Viewing profile — jpgvm
jpgvm
HN member- Joined
- Thu, Mar 14, 2013, 9:21 PM UTC
- HN karma
- 9,914
- Public activity
- 2,283 items
- HN profile
- View on Hacker News ↗
About jpgvm
Principal Software Engineer @ NVIDIA
Opinions are mine and not that of my employer, etc.
Recent public activity
-
comment
Comment #49173712
Iran is on a level so much worse than even Afghanistan. In Afghanistan, there are some mountains, but they are in the middle. The Taliban fucked off into the mountains and they cou…
-
comment
Comment #49165811
> harder to do on other architectures that themselves aren't fully deterministic though The TLDR here is that Groq LPUs are completely deterministic hardware, including for floatin…
-
comment
Comment #49154898
They can be deterministic. We did this at Groq, if you sent a request with exactly the same input token, seed and temperature value you would get precisely the same result every ti…
-
comment
Comment #49141641
If you do good work you should at least take the time to review the slop that details that work for slopiness. Otherwise it's hard to take it seriously. Especially the prefill sect…
-
comment
Comment #49106222
Damn I am not the only one then. Also entirely on STM32 now too which is rock solid and the easiest embedded programming has ever felt. probe-rs, defmt, embassy w/async HAL. Life i…
-
comment
Comment #49070405
Definitely wash the panels. However if you are using a string inverter you will want to wash all the ones on the same string in one go if you can, you don't really want them genera…
-
comment
Comment #48695063
If you want a massive MacBook anyway then it's great. They are decent for local LLMs, awesome for local image models and it's a MacBook so AppleCare+ has your back. IMO it's a no b…
-
comment
Comment #48665787
So from what little I know of the history there were some serious personal issues going on in the team that led to ~7 of the 11-12 folk on Quake eventually leaving. Sandy has gone …
-
comment
Comment #48650110
This is honestly the biggest battle with AI driven development right now. You have these extremely potent tools that can output a ton of really great code if they are wielded corre…
-
comment
Comment #48178636
So I would say the most important thing is that the APIs these are using as in mlx5 DevX (essentially direct fw access) or ibverbs are exactly the same regardless if it's CPU or GP…
-
comment
Comment #48177036
Yeah pretty much this. I would separate the knowledge into maybe 3 distinct buckets. The baseline: device/host boundary, SIMT programming etc. The intermediate: kernel architecture…
-
comment
Comment #48105001
This is a fairly crap post and the reasoning isn't sound but somehow the conclusion is still somewhat correct. You do want to use Rust with LLMs. The reason you want it is simple, …
-
comment
Comment #47955257
Don't worry though guys, climate change isn't real. /s 1200 years is a serious timescale, I think humans generally struggle reasoning about long durations or very vast distances. W…
-
comment
Comment #47947102
Github Actions is a decidedly unserious product, used largely by unserious people. It's always been poo, the YAML is bad, the reliability is bad and the cost is bad. So there is re…
-
comment
Comment #47657489
No. They have been able to destroy much more important and hardened targets, i.e multiple AN/TPY-2 radars, the extremely valuable AN/FPS-132 radar in Qatar, $Bs worth of aircraft o…
-
comment
Comment #47635455
Pre-war views were very much the status-quo was better than starting a war. Now that a war is started it has to be finished or the GCC is left far worse off with Iran in a much str…
-
comment
Comment #47494472
Thankyou for your attention to this matter!
-
comment
Comment #47246134
Sort of? My journey up from Senior/Staff was mostly on the back of proposing and then leading cross-functional projects. Which means I was doing Principal work before being granted…
-
comment
Comment #47083914
VictoriaMetrics stack. Better, cheaper, faster queries, more k8s native, etc. Easy to run with budget saved from not being on Datadog + attracts smart and observability minded engi…
-
comment
Comment #47057209
> These are not top tier talent type people, this is work that my mom could do, but Apple can pay much less by bringing people over from India, Pakistan, China to do this work inst…
-
comment
Comment #47057145
Bar was lowered massively during the COVID employment boom (with the exception of Apple really). Unfortunately resetting the bar has massive collateral damage but it will sort itse…
-
comment
Comment #47057125
I interpreted this: > I don’t see any notable difference between immigrant and non-immigrant As your main point. More educated or not doesn't actually matter, whether they meet the…
-
comment
Comment #47051723
> what % of this labor could be sourced domestically and what actually needs to be imported? None of it? The way I see it is every top tier programmer in America is already employe…
-
comment
Comment #47051436
My argument is that the H1Bs meet the bar to be hired at their respective companies. Neither pool is inherently better but availability matters. There are only so many American eng…
-
comment
Comment #47051351
You are proving my point - FAANG hires the top tier talent, and makes extensive use of both the global talent pool and sponsoring immigration in order to meet their needs. The Amer…