Sesse__

70 ms·

1021.
by Sesse__ 6y ago
> 10,000 euros isn't a lot of money. Yet, it would pay for 100TB+/month of bandwidth for 10–20 years at current market rates (which are bound to only go down). That should be ample for Debconf's streaming needs, without of
1022.
by Sesse__ 6y ago
I have two of these Hi3520D video encoders, from different manufacturers. They provide services (live-encoding H.264/HEVC to SRT or similar) you'd otherwise have to get a box that's 20x as expensive for, but they are also of
1023.
by Sesse__ 6y ago
Back when Google was still at Perforce, I used to go to perforce.com to read about how your engineers would never ever wait for Perforce, while waiting for Perforce. The Perforce servers were… large. And backed by large, extremely expensive
1024.
by Sesse__ 6y ago
Samba is another case, and a very important one at that for a lot of users.
1025.
by Sesse__ 6y ago
Ease of management.
1026.
by Sesse__ 6y ago
No, naming things is really hard. Well, giving things (functions, variables, classes, algorithms) _bad_ names is easy, giving them good names is hard. For instance, when did you last see a class called <something>Manager? Then conside
1027.
by Sesse__ 7y ago
Yes and no. If they're using a PC, yes; if not, they need to download and use the app. (Unless you can direct them to the “Use desktop page” setting in Chrome.)
1028.
by Sesse__ 7y ago
It depends on how good your hand-rolled implementation is. Will it find the shortest path in a graph faster than your Dijkstra implementation? No, leave the simple problems to hand-rolled code. Can it run e.g. TSP faster than a dedicated TS
1029.
by Sesse__ 7y ago
There's no predefined way of setting up and sharing keypairs, for one. As a company end user logging into a VPN, what you want is a place to input your username and password (and potentially 2FA credentials), not “create a keypair and
1030.
by Sesse__ 7y ago
The main idea (which has a fair amount of merit!) seems to be: If you give people too many knobs, they will invariably get confused and turn them the wrong way, creating an insecure configuration. E.g., IPsec has a “none” cipher!
1031.
by Sesse__ 7y ago
IIRC, I redid this graph in early 2019 (using Tears of Steel), and it looked pretty similar.
1032.
by Sesse__ 7y ago
Ryzen and some ARMs use perceptrons for branch prediction. It's essentially a single-layer neural network with no nonlinearity at the end. Multi-megabyte NNs for a normal branch predictor would be impossible; way too high latency to be
1033.
by Sesse__ 7y ago
I believe most of these benchmarks were set at a time when you simply couldn't run x265 on “slow” on any reasonable CPU if you ever wanted it to complete. But yes, I'd really like CPU benchmarkers to move to higher-quality presets
1034.
by Sesse__ 7y ago
I've never actually tried Twixtor myself, but assuming Twixtor hasn't just adopted the DL methods, I would assume they're significantly better for the difficult cases. Especially occlusion and non-uniform flow (think small ob
1035.
by Sesse__ 7y ago
There's a huge amount of research within this field. For a ranking list on a standardized test set, you can see e.g. http://vision.middlebury.edu/flow/eval/results/results-i1.ph... (there are many others