Underfox
@underfox3.bsky.social
Physicist, Telecom Engineering lover, HPC Enthusiast. Prog Rock/Metal fan.
---
Independent tech analyst focused on semiconductors, patent analysis and emerging technologies.
---
Independent tech analyst focused on semiconductors, patent analysis and emerging technologies.
Pinned
Underfox
@underfox3.bsky.social
· Nov 28
Underfox3 | Substack
Independent tech analyst focused on semiconductors, patent analysis and emerging technologies. Click to read Underfox3, a Substack publication. Launched a year ago.
underfox3.substack.com
I would like to invite all my followers to subscribe on my Substack. Everyone who loves technology and is hungry for new discoveries is welcome. Also, feel free to contact me!
Profile Link: underfox3.substack.com
Profile Link: underfox3.substack.com
Researchers have investigated the channel width scaling in nanoribbon transistors based on monolayer MoS2 grown on 2-inch wafers, revealing excellent electrostatic control, minimal short-channel effects, and low device-to-device variability.
arxiv.org/pdf/2601.13696
arxiv.org/pdf/2601.13696
January 22, 2026 at 5:50 AM
Researchers have investigated the channel width scaling in nanoribbon transistors based on monolayer MoS2 grown on 2-inch wafers, revealing excellent electrostatic control, minimal short-channel effects, and low device-to-device variability.
arxiv.org/pdf/2601.13696
arxiv.org/pdf/2601.13696
In this paper, researchers have demostrated that acyclic deterministic finite automatons, as well as Moore and Mealy machines, from prefix-complete instances in a binary alphabet, when the instance strings are all prefixes of a single string, is NP-complete.
arxiv.org/pdf/2601.12621
arxiv.org/pdf/2601.12621
January 21, 2026 at 9:01 AM
In this paper, researchers have demostrated that acyclic deterministic finite automatons, as well as Moore and Mealy machines, from prefix-complete instances in a binary alphabet, when the instance strings are all prefixes of a single string, is NP-complete.
arxiv.org/pdf/2601.12621
arxiv.org/pdf/2601.12621
Researchers have proposed Jet-RL, an on-policy FP8 reinforcement learning framework that enforces a unified precision workflow for training and deployment, resolving policy mismatches and ensuring stable optimization.
arxiv.org/pdf/2601.14243
arxiv.org/pdf/2601.14243
January 21, 2026 at 8:53 AM
Researchers have proposed Jet-RL, an on-policy FP8 reinforcement learning framework that enforces a unified precision workflow for training and deployment, resolving policy mismatches and ensuring stable optimization.
arxiv.org/pdf/2601.14243
arxiv.org/pdf/2601.14243
In this paper is presented DaggerFFT, a distributed FFT framework in Julia that treats highly parallel FFT computations as a dynamically scheduled task graph with each FFT stage operates on a separately defined distributed array.
arxiv.org/pdf/2601.12209
arxiv.org/pdf/2601.12209
January 21, 2026 at 7:32 AM
In this paper is presented DaggerFFT, a distributed FFT framework in Julia that treats highly parallel FFT computations as a dynamically scheduled task graph with each FFT stage operates on a separately defined distributed array.
arxiv.org/pdf/2601.12209
arxiv.org/pdf/2601.12209
In this paper is presented a detailed analysis focused on quantifying and improving #HPC resource utilization efficiency specifically within the Latin American Giant Observatory computational environment.
arxiv.org/pdf/2601.13424
arxiv.org/pdf/2601.13424
January 21, 2026 at 7:20 AM
In this paper is presented a detailed analysis focused on quantifying and improving #HPC resource utilization efficiency specifically within the Latin American Giant Observatory computational environment.
arxiv.org/pdf/2601.13424
arxiv.org/pdf/2601.13424
In this paper is proposed OPTIMUM-DERAM, a novel decentralized shared memory framework for Web3 that leverages Random Linear Network Coding to enable atomic read/write operation.
arxiv.org/pdf/2601.13146
arxiv.org/pdf/2601.13146
January 21, 2026 at 7:11 AM
In this paper is proposed OPTIMUM-DERAM, a novel decentralized shared memory framework for Web3 that leverages Random Linear Network Coding to enable atomic read/write operation.
arxiv.org/pdf/2601.13146
arxiv.org/pdf/2601.13146
In this paper is presented Memory Disorder: a timerless side-channel that uses memory re-orderings to infer activity on other processes.
arxiv.org/pdf/2601.08770
arxiv.org/pdf/2601.08770
January 16, 2026 at 11:27 PM
In this paper is presented Memory Disorder: a timerless side-channel that uses memory re-orderings to infer activity on other processes.
arxiv.org/pdf/2601.08770
arxiv.org/pdf/2601.08770
In this paper is presented a high-performance, mixed-precision solver for symmetric linear systems based on a nested recursive Cholesky decomposition that uniformly applies recursion across POTRF, TRSM, and SYRK phases.
arxiv.org/pdf/2601.08082
arxiv.org/pdf/2601.08082
January 16, 2026 at 11:06 PM
In this paper is presented a high-performance, mixed-precision solver for symmetric linear systems based on a nested recursive Cholesky decomposition that uniformly applies recursion across POTRF, TRSM, and SYRK phases.
arxiv.org/pdf/2601.08082
arxiv.org/pdf/2601.08082
In this paper is presented MatrixPIC, the first holistic co-design of the Particle-in-Cell deposition kernel, data layout, and an incremental sorting mechanism, all tailored specifically for the hybrid MPU-VPU SIMD execution model on modern CPUs.
arxiv.org/pdf/2601.08277
arxiv.org/pdf/2601.08277
January 16, 2026 at 10:41 PM
In this paper is presented MatrixPIC, the first holistic co-design of the Particle-in-Cell deposition kernel, data layout, and an incremental sorting mechanism, all tailored specifically for the hybrid MPU-VPU SIMD execution model on modern CPUs.
arxiv.org/pdf/2601.08277
arxiv.org/pdf/2601.08277
Google researchers have proposed a single-stage Huffman encoder for machine learning compression which use fixed codebooks derived from the average probability distribution of previous data batches, reducing computational, latency and data overheads.
arxiv.org/pdf/2601.10673
arxiv.org/pdf/2601.10673
January 16, 2026 at 1:57 PM
Google researchers have proposed a single-stage Huffman encoder for machine learning compression which use fixed codebooks derived from the average probability distribution of previous data batches, reducing computational, latency and data overheads.
arxiv.org/pdf/2601.10673
arxiv.org/pdf/2601.10673
In this paper, researchers have demonstrated a fully on-chip probabilistic computer for dense combinatorial optimization, using MIMO detection as a benchmark.
arxiv.org/pdf/2601.09037
arxiv.org/pdf/2601.09037
January 16, 2026 at 1:54 AM
In this paper, researchers have demonstrated a fully on-chip probabilistic computer for dense combinatorial optimization, using MIMO detection as a benchmark.
arxiv.org/pdf/2601.09037
arxiv.org/pdf/2601.09037
In this paper is presented a systematic evaluation of NVIDIA’s Blackwell consumer GPUs for production LLM inference, benchmarking four open-weight models across 79 configurations spanning quantization formats, context lengths, and three workloads.
arxiv.org/pdf/2601.09527
arxiv.org/pdf/2601.09527
January 15, 2026 at 9:26 PM
In this paper is presented a systematic evaluation of NVIDIA’s Blackwell consumer GPUs for production LLM inference, benchmarking four open-weight models across 79 configurations spanning quantization formats, context lengths, and three workloads.
arxiv.org/pdf/2601.09527
arxiv.org/pdf/2601.09527
In this paper is presented DPA-Store, an ordered in-memory KV store that uses the BlueField-3 SmartNIC and its data path accelerators to bypass OS overheads and to allow stateless clients.
arxiv.org/pdf/2601.06231
arxiv.org/pdf/2601.06231
January 14, 2026 at 10:14 PM
In this paper is presented DPA-Store, an ordered in-memory KV store that uses the BlueField-3 SmartNIC and its data path accelerators to bypass OS overheads and to allow stateless clients.
arxiv.org/pdf/2601.06231
arxiv.org/pdf/2601.06231
In this paper, researchers have proposed a SIMD-based algorithm to accelerate UTF-16 correction, leveraging ARM NEON and x64 SSE instructions to process 16-bit code units in blocks, achieving significant performance gains.
arxiv.org/pdf/2601.06349
arxiv.org/pdf/2601.06349
January 14, 2026 at 10:05 PM
In this paper, researchers have proposed a SIMD-based algorithm to accelerate UTF-16 correction, leveraging ARM NEON and x64 SSE instructions to process 16-bit code units in blocks, achieving significant performance gains.
arxiv.org/pdf/2601.06349
arxiv.org/pdf/2601.06349
Researchers demonstrated a FWM process within the single resonance bandwidth of a microresonator, revealing previously unobserved Kerr-comb dynamics, featuring parametrically driven phase multistability that can be observed directly in the temporal domain
arxiv.org/pdf/2601.07378
arxiv.org/pdf/2601.07378
January 14, 2026 at 9:58 PM
Researchers demonstrated a FWM process within the single resonance bandwidth of a microresonator, revealing previously unobserved Kerr-comb dynamics, featuring parametrically driven phase multistability that can be observed directly in the temporal domain
arxiv.org/pdf/2601.07378
arxiv.org/pdf/2601.07378
GlobalFoundries today announced its execution of a definitive agreement to acquire Synopsys’ ARC Processor IP Solutions business, including its teams of engineers and designers.
Press Release: gf.com/gf-press-rel...
Press Release: gf.com/gf-press-rel...
GlobalFoundries to Acquire Synopsys’ Processor IP Solutions Business, Expanding Capabilities to Accelerate Physical AI Applications | GlobalFoundries
GlobalFoundries (Nasdaq: GFS) (GF) today announced its execution of a definitive agreement to acquire Synopsys’ ARC Processor IP Solutions business, including its teams of engineers and designers.
gf.com
January 14, 2026 at 3:24 PM
GlobalFoundries today announced its execution of a definitive agreement to acquire Synopsys’ ARC Processor IP Solutions business, including its teams of engineers and designers.
Press Release: gf.com/gf-press-rel...
Press Release: gf.com/gf-press-rel...
Another strong sign of the end of the gold rush cycle is when resources become so scarce that outbreaks of theft and violence begin to erupt.
It's no different in the AI bubble. RAM theft is just the first sign. And believe me, it's going to get worse.
wccftech.com/ram-shortage...
It's no different in the AI bubble. RAM theft is just the first sign. And believe me, it's going to get worse.
wccftech.com/ram-shortage...
RAM Shortages Have Gone So Crazy That Thieves Are Now Actually Pulling Modules Out of PCs
RAM shortages are escalating to new extremes, and in a recent incident, a thief pulled up only to steal RAM modules from a PC.
wccftech.com
January 14, 2026 at 1:52 PM
Another strong sign of the end of the gold rush cycle is when resources become so scarce that outbreaks of theft and violence begin to erupt.
It's no different in the AI bubble. RAM theft is just the first sign. And believe me, it's going to get worse.
wccftech.com/ram-shortage...
It's no different in the AI bubble. RAM theft is just the first sign. And believe me, it's going to get worse.
wccftech.com/ram-shortage...
Researchers have demonstrated a monolithically integrated electronic-photonic WDM receiver, fabricated using GlobalFoundries 45CLO CMOS-SOI platform, that achieves an aggregate data-rate of 1.024Tb/s across 32 wavelength channels.
arxiv.org/pdf/2601.07709
arxiv.org/pdf/2601.07709
January 14, 2026 at 11:42 AM
Researchers have demonstrated a monolithically integrated electronic-photonic WDM receiver, fabricated using GlobalFoundries 45CLO CMOS-SOI platform, that achieves an aggregate data-rate of 1.024Tb/s across 32 wavelength channels.
arxiv.org/pdf/2601.07709
arxiv.org/pdf/2601.07709
In this paper is presented a chunked FFT convolution approach enabling 450K length sequence by 450K length filter convolutions on an Alveo U200 FPGA with 2.8 MB BRAM through chunking and overlap-add reconstruction. #FPGA
arxiv.org/pdf/2601.06065
arxiv.org/pdf/2601.06065
January 14, 2026 at 11:27 AM
In this paper is presented a chunked FFT convolution approach enabling 450K length sequence by 450K length filter convolutions on an Alveo U200 FPGA with 2.8 MB BRAM through chunking and overlap-add reconstruction. #FPGA
arxiv.org/pdf/2601.06065
arxiv.org/pdf/2601.06065
People are gradually discovering the most common antifragile strategy in the tech world: The refurbished market.
January 14, 2026 at 11:14 AM
People are gradually discovering the most common antifragile strategy in the tech world: The refurbished market.
For the first time, researchers have demonstrated a non-volatile field-programmable photonic gate array, implemented on a hybrid silicon-barium titanate platform.
arxiv.org/pdf/2601.07456
arxiv.org/pdf/2601.07456
January 14, 2026 at 2:50 AM
For the first time, researchers have demonstrated a non-volatile field-programmable photonic gate array, implemented on a hybrid silicon-barium titanate platform.
arxiv.org/pdf/2601.07456
arxiv.org/pdf/2601.07456
Researchers have proposed a novel non-volatile programmable photonic integrated circuit architecture utilizing MEMS with mechanical latching, enabling stable passive operation without any power connection once configured while maintaining high performance
arxiv.org/pdf/2601.06578
arxiv.org/pdf/2601.06578
January 14, 2026 at 2:35 AM
Researchers have proposed a novel non-volatile programmable photonic integrated circuit architecture utilizing MEMS with mechanical latching, enabling stable passive operation without any power connection once configured while maintaining high performance
arxiv.org/pdf/2601.06578
arxiv.org/pdf/2601.06578
It's not enough that the product offered is complete e-waste and an insult to the intelligence... Now Nvidia is doubling down on it and seeking maximum profit amidst the AI frenzy.
Remember this the next time you think about buying an NVIDIA product.
videocardz.com/newz/nvidia-...
Remember this the next time you think about buying an NVIDIA product.
videocardz.com/newz/nvidia-...
NVIDIA reportedly shifts RTX 50 supply toward RTX 5060 and 5060 Ti 8GB in 2026 - VideoCardz.com
NVIDIA said to be focusing on 8GB models now A new report from China claims NVIDIA adjusted its RTX 50 series supply plan for 2026 due to rising memory
videocardz.com
January 14, 2026 at 2:16 AM
It's not enough that the product offered is complete e-waste and an insult to the intelligence... Now Nvidia is doubling down on it and seeking maximum profit amidst the AI frenzy.
Remember this the next time you think about buying an NVIDIA product.
videocardz.com/newz/nvidia-...
Remember this the next time you think about buying an NVIDIA product.
videocardz.com/newz/nvidia-...
In this paper is presented Bgolearn, a comprehensive Python framework that makes Bayesian optimization accessible and practical for materials research, supporting both single-objective and multi-objective Bayesian optimization.
arxiv.org/pdf/2601.06820
arxiv.org/pdf/2601.06820
January 13, 2026 at 10:26 PM
In this paper is presented Bgolearn, a comprehensive Python framework that makes Bayesian optimization accessible and practical for materials research, supporting both single-objective and multi-objective Bayesian optimization.
arxiv.org/pdf/2601.06820
arxiv.org/pdf/2601.06820
