zamalek

35 ms·

721.
by zamalek 2y ago
This has been my experience too, I didn't even know that gzip could be tuned to be faster than zstd.
722.
by zamalek 2y ago
The advantage that CRDTs have is they have a complete replay. In theory you could do some clever stuff with that.
723.
by zamalek 2y ago
I think this happens because the mathematical, or causal, or entropic, notion of conflicts has been conflated with semantic conflicts. In the past I have made the same mistake, though inversely and was adamantly informed that I had no clue
724.
by zamalek 2y ago
Modals that blocks focus to an entire browser window aren't really a good idea (I'm of the opinion that they are almost always a shitty idea, but that's harder to argue). People have multiple tabs open, and what if another
725.
by zamalek 2y ago
> Depends what quality of life means to you right? I recently got laid off for the 2nd time in 4 years (probably better off than most, to be honest). I'm done with shooting for the moon. I am now strongly considering a "boring&
726.
by zamalek 2y ago
Momentum wins over superior solutions all the time. Just like SMTP, I feel like that tabs vs. spaces is a lost war.
727.
by zamalek 2y ago
> why would it be justified to dismiss a device without it as a "toy"? Which is exactly why I italicized in theory.
728.
by zamalek 2y ago
Depending on who you ask, Intel successfully conned the industry into believing that ECC isn't required for consumers - so that they could put a price on it. If you care about what you are doing in your PC in theory you should really
729.
by zamalek 2y ago
It's possible that the capture wasn't taken at the exact same frame, or that the state of the light isn't deterministic in the benchmark.
730.
by zamalek 2y ago
I think a better idea would be an NPU with slower memory, or tie it to the system DDR. I don't think consumer inference (possibly even training) applications would need the memory bandwidth offered by GDDR/HBM. Inference on my 795
731.
by zamalek 2y ago
If it works well on Linux there's a market for that. AMD are hinting that they will be focusing on iGPUs going forward (all power to them, their iGPUs are unmatched and NVIDIA is dominating dGPU). Intel might be the savior we need. Wel
732.
by zamalek 2y ago
That goes for any business/employer for what it's worth.
733.
by zamalek 2y ago
https://arxiv.org/abs/2410.05229 An 18% drop in accuracy (figure 8) is not insignificant. Even 4o suffered 10% loss (figure 6), and 4o isn't a small llm. Competent performance should have near zero performance l
734.
by zamalek 2y ago
DDD advocates for creating a shared lingo across everyone involved, including customers if they are highly embedded in the design process. This I agree with. Ultimately DDD attempts to create real-world analogies in code; you know, dog inh
735.
by zamalek 2y ago
In ANNs pruning helps prevent over-fitting. With the discovery that transformers lack reasoning capabilities this research really comes at a great time. It's a miniscule chance, but we might see this improve performance over the long t
736.
by zamalek 2y ago
Open source. I have learned a lot from all over the place, but dotnet comes heads and shoulders above the rest - those people truly take their time with you. I also contributed to CoreDNS as a learning exercise for Go, and they were notably
737.
by zamalek 2y ago
Some actor frameworks (Erlang, Orleans) would interpret each order being an actor instance, and each order line being an actor instance. To make that clearer: each order row would _logically_ have an actor running somewhere in your cluster.
738.
by zamalek 2y ago
Somewhat of a tangent: As much as us engineers hate it, I consider the fact that Excel monstrosities are commonplace high praise for the product. Being hated by engineers because your product is so common is exactly where you want to be (do
739.
by zamalek 2y ago
I call it "SQL fear", and I agree: people need to get over it. We used EFCore at my previous job, and I don't know how much time figuring out how poorly performing queries are happening. The abstraction leaks like a sieve and
740.
by zamalek 2y ago
7900XT here, previously 6900XT. Support for 6000 series and beyond is great on Linux. You have to use an override envar (Ollama has info on this in their readme). ComfyUI has instructions for anything based on torch. TensorFlow is unusable.
741.
by zamalek 2y ago
If you're using a recent Electron. Nearly everything (Discord, Slack, you name it) does not. This is all from memory, but I believe Wayland was "fixed" in Electron 19, but with a showstopper bug that was never backported. M
742.
by zamalek 2y ago
Amazon uses Hackerrank and explicitly says not to use LLMs. In that case it would be cheating. However, given that everyone is apparently using it, I now feel dumb for not doing so.
743.
by zamalek 2y ago
As I understand the situation with mosquito eradication, there are highly specific places where mosquitos are a critical source of biomass (I think the Arctic tundra was one example). In all other places they are completely and utterly wort
744.
by zamalek 2y ago
> Relatively modern, capable tech corporations that understand that their code is an asset. I strongly disagree with this. The more code you have, the more resources you have to spend maintaining it. There is a very relevant example clos
745.
by zamalek 2y ago
It's normally incredibly difficult for employees to disrupt at massive companies that would be the type which runs a data center. Disruption usually enters the corp in a sales deck, much like the one Oxide would have. It's stupid,
746.
by zamalek 2y ago
I wonder whether doing a catch without the catcher (rapid scheduled crash landing) would be feasible. Data is data.
747.
by zamalek 2y ago
Were you using WSL1? It did have some areas that needed improvement (which is what prompted to Microsoft to replace it).
748.
by zamalek 2y ago
From the archive link (blog post published 2018): > To our surprise, we found a lack of published, well-optimized, large-data hash functions. Murmur was released in 2008. Murmur3 in 2011. Release dates on higher quality functions are not
749.
by zamalek 2y ago
I remember our biggest issue being IP addresses. We had no router, or expertise, so we were at the whims of automatic addresses (254.x... as far as I recall?). Good times.
750.
by zamalek 2y ago
> person with a heavy accent Nice... Either way, a human is not a robot. Ignoring the do not call list would be up to the FTC to police. https://www.donotcall.gov/report.html A personal trick of mine was to answer those

More ›