It fits an MI250X, and the system sees it, but the drivers don't work. They tested an HPE MI250X. There's a rumor on the thread that there are two kinds of MI250X: Ones from HPEs and everyone else's. The HPEs require a special firmware, the normal ones do not. However, the majority of the MI250Xs on the secondhand market are HPE so caveat emptor.
I looked at those, the Arc 1100, the w6800, MI50, MI60, v100, v620, and basically anything with 32gb of RAM:
1. I wanted an AMD card.
2. I have an RTX 3090 that's been fun to play with, but I want to get back to using it for gaming.
3. I was looking for between 30-60 tokens/second in terms of performance on the beefier models I want to run. Looking at stock Qwen3 32B the benchmarks reported about 41 tokens/second for MI100. w6800 was 18, MI50 & MI60 could do 60s but had a lot of compromises/special things to achieve that.
4. I used FitMyLLM for some spec-based comparisons (https://www.fitmyllm.com/). The MI100 is roughly double the performance on Qwen 3.5 35B A3B Q5_K_M to the R9700 (462 token/s prefill vs 239 tokens/s, 217 tokens/s vs 118 token/s for inference)
5. I was willing to throw up to $1k at a GPU; I really wanted to throw closer to $650.
To be honest, if money was no objection I would've sprung for a MI210. I also considered the MI250 as they showed up for $1250-1400 with a whopping 128GB, but the PCIE converters for that form factor don't have working AMD drivers yet.
I also recently decided to buy a datacenter GPU and slap it into a system. Some notes from my experience that the author doesn't mention in their article:
Decommissioned NVIDIA V100s and AMD MI50s are fairly cheap, $200 for 16gb and $400-500 for 32gb, for local experimentation. They are also very old. There's an enthusiast community keeping these two cards alive and working with current platforms and models.
Nitpick, but the V100 doesn't support bfloat16. The performance hit is not a big deal if you're fiddling with local models, but the card is on it's way out in terms of hardware features.
The MI50 does support bf16, but not the current edition of AMD ROCm. Vulkan support is good and the MI50 works with most major platforms (llama.cpp, vllm, etc.), but it's not without some pain points like manual recompilation. Fortunately the open source community has already paid most of your way.
The cooling requirements for these cards cannot be understated. A consumer grade GPU may throttle if in a small case without additional fans, but if given the same treatment a datacenter GPU will overheat itself idling. You will need to buy, at least, a bunch of decent 120mm fans to prevent this or invest in some water cooling.
I ultimately went with an AMD MI100 32GB ($950). I'm an AMD fan, current ROCm editions support it, and it was low-fuss to get things working. I'm debating getting a second so I can try out bigger models like qwen3-coder-next.
Since I'm getting a lot of hate for this post and having "only 2 years"
I have 15 years of experience and counting in games and entertainment. I had 6 years of experience as a game master and software dev (in-game purchasing and balancing) before Blizzard offered me an internship. I also worked in gaming throughout all of grad school, just as a contractor instead of full time.
Those of you who have gone to grad school know stipends don't pay crap and you need a second job to make rent.
Also I continue to contract as a professor. Those of you who have worked as teachers know that teaching also doesn't pay crap.
Finally, I still work in industry. Most recent game released on Steam in 2022.
> You say in a post below that your total games industry experience was a single internship at Blizzard and then a second stint where you "quickly realized" you didn't want to be a games dev at all
That is not my entire games experience. I have 15 years total, spanning Game Master, lead gameplay engineer, game engineering director, and CTO. I was asked my route to academia, not my entire Gaming Industry CV.
> they're saying they wanted a teacher who spent time in the games industry making the sort of games they themselves would play.
Nearly all of them have played games I've worked on, and can even find my name in the credits.
> Calling your students misogynists is a shamefully harsh attack on them without any evidence to back it up
You're going to make a bunch of assumptions based on a summary of my academic career and then try to insist that misogyny doesn't exist in tech?
> I could knock on the doors of say Digipen or similar.
Seattle based? If so toss me your email or contact and I'll see if I can help you break into teaching if you're interested.
> What kind of PhD research did you do?
Assistive technology. I primarily worked with kids who had cleft lip and palate to improve their at-home speech therapy exercises. Trained some offline ASR models, built a therapy game, and automated metrics. I passed the research onto another student once I got my PhD, but the project lives on as https://spokeitthegame.com/
Dissatisfaction yes, although it doesn't manifest how you'd expect.
It comes in the form not so much in dropouts, but in bad course feedback and bad professor reviews.
"The professor made the class unfun."
"The professor said she's made games but clearly has never done that before with how she taught the class"
I'm a woman so, unsurprisingly, I experience a fair amount of misogyny from students in the class who have never made a game nor have they worked in industry but believe they know how it works.
I graduated in 2011, went straight to work at Blizzard entertainment. At the same time I had gotten accepted into graduate school so I opted for an internship at Blizzard to try both.
I went back to Blizzard in 2012, but quickly realized I wanted to do my PhD. So I left Blizzard and went full-time as a student.
I didn't have funding so I TAd classes. Eventually my advisor and I scored a big NSF grant, so she used the funds to buy out a course and have me teach it as the instructor.
From there, I wound up at a community college running a bachelor's level degree. They hired me because I was the only candidate with NSF experience. They proceeded to fire their grant manager and have me manage the whole grant without extra pay.
Actually used to hire people for exactly what you want to do: be an adjunct for night classes in tech.
If you want to go that route you need to make friends with the Dean and the head of program. It's rare that we hire someone from the general application process, because most people who work in tech do not make good instructors.
First off let me get on my high horse and say the engineering in video gaming is generally more complex than the engineering I've done working in big tech. You need a lot more creativity and ingenuity to solve the unusual problems you run into in gaming.
From there, as others have said, it's a simple supply and demand issue. Nowadays I am a university professor, nearly every student who comes in wants to pursue one of the three fields: cybersecurity, video gaming, or recently ML/AI.
This shouldn't come as a surprise, they want to work on the things that influenced them and shaped their experiences so far. There's an absolute over supply of students who want to make video games.
Gaming, like most of entertainment, is a passion-driven industry. You trade good salary for your name in the credits. You trade nights, hobbies, marriages, and your health for this opportunity. That is unless you reach that lofty 1% of developers who are too valuable to be fired.
Not all areas of gaming are like this. Gambling, like working on slot/pachinko machines, pays very well and has pretty realistic work-life balance. However every student I've talked to about this has universally said "no I don't want to make slot machines. I only want to work on GTA/Stardew Valley/Hollow Knight/Fortnite."
There's seriously no shortage of starry-eyed students who are willing to accept minimum wage to solve SDE3 level problems. I was one of them once.
Like many other financially-backed initiatives, it's been investigated. Implementation is extremely hard, and would lead to an enormous shift in the very-established culture of volunteering.
> Would it though?
Rather than immediately shoot down your idea, let's talk about logistically implementing this:
1. Cleaning/non-compliance is already fined, if it's not cleaned properly or in a timely manner. This is serious waste like blackwater spills ($500+).
2. This would impact the the self-reliance, decommodification, and leave no trace principles. Burners don't need to be expected to clean up after themselves, they can pay someone else to do it. Yes, lots of wealthy burners would do this.
3. We'd need to set up a system of accountability. Sure, we can create a new department within the org, The Waste Accountability Department. Who do we charge for the bike graveyards (https://imgur.com/a/PolJDcI)? These are bikes that get abandoned in large clusters at the end of the event. Do they get assigned to whichever camp space they end up near? Do we start to add in plenty of surveillance (human or tech-based) to see which burner left their bike? Do we add in facial or other recognition to make sure we fine the right person?
3a. Currently bike graveyards are handled by nonprofits and volunteer orgs that take those bikes, fix them, and donate them to kids in Nevada. If we continue that program, do we pay them for taking the bikes? Do we need to appraise bikes based on their value, or do some other system of cost to repair vs value? Do we just sell blocks of "1000 lbs of bike"?
4. A core element of burning man, as mentioned in 2. is "Decommodification". This would commodify cleanup, and there are loads of first and second year burners who would absolutely pay someone $500/$1000 to clean their plot. Accountability here gets hard, as the people who are willing to pay are also the people who are unlikely to verify the quality of work. There are loads of people who would prey on burners in a rush to get out, pocket the $500, and walk off. Accountability and prosecution here, again, gets hard. The decommodification principle prevents this.
5. Who would issue the fine? Burning Man is a non-profit, the fine would require legal enforcement, collections or some other method to threaten people into paying. It would also require accounting for where that money goes and how it is used. Bureau of Land Management? They already do issue fines for blackwater spills and other serious environmental hazards (see 1).
6. Currently cleanup is handled by the Department of Public Works and Playa Resto. Both are volunteer-driven. Once we start paying people to clean up, why aren't we paying the people who deploy the porto-potties, make the streets, maintain the vehicles, and operate the core infrastructure? DPW spends 3-6 months before the event preparing the site for the event.
As I hope this demonstrates, it's not as simple as just having a group of people you can pay to clean up. There's a lot of logistical challenges, not to mention a pretty big shift in the culture.
I vote we stick with making people clean up after themselves, independent of their ability to pay.
Does the purpose change how the act should be interpreted?
> A place of pristine nature is literally destroyed by humans with zero fks given
Burners give so many fucks that we willingly do a thing historically reserved for punishment in the military (de-mooping)
> in a manner where it can never be recovered from.
Environmental sustainability is an actual goal not just greenwashing.
I encourage you too look into Playa Resto https://journal.burningman.org/2022/10/black-rock-city/leavi... and note that the term volunteer is used meaning these people don't get paid.
Theme camp based on an area famous for getting hit with hurricanes and other natural disasters here.
During the rains we were one of the few places still open and where you could party, eat, and grab a solid drink. Being on Esplanade also meant we were a shelter for people to wait out the weather.
This is how camps known as "Plug n Plays" work. Charge exorbitant camp dues, provide everything for your campers, and let them lead the most privileged lifestyle out there.
Many camps did this, and were actually turning a profit at Burning Man by taking advantage of the community of volunteers.
A few years ago the Burning Man organization put a stop to this by decreasing or eliminating camps considered to be Luxury or Plug N Play. Not just because they were antithetical to the event, but because they became famous for a slew of problems.
White Ocean is one of the more famous camps in this domain. A luxury camp that charged exorbitant fees for extremely wealthy individuals to come and party without any responsibility. They had loads of sexual assaults, dosing incidents, and campers generally being shitty people. The leads also refused to pay the hired help. This led to a now-infamous vandalism incident.
White Ocean basically has a permanent ban on attending now.
You cannot incentivize people out there with money. You have to take something away that they actually care about.
There are lots of people out there who would happily pay fines or not get deposits back if they didn't have to do the less glamorous parts of the event. You have to take something away that they actually care about.
If a camp does a really bad job at moop cleanup, Burning Man organization talks to leads to understand what happened. Frequently what they will take away is the camp's placement in the event, or sometimes even the ability to attend the event as that camp at all.
For reference: I am one of the leads for a fairly large and famous Burning Man camp. We camp on Esplanade most years. We do exactly what you proposed: We have deposits, and the more people put into the camp before, during, and after the event, determines if we offer them a refund and an invitation to camp with us next year. One of the factors is if you help us during setup and strike.
An invitation to camp with us guarantees them a ticket at one of the cheaper tiers. We have plenty of campers that come in, pay the dues, do nothing for the camp, are generally useless during the event, and bail out leaving a huge mess.
Conversely, we have a very small (10-20%) team of highly dedicated individuals who stay past the event and pick every piece of string, fuzz, fluff, lag bolt, rebar, and debris out of the dust and take it out. These people get nearly their entire camp dues back. If they attend next year, the social capital that they've built doing so compounds into them becoming increasingly popular and famous on Playa.
If there's one thing that Burning Man has taught me, it is that very few people are motivated by financial incentive. If you really want to motivate someone, figure out what they genuinely desire. It's rarely money.
My colleagues that teach hard skills courses (like data structures and algorithms) either love AI and incorporate it into their teaching at every moment possible, or despise it in the same way graphing calculators were by high school math teachers when they were introduced nearly 30 years ago.
I teach soft skills classes to engineering students, and I'm unconcerned with students using AI. I write my problems in a way such that, if the student truly understands the assignment, prompting the AI to solve the problem and iterating on it takes a similar amount of time to doing the work themselves. AI is not very good at writing introspectively about the student. In other words, AI isn't going to be helpful when the homework question is "A fellow student comes to you asking for suggestions on how to maximize their chances at landing an internship. What advice do you give them that's immediately actionable?"
Try it, plug that into ChatGPT or your favorite LLM. It parrots the same generic tips everyone tells you, with very little on "how" do perform the action in an effective way. Read it, copy it into your advice document, get a poor grade. Try telling other students to take this advice. Note how they don't because the advice isn't actually actionable enough for them to take action.
LLMs are also not very good at the follow-up question "In a previous assignment you gave specific and actionable advice to a peer on the job search. Which of these suggestions were so good you are now doing them?" A number of students write a "Mental Gymnastics" essay, claiming they are following all their suggestions (because they think that's what the professor wants to hear) while the evidence they provide demonstrates they are not. A student asking an LLM to write the essay for them consistently produces a digital 'pat on the back'; a mental gymnastics essay that ultimately makes the student realize how unwilling they are to solve the #1 problem in their college career.
I've done away with exams wherever possible. I stick to project-heavy courses. What I've found to be far more concerning than AI use is the increasing loss of social skills and ability to cooperate within the younger generations. The number of students who would prefer to fail a class instead of talk to literally any human being is astounding.
The number of students who refuse to build soft skills, and believe that tech is truly a meritocracy where the only thing that matters is 'lines of code', there's no politics, and they won't work call or crunch or give code reviews, is also astounding.