baruch

34 ms·

301.
by baruch 14y ago
I've seen and used solutions of user-level threads (aka green threads, aka coroutines) to do just that. It works. It's a lot of work to get right even for a very specific use-case. In the past the power management on Linux was done by a use
302.
by baruch 14y ago
The only relevant piece I could find quickly was the Mellanox performance tuning guide: http://www.mellanox.com/related-docs/prod_software/Performan... It's also useful for any other PCIe card, sans the IB specific parts.
303.
by baruch 14y ago
There are too many workloads to consider and not enough methods to balance them all to make everyone happy. At least in Linux there were already quite a few major changes and tons of little tweaks, each tweak can improve one benchmark and c
304.
by baruch 14y ago
I don't know of specific articles but I suspect you'd find some on Intels website. In short, in current CPU generations (x86_64 is what I work with) the PCIe controller is on the CPU die so if you have multiple CPUs you have multiple PCIe c
305.
by baruch 14y ago
In order to double the profits you need to either double the sales or eliminate all costs . Or do a mix of both. Doubling sales multiple times is more likely than making profits on your infrastructure... Amazon AWS is probably the exceptio
306.
by baruch 14y ago
I don't espouse the Google interview process, just some technical questions. I've done at one point the Google interview process and actually wouldn't suggest to follow it.
307.
by baruch 14y ago
I don't quite support your position and am more in line with the parent one. It may be possible to find the "truly great" by a nice discussion but for those that are good and above it isn't so easy to separate them from the good talkers wit
308.
by baruch 14y ago
If anyone is interested in spot instance pricing there was an interesting work on the subject: http://www.cs.technion.ac.il/~ladypine/spotprice-ieee.pdf In short, it's not a perfect supply-and-demand market, but it's interesting to see th
309.
by baruch 14y ago
I have an interest in this and am trying to build up a disk survey system to try and learn about this in the open, it's still not up yet but you can subscribe for an announcement and read some thoughts at http://disksurvey.org/
310.
by baruch 14y ago
That's a really neat hack, but how many disks have you seen that do not adhere to their write-cache-disabled setting?
311.
by baruch 14y ago
At the storage level there is T10 DIF that is supposed to help with such things (among many other things), though it is used in a very limited fashion and is only supported on the higher end disks.
312.
by baruch 14y ago
Interesting, do you do disk scrubbing to make sure the data is still out there after a while? That would be the key method to ensure you can still recover the data upon need.
313.
by baruch 14y ago
That's an interesting concept but there would still be a need for the much faster and non-volatile DRAM. At which point you need to discern which is which and decide for each data structure where you want to locate it. The latency of NAND f
314.
by baruch 14y ago
You're supposed to run dm scrubs as well, if you don't it is possible that you have a bad sector somewhere on your disks. See http://en.gentoo-wiki.com/wiki/RAID/Software#Data_Scrubbing
315.
by baruch 14y ago
I've used soundcloud for the past few weeks for background programming music and the main issue I had was the need to find the next artist with enough songs for the next session. Thanks for saving me time!
316.
by baruch 14y ago
Herzliya (Glil Yam), Israel XtremIO (recently acquired by EMC) is doing Flash-based Tier 1 storage devices that run into the millions of IOPS, We are looking for sharp C developers who'd like to us build the storage and the platform and mak
317.
by baruch 14y ago
If the concern is of a mistake causing the deletion of the zone a replicated zone file will be lost just as well. To avoid this you'll need two name-servers and update each of them independently. Keeping not too long DNS caching times would
318.
by baruch 14y ago
I'd presume that any sane registrar will have it, I know my favorite registrar (Gandi) has an API for resellers. I have never used it to testify on its quality though.
319.
by baruch 14y ago
The goroutine is a user-thread, Go can create as many threads as there are cores and schedules user-threads inside these threads without the programmer managing that, you just create goroutines. When a goroutine "blocks" it schedules out fr
320.
by baruch 14y ago
I'm not much of a Go programmer and do 99% of my work in C but I've played with it a bit and there are places where the abstractions and facilities provided in it are so nice they make it worth it. The whole goroutines and channels mechanis
321.
by baruch 15y ago
You still lose x 1.2 0.8 = 0.96*x
322.
by baruch 15y ago
It's definitely a lot less painful than dealing with disks as the article suggests. The variability of disk latencies is much higher and is even less expected.
323.
by baruch 15y ago
It's an interesting analysis but still the problem for any upstart distribution technology is to get content, it's a chicken-and-egg problem. I've worked in the past for content distribution technology company and the main issue was to get
324.
by baruch 15y ago
OPML is just the start, it should also pull followers from your twitter account and projects you follow on github. I can't see myself importing them by hand, too much work and no easy benefit in sight to warrant this work.
325.
by baruch 15y ago
If this was transmission to a single client you were right but when this is used for broadcast or multicast there is less interest to minimize the transmission for each client but rather to optimize it for the entire population at once, sin
326.
by baruch 15y ago
The fountain code itself is not concerned with it, it's merely the coding layer. There needs to be a communications layer that takes care of that and the way it does that will need to consider co-habitating with TCP. In the application I wo
327.
by baruch 15y ago
Normally one would use some distribution of the number of encoded source blocks in a block. So one would send 30% single blocks, 30% double blocks, 15% triple, and so on until you get some relatively small number of high order encoded block
328.
by baruch 15y ago
It is a damn cool algorithm, I was introduced to it in a startup I worked in and we actually tried to make use of it. The algorithm as described has a problem to converge fast enough and we had to make some extra structure on top that made
329.
by baruch 15y ago
I've interviewed many people who looked great on the resume and failed to do even a trivial coding question. The good developers quickly wade through such a question without a winch which leaves plenty of time to discuss interesting project