jasomill

That's an 8GB card. For gaming (DLSS, etc.), you can still get newer 8GB cards like the 3070 which represent solid upgrades over the 1070 Ti on the used market for under $300. Still not great since similar cards were available at similar prices new a couple years ago, but at least not trading at multiples of MSRP like higher-end cards.

How would DLSS, which runs on GPU hardware — and specifically, as I understand it, GPU hardware that accelerates AI inference — eliminate the GPU?

And if a system is intended to be modular and upgradable, what would combining the GPU and CPU into a single module contribute to this goal?

Now an architecture that separates the memory controller from the CPU and gives pluggable CPUs and GPUs (and storage and network controllers, for that matter) equal access to a single pool of RAM in a secure and high-performance way, that would certainly be interesting, and not only or even primarily for gaming.

Only if you don't intend to maintain, debug, or even understand what the code is doing.

Or at least not without the benefit of more capable coding agents than you used to create it.

And only if the coding agents in question can be trusted to not deliberately or accidentally fail to provide you with critical information ("Now that you mention it, the program does occasionally POST the contents of ~/.aws/credentials to a web server. As this produces no output, I didn't consider it relevant to your question about whether it correctly outputs the string 'hello world'.").

Unplugging a PC containing a Maxwell+ card doesn't properly set its clock speed.

Even if this is an intentional nerf by Nvidia, which I doubt, then then nerf is buggy.

The minimum configuration appears to be 24GB / 18 core, which at least one vendor, Microsoft, is selling for $2,599.99, which is at least closer to $2,399 than $3,000+.

If datacenters aren't lining up to buy 64GB DGX Sparks for $5,000, what makes you think they'd be interested in 24GB RTX Sparks with fewer cores and less attractive form factors for almost half that?

And if there's a market for a DGX Spark configuration with a higher compute-to-RAM ratio, why isn't Nvidia offering it?

My best guess is that the entry-level Sparks are mostly intended for people who would otherwise be in the market for a traditional high-end laptop, who might be enticed to spend a few hundred more for at least the promise of local AI, while the higher-end models are intended for more serious AI development.

With that said, it's not hard to imagine there isn't nearly as large a built-in market for high-end SFF Linux PCs as there is for high-end Windows laptops.

If this were cost-effective, why wouldn't they just buy DGX Sparks, with similar performance at similar price points and a more convenient form factor?

Or RTX Spark and run their workloads under WSL, which as the article notes, has minimal overhead relative to running Linux directly?

This sounds more like a bug than a feature, especially given that Linux on battery runs at full clock speed and downclocks when plugged in, and Windows does the opposite, suggesting they could at least mitigate the problem with no changes to their software stack by inverting a conditional in the firmware and adding a Linux kernel parameter to control memory allocation as apparently already exists for Windows.

Also various versions of OS/2, or for something completely different, iRMX (PC versions that co-install with DOS or Windows).

For NeXT OSes, OPENSTEP 4.x also ran on x86.

I can vouch for this: whenever there's heavy snowfall (in Indianapolis, so less common and severe than Michigan, but same ballpark), nearly all cars I see stuck in snowbanks are AWD/4WD.

Personally, I'm comfortable driving anything but a RWD pick-up truck with no payload in snow and ice, try to drive at an appropriate speed for weather conditions, and rarely have trouble.

My 2012 Chrysler 200 has a similar analog clock, plus a digital clock in the radio.

The analog clock is accurate enough that I don't fuss with it other than to reset it for DST (I have no similar children).

My solution to the lack of synchronization: turn off the radio clock (which is fortunately an option).

Let's get rid of time because it kills people.

I still choose the "America/New_York" / "Eastern Time (U.S. and Canada)" time zones when setting up systems in Indianapolis because "America/Indiana/Indianapolis" / "Indiana (East)" adds nothing but the possibility of error.

Indiana started observing DST in 2006. Bad memories die hard.

The best launcher, unless you’re using something like Steam Big Picture mode with a controller, is the Windows Start Menu.

Preferably using shortcuts that launch game executables directly, not through a launcher (or, worse, two or more launchers in sequence, as is the case with, e.g., EA games launched through Steam and games with their own launchers offering equal access to the game itself and other options you’re likely to use once or twice, if ever).

But in the absence of this, leaving one bloated storefront launcher running in the background is better than leaving two or more bloated storefront launchers open in the background, and Steam’s is the best of a bad lot.

Apple's Thunderbolt FireWire 800 adapter daisy-chained with their TB2-to-TB3 adapter should work on any modern system with Thunderbolt and FireWire support, as it just presents as a bog-standard PCI FireWire card.

So worst-case, you should be able to access your old drives on a modern Mac by booting from a FireWire-compatible OS (Asahi or older macOS) on an external drive.

If history has taught us anything, it's that there's at least a limited-time market for inferior GUI alternatives to nearly anything you can easily do from a command-line. Though to be fair, there's plenty of room for value-add over curl for web requests in terms of presentation where GUIs can work well: asset viewers, syntax highlighting, pretty printing, code folding, etc., especially if combined with interactive tools for decoding and deobfuscation that feed back into the viewer.

The obvious AI integration is reverse engineering tools, assuming the trend towards locking such things down doesn't continue.

And honestly, I worry less about people paying $29 for a GUI file renaming tool than I do about the same people copy/pasting LLM-generated shell snippets, given that LLMs will inevitably have trained on the existing corpus of shell scripts, many if not most of which contain potentially dangerous bugs, especially when used outside their intended context of known-trustworthy input.

I'm not sure why I'd trust a site that gives OpenAI and example.com equal marks as "already dead" because they're "vulnerable [...] to AI-native replacement" for anything but a laugh.

This sounds exactly like the promises of putting a PC on everyone's desktop rather than a terminal connected to a locked-down mainframe, which in the end led to the lockdown migrating to PCs and the few employees who had any interest in automation forced to use tools like Excel because it was at least an improvement over VBScript and batch files and IT wouldn't let them install legit development tools unless their job description required them.

What are the odds that the average corporate IT department is going to allow the use of anything other than locked-down corporate AI with strict token quotas and stricter acceptable use policies?

That's even worse than the time I was on a conference call with the COO of a Fortune 500 retailer who referred to "the Excel spreadsheet we use to keep track of the stores we own".

The subject matter was building an auditing system to comply with the requirements of an IRS consent decree as part of a settlement for shoddy accounting practices which may or may not have included tracking billions of dollars of assets with Excel.

Corporate, sure, but painting DEC, (old) HP, and Sun with the same brush as AT&T, IBM, and Microsoft just because they sold proprietary, but generally well-documented and reliable hardware and software at whatever prices the market would bear seems unfair.

I don't get around to using it much because I have a lighter hand-me-down Dell something-or-other that works well enough for the little I do on laptops, but after two GPU replacements under AppleCare and a post-warrantly recall, my 2011 17" MacBook Pro is still running, triple booting Sequoia, Linux, and Windows 11.

Windows was fun to set up because this particular model had broken ACPI tables such that the only way to get all hardware working in Windows short of hotpatching the firmeware is via BIOS rather than UEFI boot, which surprisingly works fine if you use a the Windows 10 first-stage installer to install Windows 11, which surprisingly works fine because all the first stage does is partition and set up the disk and restore an image, and it doesn't care much about the contents of the image.

This set-up does require maintaining valid GPT and MBR partition tables, of course, and since Windows only sees MBR, it can only see four partitions, which isn't a big restriction in practice so long as I have plenty of free space in a partition accessible from all three OSes, which I do.

This is much better than my approach of acquiring obsolete versions of competitors' software to qualify for then-generous "competitive upgrades" that were often themselves available at steep discounts when the next version dropped.

For instance, there was not much of a market for the PCjr version of Lotus 1-2-3 in the early '90s when I used it to qualify for the $20 competitive upgrade to Excel 4.0 I found in the bargain bin at a local electronics retailer after Excel 5.0 released, which of course qualified me for even less expensive (not than $20) version upgrades in the future, not only to Excel, but to the full Office suite.

Not to mention the fact that Apple, unlike most of the industry at times, has never seen itself as a software company. Someone needs to build all the devices used to interact with all this AI, and there's no evidence any major "AI company" other than Google, and to a much smaller extent Meta, has any interest in doing so.

If I were Apple and wanted to make a big bet, I'd be dumping money into a top-secret robotics lab.

And presumably didn't work in shrinkwrapped retail software development, where costs for production, distribution, and retail markup were considerably higher, especially for low-cost consumer software — disk duplication costs for a $50 game were similar, if not higher, than for a $500 word processor — and for smaller developers that typically worked through a third-party publisher.

At least a handful of companies, including, incidentally, Valve: the Half-Life engine was a heavily modified version of the Quake engine.

The iPad mini was released at a price point lower than the contemporary iPad, and IIRC this continued for several generations.

Define low price, because looking at sold listings on ebay, you can get a used RTX 3070 for around $250, which is as capable as any 10-year old 8 GB card, and certainly more capable than a 660 Ti.

Still not great given that I paid about the same for a new-in-unopened-box RTX 3070 a couple years ago, but price increases no lower-end cards without enough VRAM for AI at least haven't skyrocketed like the top end of the market.

Note that Fedora CoreOS is a very low maintanance atomic Fedora system designed with fully automated updates in mind, and along with the systemd podman integration you mention, can easily be set up to automatically update both OS and containers.

Which may not be ideal for serious production workloads, but is perfect for non-mission critical appliance workloads like Pihole.

I suspect many new users get confused by the fact that lots of Nvidia hardware works poorly out of the box, rather than not at all, where the latter would at least provide a clue that a non-inbox driver might solve the problem.

By Arch, you mean that wiki with tons of information on installing and configuring software on Linux, right?

Say what you will about Arch, its community seems to have a rare gift for writing useful documentation.

I can't think of anything I've ever had trouble running on Fedora + containers, with the latter typically for proprietary software that depends on specific or ancient userlands, or simply to avoid cluttering up my base system.