#compute — Public Fediverse posts

Needle: We Distilled Gemini Tool Calling into a 26M Model

https://github.com/cactus-compute/needle

#HackerNews #Needle #Gemini #Tool #Model #AI #Distillation #Cactus #Compute

Needle: We Distilled Gemini Tool Calling into a 26M Model

https://github.com/cactus-compute/needle

#HackerNews #Needle #Gemini #Tool #Model #AI #Distillation #Cactus #Compute

Needle: We Distilled Gemini Tool Calling into a 26M Model

https://github.com/cactus-compute/needle

#HackerNews #Needle #Gemini #Tool #Model #AI #Distillation #Cactus #Compute

Needle: We Distilled Gemini Tool Calling into a 26M Model

https://github.com/cactus-compute/needle

#HackerNews #Needle #Gemini #Tool #Model #AI #Distillation #Cactus #Compute

Needle: We Distilled Gemini Tool Calling into a 26M Model

https://github.com/cactus-compute/needle

#HackerNews #Needle #Gemini #Tool #Model #AI #Distillation #Cactus #Compute

In my #homelab I have a mix of #arm ( #raspberrypi ) and #intel / #amd #compute nodes. I have never actually tested if the VPC #network works between the two architectures. Decided to test it today, seems to work fine.

Below are two #selfhosted #vm on different bare metals with a shared VPC network using iperf.

#technology #virtualization #linux

Wow, a #compute #deal with SpaceX! 🚀 Because nothing says "cutting-edge tech" like more #cloud minutes for Claude to contemplate its existence. Meanwhile, Bezos is crying into his #BezosBot at not being invited to this party of zeros and ones. 😂
https://www.anthropic.com/news/higher-limits-spacex #SpaceX #computing #tech #humor #HackerNews #ngated

Wow, a #compute #deal with SpaceX! 🚀 Because nothing says "cutting-edge tech" like more #cloud minutes for Claude to contemplate its existence. Meanwhile, Bezos is crying into his #BezosBot at not being invited to this party of zeros and ones. 😂
https://www.anthropic.com/news/higher-limits-spacex #SpaceX #computing #tech #humor #HackerNews #ngated

Wow, a #compute #deal with SpaceX! 🚀 Because nothing says "cutting-edge tech" like more #cloud minutes for Claude to contemplate its existence. Meanwhile, Bezos is crying into his #BezosBot at not being invited to this party of zeros and ones. 😂
https://www.anthropic.com/news/higher-limits-spacex #SpaceX #computing #tech #humor #HackerNews #ngated

Wow, a #compute #deal with SpaceX! 🚀 Because nothing says "cutting-edge tech" like more #cloud minutes for Claude to contemplate its existence. Meanwhile, Bezos is crying into his #BezosBot at not being invited to this party of zeros and ones. 😂
https://www.anthropic.com/news/higher-limits-spacex #SpaceX #computing #tech #humor #HackerNews #ngated

Wow, a #compute #deal with SpaceX! 🚀 Because nothing says "cutting-edge tech" like more #cloud minutes for Claude to contemplate its existence. Meanwhile, Bezos is crying into his #BezosBot at not being invited to this party of zeros and ones. 😂
https://www.anthropic.com/news/higher-limits-spacex #SpaceX #computing #tech #humor #HackerNews #ngated

OpenCL 3.1 is here.

The Khronos Group has moved several capabilities into the core spec, including SPIR-V kernels, subgroups, and integer dot products.

Also includes improvements to the memory model and synchronization, plus better alignment with Vulkan via device UUID queries.

Implementations are already underway across major vendors and open source projects.

- Full Blog: https://www.khronos.org/blog/opencl-3.1-is-here?utm_medium=social&utm_source=mastodon&utm_campaign=OpenCL_3.1_is_here&utm_content=blog
- OpenCL specification GitHub
- Khronos Discord

#OpenCL #HPC #GPU #Compute #SPIRV

"AI policy expert Lennart Heim is a useful guide to this machinery. He formerly led compute research at the RAND Center on AI, Security, and Technology and cofounded Epoch AI, which tracks the resources behind frontier AI models. His beat is where a cloud dashboard becomes a construction project—where digital demand collides with factories, transformers, chips and cables.

[An edited transcript of the telephone interview follows.]

Developers are saying the rate limits and blocked third-party tools look like a compute crunch. What does a compute shortage actually mean?

When we say “compute,” we mean computing power. For AI, training compute scales with model size: bigger neural networks need more data, and more data needs more processing power. What was underreported for years is that the same relationship holds for deployment. Running the model for users—inference—is incredibly compute-intensive because bigger models need more computing power to serve. So if more people use AI with more tokens and more intensity, you need more compute. If 10 times more people use AI 10 times more heavily, you need close to 100 times more compute.

Why does a flat-rate subscription break down for AI in a way it didn’t for earlier Internet services?

The Internet runs on flat-rate subscriptions: you pay $20 a month and get effectively unlimited use. That works when the marginal cost per user is low—a Google Workspace power user doesn’t cost Google much more than a light user. With AI, it breaks. Using AI 10 times more heavily costs the provider roughly 10 times more money. Paying per token means you literally pay for your resources; paying $20 flat means you’re often burning more compute than $20 can buy. That’s why we see rate limits mostly on monthly subscription plans. At some point, you have to rate limit."

https://www.scientificamerican.com/article/what-is-the-ai-compute-crunch-and-why-are-ai-tools-hitting-usage-limits/

#AI #GenerativeAI #DataCenters #Compute

April issue is out!

I’m tweaking the newsletter to better match what Python data engineers care about: less visualization, more data munging.

So expect a bit less Shiny and Streamlit, and more ADBC, SQLGlot, and practical ecosystem updates.

https://pythondataeng.substack.com/p/monthly-python-data-engineering-april-4e5

#python #dataeng #datascience #data #ml #machineLearning #database #compute

Stuff are connecting ! #shader #bonzomatic #compute

IWOCL 2026 is next week!

Join the global OpenCL and SYCL community in Heilbronn, Germany (May 6–8) for the premier forum dedicated to open compute languages and heterogeneous platform programming. The program includes the latest technical talks, Khronos Working Group updates, application case studies, and ample opportunity to connect with peers across industry and academia.

Registration remains open: www.iwocl.org

See you there.
#IWOCL #OpenCL #SYCL #HPC #Heterogeneous #Compute

Google controls ~25% of global AI compute, with ~3.8M TPUs and 1.3M GPUs

https://www.ft.com/content/2429f0f0-b685-4747-b425-bf8001a2e94c

#HackerNews #Google #AI #compute #TPUs #GPUs #technology

Google controls ~25% of global AI compute, with ~3.8M TPUs and 1.3M GPUs

https://www.ft.com/content/2429f0f0-b685-4747-b425-bf8001a2e94c

#HackerNews #Google #AI #compute #TPUs #GPUs #technology

Google controls ~25% of global AI compute, with ~3.8M TPUs and 1.3M GPUs

https://www.ft.com/content/2429f0f0-b685-4747-b425-bf8001a2e94c

#HackerNews #Google #AI #compute #TPUs #GPUs #technology

Google controls ~25% of global AI compute, with ~3.8M TPUs and 1.3M GPUs

https://www.ft.com/content/2429f0f0-b685-4747-b425-bf8001a2e94c

#HackerNews #Google #AI #compute #TPUs #GPUs #technology

Google controls ~25% of global AI compute, with ~3.8M TPUs and 1.3M GPUs

https://www.ft.com/content/2429f0f0-b685-4747-b425-bf8001a2e94c

#HackerNews #Google #AI #compute #TPUs #GPUs #technology

@nyc

Sam Altman already talked as #compute for currency instead of #UBI

Large Language Models (LLMs) require dramatic amount of compute resources :blobcatfearful: Nothing new, right?

Some specifics I was exposed to: while working on a coursework for my studies, have to make knowledge graph from regulatory texts. To avoid cloud, decided to run everything locally :misskey_loading:

Did not manage to launch llama4 - apparently it requires 60 GB of RAM to start. Let that sink in - 60 GB! Let alone model itself weighs more than 200 GB :blobcatheadinitshands:

Did the task with llama3 on :nvidia: GPU cooking at 70°, but at least did not have to subscribe to someone else's computer :old_man_yells_at_cloud:

#llm #aislop #compute

Intel Demos Chip to Compute with Encrypted Data

https://spectrum.ieee.org/fhe-intel

#HackerNews #Intel #EncryptedData #Compute #Chip #FHE #Technology #Privacy #Security

Ugh... this AI world is moving too fast...No wonder so many smart folks have #aianxiety

#Antigravity is a thing and apparently its pretty shmick. But I have no time for now to suck up another platform.

What the do the smart #Vibecoders do to mitigate rate limits (#Compute ceiling is something you hit very hard if you start doing serious projects)?

1. Run #Claudecode inside Antigravity for #Vibecode
2. Add automated testing agent to clean up bugs without wasting tokens. WHAT?
3. Use #Gemini3 for design and architecture
4. #Testsprite MCP for testing (Not using tokens).

I have no idea how to do this, I would have to pick up 3 new platforms and even with #Ai this is beyond me...

... I understand why #techbros are hitting #ketamine.... That or retire to write poetry.

The OpenMP ARB warmly welcomes Eric Stotzer and Ruud van der Pas to its Board of Directors!

These new appointments reinforce OpenMP ARB’s commitment to advancing a robust, portable, and forward‑looking parallel programming standard that meets the needs of modern computing.

Help us welcome Rud and Eric onboard!

Learn more: https://www.openmp.org/press-release/board-stotzer-vdpas/
#OpenMP #parellel #compute

The OpenMP ARB warmly welcomes Eric Stotzer and Ruud van der Pas to its Board of Directors!

These new appointments reinforce OpenMP ARB’s commitment to advancing a robust, portable, and forward‑looking parallel programming standard that meets the needs of modern computing.

Help us welcome Rud and Eric onboard!

Learn more: https://www.openmp.org/press-release/board-stotzer-vdpas/
#OpenMP #parellel #compute

The OpenMP ARB warmly welcomes Eric Stotzer and Ruud van der Pas to its Board of Directors!

These new appointments reinforce OpenMP ARB’s commitment to advancing a robust, portable, and forward‑looking parallel programming standard that meets the needs of modern computing.

Help us welcome Rud and Eric onboard!

Learn more: https://www.openmp.org/press-release/board-stotzer-vdpas/
#OpenMP #parellel #compute

The OpenMP ARB warmly welcomes Eric Stotzer and Ruud van der Pas to its Board of Directors!

These new appointments reinforce OpenMP ARB’s commitment to advancing a robust, portable, and forward‑looking parallel programming standard that meets the needs of modern computing.

Help us welcome Rud and Eric onboard!

Learn more: https://www.openmp.org/press-release/board-stotzer-vdpas/
#OpenMP #parellel #compute

Fetching the video to learn

Thank you 💕 for sharing

https://www.youtube.com/watch?v=Ef70x0izkFU

@geerlingguy

🦋💙❤️💋#Lobi 💙💕🌹💐💙🦋

#Raspberry #Pi5 #SBC #Compute #model #technology #notebook #Linux #OpenSource #programming #POSIX

The #US has gone too far on so many levels, but this years #CES is rage inducing.

#Europe needs an #ARM #SoC producer, stat. It needs to fully support #Linux and preferably use #OpenFirmware.

Also, the #EU should open up to importing #Chinese made #GPU in pure defiance of these yankies. #NVIDIA telling us to lease GPUs to game in the cloud is a slap in the face.

Is #YoullOwnNothingAndYoullBeHappy taken to a digital extreme. #Consumers need to own their #compute, goddamn it.

Just signed up for COMPUTE!'s Gazette print + digital magazine.

#Compute #Computesgazette #vintagecomputing #retrocomputing #homecomputers #magazine #nostalgia #computing #computers

We Must Seize the Means of Compute

https://thompson2026.com/blog/seize-the-means-of-compute/

#HackerNews #We #Must #Seize #the #Means #of #Compute #technology #computing #freedom #open-source #digital #rights

We Must Seize the Means of Compute

https://thompson2026.com/blog/seize-the-means-of-compute/

#HackerNews #We #Must #Seize #the #Means #of #Compute #technology #computing #freedom #open-source #digital #rights

We Must Seize the Means of Compute

https://thompson2026.com/blog/seize-the-means-of-compute/

#HackerNews #We #Must #Seize #the #Means #of #Compute #technology #computing #freedom #open-source #digital #rights

We Must Seize the Means of Compute

https://thompson2026.com/blog/seize-the-means-of-compute/

#HackerNews #We #Must #Seize #the #Means #of #Compute #technology #computing #freedom #open-source #digital #rights

Purchased (from a major retailer, not proud of that but it is what it is) four Mini PCs each for $129 U.S. Intel N150, 12gb ram, 2.5gbps nic, for substantially less than a 32gb #dram #sodimm kit alone! 4 whole computers (without storage or OS. The version with a 128gb ssd and Windows is $169.) These for a #cluster, so 16 cores, disturbed 48gb memory. Yes, existing old 10gbps nfs server and a small 10/2.5gbps switch.

But with #RAM prices out of control? ~$60 per core and 48gb #MPI #compute!

Purchased (from a major retailer, not proud of that but it is what it is) four Mini PCs each for $129 U.S. Intel N150, 12gb ram, 2.5gbps nic, for substantially less than a 32gb #dram #sodimm kit alone! 4 whole computers (without storage or OS. The version with a 128gb ssd and Windows is $169.) These for a #cluster, so 16 cores, disturbed 48gb memory. Yes, existing old 10gbps nfs server and a small 10/2.5gbps switch.

But with #RAM prices out of control? ~$60 per core and 48gb #MPI #compute!

Purchased (from a major retailer, not proud of that but it is what it is) four Mini PCs each for $129 U.S. Intel N150, 12gb ram, 2.5gbps nic, for substantially less than a 32gb #dram #sodimm kit alone! 4 whole computers (without storage or OS. The version with a 128gb ssd and Windows is $169.) These for a #cluster, so 16 cores, disturbed 48gb memory. Yes, existing old 10gbps nfs server and a small 10/2.5gbps switch.

But with #RAM prices out of control? ~$60 per core and 48gb #MPI #compute!

Purchased (from a major retailer, not proud of that but it is what it is) four Mini PCs each for $129 U.S. Intel N150, 12gb ram, 2.5gbps nic, for substantially less than a 32gb #dram #sodimm kit alone! 4 whole computers (without storage or OS. The version with a 128gb ssd and Windows is $169.) These for a #cluster, so 16 cores, disturbed 48gb memory. Yes, existing old 10gbps nfs server and a small 10/2.5gbps switch.

But with #RAM prices out of control? ~$60 per core and 48gb #MPI #compute!

New Cooling Tech Sparks Talk of Power Positive Data Centers

Orbis Electric has introduced a new Cooling Engine that aims to reshape how data centers manage heat as demand for AI and high performance computing continues to surge.

The system is built on the company’s HaloDrive technology, which uses an axial flux motor to capture energy that traditional cooling units waste.

By recovering power that would normally be lost, the Cooling Engine shifts cooling from a passive energy load into a source of efficiency gains.

The company says the system delivers up to twice the cooling capacity of legacy coolant distribution units while using a smaller footprint.

Orbis reports wire to fluid efficiency of up to 90 percent and projects that a single unit can generate as much as three million dollars in annual value in a 20 megawatt data hall through reduced power use, lower operating costs, and higher rack density.

The Cooling Engine is compatible with direct to chip cooling and aligns with Open Compute Project guidelines.

Orbis says the system can reduce cooling energy use by half and return up to eight percent of that energy as usable power, creating new capacity without grid expansion.

The design is purpose built for high density AI clusters. It consolidates pumping, cooling, and flow control into a single module that supports N plus one redundancy and fluid agnostic operation.

Orbis also cites environmental gains, including steep reductions in chiller load, emissions, and water use.

https://twitter.com/tbcnewsph/status/1963819060485050703

#artificialIntelligence #compute #cooling #dataCenter #energy #event #latest #news #orbis #philippines #technology

Why Sam Altman Won't Be on the Hook for OpenAI's Spending Spree
https://www.forbes.com/sites/rashishrivastava/2025/11/07/why-sam-altman-wont-be-on-the-hook-for-openais-massive-spending-spree/
#ycombinator #Sam_Altman #Infrastructure #compute #OpenAI #ChatGPT #datacenters #Sarah_Friar #government #cloud

📢🚨Blosc2 3.11.0 Released! 🚨 📢

Blosc2 also now features as a part of the NumPy array library ecosystem: https://numpy.org/

In v3.11, Blosc2 has become even more flexible; all functions can accept basically any array object! That's right:

𝚋𝚕𝚘𝚜𝚌𝟸.𝚖𝚊𝚝𝚖𝚞𝚕(𝙰, 𝙱)

will work directly for 𝙰, 𝙱 PyTorch, Jax, NumPy, Tensorflow, Zarr... arrays!

**Plot: computation (for NumPy, Blosc2, Pytorch) on NxNxN arrays from the three libraries. Machine has 32GB of RAM **

#Compute
#Tensor
#PyTorch
#Python

📢🚨Blosc2 3.9.1 Released! 🚨 📢

With this release we further extend array API standard (https://data-apis.org/array-api/latest/index.html) compliance, adding around 70 new functions, methods and attributes 🔝🔝🔝 .

This includes:
• Memory-optimised linear algebra routines: tensordot can be up to 12x 𝐅𝐀𝐒𝐓𝐄𝐑 than NumPy for large arrays 🧱!
• Optimised numerical computation routines ⏩ : functions like hypot are 10x 𝐅𝐀𝐒𝐓𝐄𝐑⚡than NumPy!

See here for benchmarks: https://ironarray.io/blog/array-api

#Compute
#Tensor
#LinearAlgebra
#Python

Alexandre Roure of the Computer & Communications Industry Association, whose members include many Big Tech groups, says the debate about blunt market access restrictions for non-EU tech companies “only distracts policymakers from the real task: finally delivering a functioning digital single market with clear, simple and practical rules”.

In private conversations, several Big Tech lobbyists and executives also express confidence in their ability to continue dominating the European market given the paucity of homegrown alternatives and the lack of urgency among many consumers.

by Barbara Moens for FT: https://archive.is/20250725082920/https://www.ft.com/content/5e25c397-61d1-4b48-b5c5-65561a4c9df2 via @Ruth_Mottram 🧵

#techSovereignty #digitalSovereignty #decoupling #EuropeAlternatives #nonUS #EuropeanAlternatives #OVH #Europe #EU #privacy #dataProtection #OVHCloud #CloudComputing #BuyEuropean #EuroStack #compute #bigTech #GAFAM #Cloud #FT

https://www.tomshardware.com/pc-components/cpus/intel-posts-flat-year-over-year-earnings-and-bleak-outlook-warns-about-macroeconomic-pressures

Intel posts flat year-over-year earnings and bleak outlook, warns about macroeconomic pressures

#Intel #INTC #nasdaq #financial #revenue #loss #macroeconomic #spending #pullbacks #competition #AMD #ARM #server #tech #technology #cio #datacenter #compute #stocks #equities #markets #stockmarket #investors #trades #trading #investing #cash #bonds #treasuries #treasury #IT #sysops #economic #economy #finance #money #earnings #trump #tariffs #business #tax #consumer #retail #spending #taxes #electronics #computing

What an excellent start to Day 1 of #ALTA2024!

In yesterday's #tutorial, Dr Nicholas I-Hsien Kuo took our participants through:

➡️ Implementing and evaluating #PEFT and quantisation techniques.
➡️ Fine-tuning and deploying #LLMs on hardware with limited resources.
➡️ Optimising workflows for real-world applications without sacrificing performance.

A huge thanks to Google Colab for our #compute requirements 👏

📷 by Taylor Liu, one of our incredible #ALTA2024 Volunteers

It doesnt look like much, its a "virtual pipes" based version of a fluid simulation. The win here is that it took me 2 days instead of a few weeks to get JFA working. This also has lot more futzing around with compute stages and differen buffers. This time I tried using a hacked up version of wgsl_to_wgpu to update the bindings. Big win there keeping everything in line for sure. Next up, hooking the two stages together so that I can use JFA to calculate the distance to water.

The original repo: https://github.com/ScanMountGoat/wgsl_to_wgpu

My hacked up version (please forgive me!) https://github.com/00alia00/wgsl_to_wgpu_bevy

#bevy #rust #rustlang #gamedev #compute #computeshader #simulation #fluids #rendering