AMD Ryzen AI Halo – $4k AI Dev Kit

Join the discussion

• Hacker News
• Was “only” $2k in its previous form but even in this updated box the mem bandwidth is woefully inadequate. There’s a few models with space for a dedicated GPU for hybrid inference but imo not worth it. Save your money for a Xeon or EPYC build
  by robotswantdata
• 32 Gb DDR4 RAM module has a bandwidth of 25 Gb/s and costs $160. If you buy 8 of these, you get 256 Gb RAM with 200 Gb/s bandwidth at $1280. And if you buy 16 x 16 Gb modules (each at $60) then you can get 400 Gb/s of bandwidth for $960.

  The only problem, you need 8 or 16 memory controllers. Memory controllers are not that expensive: Intel Core i3-14100F has 2 channel controller and costs $110, so we can estimate that 16-channel controller should cost not more than $880, and 8-channel controller should cost $440.

  So isn't it better to make a cheap CPU with 16 DRAM controllers instead of this $4K gear having only 128 Gb? Or maybe 2 CPUs each having 8 RAM channels?

  DDR5 costs 2 times more ($360 for 32 Gb) while not even having 2 times the bandwidth so it is not worth buying. It is more reasonable to make more RAM channels and stuff them with DDR4.

  by codedokode
• So what I am trying to say, industry took a wrong turn. Instead of moving to over-priced DDR5, they should just make even cheapest CPUs support 8/16 DDR4 channels. Because a 32Gb DDR5-4800 module costs $360, and two 32Gb DDR4-3200 modules cost $320, so you get twice more size, more bandwidth and it costs you less. DDR5 is just a rip off.
  by codedokode
• If you want Epyc go for it. The motherboards can be quite expensive though.
  by wmf
• Why do all similar products have a hard limit on the 128 GB VRAM part? For that price, I hoped to get at least 224 GB VRAM
  by ahmedehab_01
• All the gpu makers make all their profit selling datacenter products. They don’t want consumer/home lab stuff with lower margins to replace their data center products so they handicap the vram in those products to make them less enticing for datacenter use.
  by HDThoreaun
• The 495 is going to support 192 GB. It depends on the memory bus.

  128 bit: 96 GB?

  256 bit: 192 GB

  512 bit: 384 GB?

  1024 bit: 768 GB?

  by wmf
• Because it’s a limit of the platform

  https://community.frame.work/t/was-there-no-possible-way-to-...

  by croes
• i wish there was a system like strix halo, but with enough lanes for a dedicated PCIe 5.0 x16 slot so you can have the best of both worlds: large sparse models on CPU with unified memory, dense models on GPU with real tensors and higher bandwidth memory.
  by woodrowbarlow
• Maybe this is what Medusa Halo will turn out to be?
  by nmfisher
• AMEN! 100% agree.

  Plus a reasonably inexpensive super low-latency interconnect.

  by bronson
• 256gbs memory bandwidth is about 1/4 that of a 3090. It would be a better buy with half the memory at 4x the speed.
  by aunty_helen
• it depends

  it allows you to run smaller models much better

  imo 3090s make the most sense if you can buy at least 2x ideally 4x but of course we're talking about a completely different budget at that point

  by muyuu
• what matters is how much memory it has; with the new MTP models, Qwen3.6 with 35B MOE, it's pumping out tokens up to ~80k context with little slow down.

  It's great to get lots of tokens, but being able to handle and extent context is why it'll continue to be a great machine compared to any of the small graphics cards.

  by cyanydeez
• Any performance gains caused by the internal bandwidth of the card will evaporate once you spill into system RAM, because now your bottleneck is probably a slow PCI lane.

  And if your jobs do fit onto a 24GB card, then you are not the target user for the "AI mini PC" niche that these guys are trying to carve out

  by StableAlkyne
• Are you sure about that? High memory speed is great for dense models, or when serving at high concurrency.

  However for local single-user setups, it's often better to have access to more capable/bigger MoE models at reasonable speeds and lower concurrences, which is enabled by these platforms.

  by wolttam
• Wow the prices on these have really come up.. Got my Framework desktop mainboard (Just the motherboard + CPU + soldered 128gb RAM) in Dec 2025 for ~1900 EUR
  by ndom91
• Indeed December 2025 was the best time to buy.
  by kccqzy
• I have a Strix Halo device, and like it, but at this price, might as well buy the Nvidia-based ASUS GX 10, if you're buying it for AI. CUDA remains the stronger ecosystem. The AMD is a better desktop machine, as it has a better CPU, but the Nvidia will be a little faster and a little better supported for inference and training workloads. You can almost always do the same things with ROCm, but you're going to work a little more.

  Though, I will say that Nvidia ships a dogshit custom Ubuntu on their hardware that's hard to deal with. Nvidia is not good at software. I keep thinking they'll get better at it, but I've been dealing with their Jetson line for a couple of years now, and it still sucks. Still a clumsy custom Ubuntu, and it's not as easy as simply installing a different Linux version as it's a complicated image-based thing and no UEFI. At least, I assume they ship Ubuntu on the big devices; I've only dealt with the little embedded Jetson machines. The AMD stuff, being a regular x86_64 PC, you can install pretty much any Linux. I immediately put Fedora on mine.

  by SwellJoe
• I ... don't find the Ubuntu on my Spark to be dogshit? It's ... fine? It's just Ubuntu. Hasn't given me any grief and it's so far the only vendor I've seen that actually ships a properly supported Linux on an ARM64 device for Linux, so there's that. I use my ASUS GX10 as my daily driver, my primary workstation. Only thing that doesn't work for me on it is Spotify (probably some DRM thing). Oh, and there's no Signal ARM64 client, it seems.

  The big advantage of the DGX Spark over the Strix Halo is much faster prefill. Like 5x the speed. Also the networking hardware on it is insanely powerful, though I and 99% of other Spark users, are unlikely to use it to its full capacity.

  by cmrdporcupine
• I got a Beelink GTR 9 Pro for $1980. These Strix Halo systems were a good deal at $2k when the alternative was a DGX Spark (which is similarly memory-constrained, but has about twice the iGPU processing power of the Radeon 8060S, having as many CUDA cores as an RTX 5070) for $4k. The pitch was basically "half the GPU compute (negative), x86 instead of ARM (positive), but no CUDA (negative), for half the price (positive), but you also don't get the ConnectX-7 NIC (negative)". These more or less balanced out to being worth it if you wanted a single-node system that could also double as a generic x86 homelab server once it was obsolete for LLM workloads.

  These days, you can get a DGX Spark for $4.7k, so yes, the price has risen, but Strix Halo (with a few exceptions like the Bosgame and Corsair systems) $4k (or more!) is simply not a very good deal. If I were buying new right now, I'd 100% go DGX Spark without even thinking about it.

  Gorgon Halo (releasing this fall/winter) is allegedly coming with 3GB memory chips, enabling a 192GB maximum unified memory SKU, alongside minor (100MHz) clock bumps in the iGPU and CPU, plus memory bumping from 8000 MT/s to 8533 MT/s (matching the MBW of the DGX Spark), and is otherwise unchanged. I fear these will be $5000+. At $3000, these would be awesome. At $5000, not so much.

  by anonym29
• These devices were great when they were cheaper than the DGX Spark.

  But when they cost the same price (unless the Spark has shot up too), there's no reason to buy this over a Spark.

  The Spark is literally a faster version of this, with better software support.

  Edit: And I say that as an owner of a Ryzen AI Max 395 device.

  by Catloafdev
• yeah, if only there wasn't some global hegemony that immediately drove up the price of all memory everywhere...
  by cyanydeez
• Spark has also gone up in price, my second cost me $500 more than my first a couple months earlier (both this spring)
  by verdverm
• Yep, the only reason I bought mine (in late 2025, before hardware prices went totally crazy) was because it was half the price of a Spark. I spent a while fiddling around with the right Linux kernel, kernel firmware, ROCm installs, etc.
  by icedchai
• Depends on what you use it for.

  The CPU of Ryzen is better than that of DGX Spark, especially for modern programs that have been updated to use AVX-512 (i.e. it has a significantly higher multithreaded performance).

  Only for GPU applications the NVIDIA system is likely to be better.

  by adrian_b
• Cheapest I've been able to get a DGX Spark FE is now around $4700 just FYI. This is from multiple vendors in higher-ed.
  by grubbs
• I dropped a Framework mainboard in rack mount case and use it as a speedy low power x86 homelab as well as an inference server.
  by seemaze
• Ability to run any OS is a pretty nice benefit versus the spark.
  by kcb
• I really want a 128gb+ machine but it's brutal to be at only 256 GB/s for $4k (especially with the drawbacks of both ARM and AMD).

  I fear that by the time the RTX Spark comes out it'd have to be $6k, and by the time a 128gb or more machine with 700+ GB/s comes out it'd be at $10k, way out of most consumers' hands.

  Edit: capitalized gb/s to GB/s.

  by Tenoke