@TeksEdge Yeah Alibaba did say they would continue the mission but in March when the core Qwen leaders resigned I felt it was about trying to fence in consumers of what will be Qwen 3.7 (relevant: https://t.co/PYIDNAc0Oa)
@ethylene_66 NVIDIA has a team called RAPIDS to speed up data science workloads. Early on staffed by most of the Grandmasters on Kaggle. An investment akin to CUDA
@sokol_cc I don't know what differences are attributable to the photos, but that's okay because I don't know what kind of chip I'm looking it.
2nd chip is more dense. Top middle-left, added diagonal set of traces. Lower-middle right, removed test pad
@0xTib3rius I think old is when you stop defining a new culture and replay what you know is good.
The trick to a long life is to pay attention because your clock will tick slower and the truth is that it goes fast, real fast, and if you’re not careful it’s too late.
@davepl1968 In general LLM mistakes are very biased to the simple kind. Inverting values, swapping positions, converting units, inserting invalid assumptions based on their training data, etc. Their conviction is not the same as a human teacher
In this case it just OCRed O̅N̅ (/!ON) as ON
@jsuarez@yacineMTB I agree with you if you mean that excitement is proportional to the inputs which have increased by orders of magnitude and that it's not from closed loop feedback. But I would say that the pace of progress is impressive even if future progress is doomed by mismatched scheduling
@dhh Apple deserves a lot of credit in dedicating resources to optimizing the whole set of hardware and manufacturing towards the consumer end product.
But Apple has absolutely terrible support for external monitors and are never a good example of following a standard
@dhh It's just an LG display, it's well behaved and supported.
What you are noticing is that the Intel IO tiles/chips and their drivers are just better than alternatives like ASMedia, Broadcom, Qualcomm, Mediatek. Intel has long made it a strategy to invest in complementary hardware
@jukan05 Google is the primary competitor for the performance crown and this is about catching up to Google in photonics. Scaling out is growing faster than the system speed, and scale is now much greater than when Mellanox was acquired.
But this is also about playing financial games.