nsteel

19 ms·

271.
by nsteel 2y ago
For the record, I don't agree with the grandparent comment at all. None of this is "lame". It's a path to sustainable OSS and it's more obvious than ever that we need these pathways. But to be fair, the branding her
272.
by nsteel 2y ago
That's a very fair criticism. They should at least make an eticket version of that route a few quid cheaper and make it clear the journey doesn't include the tube part.
273.
by nsteel 2y ago
Misleading offtopic posts are a bit frustrating, yes.
274.
by nsteel 2y ago
This is about the tube so none of what you've written here is relevent. There are of course no "scanners for e-tickets" on the tube and there's no world where buying a physical ticket for a tube journey is faster than us
275.
by nsteel 2y ago
In the UK it's at 42 of the 158 stations. Up from 15 when it launched, but still nowhere near 'most'. It will get there eventually.
276.
by nsteel 2y ago
Yes: Bryan posted the video of Brendan shouting.
277.
by nsteel 2y ago
Wasn't there previously some problem with using Pico boards and you had to use a fork because PlatformIO were trying to get vendors to pay (for something they never asked for), and then kicking up a big fuss when they didn't pay u
278.
by nsteel 3y ago
Firstly, to clarify, there's nothing very special about 'gaming ram' other than the particular chunk of silicon performs better than others so they stuck a shiny sticker and an oversized heatsink on. The problem here is the l
279.
by nsteel 3y ago
You still need a memory controller, you still need to get to that controller on the edge of the die. And going to RAM more often will surely consume more power.
280.
by nsteel 3y ago
As others have said, there is nothing low latency about HBM. Renesas did have a special Low Latency HBM thing at one point, but I don't think it ever saw the light of day.
281.
by nsteel 3y ago
Of course - clamshell mode! They don't connect half the data bus from each chip. Also explains how they fit it on the card so easily without (and cheeply).
282.
by nsteel 3y ago
I did miss your point, I misunderstood the hefty price tag was just for the memory. Thank you for not biting my head off! That is pretty mad. I'm struggling to think of any explanation other than yours. Even a few supporting upgrades m
283.
by nsteel 3y ago
Perhaps I should have been clearer and said "if you want more memory than the 16GB model you need a wider bus" but this confirms what I described above. - cheap gfx die with a 128-bit memory interface. - vastly more expensiv
284.
by nsteel 3y ago
This isn't my area but won't it be quite expensive to use that GDDR? The PHY and the controller are complicated and you've got max 2GB devices, so if you want more memory you need a wider bus. That requires more beachfront an
285.
by nsteel 3y ago
I've always understood it that datacentre equipment comes with beefier control plane processors then telco equipment. And that's for three main reasons. Firstly, the data planes themselves are less powerful and have to offload mor
286.
by nsteel 3y ago
Big Chinese carriers like China Mobile, China Telecom and China Unicom rely on networking gear from Cisco, Nokia, Ericcson etc alongside Huawei, just like European/US carriers do/used to. All those vendors are going to feel losing
287.
by nsteel 3y ago
I think most people don't appreciate how hard it is for hardware companies with established products to make non-trivial architectural changes. It's risky and it usually snowballs and becomes more risky. Especially when others in
288.
by nsteel 3y ago
Re the power overhead. They probably get some power back by using smaller nodes on some chiplets than they'd otherwise be able to afford if doing the whole design with the same process. And they are betting on the benefits of backside-
289.
by nsteel 3y ago
I don't think it's really about followimg AMD. It's the only sensible thing to do given the state of scaling and economics. All interesting things Intel are doing are chiplet based. I think it's partly about showcasing t
290.
by nsteel 3y ago
I took the ambitious part to be changing how the gpu was connected to the rest of the system. Seems like a decent jump although not exactly risky. But the conclusion sums it up well: > Meteor Lake abandons the familiar Sandy Bridge clien
291.
by nsteel 3y ago
Same. I see a show advertised and I want to hear what the bands sound like, the easiest way (for me) is to search for them on Spotify and check out the top few tracks and the new release. If I like what I hear then I'll buy a ticket an
292.
by nsteel 3y ago
I'm not aware of anyone actually going to use pam8 or 16 with high-frequency links. The signal integrity is a mess and the corresponding FEC requirements undo all the benefits. I suppose a relatively high-speed chip-to-chip standard in
293.
by nsteel 3y ago
I think you've mixed your Bytes and bits here. Not sure where the power and area numbers come from but I'd expect NVLink to be worse on both counts for similar bandwidths since they shipped much, much earlier. PCIe 7.0 is 512GB&#x
294.
by nsteel 3y ago
Both the PCIe and NVLink numbers here are full-duplex bandwidth. So no, you're not missing anything, you are correct. NVLink 5.0 is far ahead (but presumably it's not actually shipping yet either). Next-gen PCIe 7.0 512GB/sec
295.
by nsteel 3y ago
I think this original question was between devices within a computer, rather than within a single chip/package. Optical interconnects within high-end servers/routers are being designed today for use in the next generation. We pre
296.
by nsteel 3y ago
Exactly this. It's already cross platform (windows and Linux, at least), extendable, looks very similar and has many of the same features. I'm not sure what killer feature is missing that would make someone reimplement the whole t
297.
by nsteel 3y ago
I like the sound of that last bit. But I honestly thought they were just a software outfit, putting pieces together, are they really providing this extra stuff? It sounds like a huge amount of work in itself.
298.
by nsteel 3y ago
How does it help with discovery? Does Roon have some (local? Cloud?) algorithm that looks at your local collection and then suggests things to stream on Qobuz/Tidal? And they've managed to do this well?
299.
by nsteel 3y ago
I'm genuinely surprised to learn Google hasn't killed Google Music (yet).
300.
by nsteel 3y ago
I really don't understand why BBC Sounds is free outside the UK. The argumement for not making it subscription based seems to be focused on UK listening via the radio. Perhaps it's just not worth the hassle to collect it. https:&

More ›