Tootfinder

Initial curve25519 accelerator refactoring update: added new MULT_AREA_OPT parameter.
0 = existing implementation (best for 7 series)
1 = resource sharing between constant and variable multipliers, but still does 32 16x32 multiplies per clock. Much slower on 7 series due to DSP cascade, but hits the same 71 MHz Fmax on Trion.
Fabric usage on Trion is slightly higher (9528 -> 10613 LE) but multiplier block usage is down from 96 to 66.
Next step will be trying to fig…

Is there a way to get hardware accelerated #VP9 support on #Skylake era #intel CPUs? Best I can tell, there's an abandoned Intel-hybrid driver with partial support that Intel doesn't want to bring…

The role of preprints in open science: Accelerating knowledge transfer from science to technology
Zhiqi Wang, Yue Chen, Chun Yang
https://arxiv.org/abs/2506.20225

The role of preprints in open science: Accelerating knowledge transfer from science to technology
Preprints have become increasingly essential in the landscape of open science, facilitating not only the exchange of knowledge within the scientific community but also bridging the gap between science and technology. However, the impact of preprints on technological innovation, given their unreviewed nature, remains unclear. This study fills this gap by conducting a comprehensive scientometric analysis of patent citations to bioRxiv preprints submitted between 2013 and 2021, measuring and acces…

A causality inspired acceleration method for the fast temporal superposition of the finite line source solutions
Marc Basquens, Alberto Lazzarotto
https://arxiv.org/abs/2507.18200

A causality inspired acceleration method for the fast temporal superposition of the finite line source solutions
We present a novel, fast method to compute thermal interactions in solids, useful for time-dependent problems involving several sources and several time and space scales such as the ones encountered in the physics of fields of closed loop borehole heat exchangers. The new method is based on the non-history temporal superposition acceleration algorithm, but presents better performance compared to the originally proposed scheme. The main idea behind it is to leverage the propagation properties of…

Tonight's quick ngscopeclient dev fix: GPU accelerating the demo scope.
It now runs quite a few times faster than before (and is faster to process subsequent filter blocks on, since the input data is now GPU resident).
Not super critical but a nice quality-of-life fix since I use the demo scope as a data source for development pretty often.

https://techcrunch.com/2025/07/07/openai-tightens-the-screws-on-security-to-keep-away-prying-eyes/
OpenAI tightens the screws on security to keep away prying eyes

OpenAI tightens the screws on security to keep away prying eyes | TechCrunch
OpenAI has reportedly overhauled its security operations to protect against corporate espionage. According to the Financial Times, the company accelerated

A detailed simulation model for fifth generation district heating and cooling networks with seasonal latent storage evaluated on field data
Manuel Kollmar, Adrian B\"urger, Markus Bohlayer, Angelika Altmann-Dieses, Marco Braun, Moritz Diehl
https://arxiv.org/abs/2506.18528

A detailed simulation model for fifth generation district heating and cooling networks with seasonal latent storage evaluated on field data
Fifth generation district heating and cooling (5GDHC) networks accelerate the use of renewable energies in the heating sector and enable flexible, efficient and future-proof heating and cooling supply via a single network. Due to their low temperature level and high integration of renewables, 5GDHC systems pose new challenges for the modeling of these networks in order to simulate and test operational strategies. A particular feature is the use of uninsulated pipes, which allow energy exchange …

Exploring Trion more as a potential FPGA option for future embedded projects.
The first thing that's hitting me hard is the lack of LUTRAM mode. Anything too small to put in a BRAM, or that you need combinatorial reads on, ends up being synthesized as DFFs.
This causes massive size explosion (and worse performance) for some of my stuff, most notably the curve25519 accelerator which balloons from 7632 LUT / 5737 FF / 32 DSP on Kintex-7 to 21625 LUT / 11528 FF / 96 MULT when sy…

Starting to think about GPU acceleration of protocol decodes (rather than just basic math blocks) in ngscopeclient.
Here's the filter graph I'm thinking of using as a benchmark: dual lane QSGMII, 20M points per channel.
The filter graph takes 944 ms end to end to run on my box (2x Xeon 6144 2080 Ti).
Major time consumers:
* Eye pattern (~345 ms)
* QSGMII (~337 ms)
* CDR PLL (~310 ms)
* 8B10B (~160 ms)
* SGMII (~120 ms)
Note that the …

Turns out the reason that the GPU accelerated level detector made the CDR slower is that I was working on test data loaded from a file, and not moving the data from CPU to GPU before starting the filter graph. So of *course* it was slower, due to the unnecessary data movement.
But that's unrealistic since on a real high-performance scope driver (e.g. thunderscope) the first thing we typically do is push data from CPU to GPU to do the int-float conversion there.
Time to repea…