QCC Echo Kernel: Passed Pauli-Error Stress Testing with 0.00% Entropy Leakage

Pushing the boundaries of efficient #LLM inference on #RISC-V! I am excited to unveil our latest work on accelerating LLM inference with hardware-aware Q4X quantization and custom kernels integrated in Llama.cpp. And yes we are showing performance gains on a real hardware (Milk-V Jupiter RISC-V board). Give our white paper a read here: https://lnkd.in/dfu-EX9B #LLM #EdgeAI #AIInference #RISC_V #LlamaCpp #AIHardware #Quantization #MachineLearning

We recently published a blog on Q4X, our compact 4-bit codebook-based quantization strategy, now integrated into LLaMa.cpp and optimized for RISC-V. Q4X maintains a tiny 64-entry codebook for dequantization at runtime, which simplifies execution compared to Llama.cpp’s default Q_K (with its multi-level scaling and mixed quantization). The result: 1. ~4% smaller model sizes 2. Faster tokens/sec throughput 3. A slight trade-off in perplexity By leveraging a simple lookup-table design, Q4X achieves efficiency without overcomplicating runtime. We are now actively working on expanding Q4X to other architectures. Check out the blog for further details: https://lnkd.in/dfu-EX9B

🚀 Allegro EU Project at Optica Quantum 2.0! We're proud to spotlight our recent contribution: “Analyzing Quantum Secure Direct Communication with Forward Error Correction” by M. Iqbal et al. This study applies FEC to a quantum secure direct communication system using a QPSK-like scheme. Through simulations on IBMQ and NetSquid, we found that forward error correction alone cannot eliminate errors unless qubit fidelity exceeds 0.981. 🔗 Learn more: https://lnkd.in/ekCGqMT2 #QuantumCommunication #QSDC #FEC #OpticaQuantum2024 #AllegroProject

Excited to introduce OpenThermo to you. OpenThermo is written in C++ (for high performance) to calculate thermochemical properties. The program can calculate thermochemical data from outputs of several quantum chemical programs such as Gaussian, Orca, NWCHEM, CP2K, GAMESS-US. Just specify your preferred temperature, concentration, pressure and the program will do the rest for you. Symmetry is fully supported; several advanced methods are implemented to treat low-frequency vibrational modes. Check it out 😉 https://lnkd.in/gfHibp_J

If you want to learn about superconducting qubits, I recommend this 2019 APR guide.

Yesterday I had a bit free time and I built a quantum multiplier simulator that follows established arithmetic designs (Cuccaro adder, shift-and-add multiplication) and provides detailed circuit metrics. Key features: - Basis-state tracking for fast measurement - Memory-aware allocation with warnings - Clear separation of logical vs. physical qubits via MCX decomposition - QASM export for use with Qiskit / Cirq The simulator reproduces correct results across test cases and exposes gate counts, depth, and qubit usage. While limited by state-vector scaling, it’s a practical tool for analyzing quantum arithmetic circuits. Code is available here: 👉 https://lnkd.in/dX834Dfh Have fun with exploring it 😉

When a nanofluid or a particle laden droplet evaporates it tends to form a shell like structure at the interface. Now if such a droplet while evaporating encounters a shock, the breakup mechanism is quite startling. Check the video. VC: Gautham. Paper available in ARXIV

📅 Day 9 – Quantum Gates & Circuits 📖 What I Learned Pauli Gates – Explored the X, Y, and Z gates as fundamental building blocks for single-qubit operations and their role in state transformations. Hadamard Gate – Understood how the Hadamard gate creates superposition states, enabling the foundation of quantum parallelism. Phase Gate – Learned how phase operations alter the relative phase of qubit states, crucial for quantum interference. CNOT Gate – Studied the controlled-NOT gate as a key two-qubit gate used to generate entanglement and perform conditional operations. Unitary Transformations – Gained clarity on how quantum gates represent unitary transformations that preserve probability and govern quantum circuit dynamics. ✅ Day 9 Challenge Completed #QuantumGates #QuantumCircuits #PauliGates #HadamardGate #PhaseGate #CNOTGate #UnitaryTransformations #QuantumComputing #PhysicsLearning #21DaysChallenge #QuCode

Happy to report on the first results from the Quantum Software Lab, School of Informatics, The University of Edinburgh collaboration with NVIDIA and Supermicro as part of the Grace Hopper Super Chip Seed Programme. We implement an ensemble decoder based on graph automorphisms using the CUDA-Q QEC library. See more details in Stergios Koutsioumpas's post below and in our blog post: https://lnkd.in/gj_DaHj2

👀LOOK: QIA member from QuTech Samuel Oslovich presenting compilation strategies for #quantumnetwork programs using Qoala in the recently held IEEE Quantum Week 2025 in New Mexico USA Here he discussed quantum network programs and possible steps a compiler can take to improve their performance. Specifically, he explored some of the different strategies a compiler can use to optimise quantum network programs for execution on Qoalam, an extension of QNodeOS which is the first operating system designed by QIA researchers for quantum networks.