>> If you're at $5,000/month, a 4.2% raise puts you at $5,210. If you're spending $600/month on gas (not unreasonable for someone that drives an SUV and lives in the suburbs instead of in the urban core), you still come out behind.
This is the problem with people treat CPI as some word from the heavens...it is not. CPI is a highly constructed figure which conveniently includes/excludes things and is really more a floor of what the inflation is. Anyone living in the real world knows experienced inflation is way higher.
> CPI is a highly constructed figure which conveniently includes/excludes things and is really more a floor
It’s an attempt at a central tendency in a complex economy with non-linear variability.
> Anyone living in the real world knows experienced inflation is way higher
Here is a map of wage changes across the U.S., 2024 to 2025 [1]. Lots of variance! If you’re on the West Coast, right now, you’re seeing above-CPI inflation. If you’re in the Northern Rockies, where I am, you’re seeing less.
ParadeDB is AGPL so not generally available on the hyperscalars. However, you can use https://github.com/timescale/pg_textsearch on Azure HorizonDB (and likely soon Flex). Disclosure: I'm the pg_textsearch maintainer and now at Azure.
I didn't quite follow your comment about vector support, are you asking for something beyond what pgvector + diskann provide (both available on Azure)?
ParadeDB maintainer here :). We would happily make it available on Azure (and all other cloud providers!) if there were a way for us to earn a living in doing so.
Fyi, we are in discussion with some hyperscalers on making this possible.
>> I didn't quite follow your comment about vector support, are you asking for something beyond what pgvector + diskann provide (both available on Azure)?
You dont support ultra-wide vectors from the largest embeddings models. We have to wierd stuff like chop up vectors across fields.
Some thing I've learned, but rarely seen explained anywhere: Storing the vectors is most likely not an issue, mostly likely you're having a problem with the indexes on top of them in which case you can use quantized vector indexes[0] (handled by pgvector) which will get past the limits imposed by PostgreSQL.
I had to switch off pgvecto.rs at some point and figured that out.
I don't have specific experience with the Azure environment here, but this probably applies if you have access to pgvector.
You could. But then you’re also building from scratch HA failover, backups, replica management, monitoring, etc - cloud vendor managed RDBMS come with lots of niceties. All of which are possible to set yourself. But a hassle, and difficult to make bullet proof.
Never ever use Azure Cosmos DB. The entire point is to lock you in. This isn’t some paranoid shit either. We use azure a lot, and I have worked with many people designing systems on Azure. Always avoid cloud providers lock in services. That’s their bread and butter. They want you to use them. They want you using Azure Cosmos DB, Azure Event Hubs, Azure Apps, Azure DataLake, etc. Same with AWS. Don’t be naive. Use Azure VMs, Azure Postgres, Azure Redis. Those are fine. You’re just paying someone for the operational cost of a service, but you can migrate of. There is no migration from Cosmos or DataLake. They tell you you can abstract your code, but that never works. They know you will be locked in. That’s the entire business model. Also resist the temptation of the offers they’ll through at you to link those services with all their other crap. Don’t be naive.
I’m not sure why you’re getting downvoted because CosmosDB doesn’t even have a local install edition. Conversely the cloud hosted offering is slower than cold molasses and costs most of your body parts… per month.
Hey! I'm a PM on the Azure PG team and work on AI features on Postgres. Wanted to address your points directly because we actually ship the capabilities you're asking about, we have made ALOT of progress in the last 3-6 months:
Hybrid search (BM25 + vector): Worth noting that ParadeDB's pg_search isn't an AWS-native feature either, you'd need to self-host it on EC2. On Azure PostgreSQL, we built pg_textsearch which provides the same BM25 ranking model (term frequency saturation, document-length normalization, IDF) natively. Fun fact, the main contributor of pg_textsearch is now on the Azure Postgres team :)
High-dimensional vectors: This is actually an area where we're ahead. pgvector with HNSW caps at 2,000 dimensions. We support pgvector for vector storage and search, and for high-dimensional / large-scale workloads we ship pg_diskann — Microsoft's graph-based vector index that supports up to 16,000 dimensions and also does advanced in-index filtering (your WHERE clauses get evaluated during graph traversal, so you don't lose recall on selective predicates).
These are available today on Azure PostgreSQL, specifically Azure HorizonDB (Preview). Happy to dig into specifics if you have a particular workload in mind.
How do we know this is due to AI usage? Perhaps it is because the students missed key in-person learning at the tail end of high school due to the pandemic lock-downs? I cant imagine learning calculus / linear algebra on my own in high school.
Absolutely: missing in-person learning due to COVID. Less attention span due to growing up in a distracting environment. A lower bar to entry due to removal of standardized testing and indirectly from No Child Left Behind. Changes in parent or student attitudes. It could be any number of things, and it's lazy to just say "with AI usage" as something that has increased at the same time.
> I cant imagine learning calculus / linear algebra on my own in high school.
I don't think they necessarily expect students to have that from high school, because the class mentioned, EECS 127, lists three college classes as prerequisites:
* Math 53 - Multivariable Calculus
* Math 54 - Linear Algebra & Differential Equations
* CS 70 - Discrete Mathematics and Probability Theory
>> Concerts became about filming a DJ twiddling a USB controller.
This is one of the worst parts of any concert, performance -- having a sea of phones in front of you recording. In a dark theatre, it is impossible to watch the actual performance when you have a screen on super-bright in front of you recording it. Also, some people literally record on ipads!
All these are reasons i've not opted to do "concert in my living room" via YouTube and a big screen tv. Not the same, but a lot less silliness around me.
When i was younger I would sometimes ride the F-train back and forth several times in the evenings just so I could think and put articulate thoughts into a notebook. This is before underground network connectivity, before smartphones, etc. The hum of the train was great and the speed of the F train at segments of Queens were exhilarating, a bit like listening to EDM while coding.
I used to do a few different versions of this for most of the 30+ years I lived in NY. I used to love walking all over manhattan alone late at night. I would do like 10 mile walks just for the heck of it with music going. Or I would hop on the bike and do similarly.
When there would be friends involved, we'd usually be at Chinatown Fair all night until close, then walk down to Elevated Acre and hang out there (this was pre-9/11) until 3 or 4, then walk our friends from Staten Island to the Ferry. If the mood was particularly good, we'd take the ferry with them and then ride it back and everyone go their separate ways.
There used to be houseboats around lower manhattan back then and it was a nice (albeit sketchy) walk from Chinatown down to the tip of the island. There was also pretty briefly this wild hole-in-the-wall DUMBO nightclub projecting porn on the walls that we would frequently stumble into on our way down there.
Things like that were honestly the best part about living in NY, but also it's long in the past.
Summer 2020 I was out there after creating as much shareholder value as I could at my then-WFH 9-5. First it was delivering postmates on rollerblades, then I did it on my bike (working better and better apps every time to offset the depreciation inherent to riding a bike), then I'd just do a 4 boro bike tour Brooklyn over the Pulaski bridge up to Astoria, get souvlaki, take the Triboro to the Bronx and then head to Manhattan on a different bridge every time (High Bridge is still my fave). Just putting up like 40–60 mile rides noodling around that town. There's no better way to see the city, and no better place to bop around on a bike, in my experience.
I'd love a way to do this locally -- pool all the PCs in our own office for in-office pools of compute. Any suggestions from anyone? We currently run ollama but manually manage the pools
If you set CPUSchedulingPolicy=idle Nice=19 IOSchedulingClass=idle in the ollama server configuration it should run in the background with lowest priority.
Seems like so much more work than "just" paying for https://huggingface.co or whichever other neocloud who already did all the setup for you and just waits for your credit card per minute/seconds/token.
Doubt this kind of workloads would agree to send data then to a cloud of randos devices, precisely when cloud providers to certify they aren't looking at clients data (Customer-managed encryption keys, CMEK).
This is the problem with people treat CPI as some word from the heavens...it is not. CPI is a highly constructed figure which conveniently includes/excludes things and is really more a floor of what the inflation is. Anyone living in the real world knows experienced inflation is way higher.
reply