tristor

6 ms·

31.
by tristor 2mo ago
RDMA is buggy and Thunderbolt only delivers 1/10th the throughput of native connectivity. 1TB of Unified Memory w/ 1.2TB/s of bandwidth with marginally ~$30k cost is a different story than 1TB of sorta Unified Memory w/
32.
by tristor 2mo ago
I would probably spend up to $30k if I could get 1TB of Unified Memory, because it would allow me a guarantee to run pretty much any local model I want, including >1T parameter models with reasonable quants. I wouldn't be surprised
33.
by tristor 2mo ago
It's a few different things. I've gone down a hardcore rabbit hole on this because I also find this exceptionally annoying. I now sometimes experience lag when typing into a terminal, where I am typing faster than characters app
34.
by tristor 2mo ago
Kinda. Strix Halo does 256GB/s of memory bandwidth, and is significantly slower than an M5 Max (614GB/s). Feels like intentional market segmentation?
35.
by tristor 2mo ago
I wish they were offering 1TB of Unified Memory for the M5 Ultra. I already have an M5 Max MBP w/ 128GB of RAM for running local models, and while there's a /few/ models that I can run in 512GB that I can't run in
36.
by tristor 2mo ago
Nice. I hope we'll see more support for jxl across the web. It's so good at this point that it's my primary output/archival format for all of my photography.
37.
by tristor 2mo ago
I found that Claude / Sol were basically useless when approaching various CTFs that allowed AI tools, but running Qwen and DeepSeek locally and GLM via OpenRouter worked fine. It's blatantly obvious to anyone that even does tange
38.
by tristor 2mo ago
At the risk of sounding arrogant, I will out myself as being incredibly smart and also incredibly unhappy in a general sense (although content with my life). The answer is very simple and mundane: I observe and understand problems that ha
39.
by tristor 2mo ago
I've had some rough moments with my Framework 13, but overall I appreciate it because when something like this happens you have a way out other than binning the entire laptop. I agree with the author that Framework should have a bette
40.
by tristor 2mo ago
I have a lot of hopes for Reticulum, but my experience has been that it is not widely deployed and really requires network-connected LoRa nodes to be worthwhile. As a pure LoRa protocol, I think MeshCore and Meshtastic are more useful, and
41.
by tristor 2mo ago
This is interesting, but is 3kW enough to be useful? What is the latest when it comes to home batteries? Has management gotten better to be smarter?
42.
by tristor 2mo ago
I'm hoping to see folks distill this with current generation Opus / Fable reasoning traces. I have had my best results locally so far from Qwopus (Qwen 3.6-27B w/ Opus 4.6 reasoning distilled). This looks GREAT and I am def
43.
by tristor 2mo ago
This is correct, and the extra insidious part is that folks who are tech-savvy and privacy-conscious are more valuable profiles than normies, because we generally work in higher income jobs/industries. So whether we click or not is ir
44.
by tristor 2mo ago
I do use this connected to a UDM Pro Max, but it's not really relevant what it goes to because the XGSPON does all of the heavy lifting, the device you insert the SFP into essentially treats it like an Ethernet connection.
45.
by tristor 2mo ago
As I noted in my other comment, AT&T Fiber doesn't actually using PPPoE, but anyway, glad to see you linked pon.wiki here. I am running XGSPON ONU stick (ONU on SFP) with the community firmware for my AT&T Fiber connection, ge
46.
by tristor 2mo ago
Quantum Fiber doesn't use PPPoE, it uses IPoE w/ VLAN tagging using VLAN 201, which is a common approach for legacy GPON networks. CenturyLink Fiber does use PPPoE, but I believe only the authentication portion of PPPoE, but I ha
47.
by tristor 2mo ago
I'm mostly replying to this line "Even in 2026, a huge number of ISPs still rely on PPPoE, including Bell Canada, AT&T Fiber, and Xfinity in the US, along with the vast majority of ISPs in Europe, Asia, and China."
48.
by tristor 2mo ago
PPPoE is pretty much only used for DSL in the US. I'm not sure about elsewhere. Xfinity and AT&T Fiber do /not/ use PPPoE. Xfinity is built on DOCSIS, and AT&T Fiber is built on XGSPON. This means neither one relies
49.
by tristor 3mo ago
I like energy drinks, but I prefer coffee. I drink my coffee black. About the only time I drink an energy drink for these past few years is when driving long distances. I don't like the chemicals used in sugar free sodas/energy
50.
by tristor 3mo ago
Since I'm on the Product side these days, I don't spend much time delving into the codebase at this point in my career. I can believe it though, as I've done quite a few personal side projects using langchain and various har
51.
by tristor 3mo ago
My observation is that AI tools generally make people /less/ productive. They are more productive in that they produce more stuff, but they are less productive in that the stuff they are producing has lower inherent value deliver
52.
by tristor 3mo ago
I have tried to switch to Darktable multiple times over the years, it never has worked for me properly in my workflows. I'm still solidly on Lightroom Classic + DxO Photo Lab + Photomatix/Zerene Stacker (for stacking/HDR&#x2
53.
by tristor 3mo ago
> There's a difference between polite pushback and being straight up disrespectful. That's a distinction without a difference. Cops see any form of noncompliance as disrespectful ("contempt of cop"). You not only le
54.
by tristor 3mo ago
The UK, per capita, is poorer than every single state in the US. https://finance.yahoo.com/economy/articles/gdp-alert-united-... As much as the current situation in the US upsets me, anyone who has spent considera
55.
by tristor 3mo ago
Companies really ought to pay for business class on transatlantic and transoceanic flights because of well-founded research showing that a seated position without movement for long periods of time creates a risk for DVT/stroke and that
56.
by tristor 3mo ago
The people who buy a Framework Desktop with 192GB of RAM (or in my case an M5 Max MBP with 128GB of RAM). I don't know what argument you're trying to make here? You are saying this product is pointless, the point of this product
57.
by tristor 3mo ago
Can it run a local LLM with 35B+ parameters using GPU acceleration for reasonable token performance? If not, then you're being obtuse.
58.
by tristor 3mo ago
You fundamentally cannot do things on the system you are envisioning that you can do on a Framework Desktop, or DGX Spark, or MBP M5 Max. Simple as that. Repairable modular laptops were a fairly rare thing when Framework came out. A repa
59.
by tristor 3mo ago
There is no socketable Unified Memory platform. The entire point of this SoC is that it has Unified Memory. The Framework Desktop has a Mini-ITX board, but you MUST use a solderable SoC with HBM stacked memory to get a Unified Memory outc
60.
by tristor 3mo ago
> IME the people I know who embody great craft in a field… they care about it in everything they do, and they don’t do a thing unless they can do it with care. Fully agreed. It's one reason I've taken to doing more things mysel

More ›