Comment Re:Not annoyed enough yet.. (Score 1) 217
But this lets you share your login with your Zune!
LoginForSure !!
And then about four years later:
LoginForSure officially dead - CNET
But this lets you share your login with your Zune!
LoginForSure !!
And then about four years later:
LoginForSure officially dead - CNET
Unless they are on a shared wafer they will get many more than 50 die off a wafer. But even so, it would require expected sales of millions of units in order to get the unit cost low enough for inclusion in a product. And all that cost would be front-end, basically it would require production ready units to even open dialog with end product manufacturers. If I were building something I would not want to include a chip from an unknown vendor, especially if they had no working samples.
In college I fabricated a custom analog die through MOSIS, and it was about $3k for 20 or so die (0.5um process). That was a shared wafer (my design and many others were placed on the reticle), which is why the units were low and the cost was low. It was also on a relatively large tech node, but it worked ok for analog. For a large digital die in a recent high-end node, maybe 40nm, it would be much more costly (mask costs go way up in deep-submicron). If someone is -really- interested they can still get quotes from MOSIS, even on high-end nodes.
I considered doing custom chip designs independently, but it is frankly a terrible business in many ways. Aside from the hardware issues, design software is a big problem. There is no viable "low-end" IC production software. There are some niche offerings, but they are all bad. The main software is either Cadence or Mentor Graphics, and they are both very expensive. The main fabs (TSMC, UMC, etc) will not offer design kit info for anything other than Cadence or Mentor (and usually what they offer requires a lot of post-processing to get it into a usable state). For digital work it would require a whole slew of stuff, simulation, synthesizer, place-and-route (and synth and P&R both assume you have a working stdcell library also), layout tools (for pad ring, chip seal, etc), DRC, LVS, etc. After digesting all that, then you need to deal with packaging, getting samples made, doing a whole slew of reliability tests on temp, corner lots, etc. In short it is a never ending stream of costs and work.
Better idea - stick with software and sell them working IP in a synthesizable form like Verilog or VHDL code (similar to what ARM does). Have them sign NDAs and all that. Big companies that do ASICs will pay for IP if they think it adds enough value. As an example, it is very common to use 3rd-party memory designs in ASICs (esp things like flash), because designing reliable memories that work at speed and don't degrade in HTOL tests is a big PITA.
Cloud Atlas is a disturbing movie to watch because some parts of it are actually plausible.
Some others I would add:
Zombieland (2009) - (I just found out this has a sequel coming this year)
I did see this article, describing same as above, but also having a DLNA backup for problem clients: https://www.howtogeek.com/303282/how-to-use-plex-media-server-without-internet-access/
I assume that DLNA implies transcoding doesn't work, as I take it to mean it just streams the original source. Overall I agree it's ridiculous such a feature cuts out if network is gone, since really they are completely unrelated.
I also found the Plex online signin requirement absurd, but there is a way to disable it (ironically you must signin to disable it):
After signin, go to Server->Settings->Network. Scroll down to "List of IP addresses and networks that are allowed without auth", and enter a subnet, such as: 192.168.1.0/24
After that anything on the subnet should not ask for a login when connecting (eg. from PC web browser). Oddly, even though it originally wanted a login when using PC, I don't recall Roku Plex app ever needing auth (perhaps it did and I don't recall).
IMO Plex is great for the equivalent of a "personal netflix" but really it is just a content/metadata management system. The real unifying device is the head unit, such as Roku since it supports all backend services. Although if I were to upgrade our Roku's I would probably opt for Nvidia Shield, as it is similar, but can also do Steam streaming.
Probably doesn't matter as Apple is just preparing the terrain for a complete ARM switch on Macs.
This is the only insightful comment on this thread. The switch has nothing to do with user "experience" or developer ease or any of that other BS nonsense. This is about them wanting to abstract their platform off of a CPU controlled by someone other than them. Why - because they are a bunch of control freak assholes who hate paying anyone else their due - see Qualcomm.
If you're working with an employer such as a military contractor on a federal contract and you bill time you didn't work, you have committed a felony and could spend five years in prison (for as little as 10 minutes mis-billing).
LOL - Yes, if anyone should be the gold standard of ethical billing, it's military contractors on a federal contract.
Warrantors cannot require that only branded parts be used with the product in order to retain the warranty.[7] This is commonly referred to as the "tie-in sales" provisions[8] and is frequently mentioned in the context of third-party computer parts, such as memory and hard drives.
And from the summary:
The new system will render the computer "inoperative" unless a proprietary Apple "system configuration" software is run after parts of the system are replaced.
So in effect they are saying "oh sure put whatever part you want into it, but it's not going to work unless we allow it". Thereby creating the onus to use "branded parts". Yeah good luck with that. I fully expect them to land in court over this.
Sadly ideology doesn't pay. The simple fact is that tools like Gimp, Lib/OO-office,video-editing and Blender are far far behind the commercial offerings and people betting on the closed source offerings will almost always make more money and have enough to offset the costs.
Out of these programs the only suite that really tries to improve things is Blender that has had pushes to improve it's UI (because even the die-hards remembers the bad learning curve they had).
The Blender of early 2000's is not the Blender of today. A couple years ago I had an opportunity to upgrade an older Cinema4D with associated cost, and I decided to try out a recent Blender just to see what its current state was. I was absolutely blown away by how advanced and powerful it is now. It was way more capable than my (admittedly old) C4D install, and some of the plugins for it are just amazing.
Although its default UI is improved, it can be even further improved (IMO) with a free addon, Sensei Format, giving it a nice streamlined workflow (the dev for that also put up a huge 100+ array of tutorial videos on YouTube). I also got a low cost addon, Zero Brush, which replaced the old C4D bodypaint (although I believe this core function is built-into Blender and it is really a UI overhaul also).
But Blender is an artistic tool, and another thing I need in my workflow is a 3D CAD modeler. In this space I don't see OSS (or low cost) alternatives. I've tried FreeCAD (frustrating workflow), and low-end tools like Sketchup just don't cut it. Closest thing I've found is DesignSpark Mechanical (which is a stripped down SpaceClaim). SpaceClaim itself has a perpetual license for $3500 (cough), which is high, but overall it is a good push/pull modeler. Autodesk Fusion 360 is there, but like other Autodesk things it is subscription based (they lure hobbyists with free non-commercial use - but frankly that could disappear overnight if they so decided). It also has a "cloud" tie-in, and requires an online connection. I'd be very interested to know if there are any other good options in this space.
What happens when Bitcoin crashes? What effects will it have on companies that accept, use, or hold it, market-makers on exchanges and futures, etc. ?
One thing I'm wondering is what is going to happen to AMD and Nvidia stock when Bitcoin crashes? I think a lot of their recent profit is driven off this craziness, and a crash might have a lot of people dumping gear to exit or cover losses. I imagine it might be a great day for buying surplus high-end video cards on eBay.
Indeed, inline one of these with the door opener.
It can probably be rigged to automatically disable at night. Even better would be to disable anytime the controlling phone is out of WiFi range (not sure if that's possible).
Well they may have the pipe dream that people would use Xbox as an HTPC, but I think it a far better option to run the reverse configuration (a true HTPC as a game machine via Steam).
I've been trying to determine alternate configurations to a PC with WMC, so as a basis:
PC running Win7/WMC: - can utilize OTA and cable tuners, PVR capability, free scheduling, expandable storage without limit - can access stored media of any type using any appropriate player (WMC, VLC, etc) - can access full web, any foreign sites, any streaming source (does not suffer "mobile web" that Android or such see, or how Hulu restricts Roku and similar) - can utilize VPN without additional hardware - can double as game machine via Steam, MAME, etc.
Now some alternatives (relative to above):
PC running Linux/Kodi - needs paid scheduling service, Steam somewhat limited
Android box (eg. FireTV) + network server (Plex) - no tuners, not sure on VPN, possibly restricted services (Hulu), simple gaming
Roku + Tablo + network server (Plex) - has tuners (need paid subscription), but no web browser, not sure on VPN, possibly restricted services (Hulu), simple gaming
Xbox or PS4 + network server (Plex?) - not sure on these (web, VPN), but almost certainly limited and annoying options, good gaming
Others?
I really don't get their stance on WMC. I have two media PCs running Win7 essentially only for WMC (more specifically the Live TV + free scheduling it has, vs Kodi/XBMC). Lacking WMC there is really no reason I wouldn't run Linux and kick them out. If they are trying to increase their presence on HTPCs in the living room it makes no sense. I would consider Roku and such but I need web browser capability.
There was some bizarre comment about people only using it for DVD playback. I don't know where they got that feedback from. I know several people who have WMC setups only for it's added Live TV PVR capability.
The issue is this: Our desktop processing requirements are actually slowing and as Linus points out, are probably ill-suited for increased parallelism.
Depends on the desktop requirements. I think he is off the mark here. Specifically to quote him from TFA:
The only place where parallelism matters is in graphics or on the server side, where we already largely have it. Pushing it anywhere else is just pointless.
Since he points out servers and graphics as largely solved, I assume he is talking about desktop usage. In this he is assuming a standard usage model for a desktop user, a set of apps - web, devel, coding, games, whatever. I think the view is that of a user who can only focus on a single-task at a time (with perhaps background OS tasks). But this is a myopic view, the rise of virtualization has enabled a convergence of hardware onto a single machine. This is only possible with the rise of multi-core/parallel computing. VMs are a huge benefit, in terms of power/area efficiency and even being able to create and destroy them on a whim.
On my desktop machine (8 core) I have two VMs running all the time. These machines used to be physical separate machines, consuming power, taking up floor space, making noise, etc. I could not have run this setup on my previous single/dual core machines. However now they are virtual, and my normal desktop usage doesn't even notice them running (even heavy 3D gaming is not lagged by these VMs).
There are compounding parallelization factors - having the whole setup on encryption means wanting the cores to handle AES in hardware, so as he points out having hordes of parallel weak cores might be pointless for that. However, multiple powerful cores, I can put those to work.
IMO the advantages are clearly obvious. Sure for a single-task desktop user, you may only want a few cores for background tasks plus the foreground task. But the ability to consolidate lots of hardware into a single box, I want as much of that as I can get. I can easily think of desktop + VM scenarios that can push beyond 4 cores.
"Protozoa are small, and bacteria are small, but viruses are smaller than the both put together."
