121–127 of 127 posts
Earlier quoted context omitted.
Didn’t somebody claim Battlemage was canceled? But they shipped it?
Battlemage was still while Pat Gelsinger was CEO, and he was the only real champion of that product, understanding its place in Intel's overall strategy. He doesn't work there anymore, and neither does a significant fraction of the GPU engineering team.
How come we don’t make chips with really small 3d printers?
Earlier quoted context omitted.
Nvidia's software support remains "best in class" so they've become the default for a lot of this stuff. To overcome that I think you'd have to get a lot of hobbyists and by extension interested young people using your hardware instead of Nvidia. So you'd have to have good software support, decent hardware, and low hardware cost. People started with Nvidia because most machines had an Nvidia card in them, and Nvidia'…
True, Given I am young, hobbyist, and have written a lot of AI libraries for myself and small companies for local deployments, and can't afford any serious Nvidia hardware I will switch to whoever can get me a system with enough VRAM for a decent price under 1000$... My best GPU is 5070Ti on which the best model I can run is Laguna S 2.1 118B with NVFP4 and offloading most experts on CPU with an expert router layer a…
Earlier quoted context omitted.
Battlemage was still while Pat Gelsinger was CEO, and he was the only real champion of that product, understanding its place in Intel's overall strategy. He doesn't work there anymore, and neither does a significant fraction of the GPU engineering team.
That didn’t answer my question. Did Battlemage ship? And didn’t that streamer in his attic definitively say it was canceled?
Intel has already stated their deal with Nvidia, this came out in December.
If you think there will be shipping Druid products, I guess you can find out sometime in 2028.
Earlier quoted context omitted.
Intel didn't do much to improve their CPUs while AMD was out of the competition with the Bulldozer disaster.
Eh, xeon 2690 was released 2012q1 at 8 cores, 2690v4 was 2016q1 at 14 cores. Same suggested customer price. Clock speed dropped a bit, but IPC was way up (AES-NI had a big throughput increase among other things). IIRC, desktop core counts were pretty stagnant, but a Haswell/Broadwell desktop chip at the same corecount and similar clocks was much faster than a Ivy Bridge/Sandy Bridge. During 2012-2016 AMD was firmly i…
Earlier quoted context omitted.
ASML is the only real tech company in the EU. Just let them have this one :)
Europe has ARM, Infinion, Siemens and co. And Trumpf and no Trump so
In the US we count these by the thousands, but you're right that there's still someone in Europe not stitching together leather handbags.
Earlier quoted context omitted.
No the beam source we know how to create; it's just that the way to do so requires a factory sized machine that better be feeding 10~30 steppers or a pair for uptime/maintenance windows and then probably more like 50 steppers. Free electron lasers are not hard at "mere EUV", and the accelerators to feed aren't either, it's just that an efficient setup requires recycling the beam which means bending it back which mean…
Uhh, you do not need an accelerator to generate EUV. It's literally just molten tin hit with a laser.
The main difficulty of that approach is that physical constraints of field strengths (exceeding those would cause stray electrons to spontaneously free themselves from any vaguely conducting surfaces they are on, and fly around the accelerator, causing arcing damage, wasting energy, and disturbing the beam/beam-control-sensors), the thing has to be fairly large; also, building it with a bit more incremental power is a cheap upgrade (at least if done at design time of the accelerator) up to around a dozen EUV scanners worth of light output.
Complicated by that you ideally want redundancy to not have EUV scanner downtime just because the accelerator needs maintenance, so in practice you'd want like 3~5 accelerators each sized for 20~30 scanners in one building with EUV-routing-flexibility to have n+1 redundancy for the accelerators.
But those 40~120 EUV scanners, especially if you include the rest of the fab to actually run them in production, is a very large investment. Like, somewhere between 1 and 5 % of TSMC's market cap. Not under 2 aircraft-carriers (the big ones)... it's in no way a "low-risk" investment. But, back before GlobalFoundries gave up on the EUV race, a free electron laser was exactly what they were working on bringing to use for EUV litho (it's mostly a energy efficiency question around recovering energy from the electron beam after it was used to generate the EUV, and the substantial difficulties of efficciently redirecting the EUV from the source to the scanners, that's still open research with FEL illumination for EUV scanners).