← Back to context

Comment by noduerme

9 hours ago

What's the big deal with having a whole liquid cooled workstation, and why is it important information for me to know what this dude's hardware is? And seriously, is there something about the rig that is necessary to chew through a dataset with a few million rows?

Liquid-cooled computers have one major benefit; usually, your computer ages over time, and there's a long period where it's still barely fast enough but you wish you had something nicer. A liquid-cooled workstation prevents you from needing to manage this grey area by catastrophically failing at unexpected intervals.

  • Also prevents you from messing with it too much, as any substantial change requires draining and refilling your loop.

  • Had me in the first half.

    I looked at using an AIO for my PC build but ultimately went with an air cooler the size of a damned rubix cube and a high airflow case.

    My room gets toasty with raytracing titles, lol

    • Wouldn't your room get equally/more toasty with liquid cooling? That heat has to get dumped somewhere, and liquid would theoretically be more effective at dumping it into your room.

      2 replies →

  • I got an Aigo AIO (AC SE 240) off of AliExpress and use it as an automated reminder that my system needs an upgrade: once it stops working (with an upper bound of maybe 4-5 years), I'll know that it's time! Didn't even need to pay extra for that feature!

He just does this with all his blog posts, don't overthink it. The tech industry is full of people with unexpected quirks.

  • We need more of this, not less. This is Hacker News. He gave us exactly what we need to know to exactly replicate his results.

    • I think it feels a little bit of an Ad for the hardware, especially the way he describes the case, telling you the exact model and how spacious it is. Bit sus but perhaps he is being OVERLY detailed and just likes telling you he has a bunch of CPU's that are well cooled in a case with two big ugly fans on the front (not into that look at all.)

      Though I can totally understand, geeky people love details. I have a habit of getting way too detailed in my writings here. So I then spend most of the time editing it down to be as clear and brief as possible. I refuse to use an LLM for my own thoughts.

    • That's how I took it too. You always provide hardware information when publishing any data set that takes a long time to compile.

I really don't think we should be shaming computer enthusiasts for being enthusiastic about their computers on HN of all places

The 9950X is an excellent CPU at a reasonable price point and works perfectly fine with an ordinary air-cooled heat sink in an ordinary case without stupid numbers of fans. The TDP is just not that high.

source: my 9950X, happily running air cooled.

(Embarrassingly, I have an M4 Max that can almost match it in the CPU-bound workload I care about while sipping some 45W. The rest of the industry really needs to catch up with Apple on power efficiency.)

Obviously he's telling you their spec incase you wish to reproduce his results. Why don't you try it and tell us how your result compares.

If you're not familiar with his blog, he also occasionally does benchmarks of databases & extensions. He starts every post by outlining his setup.

> 96 GB of DDR5 RAM

Most people drive cars worth less than this.

Why is the top comment criticising a geek for being a geek? He gave us a wealth of information including his exact methodology and queries on how he produced his results. This is an ideal approach. You want just results and "trust me, bro"?

My initial thought was that was a weird choice in this article, but I wouldn't fault someone for being thorough.

Probably a better choice as an appendix, move the good stuff up to the top. But overall its NBD.

I had the exact same thought, particularly when I read there were fewer than 4M records.

I really have to wonder if people truly know how powerful any modern computer is. Like I just assume any modern PC with sufficient storage can handle a database with a billion rows of data. I think my phone probably could.

Now if you were, say, analyzing commercial satellite imagery of the entire US and trying to find rooftop solar, matching it against the database and finding data that wasn't in the dataset, that's something where your computer power would be way more relevant.

Come to think of it, you could probably use such imagery to construct a power generation network from power plants to transmission lines to utility poles. Of course some places have underground cables but there are other datasets for that.

Another interesting project is mapping the growth of solar. This would require access to commercial satellite imagery over time. I'm sure some government agency already does it. Or used to at least. Snapshots years or even months apart are less interesting.

Anyway, I guess the point is the author's computer is capable of way more than I suspect they think it is.

  • > than I suspect they think it is.

    Because he wants to tell you about his computer it means he doesn’t know how capable it is?

  • I always make sure to downgrade my computer hardware before running a trivial analysis. Every dataset needs to redline the current configuration.

It's funny how I started skimming as soon as I saw "My Workstation" without ever consciously perceiving why I had started hitting Page Down, until you mentioned it and I went back to notice what it said there. My brain has really automated web page signal extraction.