金色周刊是金色财经推出的一档每周区块链行业总结栏目，内容涵盖一周重点新闻、行情与合约数据、矿业信息、项目动态、技术进展等行业动态。

加密世界的乱象与创新乏力，让人们不禁质疑：我们是否正在实质性迈向理想的未来

Layer2生态的繁荣使以太坊在保持安全性和去中心化的同时，显著提升了处理能力和效率，推动了应用的多样化，满足更多用户的需求。

论壇破纪录获逾2,600人參與及15万人次网上浏览，促成逾350宗融资配对

货币,普通人如何对冲货币贬值风险 金色财经,“大放水后，现金肯定会贬值”

微软在提交给美国证券交易委员会的文件中表示，其全球财务团队定期评估多种投资资产，包括加密货币，以实现多样化和风险管理。

随着链抽象概念和技术的成熟，这将推动 Crypto 行业向更加用户友好、开发者友好和高度互操作的方向发展，并催生出真正能够让普通人都能使用的杀手级应用。

​随着这几个月币安接连上线多个TON代币，市场对这一生态的热情愈发高涨。

金色财经推出《金色Web3.0日报》，为您提供最新、最快的游戏、DeFi、DAO、NFT和元宇宙行业新闻。

最好的产品不需要代币，最好的代币不需要产品。L1 是一种严肃的模因币，没有猫 / 狗图片。

According to CryptoPotato, a recent study has identified a vulnerability in Apple's M-series chips that could enable hackers to retrieve the cryptographic private keys of Mac users. The vulnerability functions as a side channel, allowing the extraction of end-to-end keys when Apple chips execute implementations of commonly used cryptographic protocols. Due to its origin in the microarchitectural structure of the silicon, direct patching is not feasible, unlike traditional vulnerabilities. Instead, the report suggests a fix that relies on integrating defenses into third-party cryptographic software. However, this approach may significantly degrade the performance of M-series chips during cryptographic tasks, particularly in earlier generations like M1 and M2. The researchers explained that the exploitation of the vulnerability occurs when both the targeted cryptographic operation and a malicious application, operating with standard user system privileges, are processed on the same CPU cluster. The latest research sheds light on an overlooked phenomenon regarding DMPs within Apple silicon. In certain cases, these DMPs misinterpret memory content, including critical key material, as the pointer value used for loading other data. As a result, the DMP frequently accesses and interprets this data as an address, leading to memory access attempts. This process, known as 'dereferencing' of 'pointers,' involves reading data and inadvertently leaking it through a side channel, representing a clear breach of the constant-time paradigm. The researchers identified this hack as a 'GoFetch' exploit, explaining that it operates on the same user privileges as most third-party applications, exploiting vulnerabilities in clusters of M-series chips. It affects classical and quantum-resistant encryption algorithms alike, with extraction times varying from minutes to hours depending on the key size. Despite previous knowledge of similar threats, the researchers said that GoFetch demonstrates more aggressive behavior in Apple's chips, posing a significant security risk.

Researchers have revealed a flaw in Apple M1 chips that allows hackers to steal cryptographic keys via CPU manipulation. source: https://decrypt.co/223062/apple-chip-flaw-hackers-steal-crypto-go-fetch

最近の学術的なブレークスルーとして、研究者はアップルのMシリーズ・チップにある深刻な脆弱性を暴露した。 この欠陥は、権威ある研究機関の学者らによって発表された論文に詳述されており、攻撃者が暗号操作中に秘密鍵にアクセスすることを可能にしている。 MacBookが暗号資産ハッキングに脆弱な理由 この問題は、アップルのM1とM2チップのマイクロアーキテクチャに深く刻み込まれている。そのため、直接パッチを当てることは不可能だ。代わりに、サードパーティの暗号ソフトウェアの調整が必要となり、パフォーマンスを損なう可能性がある。 この脆弱性の核心は、これらのチップに搭載されているデータメモリ依存型プリフェッチャ（DMP）にある。この機能は、データを予測してプリロードすることで、CPUとメモリのレイテンシを最小限に抑えることを目的としている。 source: https://jp.beincrypto.com/44ki44od44ox44or44o744oe44od44kv44ow44od44kv5pqx5y36loh55sj55ux6zuj5qiz55qe/

최근 학계에서는 애플의 M 시리즈 칩에서 주로 암호화 자산의 보안에 영향을 미치는 심각한 취약점을 발견했습니다. 권위 있는 기관의 학자들이 발표한 논문에서 자세히 설명한 이 결함은 공격자가 암호화 작업 중에 비밀 키에 액세스할 수 있게 해줍니다. 맥북이 암호화 해킹에 취약한 이유 이 문제는 Apple의 M1 및 M2 칩의 마이크로아키텍처에 깊이 뿌리박혀 있습니다. 따라서 직접적인 패치는 불가능합니다. 대신 타사 암호화 소프트웨어를 조정하여 완화해야 하므로 잠재적으로 성능이 저하될 수 있습니다. 이 취약점의 핵심은 이러한 칩의 데이터 메모리 의존 프리페처(DMP)입니다. 이 기능은 데이터를 예측하고 미리 로드하여 CPU 및 메모리 지연 시간을 최소화하는 것을... source: https://kr.beincrypto.com/base-news/51206/

According to Foresight News, a newly discovered vulnerability in Apple's M-series chips could lead to wallet key leakage. The flaw is a side channel that allows end-to-end key extraction when the Apple chip runs an encrypted wallet. However, this defect cannot be directly patched as it originates from the chip's microarchitecture design. Instead, the issue can only be mitigated by implementing defensive measures in third-party encryption software, which significantly reduces the performance of the M-series, particularly in early M1 and M2 products. The vulnerability can be exploited when targeted encryption operations and malicious applications with regular user system permissions run on the same CPU cluster.

英伟达创始人黄仁勋在美国圣何塞召开 GTC 大会上发表开幕演讲。此次演讲中，英伟达发布了他们的下一代芯片架构 Blackwell。 据介绍，Blackwell GPU 以数学家 David Harold Blackwell 的名字命名，同样沿用了英伟达此前推出的 Hopper 架构。Blackwell GPU 中包含 2080 亿个晶体管，可以支持多达 10 万亿个参数的 AI 模型。除了芯片本身之外，这一架构还采用了第五代 NVLink 高速互联、第二代 Transformer 引擎，多方位全面升级。据黄仁勋介绍，这一全新的芯片将在 2024 年晚些时间上市。 黄仁勋表示，英伟达未来计划用 Blackwell 向世界各地的人工智能公司进军，与世界各地的所有 OEM、区域云、国家主权 AI、电信公司签约。 目前，亚马逊、戴尔、谷歌、Meta、微软、OpenAI、特斯拉都已经计划使用 Blackwell GPU。值得注意的是，此前传说中的 B100 并没有出现，英伟达转而发布了由一个 Grace CPU、和两组 Blackwell GPU 组合而成的超级芯片 GB200。黄仁勋介绍，相比于 H100，GB200 的算力是前者的 6 倍，而对应处理多模态特定领域的算力表现可达 30 倍。 此外，英伟达还发布了由 36 组 Grace CPU 和 72 组 Blackwell GPU 组合而成的服务器 GB200 NVL72。 此次，英伟达将推出了一款用于机器人领域、名为 GR00T 的通用基础大模型。此外，英伟达还推出了适用于机器人的全新计算机 Thor，英伟达对于相关的性能、功耗和尺寸进行了具体的优化。 英伟达还推出了全新的 AI 推理服务器 NIM（NVIDIA INFERENCE MICROSERVICE），让所有人都可以通过这一形式自定义 AI 模型和应用。（36 氪）

According to PANews, NVIDIA announced the launch of the NVIDIA Blackwell platform to empower the new era of computing. The platform enables institutions worldwide to build and run real-time generative AI on large language models (LLM) with trillions of parameters, reducing costs and energy consumption by up to 25 times compared to the previous generation. The Blackwell GPU architecture features six revolutionary acceleration technologies that will drive breakthroughs in data processing, engineering simulation, electronic design automation, computer-aided drug design, quantum computing, and generative AI. The NVIDIA GB200 Grace Blackwell superchip connects two NVIDIA B200 Tensor Core GPUs with the NVIDIA Grace CPU through a 900GB/s ultra-low-power interconnect. Compared to the NVIDIA H100 Tensor Core GPU, the GB200 superchip achieves up to 30 times performance improvement in LLM inference workloads, with cost and energy consumption reduced by up to 25 times. In addition, NVIDIA also introduced a general-purpose large model called GR00T for the robotics field during the GTC conference in San Jose, USA. NVIDIA also launched a new computer called Thor, specifically optimized for robotics in terms of performance, power consumption, and size. NVIDIA aims to give robots smarter brains through this initiative, enabling them to naturally imitate human behavior by observing it, greatly improving coordination and flexibility in their actions.

近日，蚂蚁链（Antchain）在 Web3 infra 层的两项技术成果分别入选了“数据库国际顶会 SIGMOD 2024”和“集成电路设计国际顶会 DAC 2024”。 这两项技术为“Web3 存储 LETUS”和“zk 硬件加速”，前者可为数据流转及 Web3 infra 提供高性能存储，也可以作为 Layer1 和 Layer2 的可验证存储层，大幅改善性能、可扩展性和成本。后者能应用于可信 AI 推理、Layer2 证明等技术领域，有效提升可验证计算的效率，该技术在零知识证明的核心算法部件上实现了 328 倍于单核 CPU 的性能。

암호화폐 지갑 복구 서비스 업체 키체인X(KeychainX) 최고경영자(CEO) 로버트 로딘(Robert Rhodin)이 크립토슬레이트 기고문을 통해 "인공지능(AI) 기반 모델, 컴퓨팅 하드웨어 발전, 혁신적인 알고리즘, 사이드채널 공격 등 기술은 암호화폐 복구 서비스의 기능을 고도화시키고 있다. 이러한 기술적 혁신은 유실된 암호화폐 자산을 복구할 수 있는 새로운 희망을 제시한다"고 진단했다. 이와 관련 그는 "최근 인공지능(AI) 발전을 탄생한 GPT 기반 'PASS-GPT'는 비밀번호 추측 능력이 기존 모델보다 20% 강화됐으며, GPU 및 CPU 등 하드웨어의 성능 개선도 보다 공격적인 암호화폐 컴퓨팅 및 무작위 대입을 가능하게 만들고 있다"고 설명했다. 한편 현재까지 비밀번호 분실, 하드웨어 고장, 자산 소유자의 변고 등으로 수천억 달러 규모의 디지털 자산(암호화폐)가 유실된 것으로 알려졌다.

由清华大学大学姚期智院士领衔发起，并由清华大学交叉信息研究院孵化的一家初创企业已研发成功全球首颗零知识证明 SOC 芯片，据悉该芯片拥有完全自主知识产权，基于 12nm 先进工艺制程，可进行复杂的多标量乘法（MSM）和数论变换（NTT）运算，支持可编程设计，可支持后量子密码、多方安全计算和联邦学习的加速运算，芯片性能较传统 CPU 快千倍以上，成本可降低十倍。