I begrudgingly agree. I would like to make a case that times have moved on/the definition of preferred has changed; but I don't think that would hold up on its own as a legal argument.
Agreed, this is a difficult topic to simplify. "The grid can handle it" is almost always a * statement.
Can it handle it today?
What about next year with the approved interconnects? What about in 2030?
Does "handle it" include summer load or just average, what if temps are 5 degrees hotter than power models ran?
What if PJM can't solve it's high priority overloads this year?
People want to talk about the grid as a binary system vs what it really is.
If serving those 37 customers forces choices that increase risk or cost for real customers (citizens, critical infrastructure like hospitals), it's a bad deal.
It seems like power utilities are optimized to solve problems, and serving these 37 customers can be framed as a solvable problem more easily than "Keeping the grid affordable and reliable" can.
Nobody has proven that those 37 customers are to blame compared to the poorly structured forward pricing system which is largely to blame for some of their recent pricing increases. Not sure where this grid can handle it narrative is coming from. Most of the woes in this article are cherry picked.
If you want to pretend the only source of information on this subject is the linked article I'm sure you can find many users here to debate within your arbitrary parameters.
Same goes for proof, which I'm sure would exclude power flow models if they were available for said debate.
Edit: We can agree the article is not comprehensive, and doesn't have data PJM and utilities have not published.
I’ve linked analyses explaining why PJM’s forward capacity market has been a major driver of recent retail rate increases, not just in this one county.
If you disagree, point to the analysis or data. Suggesting I’m “pretending” doesn’t address the argument. The article cherry-picks one county while largely ignoring the broader PJM pricing mechanics affecting millions of customers. Do better. Between pricing mechanics, input cost inflation and yes increase in demand has impact but not like the article likes to weave it.
Without seeing actual power flow modeling for the lines attached to these data centers no one on the planet can accurately weigh the expense of these customers.
You can't 8760 where and when overloads happen elsewhere, unplanned downtime can't be identified in (or out of) the ISO, you can't forcast demand, you can't plan pre-purchasing of power before it's on demand. You can't speculate on if reconductoring could save money, or remove overloads.
You want to look at rate increases as the core driver. So much so you dump it on replies to others comments.
Too many variables are opaque for me to have a discussion about what is the 'largest' factor when each item on the above list could represent 10 million to 8 billion+ in cost.
But I can tell you I won't be convinced that new large loads, and who approved them are not a contributing factor because no one has "proven" it in real time. That's the kind of logical falacy that results in dismissal of real issues.
Obviously there are not easy solutions here. Has Mullvad considered offering to buy out Daniel's shares?
The mission statement that Mullvad has promised it's users does (did) come with a duty of care.
In my eyes that mission statement is compromised, it either dies here and Mullvad is just another product financing political parties by proxy, or something is done.
I hope you find a way to protect Mullvad from actions that are counter to your mission statements.
We seem to agree that what replaced religion (for profit social media) is not a basis for a strong society, or fulfilling lives. Religion seems like a close 2nd worst option though.
What pcie version are you running? Normally I would not mention one of these, but you have already invested in all the cards, and it could free up some space if any of your lanes being used now are 3.0.
If you can afford the 16 (pcie 3) lanes, you could get a PLX ("PCIe Gen3 PLX Packet switch X16 - x8x8x8x8" on ebay for like $300) and get 4 of your cards up to x8.
All are PCIe 3.0, I wasn't aware of those switches at all, in spite of buying my risers and cables from that source! Unfortunately all of the slots on the board are x8, there are no x16 slots at all.
So that switch would probably work but I wonder how big the benefit would be: you will probably see effectively an x4 -> (x4 / x8) -> (x8 / x8) -> (x8 / x8) -> (x8 / x4) -> x4 pipeline, and then on to the next set of four boards.
It might run faster on account of the three passes that are are double the speed they are right now as long as the CPU does not need to talk to those cards and all transfers are between layers on adjacent cards (very likely), and with even more luck (due to timing and lack of overlap) it might run the two x4 passes at approaching x8 speeds as well. And then of course you need to do this a couple of times because four cards isn't enough, so you'd need four of those switches.
I have not tried having a single card with fewer lanes in the pipeline but that should be an easy test to see what the effect on throughput of such a constriction would be.
But now you have me wondering to what extent I could bundle 2 x8 into an x16 slot and then to use four of these cards inserted into a fifth! That would be an absolutely unholy assembly but it has the advantage that you will need far fewer risers, just one x16 to x8/x8 run in reverse (which I have no idea if that's even possible but I see no reason right away why it would not work unless there are more driver chips in between the slots and the CPUs, which may be the case for some of the farthest slots).
PCIe is quite amazing in terms of the topology tricks that you can pull off with it, and c-payne's stuff is extremely high quality.
If you end up trying it please share your findings!
I've basically been putting this kind of gear in my cart, and then deciding I dont want to manage more than the 2 3090s, 4090 and a5000 I have now, then I take the PLX out of my cart.
Seeing you have the cards already it could be a good fit!
Yes, it could be. Unfortunately I'm a bit distracted by both paid work and some more urgent stuff but eventually I will get back to it. By then this whole rig might be hopelessly outdated but we've done some fun experiments with it and have kept our confidential data in-house which was the thing that mattered to me.
Yes, the privacy is amazing, and there's no rate limiting so you can be as productive as you want. There's also tons of learnings in this exercise. I have just 2x 3090's and I've learnt so much about pcie and hardware that just makes the creative process that more fun.
The next iteration of these tools will likely be more efficient so we should be able to run larger models at a lower cost. For now though, we'll run nvidia-smi and keep an eye on those power figures :)
You can tune that power down to what gives you the best tokencount per joule, which I think is a very important metric by which to optimize these systems and by which you can compare them as well.
I have a hard time understanding all of these companies that toss their NDA's and client confidentiality into the wind and feed newfangled AI companies their corporate secrets with abandon. You'd think there would be a more prudent approach to this.
"there are more than a dozen different types of diabetes" this feels like an ai synopsis that any human would have critiqued, but since ai said it, it made it in.
I'm not a doctor, but I've never heard of more than 5 "types" from endocrinologists in the past 20+ years.
That value (of one company) is from speculative investment. I don't think it negates that the field has a perception problem.
After seeing something like blockchain run completely afoul/used for the wrong things and embraced by the public for it, I at least agree that AI has a value perception problem.
Exactly. It is possible for a metric like DAUs to come almost entirely from marketing saturation, heavy promotion, and hype, and not from actual utility to the user. I'm not sure that's the case in particular for ChatGPT but I wouldn't be surprised.
I believe Berkshire bought (or agreed to buy) part of dominion (a public utility delivering power) on the east coast, I don't know any particulars of how it was run or if the deal even closed.
Thats the only related example I know of (non exhaustive).
