@DragonLord comment to be found some day under your answer, but processors have done all that for 15+ years, what makes this one so much faster :-) you still do not have the raw more simplistic basics that made the largest difference
the same things that were in the marketing , and on the speedo sites , when that round of processors came out. In fact nobody has covered it yet. almost as if it is written to be all about "code" from SO.
@DragonLord making the paths to everything in the processor 1/3rd smaller and having 1/3rd sized transisters doesnt do anything for speed? Putting the memory controller right on the cpu, which is completly new, and direct connecting the memory to the processor, making it fully 2X faster. Connecting PCI lanes direct to the processor. having 2x as many transisters for processing (is that that many more "execution" units?) something like that would make a gpu almost 2X faster by itself.
@DragonLord go back to the moors prediction (moors law) page on the wiki, and read Wirth's law. It is my understanding that the reduction in size and the path distance is One of the hugests advantages to speed, not power consumption.
I am talking about the most major things that have changed, perhaps you should fire up a core 2 duo Intel arhcitexture PDF and see how much of what was written by everyone was already there. Thesse PDFs have repeated that stuff. not saying i understand it, just have read it more than once, as if 1/2 of it is copy pasted from the last processors write up, add in one more cute marketing term and it is in :-0
> Increasing transistor budgets enable more execution units such as integer ALUs and floating-point units to be included in the processor core.
Also, earlier in the answer (in the superscalar section):
> The processor may have multiple functional units of a particular type (such as integer ALUs) and/or different types of functional units (such as floating-point and integer units) to which instructions may be concurrently sent.
@DragonLord Well i dont know, it brings up how do you have 2x execution units without either more piplines to them or 2x speed? Honest i really do not know, i just know that everyones write ups sound like every PDF that intel has ever written, and there were 4 major advancements specific to sandy/ivy/haswell haswell only adding in the regulators on-die which was sorta stupid.
These 4 things were 89% of the speed, some of the others , do not get implemented BY software , untill the code base (what is programmed with) and the coders utalise them, often this takes years.
So the hype that revolves around a processor (and motherboard) changes that you see at the hardware locations , is usually the most relevent things (at that time). what we read here at SU, instead is how a processor works because i am going to be programming tomorow :-)
The two could be blended together , or the most important changes that are made to speed could be at the top.
> Greater integration of parts previously external to the CPU, such as the northbridge, memory controller, and PCIe lanes, reduce I/O and memory latency. This increases throughput by reducing stalls caused by delays in accessing data from other devices.
@Bob: that said, I probably need one, or its AMD equivilent. Unless the rumours of dx12 or the next gen graphics processors letting you pool texture ram between video cards is true
> In addition to the above, greater integration of parts previously external to the CPU such as the northbridge, memory controller, and PCIe lanes reduce I/O and memory latency. This increases throughput by reducing stalls caused by delays in accessing data from other devices.
en.wikipedia.org/wiki/Moore%27s_law Find "Dennard scaling" on that page it is 1/2 way down the page. Assuming it is correct , scaling would be at the top of the list?
the wiki for Dennard scaling itself, is not so good , but the writeup on the moors law, page indicating the forgotten and all important electronics themselves, is much more interesting sounding.
Too often the web science is like watching the big bang theory , and having to hear about the long dead schodengers cat again :-)
There . . . is . . . more . . .
And as a side note, someone is working on an analog processor, a series of processors that tie together to attempt to better mimic the human brain, using aspects of analog, and claiming high-faluting speed increases.
By tossing a bit of analog in (and imperfection much like the human brain) we would no longer be limited to binary, by using it only as useful, the number of transisters themselves becomes less important , having more stages in each instead of 2.
I guess by todays definition "A classical computer has a memory made up of bits, where each bit represents either a one or a zero. A quantum computer maintains a sequence of qubits. A single qubit can represent a one, a zero, or any quantum superposition of those two qubit states;" Everything in between :-)
I believe previously "quantum" was mostly being applied to smaller and smaller and smaller. Applied to atomic changes and almost atom sized transisters and such.
I guess "crystal storage" died and didnt get on the wiki either. Some dude had a working 3D 2 divergant laser storage creation by shifting the state of the atoms or molecules in a block of crystal.
Claimed he could store 320 full length movies on this (about 1") chunk of crystal. Sort of funny that now we can store 7000 already on a chunk of aluminum
@DragonLord as a proud owner of two Samsung 850 Pros, I think they're one of the most remarkable products to come out in a long time, in terms of price and the value you get for it
the funny thing is, there are technically SSDs out there that are better than the 850 Pro, but they're so hideously expensive that the value proposition goes out the window -- and the difference is not all that remarkable for "normal people" workloads
The 850 PRO was announced in July 2014 with enterprise products even earlier.
Nobody else has even a prototype SSD with 3D NAND in it.
There have been announcements from others that they have 3D NAND available, but no SSD or other completed product has actually been announced, not even a prototype.
Samsung's first 3D NAND chips (24-layer) were announced in August 2013.
Samsung has had a consumer product with 3D NAND on the market for more than half a year while competitors are just getting started.
This shows just how far ahead of everyone else Samsung is.
Samsung leads by an entire generation. One entire generation.
Simple question. I have a server, and I want to virtualize using Vmware. Just like how Workstation does. Which software / edition do I use from the 492839823923 VMWare products?
I also tried VirtualBox, but in terms of performance, it's just horrible. (it was chewing away several cores just for no reason ... but I also had this on desktop)
Nah, it didn't get any better. Maybe the next major version, who knows
so I was pretty satisfied with KVM, but when I started setting up macvtap, the entire virtual networking went downhill. even the NAT guests lost their network access. (my biggest issue with it is that virt-manager sometimes saves settings, sometimes throws an XML error, sometimes just does nothing. of course you can get your hands dirty and just edit with virsh (that's what I did until I started to bring up bridging))
Hmm I only tried to use virt-manager (and command line) to bring it up. But once I set up the bridge - closely following steps, checking things twice, I always lost access
@JourneymanGeek Wow, I killed vncserver after pressing "run in background" and it really works. 10/10. But I will try your Bridge tutorial - I just cannot fiddle with the host's network during peak hours. :))
well. I want to order a Note 4, but there are two variants (Exynos and Snapdragon). Not sure which one should I buy. (So far I read: 1) Snapdragon got loads of roms, kernels, etc. 2) Exynos heats less, is faster, got a lot better battery life. 3) Snapdragon can overheat, but also supports 4K video recording.)
But everything I read is a bit different ... (I mean, reviews, people's posts, all contradict each other.)
> Not only will your name be listed on our site, but you will get the next version of Goto Browser 1 month prior to release! Allowing you to test it before anyone else.
I don't know why being a beta tester is such a privilege... -_-
Seriously Bob, it's just a rage material, don't take it too seriously. :)
Kickstarter and Indiegogo is literally a breeding ground for scammers and failures. Only 0.001% (or less?) turn into a real project/real proper game/real whatever. Most of them just scams/promises.
Actually there's a game I was tempted to get on kickstarter. But waiting for it to get a few reviews. Seems really cool so I don't wanna get disappointed ;p
@JourneymanGeek FIXED IT! The entire problem was due to MAC address. I didn't set one, just used an auto-generated one. Turns out the provider gave me one (which they forgot to mention in their letter). Anyway. All is good now. Not sure if I will go back to KVM, this works and its really comfy. Will check performance though. KVM offered awesome raw perf.
@OliverSalzburg i would suggest you don't think about it for a moment untill you see for yourself :-) a users description of the events could be as useless as the the text shown in a windows error. Once in your hands, and visually observing what the user had done, all will be revealed . . . in the crystal ball.
So one of the latest updates for the OnePlus One wouldn't install on my phone, so I contacted support. After some back-and-forth they suggested that I root my phone and try again
I told them that I rather not mess with the device like that, because I don't want to lose any data, but if I must do it, they should tell me how to avoid any risk
I was assured that this is a safe process and no data will be lost and pointed me to "some article" that explains how to root your phone
I find it amusing (and refreshing, really) that a phone manufacturer would recommend rooting. Now if only they provided root privs out of the box so there isn't a "rooting" process...
What I especially loved with the Samsung devices was lines like "Download the modem package that is right for your device [link to a page containing 100 different packages with cryptic names]"
And then you look into that and end up with some post on XDA where someone says "Well, some of them will result in a working phone, the other won't. Those that result in a working phone are the correct ones"
