I'm not a gamer and there is not enough memory in this thing for me to care to use it for AI applications so that leaves just one thing I care about: hardware accelerated video encoding and decoding. Let's see some performance metrics both in speed and visual quality
Intel announces Arc B-series "Battlemage" discrete graphics with Linux support
161–170 of 722 posts
Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support
#162Earlier quoted context omitted.
Nothing in my comment says about pricing it at the M4 Max level. Apple charges as much because they can (typing this on an $8000 M3 Max). 128GB LPDDR5 is dirt cheap these days just Apple adds its premium because they like to. Nothing prevents Intel from releasing a basic GPU with that much RAM for under $1k.
You're asking for a GPU die at least as large as NVIDIA's TU102 that was $1k in 2018 when paired with only 11GB of RAM (because $1k couldn't get you a fully-enabled die to use 12GB of RAM). I think you're off by at least a factor of two in your cost estimates.
Intel has Xeon Phi which was a spin-off of their first attempt at GPU so they have a lot of tech in place they can reuse already. They don't need to go with GDDRx/HBMx designs that require large dies.
Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support
#16312GB max is a non-starter for ML work now. Why not come out with a reasonably priced 24gb card even if it isn't the fastest and target it at the ML dev world? Am I missing something here?
ML is about hit another winter. Maybe intel is ahead of industry. Or we can keep asking high computers questions about programming.
Haven't people been saying that for the last decade? I mean, eventually they will be right, maybe "about" means next year, or maybe a decade later? They just have to stop making huge improvements for a few years and the investment will dry up.
I really wasn't interested in computer hardware anymore (they are fast enough!) until I discovered the world of running LLMs and other AI locally. Now I actually care about computer hardware again. It is weird, I wouldn't have even opened this HN thread a year ago.
Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support
#164Earlier quoted context omitted.
The ML dev world isn’t a consumer mass market like PC gaming is.
Launching a new SKU for $500-1000 with 48gb of RAM seems like a profitable idea. The GPU isn't top-of-the-line, but the RAM would be unmatched for running a lot of models locally.
You can’t just throw in more RAM without having the rest of the GPU architected for it. So there’s an R&D cost involved for such a design, and there may even be trade-offs on performance for the mass-market lower-tier models. I’m doubtful that the LLM enthusiast/tinkerer market is large enough for that to be obviously profitable.
Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support
#165Earlier quoted context omitted.
ML is about hit another winter. Maybe intel is ahead of industry. Or we can keep asking high computers questions about programming.
> ML is about hit another winter. I agree ML is about to hit (or has likely already hit) some serious constraints compared to breathless predictions of two years ago. I don't think there's anything equivalent to the AI winter on the horizon, though—LLMs even operated by people who have no clue how the underlying mechanism functions are still far more empowered than anything like the primitives of the 80s enabled.
Yeah… I want to think of it like mining, where you’ve found an ore vein. You have to switch from prospecting to mining. There’s a lot of work to be done by integrating our LLMs and other tools with other systems, and I think the cost/benefit of making models bigger, Bigger, BIGGER is reaching a plateau.
Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support
#166Earlier quoted context omitted.
HBM3E memory is at least 3x the price of DDR5 (it requires 3x the wafer as DDR5) and capacity is sold out for all of 2025 already... that's the price and production bottleneck. High speed, low latency server grade DDR5 is around $800-$1600 for 128GB. Triple that for $2400 - $4800 just for the memory. Still need the GPUs/APUs, card, VRMs, etc. Even the nVidia H100 with "only" 94GB starts at $30k...
Nvidia's $30,000 is a 90% margin product at scale. They could charge 1/3 that and still be very profitable. There has rarely been such a profitable large corporation in terms of the combo of profit & margin. Their last quarter was $35b in sales and $26b in gross profit ($21.8b op income; 62% op income margin vs sales). Visa is notorious for their extreme margin (66% op income margin vs sales) due to being basically a…
> And indeed that's exactly what will be required for any serious attempt to cut into their monopoly position.
You misunderstand why and how Nvidia is a monopoly. Many companies make GPUs, and all those GPUs can be used for computation if you develop compute shaders for them. This part is not the problem, you can already go buy cheaper hardware that outperforms Nvidia if price is your only concern.
Software is the issue. That's it - it's CUDA and nothing else. You cannot assail Nvidia's position, and moreover their hardware's value, without a really solid reason for datacenters to own them. Datacenters do not want to own GPUs because once the AI bubble pops they'll be bagholders for Intel and AMD's depreciated software. Nvidia hardware can at least crypto mine, or be leased out to industrial customers that have their own remote CUDA applications. The demand for generic GPU compute is basically nonexistent, the reason this market exists at all is because CUDA exists, and you cannot turn over Nvidia's foothold without accepting that fact.
The only way the entire industry can fuck over Nvidia is if they choose to invest in a complete CUDA replacement like OpenCL. That is the only way that Nvidia's value can be actually deposed without any path of recourse for their business, and it will never happen because every single one of Nvidia's competitors hate each other's guts and would rather watch each other die in gladiatorial combat than help each other fight the monster. And Jensen Huang probably revels in it, CUDA is a hedged bet against the industry ever working together for common good.
Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support
#167Earlier quoted context omitted.
ML is about hit another winter. Maybe intel is ahead of industry. Or we can keep asking high computers questions about programming.
Haven't people been saying that for the last decade? I mean, eventually they will be right, maybe "about" means next year, or maybe a decade later? They just have to stop making huge improvements for a few years and the investment will dry up. I really wasn't interested in computer hardware anymore (they are fast enough!) until I discovered the world of running LLMs and other AI locally. Now I actually care about com…
What makes local AI interesting to you vs larger remote models like ChatGPT and Claude?
Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support
#168I'm not a gamer and there is not enough memory in this thing for me to care to use it for AI applications so that leaves just one thing I care about: hardware accelerated video encoding and decoding. Let's see some performance metrics both in speed and visual quality
From what I have gathered, the alchemist av1 is about the same or sliiiightly worse than current nvenc. My a750 does about 1400fps for dvd encoding on the quality preset. I havent had the opportunity to try 1080p or 4k though.
Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support
#169Earlier quoted context omitted.
It's a chicken and egg scenario. The main problem with running inference at home is the lack of hardware. If the hardware was there more people would do it. And it's not a problem if "enthusiasts" are the only ones using it because that's to be expected at this stage of the tech cycle. If the market is small just charge more, the enthusiasts will pay it. Once more enthusiasts are running inference at home, then the l…
Mac minis are great for this. They're cheap-ish and they can run quite large models at a decent speed if you run it with an MLX backend.
mini _Pro_ are great for this, ones with large RAM upgrades.
If you get the base 16GB mini, it will have more or less the same VRAM but way worse performance than an Arc.
If you already have a PC, it makes sense to go for the cheapest 12GB card instead of a base mac mini.
Re: Intel announces Arc B-series "Battlemage" discrete graphics with Linux support
#170Earlier quoted context omitted.
Inference on very large LLMs where model + backprop exceed 48GB is already way faster on a 128GB MacBook than on NVidia unless you have one of those monstrous Hx00s with lots of RAM which most devs don't.
Because the CPU has to load the model in parts for every cycle so you're spending a lot of time on IO and it offsets processing. You're talking about completely different things here. It's fine if you're doing a few requests at home, but if you're actually serving AI models, CUDA is the only reasonable choice other than ASICs.
My comment was about Intel having a starter project, getting enthusiastic response from devs, network effects and iterate from there. They need a way to threaten Nvidia and just focusing on what they can't do won't bring them there. There is one route where they can disturb Nvidia's high end over time and that's a cheap basic GPU with lots of RAM. Like Ryzen 1st gen whose single core performance was two generations behind Intel trashed Intel by providing 2x as many cores for cheap.