Citat:
Ursprungligen postat av
raskens
Dojo ligger långt efter marknadsledande waferscale utvecklaren Cerebras. Men som inte blåser upp förväntningarna på hittepåsaker som medvetna Optimusrobotar som ska handla matvaror i affären och skruva i skruvar som genomsnittshushållet ska ha råd med. Ja dom skulle ju ha FSD klar för 5år sen nu. Nån gån borde korthuset falla ihop.
Appropå den hypeade Cybertrucken. En billig VW spöade den visst
https://www.instagram.com/reel/C0pRdT5vk24/?igshid=MTRhZmU1ODE2NA==
men det är ju klart den är ju säker mot enklare handeldvapen även om man nu bör ducka om nån riktare en AR-15 med kl 5.56 mot föraren. Ett av de populäraste vapnen i USA. Kör man med kal. 7.62 så har föraren noll chans.
Om nu Muskoten vill marknadsföra polygonbilen som skottsäker så borde han konsumentupplysa vilka kalibrar bilen är säker mot.
Som vanligt har du inte fattat poängen med saker och ting. Men hat gör ju folk blinda.
**Cerebras**
**Advantages:**
* **High performance:** Cerebras's wafer-scale architecture allows it to achieve extremely high performance, making it ideal for training large deep learning models.
* **Low latency:** Cerebras's low latency allows it to process data quickly and efficiently, which is important for real-time applications.
* **Simplicity:** Cerebras's simple architecture makes it easier to program and deploy, which can save time and money.
* **Small form factor:** Cerebras's small form factor makes it easier to deploy and manage, which can be important for organizations with limited space.
**Disadvantages:**
* **Limited scalability:** Cerebras's architecture is not as scalable as Dojo's, which can limit its ability to meet the needs of large-scale AI workloads.
* **High cost:** Cerebras's systems are very expensive, which can make them prohibitive for some organizations.
* **Limited software ecosystem:** Cerebras's systems have a limited software ecosystem, which can make it difficult to find the tools and libraries needed for specific applications.
**Tesla Dojo**
**Advantages:**
* **High scalability:** Tesla Dojo's heterogeneous architecture and custom software stack allow it to scale to meet the needs of large-scale AI workloads.
* **High performance:** Tesla Dojo is able to achieve very high performance, making it ideal for training the largest and most complex deep learning models.
* **Custom software stack:** Tesla Dojo's custom software stack is optimized for its hardware, which provides better performance and efficiency than running AI workloads on general-purpose software.
* **Wide software ecosystem:** Tesla Dojo is supported by a wide range of software libraries and tools, making it easy to integrate into existing workflows.
**Disadvantages:**
* **Complex software stack:** Tesla Dojo's custom software stack can be complex to use and manage, which can be a barrier to entry for some organizations.
* **Large form factor:** Tesla Dojo's systems are large and expensive, which can make them prohibitive for some organizations.
* **High power consumption:** Tesla Dojo's systems have a high power consumption, which can be a concern for organizations with limited power capacity.
Tesla är ute efter kostnadseffektiva och skalbara lösningar. Det är en återkommande princip.