RISC-V Motherboard For Framework Laptop Pricing Starts At $368 In Early Access

[pkese]

Can someone find some benchmark comparing JH7110 to ARM Cortex-A53 / Cortex-A55?

Most of benchmarks are comparing this CPU (JH7110) to much more powerful processors,
but technically this is on the same level as Little cores in the ARM ecosystem:

all these three have 8-stage dual issue inline superscalar (no-OOP) architecture
(think of original 1993's Intel Pentium / Pentium MMX basic architecture
but with more stages larger caches and optimized for modern silicon tech).

I'd be really interested how does it compare to the Cortex -Little series.
I'd expect that clock-for-clock RISC-V could be slightly faster in the integer computations
and Cortex in numerics (due to NEON) and memory throughput.
However Cortex-A55 does have higher clocked (2 GHz vs JH7110's 1.5 GHz) variants.

[ehansin]

AFAIK the Pro tier, which includes the FW13 shell/screen/keyboard/etc is all standard FW13 hardware & screen. You'll either like it or not so much. I'm typing this on a FW16, and I really like the system/design/ergonomics, by and large. YMMV.

There's a vibrant community of Framework folks DIYing expansion cards, what standard hardware can be swapped in, etc. They even have the full CAD of the shells/etc, all (or nearly all) board design specs/etc all posted to their github.

Regarding the first part, I do get that often things are subject to the end user. Glad people are happy though with the product, a good sign.

The second part excites me. I used to work some in electronics manufacturing. This is cool for people that like to tinker and learn. That alone is pretty awesome. Thanks fore heads up.

[Quackdoc]

--snip--

the issue is that the price the board is selling for, vs what it gives, doesn't make sense. Especially considering the form factor, you have to take your thousand dollar laptop, take the guts out, and put a barely usable, nearly 400 dollar device into it. It would be one thing if this was some chromebook esq cost laptop, but it's not, framework devices have a high entry fee, and then the motherboard has an insanely high entry fee especially considering the cost of other things in the ecosystem, and I get that the cost might make sense from a development point of view.

but for someone actually looking into a "riscv laptop" the specs just don't make sense. It's not something usable for daily driving, even if your distro did ship with compiler optimized bins, and now they have to replace, in many cases their likely daily driver. It should at least be a soc that hits the minimum bar for a daily driveable system.

[L_A_G]

For crying out loud about all the people complaining about the price... Have you never heard about something called "economies of scale"? Because that's why your consumer level stuff is so cheap while hobbyist and developer-level stuff is so much more expensive despite being made largely from the same components.

Lets explain it this way; Product A and product B are a consumer and a develop product respectively. They both cost the same to design and produce and are meant to merely break even after production costs. Product A, the consumer one, has an addressable market so large its going to sell 500.000 units while product B, the developer product probably won't sell more than 10.000. If they both cost a million to design, how do you think the need to cover this thru profit margin is going to affect the price of these two products?

[Quackdoc]

For crying out loud about all the people complaining about the price... Have you never heard about something called "economies of scale"? Because that's why your consumer level stuff is so cheap while hobbyist and developer-level stuff is so much more expensive despite being made largely from the same components.

Lets explain it this way; Product A and product B are a consumer and a develop product respectively. They both cost the same to design and produce and are meant to merely break even after production costs. Product A, the consumer one, has an addressable market so large its going to sell 500.000 units while product B, the developer product probably won't sell more than 10.000. If they both cost a million to design, how do you think the need to cover this thru profit margin is going to affect the price of these two products?

"economies of scale" does not justify charging 200 dollars for a device (no value pack stuff) which has almost identical specs selling for 40 dollars. It would be far cheaper for someone to take the Milk-V Mars CM, make some adapter boards, and plunk that in instead. or better yet, do something compatible with the lichee devices.

I only hope that DC Roma will migrate their laptops to the same form factor for their new devices, Perhapps the DC Roma III will have compatibility with framework mobos

[brucehoult]

before anyone comments, no the JH7110 is not a fast cpu, it's slow, like rpi3 levels of perf.

It's much more accurate to say it's slightly slower than a Pi 4, similar to A55 boards, unless the Arm code uses NEON.

8 GB RAM puts it completely out of Pi 3 (limited to 1 GB) class in what you can do with it, regardless of speed.

[brucehoult]

How long before the first offering comes that is in any way sane for normal use and not only developer tinkering? This is about 3 times too expensive considering what you get.

The 16 core version they've pre-announced for Q4 next year. It should be faster than Pi 5 / Rock 5 / Orange Pi 5, assuming the political situation is resolved.

[brucehoult]

Unless you specifically need RISC-V, what's the point?

None. It's for people who specifically need RISC-V.

The same company has the same SoC in the 1st gen DC-Roma laptop, for $299

https://store.deepcomputing.io/products/dc-roma-risc-v-laptop

The Frame may or may not be higher quality (I haven't touched either).

Or, an 8 core SoC with up to date RVA22+V instruction set, for $599.

https://store.deepcomputing.io/products/dc-roma-riscv-laptop-ii-with-octa-core-cpu

Framework is an expensive way to buy that capability. But will be upgradable later.

[brucehoult]

"economies of scale" does not justify charging 200 dollars for a device (no value pack stuff) which has almost identical specs selling for 40 dollars. It would be far cheaper for someone to take the Milk-V Mars CM, make some adapter boards, and plunk that in instead. or better yet, do something compatible with the lichee devices.
mobos

If you can do all that for less than $150 of your time then your time is not worth much per hour!

You can now buy the original DC-Roma laptop with the same CPU for $299. That's a very reasonable price if you don't need the higher (?) quality and upgradability of the Framework.

[Quackdoc]

It's much more accurate to say it's slightly slower than a Pi 4, similar to A55 boards, unless the Arm code uses NEON.

8 GB RAM puts it completely out of Pi 3 (limited to 1 GB) class in what you can do with it, regardless of speed.

unfortunately neon is a fairly massive killer.

You can now buy the original DC-Roma laptop with the same CPU for $299. That's a very reasonable price if you don't need the higher (?) quality and upgradability of the Framework.

This is exactly my point, This doesn't offer any good value. Frameworks are great laptops, Lots of people are wanting to, and do daily drive them. But with this you need to swap the mobo in a likely daily driver laptop, with something that is hardly usable as a daily driver except for a very slim niche. If they had elected to choose something like the TH1520, which granted, doesn't have the best upstream support, but still has some. At least now you have hardware that you can use on an almost daily basis. even with a price hike, IMO that would be worth it.

Even if it say, a 200 dollar addon board (or rather most likely boards plural) that's compatible with lichee SOMs, You can upgrade to a EIC7700... eventually, assuming lichee doesn't cancel it.

I just hope that the new laptops elect to use the same motherboard form factor assuming they can, so they can share effort between the "next gen mainboard" and whatever their new laptop will eventually be, assuming that isn't also just a k1.