• kautau@lemmy.world
      link
      fedilink
      arrow-up
      52
      arrow-down
      6
      ·
      5 months ago

      And then China popping their head out claiming Taiwan is part of China because they want to seize TSMC

        • sparkle@lemm.ee
          link
          fedilink
          Cymraeg
          arrow-up
          13
          arrow-down
          4
          ·
          edit-2
          5 months ago

          The US state department doesn’t decide which countries own or control which territory, now does it? How exactly can you say territory you don’t control (neither legally nor militarily) and likely will never control is part of your own country? Furthermore, why would the US risk ruining trade relations with China over unnecessarily pointing out reality, when it doesn’t benefit the US to recognize Taiwanese independence?

          • umbrella@lemmy.ml
            link
            fedilink
            arrow-up
            3
            ·
            5 months ago

            the us has literally asserted taiwan is part of china for decades now.

            under the kissinger term, no less.

            • sparkle@lemm.ee
              link
              fedilink
              Cymraeg
              arrow-up
              12
              arrow-down
              3
              ·
              edit-2
              5 months ago

              I’ll say it again: Why would countries risk ruining trade relations with China, one of the three most important trade powers internationally, over unnecessarily pointing out reality and thus contradicting China? And how can you seriously say territory a country doesn’t control in any capacity at all theirs? Why do you think a majority of world powers are independently trading with Taiwan if Taiwan isn’t independent from China?

              Don’t you think China would, you know, not be constantly complaining about not having control over Taiwan for the past few decades and making bluffs about invading if Taiwan were already part of China? That’s a pretty obvious sign that “no, China doesn’t own Taiwan in any capacity”.

              • ☆ Yσɠƚԋσʂ ☆@lemmy.ml
                link
                fedilink
                arrow-up
                7
                arrow-down
                14
                ·
                edit-2
                5 months ago

                What you’re doing here is called sophistry. Taiwan being part of China is a fact that’s recognized by international law. It’s really that simple. The reality is that China could remove US sponsored regime in the rogue province any time they want. However, they realize that it’s much better to remove burgerland influence in a peaceful way, and that’s what will happen. It’s incredible how people have trouble grasping such basic things.

                edit: I aboslutely love how utterly enraged lemmy radlibs get when faced with reality

                • sparkle@lemm.ee
                  link
                  fedilink
                  Cymraeg
                  arrow-up
                  8
                  arrow-down
                  1
                  ·
                  edit-2
                  5 months ago

                  What you’re doing here is called sophistry. Taiwan being part of China is a fact that’s recognized by international law.

                  Tell me you have no idea how the UN works without explicitly saying so. A majority of countries not recognizing Taiwan doesn’t mean it’s “international law” that Taiwan isn’t independent.

                  It’s really that simple. The reality is that China could remove US sponsored regime in the rogue province any time they want.

                  LMAO this is such a cope. Yeah I’m sure the extremely aggressive all-bark-no-bite and constant “you better not do <x diplomacy with Taiwan or military action in Taiwanese strait/South China Sea> again or we’ll do something about it, I swear!” empire is suuuper capable of taking Taiwan. They know if they tried full-out war against the US or its allies (Taiwan), the US navy would cut off their international trade and turn their country upside down – it’s why they’re trying so hard (and failing) to seize full control of the South China Sea.

                  However, they realize that it’s much better to remove burgerland influence in a peaceful way, and that’s what will happen.

                  Again, absolute cope. They’ve been at it for over 75 years and haven’t made any progress, considering Taiwanese have developed significantly more national identity and even more people in Taiwan support the country participating in international relations under the name “Taiwan” (80%) and consider themselves primarily Taiwanese (90%), and only 6% consider themselves more Chinese than Taiwanese (more people considered themselves primarily Chinese many decades ago but that has long since dwindled).

                  It’s incredible how people have trouble grasping such basic things.

                  It’s incredible how you have trouble grasping the situation and think China is going to “peacefully” absorb Taiwan when Taiwan is farther from China than ever in terms of national identity and international participation.

                  Several polls have indicated an increase in support of Taiwanese independence in the three decades after 1990. In a Taiwanese Public Opinion Foundation poll conducted in June 2020, 54% of respondents supported de jure independence for Taiwan, 23.4% preferred maintaining the status quo, 12.5% favored unification with China, and 10% did not hold any particular view on the matter. This represented the highest level of support for Taiwanese independence since the survey was first conducted in 1991. A later TPOF poll in 2022 showed similar results.

    • frezik@midwest.social
      link
      fedilink
      arrow-up
      11
      ·
      5 months ago

      Eh, they’ll have plenty of demand for their nodes regardless. Non-AI CPUs and GPUs are still going to want them.

    • frezik@midwest.social
      link
      fedilink
      arrow-up
      32
      arrow-down
      1
      ·
      edit-2
      5 months ago

      You can buy them new for somewhat reasonable prices. What people should really look at is used 1080ti’s on ebay. They’re going for less than $150 and still play plenty of games perfectly fine. It’s the budget PC gaming deal of the century.

      • linkhidalgogato@lemmy.ml
        link
        fedilink
        arrow-up
        22
        arrow-down
        2
        ·
        5 months ago

        it probably the best performance per dollar u can get but a lot of modern games are unplayable on it.

        • frezik@midwest.social
          link
          fedilink
          arrow-up
          13
          arrow-down
          2
          ·
          5 months ago

          Lot of those games are also hot garbage. Baldur’s Gate 3 may be the only standout title of late where you don’t have to qualify what you like about it.

          I think the recent layoffs in the industry also portend things hitting a wall; games aren’t going to push limits as much as they used to. Combine that with the Steam Deck-likes becoming popular. Those could easily become the new baseline standard performance that games will target. If so, a 1080ti could be a very good card for a long time to come.

      • umbrella@lemmy.ml
        link
        fedilink
        arrow-up
        2
        ·
        edit-2
        5 months ago

        not in my country lol. getting used cards were already the norm before, for a while you could literally only get used ones for a good price on aliexpress.

        and now our gvmnt imposed 100% tax on anything from china, so its really just not affordable.

          • umbrella@lemmy.ml
            link
            fedilink
            arrow-up
            2
            ·
            edit-2
            5 months ago

            not really a good way. we can mostly only afford used and theres no outlet to buy used from taiwan that ships internationally in a trustworthy way.

            brand new will probably be taxed too anyway.

        • Baguette@lemm.ee
          link
          fedilink
          arrow-up
          1
          ·
          5 months ago

          You can still look for used ones locally either through hardwareswap or fb marketplace (unfortunately thats the best secondhand marketplace atm). Other options include liquidation companies, sometimes universities also have a big market (from both staff and students)

          Its been pretty rough still though, good luck on your search

          • umbrella@lemmy.ml
            link
            fedilink
            arrow-up
            1
            ·
            edit-2
            5 months ago

            no such thing as hardwareswap in my country. our marketplaces are overpriced because we want to resell our shit with prices proportional to what we overpaid for it.

            importing used shit is cheaper because the value of, say, an rx580 is much lower outside the country.

            • Baguette@lemm.ee
              link
              fedilink
              arrow-up
              1
              ·
              5 months ago

              Yea lots of the secondhand tech market exists purely in the US. The market outside of there and a couple other countries (uk, ca, de) dont really exist afaik

    • kameecoding@lemmy.world
      link
      fedilink
      arrow-up
      6
      arrow-down
      1
      ·
      5 months ago

      Meanwhile I dont think I have played more than 30minutes on my ps5 this year and its june, and I have definitely not played any minutes on the 1080 sitting in my PC…

      Oh fuck scratch that I may have played about 2 hours of Dune Spice Wars

      • Geobloke@lemm.ee
        link
        fedilink
        arrow-up
        1
        ·
        5 months ago

        How are you finding dune? I watched a few let’s plays of the demo and it looked interesting…

        • kameecoding@lemmy.world
          link
          fedilink
          arrow-up
          2
          arrow-down
          1
          ·
          5 months ago

          Honestly, I don’t have an opinion on it, it didnt capture me like old world did last year, so probably not as good or I am just preferring more slow, thought out turn based stuff.

          But mostly I am just kinda over gaming as a whole, I realized it’s mostly cheap dopamine chasing for me and I don’t really enjoy it.

          • 11111one11111@lemmy.world
            link
            fedilink
            arrow-up
            1
            ·
            5 months ago

            Crack cocain is 10000x’s more rewarding and less of a come down when you realize you only spent 1/10th of the price of a console or gaming pc.

      • umbrella@lemmy.ml
        link
        fedilink
        arrow-up
        1
        ·
        edit-2
        5 months ago

        nah, its gonna become the next enshitified netflix. i stream games from my own pc.

        good thing about small screens is that you don’t need the best resolution so it does better on older computers.

        not to mention the monthly cost instead of being a once every handful of years thing.

        not judging though streaming can be fine.

  • Kalkaline @leminal.space
    link
    fedilink
    arrow-up
    55
    ·
    5 months ago

    Don’t forget AMD, good potential if they bring out similar technology to compete with NVIDIA. Less so Intel, but they’re in the GPU market too.

      • SeekPie@lemm.ee
        link
        fedilink
        arrow-up
        6
        ·
        5 months ago

        As I understand it, ARM chips are much more efficient on the same tasks, so they’re cheaper to run.

      • AdrianTheFrog@lemmy.world
        link
        fedilink
        English
        arrow-up
        5
        ·
        5 months ago

        I think its largely the chip manufacturers, but ARM is still making money on licensing fees for Nvidia’s new ai chip (with an integrated 72 core arm cpu) for example

        ARM is in the perfect place where, if a company using their architecture succeeds, they get tons of money, and if the company fails, they lose nothing.

    • r00ty@kbin.life
      link
      fedilink
      arrow-up
      40
      ·
      5 months ago

      I mean if LLM/Diffusion type AI is a dead-end and the extra investment happening now doesn’t lead anywhere beyond that. Yes, likely the bubble will burst.

      But, this kind of investment could create something else. We’ll see. I’m 50/50 on the potential of it myself. I think it’s more likely a lot of loud talking con artists will soak up all the investment and deliver nothing.

      • linkhidalgogato@lemmy.ml
        link
        fedilink
        arrow-up
        20
        arrow-down
        1
        ·
        5 months ago

        bubbles have nothing to do with technology, the tech is just a tool to build the hype. The bubble will burst regardless of the success of the tech at most success will slightly delay the burst, because what is bursting isnt the tech its the financial structures around it.

      • frezik@midwest.social
        link
        fedilink
        arrow-up
        19
        arrow-down
        2
        ·
        5 months ago

        It’s looking like a dead end. The content that can be fed into the big LLMs has already been done. New stuff is a combination of actual humans and stuff generated by LLMs. It then runs into an ouroboros problem where it just eats its own input.

        • greenskye@lemm.ee
          link
          fedilink
          English
          arrow-up
          17
          ·
          5 months ago

          I mostly agree, with the caveat that 99% of AI usage today just stupid gimmicks and very few people or companies are actually using what LLMs offer effectively.

          It kind of feels like when schools got sold those Smart Whiteboards that were supposed to revolutionize teaching in the classroom, only to realize the issue wasn’t the tech, but the fact that the teachers all refused to learn and adapt and let the things gather dust.

          I think modern LLMs should be used almost exclusively as an assistive tool to help empower a human worker further, but everyone seems to want an AI that you can just tell ‘do the thing’ and have it spit out a finalized output. We are very far from that stage in my opinion, and as you stated LLM tech is unlikely to get us there without some sort of major paradigm shift.

          • micka190@lemmy.world
            link
            fedilink
            arrow-up
            7
            ·
            5 months ago

            only to realize the issue wasn’t the tech

            To be fair, electronic whiteboards are some of the jankiest piles of trash I’ve ever had to use. I swear to God you need to re-calibrate them every 5 minutes.

        • r00ty@kbin.life
          link
          fedilink
          arrow-up
          1
          ·
          5 months ago

          Yeah, I was thinking more if there’s either an evolutionary improvement or revolutionary (or some movement toward AGI). For me it’s better if not, so I get to keep my job for a few more years. But, my general feeling is with the cash injection, there’s some chance of a breakthrough.

    • TheRealKuni@lemmy.world
      link
      fedilink
      English
      arrow-up
      10
      arrow-down
      1
      ·
      5 months ago

      I doubt it. Regardless of the current stage of machine learning, everyone is now tuned in and pushing the tech. Even if LLMs turn out to be mostly a dead end, everyone investing in ML means that the ability to do LOTS of floating point math very quickly without the heaviness of CPU operations isn’t going away any time soon. Which means nVidia is sitting pretty.

      • umbrella@lemmy.ml
        link
        fedilink
        arrow-up
        10
        ·
        5 months ago

        the WWW wasn’t a dead end but the bubble burst anyway. the same will happen to AI because exponential growth is impossible.

    • frezik@midwest.social
      link
      fedilink
      arrow-up
      5
      ·
      5 months ago

      See Sun Microsystems after the .com bubble burst. They produced a lot of the servers that .com companies were using at the time. Shriveled up after and were eventually absorbed by Oracle.

      Why did Oracle survive the same time? Because they latched onto a traditional Fortune 500 market and never let go down to this day.

    • DogWater@lemmy.world
      link
      fedilink
      arrow-up
      3
      arrow-down
      2
      ·
      5 months ago

      No they won’t, this tech isn’t going to go away Even if it plateaus. All the gpus they make will still get used.

  • ImplyingImplications@lemmy.ca
    link
    fedilink
    arrow-up
    43
    arrow-down
    13
    ·
    5 months ago

    Worst one is probably Apple. They just announced “Apple Intelligence” which is just ChatGTP whose largest shareholder is Microsoft. Figure that one out.

    • dependencyinjection@discuss.tchncs.de
      link
      fedilink
      arrow-up
      39
      arrow-down
      4
      ·
      5 months ago

      Well, most of the requests are handled on device with their own models. If it’s going to ChatGPT for something it will ask for permission and then use ChatGPT.

      So the Apple Intelligence isn’t all ChatGPT. I think this deserves to be mentioned as a lot of the processing will be on device.

      Also, I believe part of the deal is ChatGPT can save nothing and Apple are anonymising the requests too.

        • dependencyinjection@discuss.tchncs.de
          link
          fedilink
          arrow-up
          12
          arrow-down
          4
          ·
          5 months ago

          Brother I do not care about your doubts.

          I want hard facts here.

          Do you think that if you enter into a contract with a company like Apple they’ll just be like, aww shit they weren’t supposed to do that. Anyway let’s carry on.

          No. This would open OpenAi up to potential lawsuits.

          Even if they did save stuff. It gets anonymised by Apple before even being sent to ChatGPT servers.

          • Fedizen@lemmy.world
            link
            fedilink
            arrow-up
            4
            arrow-down
            2
            ·
            5 months ago

            The hard fact is OpenAI is already exposing itself to lawsuits by training on copyrighted material.

            So the proof here should be “what makes them trustworthy this time?”

            • dependencyinjection@discuss.tchncs.de
              link
              fedilink
              arrow-up
              4
              arrow-down
              1
              ·
              edit-2
              5 months ago

              Because Apples lawyers will go ham.

              I don’t want my comments here to be received as shilling Apple, more that I want them to based on actual information that is provided and not opinion pieces.

              The fact is, if they were to caught saving data then Apple would just end the contract. Is it worth it for them to lose out on that cash, for the sake of using it. When they can just use all the other sources where they are allowed to do that.

              Anyway, I don’t care what anonymised data they may or may not save. It won’t be tied to me.

              Edit: Do you have some information on this existing lawsuits and the contracts they broke?

              • Blue_Morpho@lemmy.world
                cake
                link
                fedilink
                arrow-up
                4
                arrow-down
                1
                ·
                5 months ago

                Because Apples lawyers will go ham.

                Google pays Apple $20 billion a year to keep their search on Apple devices. The subtext of “search” is Google pays Apple for your search data.

                Apple has sold your data for the right price to Google, so there should be no expectation that they won’t do the same with other companies.

                • dependencyinjection@discuss.tchncs.de
                  link
                  fedilink
                  arrow-up
                  2
                  ·
                  5 months ago

                  They sell Google the right to keep it as the default, not that they’re selling data.

                  Again, point me to some proof of it being actually selling data. As to my understanding they pay for the default engine to be Google.

            • micka190@lemmy.world
              link
              fedilink
              arrow-up
              3
              ·
              5 months ago

              There’s kind of a difference between “we scraped the internet and decided to use copyrighted content anyways because we decided to interpret copyright law as not being applicable to the content we generate using copyrighted content” (omegalul) and “we explicitly agreed to a legally-binding contract with Apple stating we won’t do that”.

            • dependencyinjection@discuss.tchncs.de
              link
              fedilink
              arrow-up
              2
              ·
              5 months ago

              Finally, a reasonable comment.

              I would concede that they want to keep it all for themselves, although a lot of anonymising of data is done.

              My point is Apple are not sharing it with every third party on the Earth.

              If you’re using Android then you don’t really have a leg to stand on, unless you’re using GrapheneOS and you’ve sandboxed Google services.

              I would rather use a device that maybe keeps it all for themselves. Rather than one where it is shared with Everyman and his dog.

              Plenty of things you can shit on Apple for, but this isn’t one of them I’m afraid.

      • Blue_Morpho@lemmy.world
        cake
        link
        fedilink
        arrow-up
        11
        arrow-down
        11
        ·
        5 months ago

        Well, most of the requests are handled on device

        Doubt.

        Voice recognition, image recognition, yes. But actual questions will go to Apple servers.

        • dependencyinjection@discuss.tchncs.de
          link
          fedilink
          arrow-up
          17
          arrow-down
          4
          ·
          edit-2
          5 months ago

          Doubt.

          Is this conjecture or can you provide some further reading, in the interest of not spreading misinformation.

          Edit: I decided to read the info from Apple.

          With Private Cloud Compute, Apple sets a new standard for privacy in AI, with the ability to flex and scale computational capacity between on-device processing, and larger, server-based models that run on dedicated Apple silicon servers. When requests are routed to Private Cloud Compute, data is not stored or made accessible to Apple and is only used to fulfill the user’s requests, and independent experts can verify this privacy.

          Additionally, access to ChatGPT is integrated into Siri and systemwide Writing Tools across Apple’s platforms, allowing users to access its expertise — as well as its image- and document-understanding capabilities — without needing to jump between tools.

          Say what you will about Apple, but privacy isn’t a concern for me. Perhaps, some independent experts will verify this in time.

          • Blue_Morpho@lemmy.world
            cake
            link
            fedilink
            arrow-up
            5
            arrow-down
            2
            ·
            5 months ago

            Which is exactly what I said. It’s not local.

            That they are keeping the data you send private is irrelevant to the OP claim that the AI model answering questions is local.

            • dependencyinjection@discuss.tchncs.de
              link
              fedilink
              arrow-up
              3
              ·
              5 months ago

              OP here being me.

              Well, most of the requests are handled on device with their own models. If it’s going to ChatGPT for something it will ask for permission and then use ChatGPT.

              I feel I was pretty explicit in explaining how some requests will go to ChatGPT.

        • AdrianTheFrog@lemmy.world
          link
          fedilink
          English
          arrow-up
          3
          ·
          5 months ago

          Apple has published papers on small LLM models and multimodal models already. I would be surprised if they aren’t using them for on-device processing.

      • frezik@midwest.social
        link
        fedilink
        arrow-up
        6
        ·
        5 months ago

        Yeah, if anything, Apple is behind the curve. Nvidia/AMD/Intel have gone full cocaine nose dive into AI already.

    • Ken27238@lemmy.ml
      link
      fedilink
      arrow-up
      8
      ·
      5 months ago

      Not true. Most if not all requests are handled by apples own models on device or on their own servers. When it does use OpenAI you need to give it permission each time it does.

    • photonic_sorcerer@lemmy.dbzer0.com
      link
      fedilink
      English
      arrow-up
      9
      arrow-down
      2
      ·
      5 months ago

      That’s just not true. Most requests are handled on-device. If the system decides a request should go to ChatGPT, the user is promped to agree and no data is stored on OpenAI’s servers. Plus, all of this is opt-in.

      • Blue_Morpho@lemmy.world
        cake
        link
        fedilink
        arrow-up
        5
        arrow-down
        8
        ·
        5 months ago

        Most requests are handled on-device.

        Literally impossible.

        “Hey Siri, what’s the weather forecast for tomorrow.”

        < The Farmer’s Almanac that is in my local model says it will rain tomorrow. >

        • PassingThrough@lemmy.world
          link
          fedilink
          English
          arrow-up
          12
          arrow-down
          1
          ·
          5 months ago

          I think there’s a larger picture at play here that is being missed.

          Getting the weather is a standard feature for years now. Nothing AI about it.

          What is “AI” is, Hey Siri, what is the weather at my daughter’s recital coming up?

          The AI processing, calculated on-device if what they claim is true, is:

          1. the determination of who your daughter is
          2. What is a recital? An event? Are there any upcoming calendar events that match this concept?
          3. Is the “daughter” associated with this event by description or invitation? Yes? OK, what’s the address?
          4. Submit zip code of recital calendar event involving the kid to the weather API, and churn out a reply that includes all this information…

          Well {Your phone contact name}, it looks like it will {remote weather response} during your {calendar event from phone} with {daughter from contacts} on {event date}.

          That is the idea between on-device and cloud processing. The phone already has your contacts and calendar and does that work offline rather than educating an online server about your family, events and location, and requests the bare minimum from the internet, in this case nothing more than if you opened the weather app yourself and put in a zip code.

          • Blue_Morpho@lemmy.world
            cake
            link
            fedilink
            arrow-up
            2
            ·
            5 months ago

            Nothing AI about it.

            Voice processing is AI and was done by Apple servers. Previously, only the keyword “Hey Siri” was local. Onboard AI chips will allow this to be local. The actual queries will go to the servers. Phones do not have the power to run useful LLM locally- at least not with the near instantaneous response times phone users expect. A 56 Watt 128GB RAM M3 Max does around 8.5 tokens/second.

            https://www.nonstopdev.com/llm-performance-on-m3-max/

            • PassingThrough@lemmy.world
              link
              fedilink
              English
              arrow-up
              1
              ·
              edit-2
              5 months ago

              Onboard AI chips will allow this to be local.

              Phones do not have the power to ~~~

              Perhaps this is why these features will only be available on iPhone 15 Pro/Max and newer? Gotta have those latest and greatest chips.

              It will be fun to see how it all shakes out. If the AI can’t run most queries on the phone with all this advertising of local processing…there’ll be one hell of a lawsuit coming up.

              EDIT: Finished looking for what I thought I remembered…

              Additionally, Siri has been locally processed since iOS 15.

              https://www.macrumors.com/how-to/use-on-device-siri-iphone-ipad/

              • Blue_Morpho@lemmy.world
                cake
                link
                fedilink
                arrow-up
                2
                ·
                5 months ago

                Perhaps this is why these features will only be available on iPhone 15 Pro/Max and newer?

                I’m not guessing. I linked to the article about the M3 which is much more powerful than the a17 pro in the 15 pro and has the same NPU.

                • PassingThrough@lemmy.world
                  link
                  fedilink
                  English
                  arrow-up
                  1
                  ·
                  5 months ago

                  Forgive me, I’m no AI expert to fully compare the needed tokens per second measurement to relate to the average query Siri might handle, but I will say this:

                  Even in your article, only the largest model ran at 8/tps, others ran much faster, and none of these were optimized for a task, just benchmarking.

                  Would it be impossible for Apple to be running an optimized model specific to expected mobile tasks, and leverage their own hardware more efficiently than we can, to meet their needs?

                  I imagine they cut out most worldly knowledge etc/use a lightweight model, which is why there is still a need to link to ChatGPT or Apple for some requests, would this let them trim Siri down to perform well enough on phones for most requests? They also advertised launching AI on M1-2 chip devices, which are not M3-Max either…

        • MotoAsh@lemmy.world
          link
          fedilink
          arrow-up
          3
          arrow-down
          1
          ·
          edit-2
          5 months ago

          Literally not what people are talking about. It’s the “AI” part of the task that doesn’t leave the device (unless it prompts to ask chat gpt). Not that it can magically gleam live info without making any request to the web…

          Jeeze, fucking… get your shit straight, making me defend Apple… Fucking do better.

        • lolcatnip@reddthat.com
          link
          fedilink
          English
          arrow-up
          3
          arrow-down
          1
          ·
          5 months ago

          The “AI” parts are what they’re saying happens on the device. This isn’t a gotcha.

    • moshtradamus666@lemmy.world
      link
      fedilink
      arrow-up
      9
      arrow-down
      1
      ·
      edit-2
      5 months ago

      I don’t think it’s a problem, more like a situation. You are not doing anything wrong or stupid, just interested in something new and promising and have the resources to pursue it. Good for you, may you find gold.

    • Naz@sh.itjust.works
      link
      fedilink
      arrow-up
      20
      arrow-down
      2
      ·
      edit-2
      5 months ago

      I’m an AI Developer.

      TLDR: CUDA.

      Getting ROCM to work properly is like herding cats.

      You need a custom implementation for the specific operating system, the driver version must be locked and compatible, especially with a Workstation / WRX card, the Pro drivers are especially prone to breaking, you need the specific dependencies to be compiled for your variant of HIPBlas, or zLUDA, if that doesn’t work, you need ONNX transition graphs, but then find out PyTorch doesn’t support ONNX unless it’s 1.2.0 which breaks another dependency of X-Transformers, which then breaks because the version of HIPBlas is incompatible with that older version of Python and …

      Inhales

      And THEN MAYBE it’ll work at 85% of the speed of CUDA. If it doesn’t crash first due to an arbitrary error such as CUDA_UNIMPEMENTED_FUNCTION_HALF

      You get the picture. On Nvidia, it’s click, open, CUDA working? Yes?, done. You don’t spend 120 hours fucking around and recompiling for your specific usecase.

      • barsoap@lemm.ee
        link
        fedilink
        arrow-up
        5
        ·
        edit-2
        5 months ago

        Also, you need a supported card. I have a potato going by the name RX 5500, not on the supported list. I have the choice between three rocm versions:

        1. An age-old prebuilt, generally works, occasionally crashes the graphics driver, unrecoverably so… Linux tries to re-initialise everything but that fails, it needs a proper reset. I do need to tell it to pretend I have a different card.
        2. A custom-built one, which I fished out of a docker image I found on the net because I can’t be arsed to build that behemoth. It’s dog-slow, due to using all generic code and no specialised kernels.
        3. A newer prebuilt, any. Works fine for some, or should I say, very few workloads (mostly just BLAS stuff), otherwise it simply hangs. Presumably because they updated the kernels and now they’re using instructions that my card doesn’t have.

        #1 is what I’m actually using. I can deal with a random crash every other day to every other week or so.

        It really would not take much work for them to have a fourth version: One that’s not “supported-supported” but “we’re making sure this things runs”: Current rocm code, use kernels you write for other cards if they happen to work, generic code otherwise.

        Seriously, rocm is making me consider Intel cards. Price/performance is decent, plenty of VRAM (at least for its class), and apparently their API support is actually great. I don’t need cuda or rocm after all what I need is pytorch.

    • ShrimpCurler@lemmy.dbzer0.com
      link
      fedilink
      arrow-up
      4
      ·
      5 months ago

      I think it’s in the pipeline. AMD has bought Xilinx, which builds FPGAs and already had some AI specific cores in their processors. I believe they’re developing that further and integrating it in their GPUs now.

    • megopie@beehaw.org
      link
      fedilink
      arrow-up
      4
      ·
      5 months ago

      So, AMD has started slapping the AI branding on to some of their products, but they haven’t leaned in to it quite as hard as Nvidia has. They’re still focusing on their core product line up and developing the actual advancements in chip design.

  • urska@lemmy.ca
    link
    fedilink
    arrow-up
    1
    arrow-down
    1
    ·
    5 months ago

    Nvidia and IA technology are both legit. Those companies need nvidia GPU for their development.