VHPC Workshop
@vhpc.bsky.social
Workshop on Virtualization in High-Performance Cloud Computing. News and views on #HPC #virtualization #VHPC #VMs #containers #cloud #ai
VHPC '25 Program is now live: vhpc.org
Featured Keynote Tutorial:
"Writing a hypervisor from scratch"
Join Seiya Nuta from Vercel Inc. for an hands-on tutorial covering fundamental concepts and practical implementation techniques.
Featured Keynote Tutorial:
"Writing a hypervisor from scratch"
Join Seiya Nuta from Vercel Inc. for an hands-on tutorial covering fundamental concepts and practical implementation techniques.
VHPC 2025 - 20th Workshop on Virtualization in High-Performance Cloud Computing
VHPC 2025 - The 20th Workshop on Virtualization in High-Performance Cloud Computing, co-located with Euro-Par 2025 in Dresden, Germany
vhpc.org
August 23, 2025 at 4:04 AM
VHPC '25 Program is now live: vhpc.org
Featured Keynote Tutorial:
"Writing a hypervisor from scratch"
Join Seiya Nuta from Vercel Inc. for an hands-on tutorial covering fundamental concepts and practical implementation techniques.
Featured Keynote Tutorial:
"Writing a hypervisor from scratch"
Join Seiya Nuta from Vercel Inc. for an hands-on tutorial covering fundamental concepts and practical implementation techniques.
The Euro-Par Dresden VHPC workshop will emphasize containers/virtualization for LLM training/inference workloads
May 10, 2025 at 1:54 AM
The Euro-Par Dresden VHPC workshop will emphasize containers/virtualization for LLM training/inference workloads
VHPC 2025 : 20th Workshop on Virtualization in High-Performance Cloud Computing
Deadline Extension Dresden Workshop to May 25, 2025
vhpc.org
Deadline Extension Dresden Workshop to May 25, 2025
vhpc.org
May 9, 2025 at 7:52 PM
VHPC 2025 : 20th Workshop on Virtualization in High-Performance Cloud Computing
Deadline Extension Dresden Workshop to May 25, 2025
vhpc.org
Deadline Extension Dresden Workshop to May 25, 2025
vhpc.org
reward = compute_reward(result)
if reward > self.last_reward -> new_interpreter
would likely converge on fitting a sequence using attention
scary stuff
if reward > self.last_reward -> new_interpreter
would likely converge on fitting a sequence using attention
scary stuff
January 7, 2025 at 10:26 AM
reward = compute_reward(result)
if reward > self.last_reward -> new_interpreter
would likely converge on fitting a sequence using attention
scary stuff
if reward > self.last_reward -> new_interpreter
would likely converge on fitting a sequence using attention
scary stuff
a homoiconic transformer in the form of
understanding = torch.matmul(interpreter, sequence.transpose(-2, -1))
new_interpreter = torch.matmul(understanding.softmax(-1), interpreter)
result = torch.matmul(new_interpreter, sequence.transpose(-2, -1))
understanding = torch.matmul(interpreter, sequence.transpose(-2, -1))
new_interpreter = torch.matmul(understanding.softmax(-1), interpreter)
result = torch.matmul(new_interpreter, sequence.transpose(-2, -1))
January 7, 2025 at 10:23 AM
a homoiconic transformer in the form of
understanding = torch.matmul(interpreter, sequence.transpose(-2, -1))
new_interpreter = torch.matmul(understanding.softmax(-1), interpreter)
result = torch.matmul(new_interpreter, sequence.transpose(-2, -1))
understanding = torch.matmul(interpreter, sequence.transpose(-2, -1))
new_interpreter = torch.matmul(understanding.softmax(-1), interpreter)
result = torch.matmul(new_interpreter, sequence.transpose(-2, -1))
if booth adapt autonomously in response
January 6, 2025 at 3:10 PM
if booth adapt autonomously in response
so an engineering problem, what if science delivers a few more incremental breakthroughs in addition.
We are now confident we know how to build AGI as we have traditionally understood it.
blog.samaltman.com/reflections
#agi #singularity
Reflections
The second birthday of ChatGPT was only a little over a month ago, and now we have transitioned into the next paradigm of models that can do complex reasoning. New years get people in a reflective...
blog.samaltman.com
January 6, 2025 at 3:08 PM
so an engineering problem, what if science delivers a few more incremental breakthroughs in addition.
We are now confident we know how to build AGI as we have traditionally understood it.
blog.samaltman.com/reflections
#agi #singularity
Altman clearly says near singularity
January 6, 2025 at 12:19 PM
Altman clearly says near singularity
this is a nice ghostty.org config based on Aardvark Blue with an underline cursor for macOS/zsh.
gist.github.com/pacedproton/...
#ghostty
gist.github.com/pacedproton/...
#ghostty
ghostty.org
December 29, 2024 at 10:10 PM
this is a nice ghostty.org config based on Aardvark Blue with an underline cursor for macOS/zsh.
gist.github.com/pacedproton/...
#ghostty
gist.github.com/pacedproton/...
#ghostty
[xAI Colossus] "Across all three tiers of the network fabric, the system has experienced zero application latency degradation or packet loss due to flow collisions".
0 flow collisions across all fabrier tiers seems improbable for RoCE.
nvidianews.nvidia.com/news/spectru...
#roce #infiniband #hpc
December 25, 2024 at 9:08 PM
[xAI Colossus] "Across all three tiers of the network fabric, the system has experienced zero application latency degradation or packet loss due to flow collisions".
0 flow collisions across all fabrier tiers seems improbable for RoCE.
nvidianews.nvidia.com/news/spectru...
#roce #infiniband #hpc
latent non-language domain-specific spaces probably abound - wavelet coefficients, eigenspaces, but this is bytes only.
7/ BLT has direct access to bytes but at its core still has a Transformer that operates over patches - the best of both worlds. But it’s just the beginning! We are releasing the training and inference code at github.com/facebookrese...
GitHub - facebookresearch/blt: Code for BLT research paper
Code for BLT research paper. Contribute to facebookresearch/blt development by creating an account on GitHub.
github.com
December 15, 2024 at 7:27 PM
latent non-language domain-specific spaces probably abound - wavelet coefficients, eigenspaces, but this is bytes only.
LLM formal specifications
in training or not: in-context specifications and operational semantics!
in training or not: in-context specifications and operational semantics!
December 15, 2024 at 3:04 PM
LLM formal specifications
in training or not: in-context specifications and operational semantics!
in training or not: in-context specifications and operational semantics!
Reposted by VHPC Workshop
Linux 6.13 KVM Eliminates An "Awful Idea", Many x86_64 Improvements - https://www.phoronix.com/news/Linux-6.13-KVM
Linux 6.13 KVM Eliminates An "Awful Idea", Many x86_64 Improvements
The KVM changes were merged yesterday for Linux 6.13 in further enhancing the open-source virtualization stack...
www.phoronix.com
November 25, 2024 at 7:59 PM
Linux 6.13 KVM Eliminates An "Awful Idea", Many x86_64 Improvements - https://www.phoronix.com/news/Linux-6.13-KVM
Virtualization strikes new ground in this paper by software-virtualizing QEC logical qubits by software-based detecting and correcting errors to abstract noisy physical qubits.
arxiv.org/html/2411.11...
arxiv.org/html/2411.11...
Logical computation demonstrated with a neutral atom quantum processor
arxiv.org
November 20, 2024 at 9:06 PM
Virtualization strikes new ground in this paper by software-virtualizing QEC logical qubits by software-based detecting and correcting errors to abstract noisy physical qubits.
arxiv.org/html/2411.11...
arxiv.org/html/2411.11...
Reposted by VHPC Workshop
November 18, 2024 at 9:56 PM
VHPC is now active on Mastodon as @vhpc@[email protected]@[email protected] and on Bluesky under the handle @vhpc.bsky.social.
November 16, 2024 at 7:40 PM
VHPC is now active on Mastodon as @vhpc@[email protected]@[email protected] and on Bluesky under the handle @vhpc.bsky.social.