Ashes of the Singularity does 16 threads (and Oxide Games can probably make it use even more threads with little difficulty), but that's an exception rather than the rule as most games are not this amenable to parallelization.
ra3: q to select all units of all types across map
units show up rouped on the right. right click a group to deselect
select all with q, deselect engineers => all fighting units
very smooth
even trying to click on a flying (circling) unit was nearly impossible
and you have to double-click to select all of type... try doing that on a fighter plane
@bwDraco aware of that, and that's the most rudimentary form of multiselect possible. it technically prevents it from being totally unplayable, but as far as RTSes go it's pretty poor.
what if I want to select one bomber and use that to select all other bombers w/ hotkey? I did that a lot in both red alert and age of mythology. very easy.
in aots? might as well send your fighters on a bombing run cause you can't easily select just one
> In Command & Conquer there are two hotkeys that I miss a lot on Ashes: one which selects absolutely every military unit you've got across the map. The other selects every unit on SCREEN of the same type of ones you currently selected. That is better than double clicking because not only it's easier, but it opens to possibilitys like selecting some brutes and arches and then pressing the hotkey to select both the brutes and archers around the screen. That's not possible with double cliking. Additionaly, when you press that same hotkey again having selected already every unit of that kind o…
> Being grown up is very rare and you might only see it a few times in your life. [...] Being [an adult] requires "the look." When you look into an adult's eyes and see wariness, regret, stern impassiveness, but most of all, intelligence.
The most important thing is, Intel is no longer able to price their HEDT processors as high as the market will bear like they did with the $1700 i7-6950X. Competition brings down prices. $2000 for 18C/36T is not excessively high IMO. The real stars of the show are the 8C, 10C, and 12C processors.
The 8C part has some PCIe lanes disabled, but 28 is plenty and the $599 price point is clearly targeted at Ryzen 7 1800X.
You're still paying lots of money for these extra cores, but Intel isn't free to charge excessive prices anymore.
This is a textbook example of how monopolies are bad for the consumer: competition keeps prices in check, and when there's no competition, there's no need for the monopoly to charge sane prices.
> The robot has a touchscreen chest, two arms and a head. For the past 10 days it has offered blessings in a choice of German, English, French, Spanish or Polish. Worshippers can choose between a male or female voice.
The robot raises its arms, flashes lights, recites a biblical verse and says: “God bless and protect you.” If requested, it will provide a printout of its words. A backup robot is available in case of breakdown.
So... with Skylake-X a reality, I was wondering if I could cost-optimize Lucifer without sacrificing scalability. This X99 build simulates an X299 build with an 8C processor and high-end EATX motherboard, but there's precious little room left for cost-cutting and I'd rather not have to replace the system SSD later in the system's life.
For the life of me, I cannot bring the cost down to inside of $3000.
Lucifer is reserved for an Intel HEDT build (possibly up to the LPC class)
Regardless of the cost optimizations, the scalability has to be there, and I do not want to wind up replacing the motherboard, case, PSU, or boot SSD absent hardware failure.
I don't take assigning the name Lucifer lightly, so even if the performance isn't there, it needs to be readily upgradeable to that level without the need to replace the foundation it's built on.
@bwDraco I thought the same when I picked the system. But even when I add another GPU (which is actually highly unlikely), 750W is plenty. Not saying that this is true for you, but I made the same overestimation in the past and it was never worth it
I really feel like that it would've been a nice question at least related to computer science. Should've withheld my comment? :/ superuser.com/questions/1214293/…
...I suppose I could just drop to the i7-7800X (6C/12T Skylake-X @ 3.5-4.0 GHz, $389) if the plan is to replace the CPU in the future. Perhaps I could also reduce RAM to 16GB of DDR4-2400.
The cooler and radiator is always one unit. The pump gets integrated and you don't need a reservoir (which kind of is optional anyway ... if you're hardcore).
Minor difference but important ... last build I planned had liquid cooling until I thought about maintenance.
EKWB has an intermediate solution that is preassembled but can be expanded to a custom loop. The Predator line seems to have been discontinued, but a replacement is probably on the horizon.
The downside to AIOs is that you don't see if you have any build up etc. except if you monitor your temps constantly. The upside is that (assuming they are fine) they require less maintenance.
optional not optimal. Filling a loop without a res is going to be a really painful process but would probably be possible. Hence it's kind of optional. I'd guess you'd only do it for a show build or something. Essentially you'd be going for a closed loop DIY.
You could do it that way (though from what I read it's better to have a actual valve near the lowest point) but primarily it's easier to fill as you have a spot in your system for the exchange of liquid and air that has big enough holes for it to work easily. Usually you'd fill your res with liquid (say 500 ml?), run the pump half a second, fill it again, run the pump, fill it and so on until you got enough liquid in the loop in order to let the pump run constantly.
After that you can do all kinds of Voodoo to get the rest of the air out of your system and top off with liquid.
omfg.... we have an email form on a customers site. it gets sent to their default group email (info@) and for reasons only known to them they insist on also having individuals in the to field as well. they asked for one user to be removed, we did it two weeks ago now they are complaining the user is still receiving emails
Well, well. The SP3 socket is going to be tricky to work with. The cover is secured with Torx screws. Then again, it was originally designed for server processors...
> Measure and report on the performance of the entire AMD Radeon graphics product line, plus relevant competitors, using industry standard benchmarks, 3D games, and other relevant tests.
yes it means moving to canada but on the plus side it means moving to canada
even for actually downloading Linux distros I don't use BT because it's usually stupidly slow on my connection; my uplinks generally can't handle the huge number of open connections that BT uses, compared to HTTP, which is very efficient
because the literal problem I've encountered with some of my uplinks is that the number of open TCP sockets exceeding about 500 causes massive slowdowns; I guess they don't handle TCP flows very efficiently or don't have much space in their stateful firewall for them
@Avery then it's not true BT; true BitTorrent would require you to have many connections to many different peers. If you just have one connection to one peer (or to the distro's super-seed or whatever), you're basically doing exactly the same thing as downloading cdimage.ubuntu.com/...whatever, but using a thick client instead of your web browser.
any HTTP download manager worth its salt supports resume natively; I know some browsers still don't, but there are extensions that support it too. Anyway my connection has fortunately been stable enough in recent years that I almost never need to resume even large downloads.
As for verification, doesn't basically every (good) distro ship a built-in verify command in the GRUB bootloader of the CD image now?
you'll need that anyway even if the download is fine, because of the chance of corruption during transfer to the install media.
as for only downloading corrupted parts... meh. TCP is a reliable protocol with built-in checks that aren't infallible but eliminate the vast majority of data corruption (it automatically requests a resend if the checksum fails), especially if you aren't resuming a download (resume is somewhat dodgy in that respect). I think I last had a corrupted download over TCP that wasn't the fault of the server sending bad data, back when I had a dial-up modem.
maybe if FF or Chrome implement some kind of a binary diff or chunk based download protocol (whether "BT" or something else entirely) out of the box, it could be an improvement for everyone, as you'd get "free" (automatic) file-level checksumming and easy self-healing. But it's not worth dragging around an extra thick client for me, when I only use it for downloading ISOs once or twice a year, and no other sites support that.
Kinda reminds me of those awesome new image formats that have come out since PNG/JPEG were invented. BPG and WebP. If anybody actually adopted and used them besides 0.001% of websites, they'd be fantastic. Chicken and the egg problem for adopting new, (better) standards.
