141–150 of 355 posts
Could someone explains to me what the secret is here? Apart from the fancy marketing, is it the full integration? The hardware? It took me a while to find an actual picture of one of the modules.
Related question: Are services like AWS Outpost from public clouds the main competitor for Oxide?
Earlier quoted context omitted.
Yep, it’s basically slick hyperconverged infrastructure that you can buy a rack at a time without a subscription license.
Do you get updates to the proprietary software stack if you go without a subscription license? If the answer is no, then you might own the hardware on paper, but you don't control any of the software that makes said hardware useful. If the answer is yes, on the other hand, then one must ask who is paying for those updates, because that can't be sustainable.
Has anyone purchased/enquired about an Oxide rack or currently running one ? any info on pricing ?
Prices are not public, but comments over the years have hinted it’s in the $500K to $1mm range. Their hardware is multiple generations behind at this point, however. I wonder if they’re starting to reduce the price because it’s hard to justify paying so much for old hardware. They could just be targeting customers who don’t care as much about performance or efficiency as they do the software stack.
Earlier quoted context omitted.
There is a gold-rush going on. It needs plenty of compute besides GPU's, this is definitely the time to scale as quickly as possible.
I don't think any Oxide racks come with GPU's at present, and the power density of modern GPU-centric AI compute is on a rather unprecedented level. Oxide racks are very well cooled but are no match for the racks in an AI datacenter that's literally burning a full gigawatt of power.
Databases, K/V stores, crawlers, services, etc all still necessary besides GPU's. The closer to GPU's the better and if you have GPU's in your own DC.
Earlier quoted context omitted.
I think https://oxide.computer/product/specifications makes it pretty clear: Compute Sleds (Total) 16, 24, or 32 CPU Cores 1024–2048 Memory (DRAM) 8–32 TiB Storage 465.75–931.5 TiB Network Switches 2 Switching Capacity 12.8 Tbit/s Power Shelves Up to 2 Power Supplies per Shelf 6 (5+1 or 3+3) Typical / Max Power Draw 12 / 15 kW Dimensions H × W × D 2354mm (92.7”) × 600mm (23.7”) × 1060mm (41.8”) Weight Up to ~2,518 lb…
Unfortunately you have to read further down in the specs to see the actual hardware details. They’re on older generation hardware, so going by cores and RAM alone isn’t enough to tell you about the speed of those cores and the DDR4 RAM. Hopefully raising money helps them iterate faster on their hardware so they’re not so far behind.
Earlier quoted context omitted.
I actually did apply, The mere application takes hours upon hours, and for what a generic rejection email. This isn't the worst though, I recently went through an interview with another startup company, and after six interviews and a take-home project I found myself getting the same generic rejection. The CEO went out of his way to tell me he didn't like my resume since I've had to hop around a little bit to stay emp…
If you went through multiple rounds it likely means they were seriously considering you but ultimately they didn’t get to a yes. If it’s any comfort that means you did pretty well. The short stints on a resume is likely not the only reason you didn’t get to 100%, but unfortunately you should know that it’s seen as a pretty bad signal. The general expectation is 2 years minimum at a gig. If you have multiple short non…
Company A got their funding pulled and shut down. Company B, where I was actually at for about a year and a half, switched owners and shutdown my entire office. Company C merged into it's main competitor and effectively fired most of us.
I will admit I was at one fantastic job and after around 3 years I probably could of stayed indefinitely. But back then I didn't recognize the value of a solid job. If you land somewhere and you're well liked by people, and able to do quality work, you really should just stay there instead of chasing slightly more money.
Earlier quoted context omitted.
Prices start around 800k last time I heard, I don't know if that fits within what you consider a premium or not :-)
Hah, that might be slightly above what I'm ready to pay for a at-home server yes :) But given the right specifications and software integration, I'd probably be ready to pony up up to somewhere around 10K for a complete solution if it could replace all my existing hardware, even if it was more expensive than other options with worse tradeoffs.
Earlier quoted context omitted.
Yep, it’s basically slick hyperconverged infrastructure that you can buy a rack at a time without a subscription license.
Do you get updates to the proprietary software stack if you go without a subscription license? If the answer is no, then you might own the hardware on paper, but you don't control any of the software that makes said hardware useful. If the answer is yes, on the other hand, then one must ask who is paying for those updates, because that can't be sustainable.
what proprietary software stack? they just publish it all on https://github.com/oxidecomputer/ .
Oxide is the only company where I check the careers page hoping that they have a position which I can apply to. Happy to see their success. Especially so if you've been following their journey through their podcasts (easily the best tech podcast out there if you care about your craft; no filler, all killer).
I really tried to like the podcast. It’s been a few years, so maybe it’s improved. The topics were good. The guests were great. But Bryan Cantrill was just terrible at letting his guests actually talk. Bryan, if you’re listening, please let your guests talk. We have a large amount of content on YouTube if we want to hear the Bryan Cantrill take on, well, anything and everything. And it’s often amusing and sometimes r…