{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-07-28T06:20:51.496Z","headline":"AMD 携手 Cerebras 押注低时延 AI 推理：每瓦每秒 Token 吞吐提升 5 倍","description":"AMD 在 Advancing AI 2026 活动中宣布与 Cerebras 合作，联合开发针对超低延迟 AI 推理的解决方案。该方案整合 AMD 的 Helios 机架与 Cerebras 的 Wafer-Scale Engine，分别处理提示词与 Token 生成环节，预估可将每瓦每秒 token 数量提高 5 倍。","url":"https://www.aioga.com/news/cmryitqwt01k1rolgn6566and/","mainEntityOfPage":"https://www.aioga.com/news/cmryitqwt01k1rolgn6566and/","datePublished":"2026-07-24T05:49:28.000Z","dateModified":"2026-07-24T05:49:28.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://www.ithome.com/0/981/112.htm","https://aihot.virxact.com/items/cmryitqwt01k1rolgn6566and"],"canonicalUrl":"https://www.aioga.com/news/cmryitqwt01k1rolgn6566and/","directAnswer":{"@type":"Answer","text":"Aioga 编辑摘要：AMD 在 Advancing AI 2026 活动中宣布与 Cerebras 合作，联合开发针对超低延迟 AI 推理的解决方案。 Aioga 将其归入「行业动态」方向，重点关注它对真实使用和行业竞争的影响。","url":"https://www.aioga.com/news/cmryitqwt01k1rolgn6566and/","dateCreated":"2026-07-24T05:49:28.000Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"IT之家 source article","url":"https://www.ithome.com/0/981/112.htm","datePublished":"2026-07-24T05:49:28.000Z","provider":{"@type":"Organization","name":"IT之家","url":"https://www.ithome.com/0/981/112.htm"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmryitqwt01k1rolgn6566and","datePublished":"2026-07-24T05:49:28.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmryitqwt01k1rolgn6566and"}}],"aggregationSource":"IT之家（RSS）","originalPublisher":{"name":"IT之家","url":"https://www.ithome.com/0/981/112.htm"},"article":{"id":"cmryitqwt01k1rolgn6566and","slug":"cmryitqwt01k1rolgn6566and","url":"https://www.aioga.com/news/cmryitqwt01k1rolgn6566and/","title":"AMD 携手 Cerebras 押注低时延 AI 推理：每瓦每秒 Token 吞吐提升 5 倍","title_en":"","summary":"AMD 在 Advancing AI 2026 活动中宣布与 Cerebras 合作，联合开发针对超低延迟 AI 推理的解决方案。该方案整合 AMD 的 Helios 机架与 Cerebras 的 Wafer-Scale Engine，分别处理提示词与 Token 生成环节，预估可将每瓦每秒 token 数量提高 5 倍。","source":"IT之家（RSS）","sourceUrl":"https://www.ithome.com/0/981/112.htm","aiHotUrl":"https://aihot.virxact.com/items/cmryitqwt01k1rolgn6566and","publishedAt":"2026-07-24T05:49:28.000Z","category":"行业动态","score":45,"selected":false,"articleBody":["IT之家：https://www.ithome.com/ 7 月 24 日消息，科技媒体 Wccftech 今天（7 月 24 日）发布博文，报道称在 Advancing AI 2026 活动中， AMD 宣布携手 Cerebras 公司，联合开发 AI 推理解决方案。","根据官方新闻稿，该方案主要针对超低延迟 AI 推理场景，将整合 AMD 的 Helios 机架级解决方案与 Cerebras 的晶圆级引擎，从而应对英伟达和 Groq 合作。","按照双方披露的路线，Helios 机架承担高性能、可扩展吞吐引擎角色，主要处理提示词与大上下文窗口。","Cerebras 的 Wafer-Scale Engine 则负责对内存带宽要求更高的 Token 生成与解码环节，强调超低时延表现， 结合使用 2 个计算引擎，预估可以将每瓦每秒 token 数量提高 5 倍。 IT之家附上相关截图如下：","Cerebras 介绍称，Wafer-Scale Engine 在单块互连硅片上无缝连接数十万个计算核心和数十 GB 的 SRAM，减少外部网络瓶颈对数据传输的影响。"],"articleImages":[{"sourceUrl":"https://img.ithome.com/newsuploadfiles/2026/7/d898dce1-a643-41e8-9f34-31dfb785ae2b.jpg?x-bce-process=image/format,f_auto","alt":"","afterParagraph":0,"url":"/media/articles/cmryitqwt01k1rolgn6566and/6c09a54e15314227.jpg"},{"sourceUrl":"https://img.ithome.com/newsuploadfiles/2026/7/559774b7-fb1d-4bdf-9d8f-1c94d6d70632.jpg?x-bce-process=image/format,f_auto","alt":"","afterParagraph":1,"url":"/media/articles/cmryitqwt01k1rolgn6566and/f474bb280235ee5d.jpg"},{"sourceUrl":"https://img.ithome.com/newsuploadfiles/2026/7/eccc9494-34a3-4cd9-b9b5-5f0c50c2b47f.jpg?x-bce-process=image/format,f_auto","alt":"","afterParagraph":3,"url":"/media/articles/cmryitqwt01k1rolgn6566and/ce50e8f810f2c4ae.jpg"},{"sourceUrl":"https://img.ithome.com/newsuploadfiles/2026/7/cc070fc0-6ee7-4a80-b7b4-579bd9777f17.jpg?x-bce-process=image/format,f_auto","alt":"","afterParagraph":3,"url":"/media/articles/cmryitqwt01k1rolgn6566and/2d5070195e011a00.jpg"}],"mediaStatus":"ok","articleBodyZh":[],"translationStatus":"","bodyOrigin":"source-page","editorial":{"summary":"Aioga 编辑摘要：AMD 在 Advancing AI 2026 活动中宣布与 Cerebras 合作，联合开发针对超低延迟 AI 推理的解决方案。 Aioga 将其归入「行业动态」方向，重点关注它对真实使用和行业竞争的影响。","background":"背景分析：公司与行业类动态需要放在竞争格局、商业化路径、资本信号和监管环境中观察，单条公告不能代表最终结果。","viewpoint":"Aioga 判断：这条动态更适合作为行业观察信号，当前信息足以建立线索，但不足以推导长期结论。","implications":"影响分析：对相关团队而言，短期应先核对来源、可用范围和实际成本，再判断是否值得接入或跟进。","nextStep":"后续观察：继续观察官方文件、合作落地、收入或用户信号、竞品动作和监管后续。","evidenceRefs":["title","summary","articleBody"],"confidence":"medium","status":"published","aiGenerated":false,"autoApproved":true,"generatedBy":"rule-safe-fallback","generatedAt":"2026-07-28T06:29:10.555Z","sourceHash":"7bc96b5e92074af0","validation":{"passed":true,"mode":"rule-safe-fallback","checks":["schema","length","source-attribution","no-html"]}},"tags":["行业动态","IT之家（RSS）"],"translations":{"zh-CN":{"title":"AMD 携手 Cerebras 押注低时延 AI 推理：每瓦每秒 Token 吞吐提升 5 倍","summary":"AMD 在 Advancing AI 2026 活动中宣布与 Cerebras 合作，联合开发针对超低延迟 AI 推理的解决方案。该方案整合 AMD 的 Helios 机架与 Cerebras 的 Wafer-Scale Engine，分别处理提示词与 Token 生成环节，预估可将每瓦每秒 token 数量提高 5 倍。","category":"行业动态","source":"IT之家","aggregationSource":"IT之家（RSS）","pageTitle":"AMD 携手 Cerebras 押注低时延 AI 推理：每瓦每秒 Token 吞吐提升 5 倍 - Aioga AI资讯","description":"AMD 在 Advancing AI 2026 活动中宣布与 Cerebras 合作，联合开发针对超低延迟 AI 推理的解决方案。该方案整合 AMD 的 Helios 机架与 Cerebras 的 Wafer-Scale Engine，分别处理提示词与 Token 生成环节，预估可将每瓦每秒 token 数量提高 5 倍。","url":"https://www.aioga.com/news/cmryitqwt01k1rolgn6566and/"},"en":{"title":"AMD Teams Up with Cerebras to Bet on Low-Latency AI Inference: Token Throughput per Watt per Second Increased 5-Fold","summary":"AMD announced a collaboration with Cerebras at the Advancing AI 2026 event to jointly develop solutions for ultra-low latency AI inference. The solution integrates AMD's Helios rack with Cerebras' Wafer-Scale Engine, handling the prompt and token generation stages respectively, and is estimated to increase tokens per second per watt by five times.","category":"Industry","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD Teams Up with Cerebras to Bet on Low-Latency AI Inference: Token Throughput per Watt per Second Increased 5-Fold - Aioga AI News","description":"AMD announced a collaboration with Cerebras at the Advancing AI 2026 event to jointly develop solutions for ultra-low latency AI inference. The solution integrates AMD's Helios rac...","url":"https://www.aioga.com/en/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:22:29.604Z"},"ja":{"title":"AMD、Cerebrasと提携し低遅延AI推論に賭ける：ワットあたり毎秒トークンのスループットが5倍に向上","summary":"AMDは、Advancing AI 2026イベントでCerebrasとの提携を発表し、超低遅延AI推論向けのソリューションを共同開発することを明らかにしました。このソリューションは、AMDのHeliosラックとCerebrasのWafer-Scale Engineを統合し、それぞれプロンプトとトークン生成の段階を処理し、ワット当たりの毎秒トークン数を5倍に増加させると見込まれています。","category":"業界動向","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD、Cerebrasと提携し低遅延AI推論に賭ける：ワットあたり毎秒トークンのスループットが5倍に向上 - Aioga AIニュース","description":"AMDは、Advancing AI 2026イベントでCerebrasとの提携を発表し、超低遅延AI推論向けのソリューションを共同開発することを明らかにしました。このソリューションは、AMDのHeliosラックとCerebrasのWafer-Scale Engineを統合し、それぞれプロンプトとトークン生成の段階を処理し、ワット当たりの毎秒トークン数を5倍に...","url":"https://www.aioga.com/ja/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:22:41.135Z"},"ko":{"title":"AMD, Cerebras와 손잡고 저지연 AI 추론에 도전: 와트당 초당 토큰 처리량 5배 향상","summary":"AMD는 Advancing AI 2026 행사에서 Cerebras와의 협력을 발표하고, 초저지연 AI 추론을 위한 솔루션을 공동 개발한다고 밝혔습니다. 이 솔루션은 AMD의 Helios 랙과 Cerebras의 Wafer-Scale Engine을 통합하여 각각 프롬프트와 토큰 생성 단계를 처리하며, 예상되는 전력당 초당 토큰 수를 5배 증가시킬 수 있다고 합니다.","category":"업계 동향","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD, Cerebras와 손잡고 저지연 AI 추론에 도전: 와트당 초당 토큰 처리량 5배 향상 - Aioga AI 뉴스","description":"AMD는 Advancing AI 2026 행사에서 Cerebras와의 협력을 발표하고, 초저지연 AI 추론을 위한 솔루션을 공동 개발한다고 밝혔습니다. 이 솔루션은 AMD의 Helios 랙과 Cerebras의 Wafer-Scale Engine을 통합하여 각각 프롬프트와 토큰 생성 단계를 처리하며, 예상되는 전력당 초당 토...","url":"https://www.aioga.com/ko/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:23:34.079Z"},"es":{"title":"AMD se une a Cerebras para apostar por la inferencia de IA de baja latencia: el rendimiento de tokens por vatio por segundo aumenta 5 veces","summary":"AMD anunció en el evento Advancing AI 2026 su colaboración con Cerebras para desarrollar conjuntamente soluciones de inferencia de IA de ultra baja latencia. La solución integra el rack Helios de AMD y el Wafer-Scale Engine de Cerebras, manejando respectivamente la etapa de prompts y la generación de tokens, estimando un aumento de hasta 5 veces en la cantidad de tokens por vatio por segundo.","category":"Industria","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD se une a Cerebras para apostar por la inferencia de IA de baja latencia: el rendimiento de tokens por vatio por segundo aumenta 5 veces - Aioga Noticias de IA","description":"AMD anunció en el evento Advancing AI 2026 su colaboración con Cerebras para desarrollar conjuntamente soluciones de inferencia de IA de ultra baja latencia. La solución integra el...","url":"https://www.aioga.com/es/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:23:22.915Z"},"fr":{"title":"AMD s'associe à Cerebras pour miser sur l'inférence IA à faible latence : le débit de tokens par seconde et par watt est multiplié par 5","summary":"Lors de l'événement Advancing AI 2026, AMD a annoncé sa collaboration avec Cerebras pour développer conjointement des solutions pour l'inférence AI ultra-faible latence. Cette solution intègre le rack Helios d'AMD et le Wafer-Scale Engine de Cerebras, prenant respectivement en charge les étapes de l'invite et de la génération de tokens, et devrait permettre de multiplier par 5 le nombre de tokens par watt et par seconde.","category":"Industrie","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD s'associe à Cerebras pour miser sur l'inférence IA à faible latence : le débit de tokens par seconde et par watt est multiplié par 5 - Aioga Actualités IA","description":"Lors de l'événement Advancing AI 2026, AMD a annoncé sa collaboration avec Cerebras pour développer conjointement des solutions pour l'inférence AI ultra-faible latence. Cette solu...","url":"https://www.aioga.com/fr/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:24:21.334Z"},"de":{"title":"AMD geht eine Partnerschaft mit Cerebras ein, um auf latenzarme AI-Inferenz zu setzen: Durchsatz von Tokens pro Watt und Sekunde steigt um das Fünffache","summary":"AMD kündigte auf der Advancing AI 2026-Veranstaltung eine Zusammenarbeit mit Cerebras an, um gemeinsam eine Lösung für ultraschnelle KI-Inferenz zu entwickeln. Die Lösung integriert AMDs Helios-Rack mit Cerebras' Wafer-Scale Engine, wobei jeweils die Verarbeitung von Prompts und die Token-Generierung übernommen werden. Es wird geschätzt, dass die Anzahl der Tokens pro Watt pro Sekunde um das Fünffache erhöht werden kann.","category":"行业动态","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD geht eine Partnerschaft mit Cerebras ein, um auf latenzarme AI-Inferenz zu setzen: Durchsatz von Tokens pro Watt und Sekunde steigt um das Fünffache - Aioga KI-News","description":"AMD kündigte auf der Advancing AI 2026-Veranstaltung eine Zusammenarbeit mit Cerebras an, um gemeinsam eine Lösung für ultraschnelle KI-Inferenz zu entwickeln. Die Lösung integrier...","url":"https://www.aioga.com/de/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:24:20.019Z"},"pt-BR":{"title":"AMD se une à Cerebras para apostar em inferência de IA de baixa latência: aumento de 5 vezes na taxa de tokens por watt por segundo","summary":"A AMD anunciou na conferência Advancing AI 2026 uma parceria com a Cerebras para desenvolver conjuntamente soluções de inferência de IA com latência ultra baixa. A solução integra o rack Helios da AMD com o Wafer-Scale Engine da Cerebras, lidando respectivamente com os estágios de prompt e geração de tokens, estimando-se um aumento de 5 vezes na quantidade de tokens por watt por segundo.","category":"行业动态","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD se une à Cerebras para apostar em inferência de IA de baixa latência: aumento de 5 vezes na taxa de tokens por watt por segundo - Aioga Notícias de IA","description":"A AMD anunciou na conferência Advancing AI 2026 uma parceria com a Cerebras para desenvolver conjuntamente soluções de inferência de IA com latência ultra baixa. A solução integra...","url":"https://www.aioga.com/pt-BR/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:25:05.343Z"},"ru":{"title":"AMD совместно с Cerebras делает ставку на AI-инференс с низкой задержкой: пропускная способность токенов на ватт в секунду увеличена в 5 раз","summary":"AMD на мероприятии Advancing AI 2026 объявила о сотрудничестве с компанией Cerebras для совместной разработки решений для сверхнизкой задержки AI-инференса. Решение объединяет стоечную систему Helios от AMD и Wafer-Scale Engine от Cerebras, обрабатывающие соответственно этапы подсказок и генерации токенов, что, как ожидается, может повысить количество токенов в секунду на ватт в 5 раз.","category":"行业动态","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD совместно с Cerebras делает ставку на AI-инференс с низкой задержкой: пропускная способность токенов на ватт в секунду увеличена в 5 раз - Aioga Новости ИИ","description":"AMD на мероприятии Advancing AI 2026 объявила о сотрудничестве с компанией Cerebras для совместной разработки решений для сверхнизкой задержки AI-инференса. Решение объединяет стое...","url":"https://www.aioga.com/ru/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:25:10.969Z"},"ar":{"title":"AMD تتعاون مع Cerebras للمراهنة على الاستدلال بالذكاء الاصطناعي منخفض الكمون: زيادة معدل معالجة الرموز لكل واط لكل ثانية بمقدار 5 أضعاف","summary":"أعلنت شركة AMD في حدث Advancing AI 2026 عن التعاون مع شركة Cerebras لتطوير حلول مشتركة للاستدلال بالذكاء الاصطناعي منخفض التأخير للغاية. يدمج هذا الحل رف Helios من AMD مع محرك Wafer-Scale من Cerebras، حيث يتعامل كل منهما مع مراحل معالجة الكلمات المفتاحية وتوليد الرموز، ومن المتوقع أن يزيد عدد الرموز لكل واط في الثانية بمقدار 5 أضعاف.","category":"行业动态","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD تتعاون مع Cerebras للمراهنة على الاستدلال بالذكاء الاصطناعي منخفض الكمون: زيادة معدل معالجة الرموز لكل واط لكل ثانية بمقدار 5 أضعاف - Aioga أخبار الذكاء الاصطناعي","description":"أعلنت شركة AMD في حدث Advancing AI 2026 عن التعاون مع شركة Cerebras لتطوير حلول مشتركة للاستدلال بالذكاء الاصطناعي منخفض التأخير للغاية. يدمج هذا الحل رف Helios من AMD مع محرك Wafe...","url":"https://www.aioga.com/ar/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:25:58.625Z"},"hi":{"title":"AMD ने Cerebras के साथ मिलकर कम विलंबता वाले AI अनुमान पर दांव लगाया: प्रति वाट प्रति सेकंड टोकन थ्रूपुट 5 गुना बढ़ा","summary":"AMD ने Advancing AI 2026 कार्यक्रम में Cerebras के साथ साझेदारी की घोषणा की, जो अल्ट्रा-लो लेटेंसी AI इनफेरेंस के लिए एक समाधान विकसित करेगा। यह समाधान AMD के Helios रैक और Cerebras के Wafer-Scale Engine को एकीकृत करता है, जो क्रमशः प्रॉम्प्ट और टोकन जनरेशन चरणों को संभालते हैं, और अनुमान है कि यह प्रति वाट प्रति सेकंड टोकन की संख्या को 5 गुना बढ़ा सकता है।","category":"行业动态","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD ने Cerebras के साथ मिलकर कम विलंबता वाले AI अनुमान पर दांव लगाया: प्रति वाट प्रति सेकंड टोकन थ्रूपुट 5 गुना बढ़ा - Aioga AI समाचार","description":"AMD ने Advancing AI 2026 कार्यक्रम में Cerebras के साथ साझेदारी की घोषणा की, जो अल्ट्रा-लो लेटेंसी AI इनफेरेंस के लिए एक समाधान विकसित करेगा। यह समाधान AMD के Helios रैक और Cerebra...","url":"https://www.aioga.com/hi/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:26:04.579Z"},"it":{"title":"AMD collabora con Cerebras puntando su inferenza AI a bassa latenza: rendimento di token per watt al secondo aumentato di 5 volte","summary":"AMD ha annunciato durante l'evento Advancing AI 2026 una collaborazione con Cerebras per sviluppare congiuntamente soluzioni di inferenza AI a latenza ultra-bassa. La soluzione integra il rack Helios di AMD con il Wafer-Scale Engine di Cerebras, gestendo rispettivamente le fasi di prompt e generazione dei token, e si prevede che possa aumentare di 5 volte il numero di token per watt al secondo.","category":"行业动态","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD collabora con Cerebras puntando su inferenza AI a bassa latenza: rendimento di token per watt al secondo aumentato di 5 volte - Aioga Notizie IA","description":"AMD ha annunciato durante l'evento Advancing AI 2026 una collaborazione con Cerebras per sviluppare congiuntamente soluzioni di inferenza AI a latenza ultra-bassa. La soluzione int...","url":"https://www.aioga.com/it/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:26:46.936Z"},"nl":{"title":"AMD werkt samen met Cerebras om in te zetten op AI-inferentie met lage latentie: tokenverwerking per watt per seconde verhoogd met 5 keer","summary":"AMD kondigde tijdens het Advancing AI 2026-evenement een samenwerking met Cerebras aan om gezamenlijk oplossingen voor AI-inferentie met ultralage latency te ontwikkelen. Deze oplossing integreert AMD's Helios-rack met Cerebras' Wafer-Scale Engine, waarbij respectievelijk promptverwerking en token-generatie worden afgehandeld, en naar verwachting het aantal tokens per seconde per watt met vijfvoud kan verhogen.","category":"行业动态","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD werkt samen met Cerebras om in te zetten op AI-inferentie met lage latentie: tokenverwerking per watt per seconde verhoogd met 5 keer - Aioga AI-nieuws","description":"AMD kondigde tijdens het Advancing AI 2026-evenement een samenwerking met Cerebras aan om gezamenlijk oplossingen voor AI-inferentie met ultralage latency te ontwikkelen. Deze oplo...","url":"https://www.aioga.com/nl/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:26:50.756Z"},"tr":{"title":"AMD, Cerebras ile iş birliği yaparak düşük gecikmeli AI çıkarımına yatırım yapıyor: Watt başına saniyede Token işleme kapasitesi 5 kat artıyor","summary":"AMD, Advancing AI 2026 etkinliğinde Cerebras ile iş birliği yaparak ultra düşük gecikmeli AI çıkarımına yönelik çözümler geliştireceğini duyurdu. Bu çözüm, AMD'nin Helios rafını ve Cerebras'ın Wafer-Scale Engine'ini entegre ediyor ve sırasıyla istem ve token üretim aşamalarını işliyor; her watt başına saniye token sayısını 5 kat artırması öngörülüyor.","category":"行业动态","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD, Cerebras ile iş birliği yaparak düşük gecikmeli AI çıkarımına yatırım yapıyor: Watt başına saniyede Token işleme kapasitesi 5 kat artıyor - Aioga AI Haberleri","description":"AMD, Advancing AI 2026 etkinliğinde Cerebras ile iş birliği yaparak ultra düşük gecikmeli AI çıkarımına yönelik çözümler geliştireceğini duyurdu. Bu çözüm, AMD'nin Helios rafını ve...","url":"https://www.aioga.com/tr/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:27:35.766Z"},"vi":{"title":"AMD hợp tác với Cerebras đặt cược vào AI suy luận độ trễ thấp: Thông lượng Token mỗi watt mỗi giây tăng gấp 5 lần","summary":"AMD đã công bố hợp tác với Cerebras tại sự kiện Advancing AI 2026 để cùng phát triển giải pháp suy luận AI độ trễ cực thấp. Giải pháp này tích hợp giá đỡ Helios của AMD và Bộ xử lý Wafer-Scale của Cerebras, lần lượt xử lý các bước gợi ý và tạo Token, dự kiến có thể tăng số lượng token trên mỗi watt mỗi giây lên gấp 5 lần.","category":"行业动态","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD hợp tác với Cerebras đặt cược vào AI suy luận độ trễ thấp: Thông lượng Token mỗi watt mỗi giây tăng gấp 5 lần - Tin tức AI Aioga","description":"AMD đã công bố hợp tác với Cerebras tại sự kiện Advancing AI 2026 để cùng phát triển giải pháp suy luận AI độ trễ cực thấp. Giải pháp này tích hợp giá đỡ Helios của AMD và Bộ xử lý...","url":"https://www.aioga.com/vi/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:27:35.522Z"},"id":{"title":"AMD bekerja sama dengan Cerebras bertaruh pada inferensi AI latensi rendah: throughput token per watt per detik meningkat 5 kali lipat","summary":"AMD mengumumkan kerja sama dengan Cerebras di acara Advancing AI 2026 untuk mengembangkan solusi inferensi AI dengan latensi sangat rendah secara bersama-sama. Solusi ini mengintegrasikan rak Helios milik AMD dengan Wafer-Scale Engine milik Cerebras, masing-masing menangani tahap prompt dan generasi token, dan diperkirakan dapat meningkatkan jumlah token per detik per watt hingga 5 kali lipat.","category":"行业动态","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD bekerja sama dengan Cerebras bertaruh pada inferensi AI latensi rendah: throughput token per watt per detik meningkat 5 kali lipat - Berita AI Aioga","description":"AMD mengumumkan kerja sama dengan Cerebras di acara Advancing AI 2026 untuk mengembangkan solusi inferensi AI dengan latensi sangat rendah secara bersama-sama. Solusi ini menginteg...","url":"https://www.aioga.com/id/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:28:16.973Z"},"th":{"title":"AMD จับมือ Cerebras ลงเดิมพัน AI การให้เหตุผลความหน่วงต่ำ: การประมวลผล Token ต่อวัตต์ต่อวินาทีเพิ่มขึ้น 5 เท่า","summary":"AMD ประกาศความร่วมมือกับ Cerebras ในงาน Advancing AI 2026 เพื่อพัฒนาวิธีแก้ปัญหาด้าน AI inference ที่มีความหน่วงต่ำอย่างยิ่ง โซลูชันนี้รวมเครื่อง Helios ของ AMD กับ Wafer-Scale Engine ของ Cerebras ซึ่งจัดการกับขั้นตอนของ prompt และการสร้าง Token ตามลำดับ คาดว่าสามารถเพิ่มจำนวน token ต่อวัตต์ต่อวินาทีได้ 5 เท่า","category":"行业动态","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD จับมือ Cerebras ลงเดิมพัน AI การให้เหตุผลความหน่วงต่ำ: การประมวลผล Token ต่อวัตต์ต่อวินาทีเพิ่มขึ้น 5 เท่า - ข่าว AI Aioga","description":"AMD ประกาศความร่วมมือกับ Cerebras ในงาน Advancing AI 2026 เพื่อพัฒนาวิธีแก้ปัญหาด้าน AI inference ที่มีความหน่วงต่ำอย่างยิ่ง โซลูชันนี้รวมเครื่อง Helios ของ AMD กับ Wafer-Scale Eng...","url":"https://www.aioga.com/th/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:28:25.404Z"},"pl":{"title":"AMD współpracuje z Cerebras, stawiając na niskolatencyjne wnioskowanie AI: przepustowość tokenów na wat na sekundę zwiększona 5-krotnie","summary":"AMD ogłosiło na wydarzeniu Advancing AI 2026 współpracę z Cerebras w celu wspólnego opracowania rozwiązań do ultra-niskolatencyjnego wnioskowania AI. Rozwiązanie łączy stojak Helios firmy AMD z Wafer-Scale Engine firmy Cerebras, obsługując odpowiednio etapy podpowiedzi i generowania tokenów, co według szacunków może zwiększyć liczbę tokenów na wat na sekundę pięciokrotnie.","category":"行业动态","source":"IT之家（RSS）","aggregationSource":"IT之家（RSS）","pageTitle":"AMD współpracuje z Cerebras, stawiając na niskolatencyjne wnioskowanie AI: przepustowość tokenów na wat na sekundę zwiększona 5-krotnie - Aioga Wiadomości AI","description":"AMD ogłosiło na wydarzeniu Advancing AI 2026 współpracę z Cerebras w celu wspólnego opracowania rozwiązań do ultra-niskolatencyjnego wnioskowania AI. Rozwiązanie łączy stojak Helio...","url":"https://www.aioga.com/pl/news/cmryitqwt01k1rolgn6566and/","contentTranslated":true,"sourceHash":"7d7cf65ce111eb63","translatedAt":"2026-07-27T06:29:11.628Z"}}}}