less than expected. He also mentioned that the social trading application FOMO still relies on Relay.Alon stated that the self-developed cross-chain bridge Pump Bridge has been opened to all Pump.fun application users. Compared to Relay, the purchase landing speed has improved by 2 times, and the error rate has decreased by 92%, dropping from over one-ninth to less than one percent. Users can obtain more tokens on average with the same amount of funds, and currently, no additional cross-chain fees are charged. He mentioned that due to the limitations of light speed and target chain confirmation time, the theoretical fastest landing time for purchases is about 1.1 seconds, and the team is still working to reduce the remaining gap.The above comparison comes from random tests conducted in the Pump.fun mobile application from September 28 to October 7, covering Robinhood, BSC, Base, Arc, and Ink, excluding Ethereum, with a total of 416,669 transaction results recorded. Among them, 383,326 transactions recorded the time taken from click to successful prompt, and an additional 35,333 transactions of completed USDC purchases under $500 were used for comparison. Alon also stated that the platform fee rate is 0.45%, with Callout rewards of at least $300,000 daily, totaling about $20 million over two months, and the platform does not retain platform fees.

first_img Amazon launched the open-source tool TorchNeuron, lowering the threshold for transitioning to self-developed chips

Financial Associated Press, October 20 - Amazon launched the open-source tool TorchNeuron on the 19th local time, aiming to lower the barrier for developers to transition from NVIDIA chips to its self-developed chips, challenging NVIDIA's software barrier built on CUDA. According to reports, this tool allows AI code written in PyTorch to run on Amazon's self-developed AI chips without the need for rewriting. PyTorch is maintained primarily by Meta and is described as an open-source framework used by most AI developers.Amazon's chip design department, Annapurna Labs, introduced that with TorchNeuron, developers only need to modify one configuration to run the same code on the self-developed Trainium chip. Amazon announced the launch of this tool on the 19th local time and plans to publicly release it in a free open-source format at the PyTorch conference in San Jose, California on the 20th.Colin Brace, Vice President of Software at Annapurna Labs, stated in an interview that the launch of TorchNeuron is intended to reduce the steps required for developers to try Amazon's self-developed chips, starting from the customers' existing workflows. He revealed that over 30 early customers are already using it, covering fields such as robotics, world models, and voice AI, and mentioned that switching to Amazon chips can save costs.

first_img Microsoft launched its self-developed AI chip Maia 300, with the key small chip using TSMC's 2-nanometer technology

According to a report by Central News Agency citing Reuters, Microsoft announced the launch of its third-generation self-developed AI chip, Maia 300, with plans to deploy over 1 million units next year to support internal AI services. The small chip responsible for key computing functions uses TSMC's 2-nanometer process. The Maia 300 can handle code writing and perform complex business tasks on behalf of users. The previous generation of chips has been used to run Microsoft's internal AI models and the OpenAI models provided through Microsoft Azure cloud computing services.Microsoft is taking a different technological approach compared to NVIDIA. The NVIDIA AI accelerator uses a large computing chip surrounded by high-bandwidth memory chips; the Maia 300 core processing unit is composed of small chips, with the small chips responsible for key computing functions using TSMC's 2-nanometer process, while other small chips use the lower-cost 3-nanometer process to handle data transmission out of the processor and transfer data to other locations within the system.The Maia 300 will be installed in cabinets, with each cabinet accommodating 96 chips and consuming about 167 kW of power. In contrast, NVIDIA's latest GB300 AI server cabinet is configured with 72 GPUs and consumes about 142 kW of power. Microsoft stated that it plans to deploy the Maia 300 on a large scale starting next year, reaching several gigawatts. Microsoft previously indicated that it expects its capital expenditures to exceed $175 billion this year.

first_img Thomson Reuters launched its self-developed AI model Thomson-1, based on Alibaba's open-source Qwen

According to Business Insider, Thomson Reuters launched its first self-developed AI model, Thomson-1, on Monday. This model is based on Snowdon and constructed through "re-alignment" of Alibaba's open-source Qwen model. Open-source means anyone can download and modify the model for free. Thomson-1 will take over some tasks previously handled by Anthropic's Claude, but it is not intended to completely replace collaboration with Anthropic and other labs, initially focusing on the company's areas of expertise, starting with document review.This move aims to address the high AI costs brought by models like Claude and OpenAI Codex, and reflects the use of cheaper Chinese open-source AI. CTO Joel Hron stated that the main reason is to better leverage Thomson Reuters' own expertise and control costs. The company expanded its collaboration with Anthropic in May this year for the AI legal assistant CoCounsel, which still primarily relies on Claude. Hron said, "Our main goal is to gradually make Thomson the model that drives more and more capabilities for CoCounsel."Thomson Reuters, in collaboration with a team from Imperial College London, spent months transforming Qwen to build Snowdon and ensured it is "ethically and politically bias-processed and safe to use." Hron pointed out that having a proprietary model allows for development based on its own intellectual property rather than continuously paying external AI companies, comparing it to renting versus buying a house: renting provides shelter but does not accumulate long-term equity.
app_icon
ChainCatcher Building the Web3 world with innovations.