07.03.2026 14:10
👍 1
🔁 0
💬 0
📌 0
Cuda. I tried a long time ago to build something with opencl and found out that there is no support. AMD has had plenty of time to work with developers since then but nothing has happened.
I guess NVIDIA figured out that Ballmer was right, it’s all about the developers
Coming next week
ok, that figure looked fine in the preview but was transparent, here's a better version
The only downside to this work is that now I feel frustrated having to wait minutes for classical kallisto to finish
like dealing with sparse arrays, variable sized lists and things we would solve with dynamic memory allocation on the CPU.
and over again. Prefix scan. If you are interested in learning more about this then read this paper www.cs.cmu.edu/~guyb/papers... by Guy Blelloch from 1990. It is one of the most clearly written papers I've read and it gives you the algorithmic building blocks to solve problems on GPUs
Contrary to regular scientific programming, where the state consist of a high dimensional vector and most operations are matrix multiplication, string processing does not have obvious programming paradigms that port easily to GPUs. And yet in this work we always wind up using the same tool over ...
The key insight (aside from the speed) is that we need to reconsider all of the algorithmic framework if we want to use GPUs for large scale processing of sequencing datasets. It's nontrivial but I hope this paper gives some insight into how it's possible.
For a large dataset (295M reads) the GPU version took 50 seconds. Just running zcat on the files takes 10 minutes! Decompression and parsing FASTQ is a major bottleneck . Instead of using kseq we moved this work parsing to the GPU which delivers amazing throughput.
Figure 1 shows the main results, we can run 30M paired end reads in under 10 seconds using a RTX 5090 NVIDIA GPU. The average speedup is 30x for smaller reads and that includes startup time, on average we can process about 3.6M paired end reads per second.
Excited to share this preprint that describes my latest work on using GPUs to accelerate processing of RNA-seq data.
The title says it all: "RNA-seq analysis in seconds using GPUs" now on biorxiv www.biorxiv.org/content/10.6... and github github.com/pachterlab/k...
Figure 1 shows they key result
2015: growing almonds consumes too much energy and water
2026: growing kids consumes too much energy and eater
bsky.app/profile/x2y....
Jafngóð og bókin?
In pretty sure I’m thoroughly radiated after this
I’ve been exploring different models in cursor and opus 4.5 is amazing. I often ask it to sketch out code with the prompt “do not make any modifications to the code” and ask about my assumptions and poke holes, only when I agree can it proceed
Spotify wrap up is so last year
I also learned today that Nirvana is not “niche”. But then again this is the generation that invented 6-7
14yo: I need a niche music recommendation
Me: how about Godspeed You! Black Emperor, let’s find it
*not found on Spotify*
Me: That’s how nice they are
Today I found out that the Pacbio tools for working with barcodes and adapters in Kinnex are called lima and skera. To whomever came up with that joke, I salute you, you’ve reached 50% of the audience
Retracting a paper is going to hurt
All right, I posed this question to my PhD student this morning and he looked at me funny and asked if it was a trick question
What is the running time of this code
a, b = 0, 1
for _ in range(2, n+1):
a, b = b, a + b
return b
Happy Q4 for those who celebrate
Today I fixed a humming converter and an intercom receiver set from the 70’s. I’m surprised at how good ChatGPT is at diagnosing problems and coming up with fixes for hardware that is older than me. Since I didn’t have to replace the units I can justify the cost and environmental impact of AI
The Genome Informatics conference (@ Cold Spring Harbor Lab, Nov 5 - 8) abstract deadline is **today**. We welcome your submissions! Topics include:
- PanGenomes
- Genome Assembly & Seq. Algos.
- Algorithmic Evo. Bio
- Single Cell & Spatial Omics
- Microbial Genomics
- AI/ML & Integrative Omics
🙏🙏🙏
Perfect marketing doesn’t exi…
Glengarry Glen Ross
It is by the juice of coffee that thoughts acquire speed …
Job alert: Join us for a postdoc in AI in genetics at @EMBL Heidelberg! Great collaboration with @Adrian Cortes @GSK, aiming to develop new tools to elucidate genetic effects using population-scale cohorts and single-cell readouts. Please share!
embl.wd103.myworkdayjobs.com/de-DE/EMBL/d...
