Slashdot Mirror
A 2nd Core to Keep Windows Chugging Along?
Eh-Wire writes "Almost every hardware junkie I know would give most anything to take a spin in the new dual core hot rods from Dell or one of the custom system builders. But what if you actually needed that second core to run your anti-virus, spyware detection software and firewall just to get a little gaming or Internet surfing done on the first core. Would that really be a good reason to bring home a shiny new machine? I can think of a couple of different things I could use a second core for but running an iron lung on it just to keep the machine chugging along just isn't one of them. Curiously enough, PCMag thinks that's a perfectly good reason."
Wonder what Tiger would be like on a dual core processor......
It's either on the beat or off the beat, it's that easy.
I moderate therefore I rule!
--
Its making me download the page and view it locally. The headers look messed up, might be due to a poorly written script thats sending out incorrect headers.
Content-Type: text/html
; charset=ISO-8859-1
That line, although valid, though not be two lines. The line break is throwing Firefox off.
In America we are imprisoned by our fear of them.
It's not news. It's a plug for his website. That first link belongs to the submitter.
yes, but to get a dual core computer, you wont have to spend an extra $400 on a special mother board, you won't have to spend the full price for a second CPU.
Instead you will pay the usual price for the motherboard, and around $80 more than the cost of a single CPU.
Intel and AMD need to sell the dual core CPU's cheap to get them in the market fast, so that all those lazy programmers will actually take advantage of the new hardware out there.
George Bush + Linux = "I will not let information get in the way of the fight against Windows"
That was my thought when I saw the first mention of using the second core for a virus scan. However, a little later in the article, the author devotes a paragraph to the subject:
One of the complaints we've heard from readers is that "protection" programs, like Norton Internet Security, are useful for safeguarding their systems. but slow their computers to a crawl. Dual-core Hyper-Threaded processors, such as the Pentium EE 840, can help, improving your computing experience because the processor's dual cores can process tasks simultaneously. While most of the system is "concentrating" on making sure your Internet or gaming experience is fulfilled in the foreground, the reserve power that the dual cores provide protects you in the background, running Norton or other antivirus or firewall programs.
This looks to me as a little bit more substantial than a mere example. The article doesn't devote a paragraph to any other specific application.
Not saying this is worth putting up on Slashdot's front page, mind you. It's just somebody's article; it might be conversation fodder, but it's not news.
I'm running Firefox 1.0.2 on FreeBSD and I get asked if I want to download an .html file. I've seen this before on a couple of occasions with Firefox - even on Windows. Their Apache is misconfigured.
I don't always use unix-like operating systems; but when I do, I prefer FreeBSD.
According to the article linked to higher in the thread, the upcoming IBM chip will have two cores, each with a separate cache. It also opined that to the OS, it will appear as if the machine has two processors. (Actually the example was discussing two multicored procs appearing as four processors, but you get the idea.)
"Ladies and gentlemen, my killbot features Lotus Notes and a machine gun. It is the finest available."
Sounds like something that could actually get .net apps running in the near vicinity of fast, as opposed to downright hang-dog slow.
As I've seen over the years, the more CPU(s) you throw at developers (myself included) the more difficult tasts suddenly fall into the realm of 'possible' because you simply couldn't without the extra resources. However... tools which were absolutely hideously slow suddenly look acceptable (where PHB's are concerned) because they're blind to how much faster and requiring less resources something actually developed efficiently could run.
A feeling of having made the same mistake before: Deja Foobar
In point of fact, a dual-processor machine is always going to be faster than a single-processor dual-core machine, all other things being equal. You've got twice the cache and twice the bandwidth from CPU to memory on a dual-processor system.
There has never not been a dual-processor Power Mac G5. The first generation of G5s we shipped included a single 1.6 GHz, single 1.8 GHz and dual 2.0 GHz.
Heh, they've already done that. Remember the ATX 12V connector? 4 pins of juice that basically plug in directly for powering the hungriest of Intel processors (I've noticed the AMD boards have stopped using them, for the most part, as the extra juice was never really needed). Also, remember that the power has to go through connections much smaller than motherboard traces--namely, the CPU itself! In fact, something like 50% of the pins on a CPU are for power; they can't put it on just a few pins, because the current would cause the things to melt. This is basically how they "solve" the power distribution power: using lots and lots more wires.
Funny you should say that on today of all days. I spent a big chunk of the afternoon finalizing some of the documentation for launchd.
The traditional UNIX startup model calls for a lot of tasks to be fired off at boot time, one after the other. Whether you use init scripts or rc scripts or whatever, the model is the same.
In Panther, we created a fairly sophisticated system for firing off these tasks in parallel instead of serially. The net result was a decrease in cold-start times of about 100%.
Now we've got launchd. The idea now is that instead of making the user wait for a bunch of services to start, we let launchd fire them both in parallel and asynchronously.
I don't want to get extremely specific here for reasons I hope are obvious, but on modern (i.e., dual-G5) hardware, the time from the end of power-on tests and the initialization of Open Firmware to the menu bar and dock appearing and the system accepting user input is as little as four seconds.
Four seconds to cold-boot the operating system.
Pretty impressive, no? All it takes is a willingness to look at the traditional way of doing things, recognize massive stupidity, and correct it.
In fact, Windows XP SP1 with AVG *and* a software firewall ran office and home apps faster on my old C433/256 than Mandrake 9.2 *or* FreeBSD 4.3 with no A/V or firewall. But, since I dare say so on Slashdot, I'm either a liar or a paid Microsoft shill.
-
Inventor of the term 'pardon my French'.
"Of about 100%?" I'm way too sleepy. Obviously I meant of about 50%.
Rather than bitching, why not spend a little time figuring it out? It's pretty obvious, if you think about it. Here we go. First, choose the Tools menu, because Tools always contains configuration menu options. Next, choose Customize under tools, because Customize in is where you customize menus and toolbars in Office applications (and many other Microsoft apps as well). Click over to the Options tab, because you're looking for options (the other two, Toolbars and Commands, are obviously not what you want). Looky there! I see a checkbox for "Always show full menus"! I wonder what that could do?
Yes, it's "buried", but it's buried in a logical place if you're familiar with Office products. (disclaimer: The above steps are for Word 2003. They may be different on older versions, but probably not.)
The 244 watts figure is the power consumption of the entire system, not the chip itself.
In point of fact, you're wrong. It's not always going to be better. It very much will depend on the work load and the system architecture.
Last I checked, Intel systems share a bus to memory. That means you do NOT get extra bandwidth to memory for each additional CPU. You do on AMD systems though.
That aside, however, there are loads in which dual core with shared cache would be TREMENDOUSLY better than a dual CPU setup would be.
Here's an example:
Run the network stack on one core, run the consumer application on the other. Network stack loads the data to be worked on from the network -- it's now in the shared cache. It does its thing, and sends the data to the userspace consumer application. Because it's in the shared cache, the other core doesn't have to go to memory to get the data before working on it, saving a bunch of clock cycles. The consumer level application does its thing, and sends back a reply. The reply is already in the cache, so the other core doesn't have to read from memory, saving another load. It manipulates the data, and writes the result to the network device.
On a dual processor system? The stack loads the data, does its thing. It writes the data back to memory, and hands it off to the application. On the other CPU, the application loads the data, does its thing, and writes the response to memory. The stack on the other CPU now has to read the response from memory before working on it and writing it to the network device.
Tally it up:
Dual core: 1 read from main memory, 1 write to main memory.
Dual processor: 3 reads from main memory, 3 writes.
This is a little simplified, but is a reasonable way to think about it.
For some loads, the dual processor will be better. But not for all of them, that's for sure.
Wouldn't the majority of the time spent switching from one high-ram-usage app to another be spent paging data in and out to disk (virtual memory)?
Sounds I/O bound to me. Extra RAM will make more of a difference than dual cores, since you avoid paging as much stuff.
Now if the stupid app is recalculating everything just because it got window focus... hmm, I'd call that a crappy application.
how from a usage standpoint is a dual-core machine any different than a dual-processor one? Obviously from a design standpoint it's much nicer to have one chip with two cores rather than two separate chips, especially in terms of cost, but does a dual-core processor ACT any differently to the software than two separate processors would?
There is lower latency between the 2 CPUs too... multiple CPU do need to communicate with each other often to maintain cache coherence and other stuff...
Online backup with Mozy, sounds like Ozzie, but more!
Windows XP is multi-threaded.
Hey, look! I'm replying twice to the same post!
What you likely mean is that Windows XP Professional (as well as Windows 2000 Professional) can "see" a second processor and make use of it. XP Home and the regular Windows 2000 cannot do this.
"Multi-threading" refers to the individual application's ability to execute more than one instruction at a time if it has access to two processors. AutoCAD (from what I understand) cannot do this. Microstation, 3D Studio and Photoshop can because they were written to take advantage of more than one CPU.
not to mention that many of the programs that allow you to turn them "off" at startup just add a switch to the registry key so that they are actually just hidden. or at the very least they still slow down start up so that they can at least start to run the program, only to be shut off by the switch...
"Alcohol, cause of, and solution to, all of life's problems" -Homer Simpson
Have to call you on that one ,Gnome and KDE will run comfertably on low end systems and if you dont want the bloat then dont install it.
/gnome devs or linux . With linux and Xfree you dont have to install these things ,with windows you have very little other option. .
I understand your point however i belive what your thinking about is the default install of KDE or Gnome that comes with alot of distros.
Gnome and KDE can be rather efficent if done properly. Also if you really can't stand the bloat , a default install of Xfce is rather clean not to mention other window managers such as fluxbox.
Fair enough the default is what most people see , however this is not the fault of the KDE devs
Also if you take a little time with KDE or gnome i think you can find yourself with a rather homogeneous enviroment
I use linux as i love the freedom it allows me in deciding how i want my system , and i assure you i can have a wonderfull Desktop enviroment on linux that would run comfertably on very old machines and be homogenous to my needs
The only things certain in war are Propaganda and Death. You can never be sure which is which though
Using the second CPU for background tasks makes sense, whatever the reason the background tasks run. The reality is simply that if you have independant processes they will benefit from a second CPU - it has nothing to do with not caring about quality or going for more profit. There is also a lot of quality stuff outside of GNU as well, for example: BSD, linux, solaris, AIX etc etc - and then a lot of applications.
Why aren't you just running X instead of VNC if you want it to go fast? A few signals is going to be a lot less traffic than sending bitmaps down the wire at a high refresh rate. VNC is what you use when things that can't do X are involved. There are extensions to X that have been around for a long time that let you compress the traffic, and if your X server is old then ssh can do the compression for you.A decent operating system can run multiple processes at once efficiently on the main processor (and if it's got multiprocessor support, either with discrete processor chips or just multiple cores, it can do a reasonable job of spreading the load.) Doing the job right includes managing the caches of user programs and user data and the caches of system-utility programs and data, and the right way to do that is to use an operating system that's good at managing such things. And if monitoring the user's application for safety takes as much horsepower as running the user's application, that's sometimes an indication that either the user is running really really simple applications, but more often an indication that the operating system is fundamentally not very good at protecting processes from each other and needs all the help it can get.
There may be occasional interesting research applications where it's worth wasting most of the horsepower of the second core or second processor having it monitoring the rest of the system by having it run as a trusted security monitor that's outside the primary operating system. Some of the DRM systems do things like that, though their trust-enforcement chip is a lot lower in horsepower than the main CPU, because it's basically just checking on file I/O and running checksums on the IOS and the operating system used to boot the machine.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
See CSS Zen Garden for proof of that...
(and for the web illiterates out there: there are no tables in CSSZG, and the only thing that changes between two designs is the stylesheet associated with the page, the HTML file doesn't change anywhere but where it links the aforementioned stylesheets)
"The way we can tell it's C# instead of Haskell is because it's nine lines instead of two." -- wadler
Maybe you should read up on .NET a bit more :)
.NET structs are always created on the stack. Only classes get created on the heap and need garbage collecting.
Points are structs, under
Better yet, combine Adblock with Flashblock.
Flashblock automatically replaces all flash elements with an icon you can click on to start the flash.
This means I don't have to universally block flash, but I won't have any flash crap wasting my time unless I specifically request it.
Adblock is still useful to remove other offending items, but I don't end up blocking every flash item I see anymore.
Give a man a fish, he'll eat for a day, but teach a man to phish...
Adblock already has that functionality on its own.
Jan
As for spreadsheets, I see them more as a rapid prototyping tool (if even that). When I want to get anything done that involves large lists of data, I write a Perl script to do the job. Mind you, Perl is a lot more powerful than spreadsheet programs, and it, too, takes a lot less system resources than any given contemporary spreadsheet program.
Of course, every (wo)man has his/her own preferences, and I don't write this to encourage everyone to use emacs/LaTeX/perl, but rather to spread the fact that you don't need even a 350 MHz PII or even 64 MBs of RAM to be productive, and that it is most certainly program design that makes Open/Microsoft Office take much more resources than really necessary. While you may not need a 2 GHz machine like the GP said, you do certainly need a lot more because of the fancy GUIs and stuff.