[HN Gopher] The Framework Desktop is a beast
___________________________________________________________________
The Framework Desktop is a beast
Author : lemonberry
Score : 93 points
Date : 2025-08-08 20:19 UTC (1 days ago)
(HTM) web link (world.hey.com)
(TXT) w3m dump (world.hey.com)
| Jtsummers wrote:
| > In some ways, the Framework Desktop is a curious machine.
| Desktop PCs are already very user-repairable! So why is Framework
| even bringing their talents to this domain? In the laptop realm,
| they're basically alone with that concept, but in the desktop
| space, it's rather crowded already. Yet it somehow still makes
| sense.
|
| And even more curious, Framework Desktop is deliberately less
| repairable than their laptops. They soldered on the RAM. Which
| makes it a very strange entry for a brand marketing itself as the
| DIY dream manufacturer. They threw away their user-repairable
| mantra when they made the Desktop, it's _less_ user repairable
| than most other desktops you could go out and buy today.
| sethops1 wrote:
| The RAM is soldered on all Halo Strix platforms because physics
| is getting in the way. With pluggable DIMMs the memory
| bandwidth would be halved, at best.
| wishinghand wrote:
| Why is that? Why would soldering the connections vs plugging
| them in affect how much data per second they transfer?
| colejohnson66 wrote:
| Sockets have resistance and crosstalk, which affects signal
| integrity.
| aeonik wrote:
| Wait, your telling me, I should have been desoldering the
| sockets off my motherboard, and directly soldering my RAM
| to the leads this entire time?
| ElectricalUnion wrote:
| Compression Attached Memory Module (CAMM) tries to be a
| middle-term solution for that, by reducing how crappy
| your average RAM socket is to latency and signal
| integrity issues. But, at this point, I can see CAMM
| delivered memory being reduced to a sort of slower,
| "CXL.mem" device.
| aeonik wrote:
| Seriously though,
|
| Would desoldering the sockets help?
|
| Why are the sockets bad?
| wmf wrote:
| Yes. (That isn't actually possible because the pinouts
| are different but soldered RAM is faster.)
| tejtm wrote:
| mind the gap
| onli wrote:
| He is still right. It is a desktop PC that is less repairable
| than all other desktop PCs, from a brand that is known to
| champion repairability. They had a reason for it, but
| could've chosen to not create more throwaway things.
| trenchpilgrim wrote:
| The 128GB version wouldn't be throwaway, since that's the
| maximum the platform as a whole supports anyway- more
| memory than that would require a new mainboard and CPU at
| the same time.
| nemomarx wrote:
| Not for upgrade reasons but what if you have a fault in
| one of the dimms or etc? Now you can't just drop in a
| replacement without changing everything.
| cyanydeez wrote:
| it's basically the same as asking what happens when your
| M4 Apple has a fault. It's soldered based on the desire
| to use the ram as part of the GPU.
|
| Without that, it's really not a interesting solution.
|
| demanding replaceable ram means also not wanting the
| benefits of the integrated memory
| beeflet wrote:
| the platform should support more memory
| richardw wrote:
| That applies to all computers when you buy the fully
| specced versions on day 1. A maxxed out iPad isn't
| throwaway, but framework represents a higher standard of
| upgradability.
| epistasis wrote:
| I have been continuously baffled by the people that think
| that soldered on RAM is somehow "throwaway". My last
| desktop build is eight years old and I have never upgraded
| the ram. Never will. My next build will have an entirely
| new motherboard, ram, and GPU, and the last set will end up
| at the ewaste recycler, because who could I find that wants
| that old hardware?
|
| Soldered RAM, CPU, and GPU, that give space benefits and
| performance benefits is exactly what I want, and results in
| no more ewaste at all. In fact less ewaste, because if I
| had a smaller form factor I could justify keeping the older
| computer around for longer. The size of the thing is a
| bigger cause of waste for me than the ability to upgrade
| RAM.
|
| Not everybody upgrades RAM, and those people deserve
| computers too. Framework's brand appears to be offering
| something that other suppliers are not, rather than expand
| ability. That's a much better brand and niche overall.
| onli wrote:
| > _Not everybody upgrades RAM, and those people deserve
| computers too._
|
| No. It's end of the line with consumerism and we either
| start repairing and recycling or we die. Framework
| catered to people who agree with that, and this product
| is not in line.
|
| I have no idea why you would not upgrade your memory, I
| have done so in all PCs I ever owned and all laptops, and
| it's a very common (and cheap) upgrade. It reduces waste
| because people can then use their system longer, which
| means less garbage over the lifetime of a person. And as
| was already commented, it is not only about upgrades, but
| also about repairs. Ram breaks rather often.
| epistasis wrote:
| I have had the system for eight years and at no point
| would upgrading RAM have increased performance.
|
| Upgrading the RAM would have created _more_ waste than
| properly sizing the RAM to COU proportion from the
| beginning.
|
| It is very odd to encounter someone who has such a narrow
| view of computing that they cannot imagine someone not
| upgrading their RAM.
|
| I have not once, literally not once have RAM break
| either. I have been part of the management of clusters of
| hundreds of compute nodes, that would occasionally each
| have their failures, but not once was RAM the cause of
| failure. I'm fairly shocked to hear that anybody's RAM
| has failed, honestly, unless it's been overlocked or
| something else.
| onli wrote:
| > _It is very odd to encounter someone who has such a
| narrow view of computing that they cannot imagine someone
| not upgrading their RAM._
|
| Uncalled for and means the end of the discussion after
| this reaction. Ofc I can imagine that, it's just usually
| a dumb decision.
|
| That you did not have to upgrade the ram means one of two
| things: You either had completely linear workloads, so
| unlike me did not switch to a compiled programming
| language or experimented with local LLMs etc. Or you
| bought a lot of ram in the beginning, so 8 years ago with
| a hefty premium.
|
| Changes nothing about the fundamental disagreement with
| the existence of such machines. Especially from a company
| that knows better. I do not expect ethical behaviour from
| a bottom of the barrel company like Apple, but it was
| completely reasonable to expect better from framework.
| v5v3 wrote:
| >I have been continuously baffled by the people that
| think that soldered on RAM is somehow "throwaway"
|
| One of the primary objections to soldered RAM was/is the
| cost to purchase. As the likes of Apple priced Ram
| upgrade at a hefty premium to retail prices.
| epistasis wrote:
| I can see that objection too, and it seems far more
| reasonable than assuming that soldered RAM automatically
| means a reduced lifespan machine.
|
| But are Framework's RAM prices unreasonable? $400 for
| 64GB more of LPDDR5x seems OK. I haven't seen anybody
| object to Framework's RAM on those grounds.
| linotype wrote:
| The only part I've ever had not fail on a PC or Laptop is
| RAM.
| beeflet wrote:
| Why not make a platform with a greater number of channels
| aDyslecticCrow wrote:
| That's a question for AMD and TCMC. They only have so much
| space on the silica. More memory channels means less of
| something else. This is not a "framework platform" issue,
| it's the specification of that CPU.
| beeflet wrote:
| Well they chose to use this hardware platform. It all
| sounds like market segmentation to me, now that AMD is on
| top.
| wmf wrote:
| To be clear, AMD is giving you 2x the bandwidth of
| competing chips and you're complaining that it isn't 4x.
| beeflet wrote:
| My complaints are the maximum RAM of the system and the
| modularity of the RAM.
|
| With an increased number of channels, you could have a
| greater amount of RAM at a lower frequency but at the
| same bandwidth. So you would at least be able to run some
| of these much larger AI models.
| v5v3 wrote:
| Risk cannabalising sales from their other products?
|
| For example Nvidia seek to ban consumer GPU use in
| datacenters as they to sell datacentre GPUs.
|
| If they made consumer platforms that can take 1tb of ram
| etc, then people may choose to not buy EYPC.
|
| Afterall many cloud providers already offer Ryzen VPS's.
| undersuit wrote:
| I wonder if there were similar complaints when cache moved
| from motherboards to soldered on the cpu package.
| beeflet wrote:
| The difference in performance between DRAM and flash memory
| is far greater than SRAM and DRAM. The total RAM of a
| system is a hard limit on the type of programs you can
| practically run because swapping is slow.
| antonvs wrote:
| > And even more curious
|
| It's easy to find out the reason for this. And the article's
| benchmarks confirm the validity of this reason. Why comment
| from a place of ignorance, unless you're asking a question?
| aniviacat wrote:
| Your comment is unnecessarily hostile.
|
| There are plenty of components to choose from which do not
| need soldered-on RAM. Giving up modularity to gain access to
| higher memory bandwidth is certainly a trafeoff many are
| willing to make, but to take that tradeoff as a company known
| for modularity is, as the parent comment put it, curious.
| antonvs wrote:
| Every single description of the Framework desktop that I've
| seen has addressed this issue. To comment as though it's
| some sort of mystery is disingenuous at best. My comment
| was precisely as friendly as the commenter deserved.
|
| And as I said, if you read the article, you'll see that the
| tradeoff in question has paid off very well.
| suspended_state wrote:
| > They soldered on the RAM. Which makes it a very strange entry
| for a brand marketing itself as the DIY dream
|
| This was also my first thought when discovering this new model,
| but I think it was a pragmatic design decision.
|
| The questions you should ask yourself are:
|
| - which upgradable memory module format could be used with the
| same speed and bandwidth as the soldered in solution,
|
| - if this solution exists, how much would it cost
|
| - what's the maximum supported amount of ram for this CPU
| beeflet wrote:
| >which upgradable memory module format could be used with the
| same speed and bandwidth as the soldered in solution
|
| CAMM perhaps? The modular memory is important, because they
| are selling them to two different markets: gamers that want a
| small powerful desktop, and people running LLMs at home. The
| modularity of the RAM allows you to convert the former into
| the latter at a later date, so it seems pretty critical to
| me.
|
| For this reason alone, I am going to buy a used epyc server
| instead of one of these desktop things. I will be able to
| equip it with a greater amount of RAM as I see fit and run a
| greater range of models (albeit at lower speed). The ability
| to run larger models slowly at a later date is more important
| than the ability for me to run smaller models faster now. So
| I am an example of a consumer who does not like framework's
| tradeoff.
|
| You would think that they would at least offer some type of
| service where they take it into the factory and refit it with
| new ram chips. Perhaps they could just buy used low-ram
| boards at a later date and use them to make refurbished high-
| ram boards.
|
| Another solution is to make it so that it supports both
| soldered and unsoldered ram (but at a lower frequency).
| Gaming is frequency-limited but does not require much ram,
| but a lot of workloads like AI are bandwidth limited. Hell,
| if you're going to have some high-frequency RAM irreplacibly
| soldered to the motherboard, it might as well be a chiplet!
| trenchpilgrim wrote:
| The CEO mentioned in an LTT video that they worked with AMD to
| try to make CAMM memory work and hit some technical problems.
| bluescrn wrote:
| Gotta chase that AI bubble at any cost
| jeroenhd wrote:
| According to the Framework CEO on the Linus Tech Tips video
| about this thing [1], they tried and AMD assigned an engineer
| on getting modular memory to work and decided it's not
| possible.
|
| Unless there's another company out there shipping this CPU with
| replaceable memory, I'll believe them. Even with LPCAMM,
| physics just doesn't work out.
|
| Of course, there are plenty of slower desktops you can buy
| where all the memory can be swapped out. If you want a
| repairable desktop like that, you're not stuck with Framework
| in the same way you would be with laptops. You'll probably
| struggle to get the same memory bandwidth to make full use of
| AMD's chips, though.
|
| [1]: https://youtu.be/-lErGZZgUbY?feature=shared&t=446
| cyanydeez wrote:
| there isn't. This thing was designed the same way apple
| designed their unified memory. This thing is meant to work
| hand in hand with it's iGPU.
| tgma wrote:
| Even on regular AMD 7000 and 9000 series the DDR5 memory
| controller is very sensitive and hard to get a stable system
| with fast RAM on many motherboards when all 4 modules are
| present. At today's RAM speeds, I definitely think a stable
| soldered system is increasingly a better trade-off.
| pzmarzly wrote:
| HN admins: can the domain extractor be changed to say
| "world.hey.com/dhh" here instead of just the domain name? From
| what I see, Hey World is a blogging platform, similar to Medium
| but markdown and email based. And the username (blog name) is in
| the second part of the URL.
| mtmail wrote:
| Best to email the moderators (link in footer). I've made
| similar suggestions about other blogging platforms and got a
| positive reply.
| ahmedfromtunis wrote:
| It fun to see that, in an era where most CEOs are all-in with AI
| both on a personal level and at their companies, dhh chose to
| rather take a deep dive into the world of linux and config files
| and indie computer brands.
|
| Curious what will the long term impact of this be on the longtime
| viability of Basecamp and its sister/daughter brands.
| OrvalWintermute wrote:
| very impressive...
|
| Max+ 395 specced with: 128GB of non-upgradeable LPDDR5x WD_BLACK
| SN850X NVMe - M.2 2280 - 8TB Noctua fan 3x + 3x extra USB-A &
| USB-C ports No OS option.
|
| only $2,776.00!!!
| micromacrofoot wrote:
| everything's cheap when you're rich
| OrvalWintermute wrote:
| A mac studio configured similarly With 128 GB of memory and
| an 8 TB SSD is $5799
| mft_ wrote:
| One detail is that the memory bandwidth on the M4 Max and
| M3 Ultra (especially) is considerably higher than the 395+.
| jeroenhd wrote:
| The Mac Studio (at least the M3 Ultra model) blows the
| Framework out of the water when it comes to AI performance,
| though, at least according to this (Dutch language)
| benchmark: https://tweakers.net/reviews/13614/framework-
| desktop-de-fram...
|
| Paying twice the price for twice to seven times the
| performance may not be such a bad thing. Then again, with
| Apple you're kind of stuck with macOS and the like, so
| Framework may still be the better option depending on your
| use case.
| everfrustrated wrote:
| Bear in mind the gpu has access to all of that 128GB as well,
| so for AI thats very very cheap.
| cyanydeez wrote:
| yeah, slap the Apple brand on this and it's basically the
| same thing.
|
| People seem to really not understand the limits of wanting
| unified memory architecture.
| ashleyn wrote:
| How is AMD GPU compatibility with leading generative AI
| workflows? I'm under the impression everything is CUDA.
| wolfgangK wrote:
| Indeed, recent Flash Attention is a pain point for non CUDA.
| trenchpilgrim wrote:
| Certain chips can work with useful local models, but
| compatibility is far behind CUDA.
| pja wrote:
| llama.cpp combined with Mesa's Vulkan support for AMD GPUs has
| worked pretty well with everything I've thrown it at.
| throwdbaaway wrote:
| https://llm-tracker.info/_TOORG/Strix-Halo has very
| comprehensive test results for running llama.cpp with Strix
| Halo. This one is particularly interesting:
|
| > But when we switch to longer context, we see something
| interesting happen. WMMA + FA basically loses no performance
| at this longer context length!
|
| > Vulkan + FA still has better pp but tg is significantly
| lower. More data points would be better, but seems like
| Vulkan performance may continue to decrease as context
| extends while the HIP+rocWMMA backend should perform better.
|
| lhl has also been sharing these test results in
| https://forum.level1techs.com/t/strix-halo-ryzen-ai-
| max-395-..., and his latest comment provides a great summary
| of the current state:
|
| > (What is bad is that basically every single model has a
| different optimal backend, and most of them have different
| optimal backends for pp (handling context) vs tg (new text)).
|
| Anyway, for me, the greatest thing about the Strix Halo +
| llama.cpp combo is that we can throw one or more egpu into
| the mix, as echoed by level1tech video
| (https://youtu.be/ziZDzrDI7AM?t=485), which should help a lot
| with PP performance.
| kristianp wrote:
| > There's at least a little flexibility with the graphics card if
| you move the board into a different case--there's a single PCIe
| x4 slot on the board that you could put an external GPU into,
| though many PCIe x16 graphics cards will be bandwidth starved.
|
| https://arstechnica.com/gadgets/2025/08/review-framework-des...
| conradev wrote:
| You can also use an adapter to repurpose an M.2 slot as PCIe
| x16, but the bandwidth is the same x4
| tgma wrote:
| That's just called a PCIe x4 [1]. Each PCIe lane is an
| independent channel. The wider slot will simply have
| disconnected pins. You can actually do this with regular
| motherboard PCIe x4 slots by cutting the plastic at the end
| of the slot so you can insert a wider card and most cards
| work just fine.
|
| [1]: It sounds like a nitpick but a PCIe x16 with x4
| effective bandwidth can exist and is a different thing: if
| the actual PCIe interface is x16, but there is an upstream
| bottleneck (e.g. aggregate bandwidth from chipset to CPU is
| not enough to handle all peripherals at once at full rate.)
| wolfgangK wrote:
| For LLM inference, I don't think the PCIe bandwidth matters
| much and a GPU could improve greatly the prompt processing
| speed.
| ElectricalUnion wrote:
| Only if your entire model fits the GPU VRAM.
|
| To me this reads like "if you can afford those 256GB VRAM
| GPUs, you don't need PCIe bandwidth!"
| jgalt212 wrote:
| Yeah, I think so. Once the whole model is on the GPU
| (potentially slower start-up), there really isn't much
| traffic between the GPU and the motherboard. That's how I
| think about it. But mostly saying this as I'm interested in
| being corrected if I'm wrong.
| monster_truck wrote:
| There are no situations where this matters yet. You have to
| drop down to an 8x slot on PCIe 3.0 to even begin to see any
| meaningful impact on benchmarks (synthetic or otherwise)
| chvid wrote:
| What is it with this guy and his bogus anti Apple benchmark?
| micromacrofoot wrote:
| he's mad they take a cut of his app store subscriptions, this
| is a huge saga that's been spanning years for him
|
| (not sure why I'm being downvoted, it's true...
| https://x.com/dhh/status/1747697778455962014)
| linotype wrote:
| Not sure why you're being downvoted. He's also the guy that
| inflicted Ruby on Rails upon us.
| TheRealDunkirk wrote:
| Sorry, but what's "bogus" about benchmarking your specific
| workloads? What benchmark do you think he should have run?
| Rohansi wrote:
| One more that favors Apple, obviously. 2/3 of the benchmarks
| having Apple Silicon on top is not enough!
| OrvalWintermute wrote:
| I don't know about pro or anti Apple, but the framework
| optional 8TB NVME SSD is only +$699 whereas Apple charges $2200
| for a similar 8TB NVME SSD
| dlivingston wrote:
| Context on DHH vs. Apple: https://youtu.be/mzDi8u3WMj0
|
| I find myself agreeing with much of what he says. The Apple of
| today is not the same as the Apple of 15 years ago.
| beeflet wrote:
| is the apple fief facing resistance? I was under the
| impression that Epic v. Apple was a nothingburger
| andrepd wrote:
| What could you possibly be talking about? The only benchmark I
| see in the article where Apple M4 isn't near the top is the
| rails test suite.
| dustingetz wrote:
| for $2k that is a lot of computer
| Xenoamorphous wrote:
| Oh, to be young again and care about benchmarks, bogomips and
| stuff like that.
| beeflet wrote:
| Do you even bench, mark?
| Terr_ wrote:
| Or to generally be back in the era where such a thing would be
| useful for LAN-parties...
| TiredOfLife wrote:
| Note that the main benchmark in the post is practical. How long
| a test suite for the product he makes runs.
| ThinkBeat wrote:
| Reads like an advertisement to me.
| coolgoose wrote:
| I doubt that Dhh if all people would do advertisement (as paid)
| for this. He genuinely seems to enjoy his macos break
| echelon wrote:
| Apple's walled garden and mob tactics pissed DHH off [1-3].
| Of course he's going to advocate for alternatives.
|
| Tim Sweeney is in the same camp.
|
| [1] https://www.hey.com/apple/
|
| [2] https://world.hey.com/dhh/apple-rejects-the-hey-calendar-
| fro...
|
| [3] https://world.hey.com/dhh/hey-is-finally-for-sale-on-the-
| iph...
| jamesgill wrote:
| I like Framework and own one of their laptops. But the desktop
| seems more a triumph of gimmicky marketing than a desktop that's
| meaningfully different. And, it seems significantly overpriced.
| zozbot234 wrote:
| It's taking a newly released mobile- and mini-PC-focused
| platform that's usually paired with proprietary technology, and
| building something that's as close as possible to a standard
| desktop with it. Seems very much in the Framework spirit once
| you account for that side of it.
| damonll wrote:
| You need to use Orbstack as the engine on Mac, otherwise it's not
| a fair fight. It's at least 2x as fast as Docker
| fmajid wrote:
| I cancelled my Frameword Desktop order and ordered a HP Z2 Mini
| G1a instead, the goal being to replace my Mac Studio as I've had
| it with Apple's arrogance and lousy software quality. The HP is
| much smaller, has ECC RAM and 10G Ethernet. Significantly more
| expensive, however.
___________________________________________________________________
(page generated 2025-08-09 23:02 UTC)