puzzle

64 ms·

601.
by puzzle 9y ago
False. Years ago I had a business credit card charge from either Target or Kmart where you could see the individual items. So some merchants do share that kind of detail.
602.
by puzzle 9y ago
Google also has/had Borgmon, Sawzall, GCL and a few Python derivatives.
603.
by puzzle 9y ago
Hopefully, for customers, it also means not having to deal with proprietary Linux drivers and associated headaches.
604.
by puzzle 9y ago
Maybe a paper is forthcoming? At least Jeff Dean has confirmed that it's some kind of FP, not e.g. decimal arithmetic. :-)
605.
by puzzle 9y ago
The hostname is static.corp and the service is static-on-bigtable, so...
606.
by puzzle 9y ago
Part of the problem is that computing the downtime at a higher resolution, in a way that scales from low-traffic users (the quoted minimum of 60 requests per minute is also not a coincidence) to very large ones, is not trivial. You have to
607.
by puzzle 9y ago
Cloud Bigtable is similar.
608.
by puzzle 9y ago
No, they worked on Colossus, aka GFS2, which Spanner uses to store data.
609.
by puzzle 9y ago
This is funny. Google used to have clusters in Virginia. Not in Ashburn, but close, in a colo facility. You wanted to use them if you cared about latency to e.g. Philly, NYC and Boston. Then they were turned down because of all the capacity
610.
by puzzle 10y ago
Google Maps' routing weighs simplicity of directions, such as fewer turns, more than Waze does, even if it adds an extra minute or two. In most cases, that kind of heuristics means taking main roads.
611.
by puzzle 10y ago
When I was at Google, I kept hearing that California laws in particular are not employee-friendly and that you should ask for an opinion from a personal lawyer. As to why this hard stance, I always suspected that sweeping that kind of stuf
612.
by puzzle 10y ago
Awesome! Happy to hear that more containers will run without unneeded privileges.
613.
by puzzle 10y ago
My main point is that there is work behind the scenes happening that you don't have to worry about, on an undefined number of machines that you don't have to care about. Yes, there is still a cluster which does some kind of planni
614.
by puzzle 10y ago
It's not, but it resembles BQ a lot in the central part of the query execution, where it transparently performs work on your behalf on thousands of other machines that you don't have to be aware of. Is that correct?
615.
by puzzle 10y ago
At a first read, it sounded like you didn't need to run a cluster ("Spectrum scales to thousands of instances"), which is BigQuery's big advantage. Later, it says that you still need a cluster, but it's only used in
616.
by puzzle 10y ago
You can assume it's only a matter of time.
617.
by puzzle 10y ago
Amazon bought Cloud9 last year...
618.
by puzzle 10y ago
Ah, wait: https://github.com/madeden/blogposts/blob/master/k8s-gpu-clo... You don't need to mount the /dev entries into the container at all. The experimental support creates them automatically
619.
by puzzle 10y ago
That's weird, because all the times I tried the experimental support, it didn't need privileged containers. From the YAML files, it looks like it's using hostPath directories, but those don't require special privileges,
620.
by puzzle 10y ago
The Kubernetes docs don't say anything about having to use privileged containers for GPU support. Privileged containers are given tens of Linux capabilities; which of those are actually needed in your setup? Or, conversely, which speci
621.
by puzzle 10y ago
What is the requirement for privileged containers? The post never explains it.
622.
by puzzle 10y ago
Correction: the ASIC processors that Google used in 2015 and just talked about, two years later, were for inference. We don't know what they use now and how. The paper hints that training might happen in a later revision. You have to w
623.
by puzzle 10y ago
It's not necessarily AMD or something new. Last year I was wondering why there was no party. In 2015, they had a big one, with a couple of trendy bands on stage. Somebody has probably been looking at return on investment for these even
624.
by puzzle 10y ago
You forgot that Intel bought Nervana Systems last year, right before IDF. They develop custom ASIC processors for machine learning.
625.
by puzzle 10y ago
To give you an idea, last year you had talks about anything from silicon photonics, memory validation or Thunderbolt, all the way to indie games.
626.
by puzzle 10y ago
True. That's the main reason, for example, that I didn't get a ticket to I/O. GCP Next had a lot more topics relevant to my interests, despite it having a more "commercial" or "salesy" feel to it. I/O
627.
by puzzle 10y ago
Interesting. Was anyone even remotely picturing something like this? Congratulations to the Wercker team, but for anyone on the "Never Oracle (again)" train, this means one less option for CD/CI. My only exception is VirtualB
628.
by puzzle 10y ago
To be fair, IDF was all over the place. I suspect I'm not the only one who felt that only a fraction of the talks were relevant to my interests. On the other hand, it was great for networking.
629.
by puzzle 10y ago
And to be clear: it is your product and it's your choice how to develop it or make it available on the market. The only reason that I took the time to vent is that I actually like Teleport a lot. You won't see me complain about Or
630.
by puzzle 10y ago
An aside, which doesn't necessarily affect your reasoning about Books... You didn't mention that GOOG-411 (I still have the t-shirt and other schwag) also had rampant abuse, had seen legitimate traffic shift to smartphones and, la

More ›