obidan

This is untrue. It is very ordinary. How much do you know about GPU drivers that you state this so assuredly? Drivers are software. Software can be improved. Do you believe there are no prior examples of GPU drivers being improved such that particular compute patterns go up in performance by more than 15%? This driver improved Total War performance by 71% https://www.nvidia.com/download/driverResults.aspx/74714/en-... Also note you can go ahead and improve any open source driver right now, most likely. Compile it for your specific card and remove all other architecture specific if-cases and you can get an improvement.

Edit: also here’s a opencl 30% compute perf increase documented here : https://m.hexus.net/tech/news/graphics/74425-haswell-systems... that i just googled for

Kagi does not use their own indexes, aside from perhaps very small things.

They use both Bing and Google indexes. It is why the cost per search is relatively high.

This change will impact them and it's currently a discussion topic on their Discord as well.

Pretty sad. I just started using it a week ago and am still planning on not going back.

I worked in the automotive industry and I can totally confirm. The worse thing is, there are now AUTOSAR experts and AUTOSAR tools and AUTOSAR Tool Experts and within that tool there’s a ARXML generator that’s generated with another tool …

There’s no way this can ever become safe, robust, software. The worse part is, there’s so many careers that depend on this obscure skillset that I am unsure a change can come from existing companies.

I mean, no it isn’t.

At the same time, an operation which works like this is certainly extremely flawed. I doubt there’s one engineer working on GDrive that doesn’t think a folder size feature wouldn’t be useful and relatively simple to implement. So that’s more of an argument towards the operation going bad, instead of the engineers.

Maybe this slipped through the cracks but this thread already contains a bunch of examples of stuff like this happening.

They can be architecting Tensorflow really well and that certainly is impressive but when GMail takes 3 seconds to load the main page and downloads 10MB during that process you know something is going wrong.

Interestingly this is written in Golang. I wonder what the main considerations for choosing it were.