Wikipedia:Reference desk/Archives/Computing/2011 May 10

= May 10 =

Installing Chameleon (the EFI emulator/bootloader) with Linux
I am using VMware and I have an installation of Ubuntu 11.04 on one virtual disk and a blank one on another. I want to use Ubuntu 11.04 to install Chameleon on the blank VMDK. I either want to have Chameleon as the sole bootloader or use GRUB2 to chain boot into Chameleon. I have tried for the past several days but no luck. Can I have command-by-command instructions on how to do this? --Melab±1 &#9742; 02:26, 10 May 2011 (UTC)
 * Is your VM able to actually emulate EFI? I do not think VMware Player can do this.  Check for VMWare compatibility.  I am pretty sure none of the commercial VM emulation environments are UEFI aware, let alone any of their free tools.  Some VMware hypervisors can boot as a UEFI-aware host OS; but don't expose that hardware standard to the guest OS.  If anyone is aware otherwise, please feel free to correct me.  Nimur (talk) 19:15, 10 May 2011 (UTC)


 * Virtual Box has experimental EFI support (ref), with sufficient support for very determined people to reportedly get OS-X Snow Leopard working in it (apparently using Chameleon as the bootloader). -- Finlay McWalter ☻ Talk 01:37, 11 May 2011 (UTC)
 * Well, I want to make a pre-made VM that has Chameleon pre-installed to distribute. --Melab±1 &#9742; 01:55, 11 May 2011 (UTC)
 * But, your VM image needs a virtual machine emulator to run it. It doesn't matter whether your motherboard actually supports UEFI or not; what matters is whether the virtual "motherboard" (the hardware emulation exposed by the VM player) supports UEFI.  You will need to investigate the VirtualBox.  This technology isn't mature, so if you aren't an EFI firmware hacker, you might find it a little less than user-friendly.   Nimur (talk) 03:23, 11 May 2011 (UTC)
 * I know that it needs a program like VirtualBox to run. I have tried things like sudo dd=boot0 of=/dev/hdb and nothing works. --Melab±1 &#9742; 21:28, 11 May 2011 (UTC)

Max Theoretical Overclock speed
Hypothetically, if one could use some sort of magical coolant that keeps the CPU speed of a computer at a safe temperature (say 50C), how fast can one overclock a CPU (Pentium 4 or one of the newer Intel i7's)? In other words, aside from heat, what is the next limiting factor in CPU overclock performance? ThanksAcceptable (talk) 03:23, 10 May 2011 (UTC)
 * The CPU or the motherboard? You can look at world record overclocks, AFAIK heat isn't a problem in most of these. E.g. . 50C is probably way too high for a good overclock though. Even with aircooling it often isn't hard to achieve such temperatures and still run in to a wall even with increasing the Vcc (presuming you don't just kill the CPU with the high Vcc for reasons mostly unrelated to heat which I'm pretty sure isn't uncommon). Note that most world record overclocks undoubtedly start with specific components (including CPUs) that are already known to be good overclockers. Nil Einne (talk) 06:42, 10 May 2011 (UTC)
 * Capacitance between conductors will limit max frequency. Usually 1,5 times the stamped frequency is ok. Any more and instability occurs. Electron9 (talk) 09:46, 10 May 2011 (UTC)


 * The limiting factor for clock frequency in a digital circuit will be clock skew, if thermal constraints are a non-issue. There is no way to solve clock-skew except to re-design, re-layout, and re-fabricate the CPU circuitry.  If clock-skew is solved, the next limiting factor will be transition time; to solve this problem will require using a different semiconductor process or a different type of material; or, if we're really diving off the deep-end, replacing the physical representation of digital information storage, which currently uses electric voltage (and is therefore fundamentally limited by physics related to properties of the fundamental unit of mobile charge, the electron).  Nimur (talk) 16:57, 10 May 2011 (UTC)

Intel Core i7-2600 underclocking
With proper (P67) motherboard, is Intel Core i7-2600's underclocking and undervolting feature as good as Intel Core i7-2600K's? I heard that its multiplier is locked. Does that mean that I can't underclock it? Many thanks.

P.S. In case anyone wonders, I want to underclock and undervolt an i7-2600 simply as a temporary fix in case of cooling problems (things like failed case fan, dusty CPU fan, or dried thermal paste) happening at unfortunate times. 118.96.163.70 (talk) 08:34, 10 May 2011 (UTC)
 * I can't speak specifically about the Core i7-2600 but most modern CPUs when they have their multiplier locked only lock it upwards. Downwards isn't unlocked because it's used by power saving features like Enhanced SpeedStep or Cool'n'Quiet. Voltage should depend solely on the motherboard. Nil Einne (talk) 12:36, 10 May 2011 (UTC)

WebCite
Is webcitation.org working for anyone else? Pages that I have archived in the past work fine but anything I've tried to archive in the last few days has failed. doom gaze  (talk)  10:23, 10 May 2011 (UTC)
 * www.downforeveryoneorjustme.com is your friend.  Kingsfold  (Quack quack!)  15:14, 10 May 2011 (UTC)
 * Thanks for the link, that may come in handy one day. I like how it's really matter-of-fact and tells me "It's just you" haha. However, it doesn't really help me in this case. WebCite seems to do everything it should, but the final archived versions of any page I try don't work (the blue webcite bar on the top loads, but not the rest of the page). Could someone give it a try? Archiving something takes about 10 seconds, tops. doom gaze   (talk)  16:00, 10 May 2011 (UTC)
 * For future use save the following line as a bookmarklet, which allows you one-click access to DFEOJM, automatically checking the current web address you're on:
 * javascript:(function{location="http://downforeveryoneorjustme.com/"+location.host;})
 * Regards. Zunaid 10:43, 11 May 2011 (UTC)

Powerpoint conversion
How do I change the file extension of a Powerpoint presentation project so that it works on a Macintosh computer/laptop? 72.235.230.227 (talk) 16:59, 10 May 2011 (UTC)


 * Microsoft Office, LibreOffice ¦ Reisio (talk) 17:16, 10 May 2011 (UTC)


 * Simply renaming the file extension is not sufficient. You need to either:
 * Get a Mac that is able to read Microsoft Powerpoint (PPT or PPTX) formats. Reisio linked a few options - you can actually get Microsoft Powerpoint for Mac OS X.
 * Or, export/convert your file to a different format, such as PDF or Keynote. Modern versions of Microsoft Powerpoint can save a document as a PDF; this will either be in your "print", "save", or "export" menu item, depending on your version of Powerpoint.  Nimur (talk) 19:05, 10 May 2011 (UTC)


 * Does the Mac have MS Office? If not, it won't be able to open Powerpoint files, the same as a Windows machine. Other Mac programs that can open Powerpoint files (with reasonable success) include Keynote, which is the iWork presentation software. --Mr.98 (talk) 20:00, 10 May 2011 (UTC)


 * The mac has MS Office, and there is probably a 30 day trial. Keynote also has a 30 day trial. Also, without downloading any software, you can use the mac Quick Look feature. Just select the powerpoint in the Finder, and press the space bar. It won't have animations or transitions, but you can see the slides in this mode. --Thekmc (Leave me a message) 19:52, 11 May 2011 (UTC)

Internet time not accurate
I have WinXP. I've just tested internet time accuracy by updating my computer clock time though Windows with time.nist.gov Then a couple of minutes later I listened for the time "beeps" on BBC Radio 4, analogue. Note that the analogue radio is not delayed at all as digital radio may be, and the BBC take great care to have very accurate beeps. They even take account of the time the signal takes to reach London from their transmitter - see Greenwich Time Signal.

My computer clock was three seconds slow! Why is this? I'd only reset it a couple of minutes previously. 2.97.218.79 (talk) 18:10, 10 May 2011 (UTC)


 * Let's clarify some technology: was your clock 3 seconds slow or 3 seconds behind? If your hardware clock is actually slow, in other words loses a few seconds over the span of a few minutes, there's nothing that can be done about this (except to upgrade to higher-quality hardware).  Clock drift is very real; for this reason, you might want to set your computer to auto-sync its time more often (say, 12 times per day, or whatever is needed to remain accurate).  The other possibility is that you have a very bad, (pathologically bad) network connection, and even the technology behind Network Time Protocol (used by time.nist.gov) can not compensate accurately, resulting in an error of three seconds during your synchronization.  Nimur (talk) 18:14, 10 May 2011 (UTC)


 * (ec) I'd say there was some delay in getting the current time and having your computer reset it's time to match. The Internet is subject to delays, as is your computer.  But, for most purposes, 3 seconds doesn't much matter.  Computer clocks don't seem to keep time very accurately anyway, so there's not much point in starting much closer than that. StuRat (talk) 18:18, 10 May 2011 (UTC)

My computer clock is three seconds slow - when the actual time was 19.00.00, the computer clock was at 18.59.57. The clock is not losing time, its just that Windows reset it at the wrong time. I'd be interested to know what the proceedure is for resetting across the internet, given the usual delays. 2.97.218.79 (talk) 18:25, 10 May 2011 (UTC)


 * That's a clock which is "behind", not "slow". StuRat (talk) 18:45, 10 May 2011 (UTC)


 * World English Dictionary "slow", meaning 5: "(of a clock, etc) indicating a time earlier than the correct time" http://dictionary.reference.com/browse/slow -- Finlay McWalter ☻ Talk 21:35, 10 May 2011 (UTC)


 * Well, that certainly is the result of a clock which is slow, and in common speech, it may not be necessary to make that distinction. I suppose that's similar to "weight", which, in common speech, may refer to either weight or mass. StuRat (talk) 22:41, 10 May 2011 (UTC)
 * I now realize that in my attempt to clarify terminology, I mistyped technology... I apologize for the additional, unnecessary obfuscation. Human communication is more complicated than computer communication. Nimur (talk) 00:15, 11 May 2011 (UTC)
 * The procedure is Network Time Protocol, as I linked above. The details are a bit technical; but as I said, it can lose accuracy if your internet connection is very bad (and by bad, I specifically mean has a large asymmetric variance in the packet transit time between you and an NTP server).  You don't usually notice this technical detail, because it doesn't affect most "web browsing" experience; except that it affects clock synch accuracy.  Nimur (talk) 18:30, 10 May 2011 (UTC)


 * Also, I see that you are in the U.K.; and you are trying to synch to an NTP server located in the United States - (time.nist.gov)! This will explain the symptom you are seeing.  Consider synchronizing to a server on your side of the ocean: e.g., uk.pool.ntp.org - and I bet you'll get better results!  Here are some instructions, NTP Servers in United Kingdom.  The issue is not technically due to the distance between you and NIST; it's due to the asymmetric latency, which is probably due to traffic patterns on the trans-Atlantic routes.  Nimur (talk) 18:34, 10 May 2011 (UTC)

I've just had a series of several edit conflicts trying to write things which you have now covered. I'm going to look for better free software. Is there a simple explaination of the protocol anywhere? Does the software just note the time the request is sent, note the time it get the reply, and adjust the given time by half the difference? Thanks. 2.97.218.79 (talk) 18:37, 10 May 2011 (UTC)


 * In a word, no, there is no "simple" explanation of NTP, and it isn't merely dividing the round-trip-time in half; in fact, NTP is actually very complicated. Our article does the best possible job to explain it in simple terms, but if you want to know how it really works, it's a very math-heavy topic.  It uses estimation theory, the Marzullo's algorithm, and engineering analysis of Internet Protocol, to first set your clock, and then compensate for the estimated error between your clock and the canonical time at the server.  The diagrams in the Marzullo Algorithm article may help you intuit the concept behind the error estimation, but the mechanism is nontrivial.  As a user of NTP, your best solution for more accurate time is to select a closer NTP server (such as uk.pool.ntp.org), not to replace your NTP client software.  Nimur (talk) 18:51, 10 May 2011 (UTC)


 * Actually Windows only provides a partial NTP implementation, with an estimated error of 1-2 seconds for the typical user. There is a good chance he could improve by using a better client (in addition to moving to a closer server).  Dragons flight (talk) 21:28, 10 May 2011 (UTC)


 * That appears to only be partially true. Although the OP is using Windows XP so this doesn't apply to them, since Windows 2003, the full NTPv3 protocol is implemented. However Microsoft still says it's not a "full-featured NTP solution" and that they can only reliably maintain the accuracy to within 1-2 seconds, greater accuracy is outside the design spec. (Guaranteed supported accuracy is only 300 second.) See Network Time Protocol and . In other words, while other NTP clients could be more accurate, they could also not be. If they are more accurate, the vendor should be able to say why. Since support for the protocol may not be different (to be fair it's only NTPv3 not NTPv4), questions of how much better do they handle things like internal hardware latency and whatever else is I presume necessary for such great accuracy. has some discussions including from Microsoft and a vendor who actually seems to know what they are talking about and  which has someone who actually tested the Windows 2003 implementation. A perhaps important point which others have hinted above is if you really want such great accuracy you may want to consider whether your bog standard RTC is good enough. Microsoft's POV appears to be there's point engineering for or supporting greater accuracy then 1-2 seconds since they don't trust the RTC to be able to maintain such great accuracy. (Another important thing is the difference between the server and the client, the server may be worse then ntpd, how much worse the client is isn't clear to me.) P.S. I do agree using a local based time server is a good idea. I always change time.windows.com or whatever the default is to nz.pool.ntp.org or my.pool.ntp.org or whatever. P.P.S. I read  and  may be helpful if you really care about obtaining such high accuracy on Windows. Nil Einne (talk) 08:42, 11 May 2011 (UTC)

Following the link given by Nimur, I found this http://www.meinberg.de/english/sw/ntp.htm as the least user un-friendly choice. That software disables the Windows time software. I have not come across any way to get the Windows time software to use time servers on this side of the Atlantic. My computer clock is now accurate. Thanks 92.28.240.189 (talk) 12:07, 12 May 2011 (UTC)
 * To change the time server on Windows XP, double click on the clock, click on internet time and then enter the new server, for example uk.pool.ntp.org as suggested above, then click apply or ok. Windows Vista and 7 are similar but things have been moved around slightly Nil Einne (talk) 15:40, 12 May 2011 (UTC)

Thanks, I stopped the Meinberg program running, then entered uk.pool.ntp.org into menu, which I did not realise could be done before. But when I ask it to update the time, it just says an (unspecified) error occurs. 92.28.240.189 (talk) 16:05, 12 May 2011 (UTC)
 * Try a few times. If that doesn't help, you may need to re-enable the w32time service. I believe it is disabled by Meinberg from when I saw their website a few days ago. Note that since you are using Windows XP unless it is Windows XP x64, your w32time service probably has a poor implementation of the simple NTP protocol rather then the full NTPv3 protocol implemented in newer versions of Windows as I strongly suspect Microsoft did not bother to update it so it possible Meinberg would perform better. Nil Einne (talk) 08:32, 13 May 2011 (UTC)

Interpreting CGI to mean PHP?
I have PHP scripts on a web server that has required the file extensions to be .CGI and each script to start with the line "#!/usr/local/bin/php". Somehow the server then knows to process these as PHP. (It is a university server, hence the completely backasswards nature of its settings.)

I'm porting the site to a new server, and I'd like to be able to keep the old filenames, etc. On my new server, when I leave the extension as .CGI, all I get is a 500 Server Error. When I change the extension to .PHP, it processes them correctly, but I'd like to be able to keep them as .CGI.

What kind of setting would I need to do to get the server to process the CGI as PHP? I have a reasonable amount of control over the new server (it is with Bluehost, and they have a pretty extensive control panel). I'm just not sure what sort of thing I'm looking for on this. Is it an .htaaccess sort of thing, or...? --Mr.98 (talk) 20:15, 10 May 2011 (UTC)


 * If it's Apache there are several ways:
 * Designate a script dir with ScriptAlias (which I think relies on the #! line being correct) with the access set appropriately
 * You can use the CGIMapExtension directive
 * Install mod_php5, which should create a .conf file in apache2/mods-available which has a FilesMatch directive for the files it handles (for example mine reads   .
 * Other servers will differ -- Finlay McWalter ☻ Talk 21:07, 10 May 2011 (UTC)


 * This page in the Joomla docwiki explains it much better than I have. -- Finlay McWalter ☻ Talk 21:11, 10 May 2011 (UTC)


 * Also note that there are a few subtle differences between PHP processed as a CGI (akin to a "command-line" program) as opposed to a linked Apache module. I think, for example, using the PHP header will behave differently when using CGI.  Take a quick look at common gateway interface for a refresher - CGI programs are expected to (and are able to) deliver the entire HTTP response, not just the content portion.  Nimur (talk) 21:59, 10 May 2011 (UTC)


 * Thanks! I actually figured out an even easier way. In the Bluehost interface I found a section on "Apache handlers," and was able to make Apache handle .cgi as .php files from there very easily by just playing with the settings a bit. --Mr.98 (talk) 22:29, 10 May 2011 (UTC)