Wikipedia:Reference desk/Archives/Computing/2010 August 31

Welcome to the Wikipedia Computing Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

August 31

CD ripping -- byte perfect?

Let's say I take a copy of a CD, put it into a computer, and rip it into MP3s using iTunes.

I then take the CD out, put it into an identical or similar computer, and rip it into MP3s using iTunes, with the same settings.

Will the subsequent MP3s be identical? If I made md5 hashes of them, would they match?

I know that obviously if you change the settings, and probably even use different rippers, you would definitely get different results. But what about same rippers, same settings? (I guess you don't even need two computers to test this — you could just rip the same song twice and compare.) Is there some kind of random influence here or would it be identical? --Mr.98 (talk) 00:27, 31 August 2010 (UTC)

The simple answer is it depends on the software used, quality of the CD (whether it's badly scratched or not) and drive. Probably other factors too. Red Book (audio Compact Disc standard) wasn't exactly designed for bit perfect data transmission. For example, it has very limited error correction compared to a mode 1 CD-ROM. During playback, the reader justs reads and usually averages any errors to avoid glitches which works okay if the CD isn't too badly scratched. Not all drives are equally good in how they rip audio. For example some drives don't report errors in the ripping, some even say they don't but don't do it properly. Another problem is that precisely whether to start reading a track is something drives don't agree on. Some even vary from read to read. For this reason, some CRC implementations designed for use with CD audio ignore null samples at the beginning and end, i.e. they only work with the actual audio data. Of course some CDs are semi continuous and the sound from one track flows on to the next in which case this isn't going to help. My impression is drives are better then they used to be and there are good software programs like Exact Audio Copy which will enable you to get generally get bit identical streams compared to 12+ years ago (about when I was first interested in such things). Of course having an unscratched CD also helps. Note that any good (IMHO) CD ripping programme already has some sort of CRC or other double rip comparison function/option built in. There are also databases where you can compare your CRC to user submitted CRCs online e.g. [1]. See also [2] [3] [4]. Nil Einne (talk) 09:12, 31 August 2010 (UTC)

Interesting, thanks. --Mr.98 (talk) 00:54, 1 September 2010 (UTC)

iTunes does not do this sort of checking. For Windows you'd want EAC, as Nil Einne mentioned, for Mac OS you'd want XLD, for other Unixes you could use rubyripper. ¦ Reisio (talk) 02:47, 3 September 2010 (UTC)

Opening image in new tab in same window in Internet Explorer by middle mouse

I am having trouble in opening image in new tab in Intenet explorer by using middle mouse. In mozilla, if I click on image by middle mouse then it automatically opens in new tab under the same window. But in IE 8 it opens in same tab. I want image to open in new tab by using middle mouse. Is there any way to do it?--180.234.22.214 (talk) 00:30, 31 August 2010 (UTC)

Is tabbed browsing turned on in your IE 8? Go to Tools->Internet Options and pick "Tabs" near the bottom of the "General" tab. There you can configure how IE 8 handles tabbed browsing. It has always worked for me, but I do have a vague recollection of having to put a tick in a box to enable tabbed browsing when I upgraded from IE 7 (or was it IE 6?). Astronaut (talk) 10:02, 31 August 2010 (UTC)

Motherboard: The new 'chairman'?

I just got out of a meeting where the person leading it said the word 'motherboard' and then quickly admonished themselves and muttered something about not using that word. I asked around the office here and the opinion from my co-workers is that it's probably some political correctness thing. I've never heard any objections to motherboard before, so is this a new wave of overzealous PC or what?

And yes, I thought about putting this on the Lan. desk but figured there would be more input from IT industry people here. Dismas|^(talk) 00:31, 31 August 2010 (UTC)

I'm pretty sure concerns about the word motherboard aren't new and have existed for 10+ years. It's the primary reason AFAIK why some vendors started using the term mainboard and that's been going on for quite a while. If anything I think people care less nowadays. I seem to recall Asus made the switch circa 1997 but they seem to have gone back in recent year. Nil Einne (talk) 02:07, 31 August 2010 (UTC)

Wow, that's pretty lame. I had never heard of this. It reminds me of the "controversy" over using the terms "master" and "slave" for IDE hard disks. Comet Tuttle (talk) 15:08, 31 August 2010 (UTC)

Eh, "master" and "slave" are a bit more problematic. (Who minds being a mother?) "Master/slave" smacks of "white dudes named this," whereas "motherboard", I don't know, maybe sounds like "dude named this"? I'm not even sure it's objectionable that the main, important board be the "mother." I don't think I would have a problem with it if it was the "fatherboard." "Master/slave" is an imprecise metaphor anyway. I don't personally find it that much better (or any more convincing) when people say, "but there is also slavery outside of the USA, it doesn't have to be a race thing." Still an ugly metaphor, and not a very good one. There are easier and less leaden ways to indicate that one is in charge of the other. "Boss/Servant" is really no worse, and doesn't invoke exploitation to the same degree, much less the race stuff.

My favorite "PC engineering substitution" was once having to replace every instance of the term "Trip Cock" with "Trip Valve" in a technical manual (designed to be used only by city engineers) in order to conform with New York City municipal standards. Lest we offend the one engineer who will actually reads it and doesn't know that "cock" is a technical term in certain contexts... --Mr.98 (talk) 00:43, 1 September 2010 (UTC)

Wait, also? I'd rather hope there is only slavery outside of the USA, I'm pretty sure you abolished that stuff. To my mind it's a historical term, although I suspect it's not deliberately invoking any concepts about people, but is just carried on from some precedent set by a previous engineering term. Perhaps it comes from hydraulics, where a master cylinder has slave cylinders. (There is also the engineering term Governor (device), but there's no specific word for the thing it governs, and there are also master and slave clocks.) Plenty of people mind being mothers - typically because the fathers also mind being fathers, and have taken some unilateral action to address this - but as you say, since the board is an inanimate bit of laminated fiberglass, it's entirely unclear whether the social position implied by "mother" is one of importance or servility. 213.122.52.190 (talk) 08:47, 1 September 2010 (UTC)

Well, prohibition doesn't mean abolishment. There's still slavery in the USA, it's just illegal and we call it Human trafficking now. I know for some people these days, the "slave/master" terms are more associated with D/S fetishists. :) Indeterminate (talk) 16:32, 2 September 2010 (UTC)

Depending on context he may simply have been admonishing himself for using 'technical' words with a non-technical audience. APL (talk) 14:20, 1 September 2010 (UTC)

Preventing Windows Mail from marking email as phishing

I've just begun grad school, and one of the non-scholastic requirements of this program is that I subscribe to a school-wide listserv for announcements. Two of today's emails from professors to us students have been marked as phishing upon arriving in my Windows Mail inbox, for a reason that I can't understand. Is there any way to tell Windows Mail to whitelist all email coming in from a listserv? I've found instructions on turning off the filter and on whitelisting specific email addresses, but I want the filter to keep working, and as many professors and department staff post to this listserv, it would be a waste of time to try to whitelist all of their addresses. Nyttend (talk) 01:30, 31 August 2010 (UTC)

Could you link to the resources you found to whitelist specific email addresses? Nil Einne (talk) 02:20, 31 August 2010 (UTC)

I went into the Help dropdown menu, selected View Help, typed "phishing" (without the quotes) into the "Search Help" bar, and got the results of what I speak. The most relevant page that I found is one called "Security and privacy in Windows Mail", which says that one can create a list of "Safe Senders" to ensure that your friends and relatives don't become blocked by the phishing filter. Nyttend (talk) 02:28, 31 August 2010 (UTC)

I've looked in to this a bit more and I'm not sure if it's possible. Am I right you get a red shield warning you it's phishing? I was wonder if the 'safe sender' thing was just telling you how to set up a filter but it appear it's a specialised function seperate from the Windows Mail filter ('mail rule') option so I don't think you can use this to help. I've looked in the filter options and can't see anything there to help either. If I am correct, the only options I can think of are either use a mail client other then Windows Mail or use Windows Mail but disable the built in filters, perhaps using some other filter addon or external filters if necessary. Nil Einne (talk) 04:58, 31 August 2010 (UTC)

Some lists are also available as digests. While a digest won't arrive in realtime, it usually comes from the same e-mail address, so switching to the digested form of the list, if available, might help. -- 78.43.71.155 (talk) 08:44, 31 August 2010 (UTC)

This list isn't available as a digest, but it's rather important to me that emails come in as soon as possible. Nil Einne, you're right about the message: I get a shield, and every column of text for an email marked as phishing displays as red in the Inbox. It's not a big deal, so if there's no easy way to disable it, I don't think I'll bother. Thanks for the help! Nyttend (talk) 12:19, 31 August 2010 (UTC)

Problem on saving DDS images with alpha

OK, so I'm trying to mod an HUD texture in Mafia II, and whenever I try to save it as DDS, the alpha channel gets borked as 1-bit instead of an explicit one, even if I try to use the DXT5 option. I'm using Photoshop CS3, along with Nvidia's latest DDS tools. Blake Gripling (talk) 02:17, 31 August 2010 (UTC)

Must I retain Acrobat update files?

After upgrading my office computer to Acrobat Standard v9 a while back, I now seem to be accumulating a shitload of files in C:\PF\Adobe\Acrobat\Update. Standard name is of the form X_modulename_nnn.rtp, where X is mostly the letter H, but some A's and M's too, and nnn corresponds to the release number, 910 for 9.1, 934 for 9.3.4, etc.

By this point, I have up to five copies of each file (with increasing release numbers), taking up a couple hundred megabytes.

• Can I safely get rid of all but the latest version of each file?
• Having installed the update, and verified that it works, do I even need to keep the latest version?

Thanks, DaHorsesMouth (talk) 02:31, 31 August 2010 (UTC)

• It's quite likely that all of these are just temporary files used for installing the update. If you can't find any documentation suggesting otherwise, it might be an idea to back the folder up somewhere, then delete the original. If Acrobat continues to work as before, then you've answered the question - if it stops working, just restore the folder from the backup. ~ mazca ^talk 14:20, 31 August 2010 (UTC)

Error message on XBox

I keep getting error message on my xbox. It says w-0000-0000,x-8000-0013,y-0000-0000,z-0000-0000,ID-FFFF-FFFF. It is on my wireless system and I have tryed two devices. I get on for less than a minute then it go to blimking. To me this all started when they came and made it wpa2p from wep. —Preceding unsigned comment added by 97.113.213.249 (talk) 04:01, 31 August 2010 (UTC)

Does the error message go away if you switch back to WEP? Astronaut (talk) 09:54, 31 August 2010 (UTC)

spam filter "throwing out" political messages

Hi, I have an email account with a big internet company. I support a particular candidate for federal office. I subscribe to the candidate's email newsletter, which I just realized is being put into the spam file. I'll keep an eye out for it, so the problem is solved for me, but is it possible or likely that major internet companies are intentionally favoring some candidates by marking newletters as spam? If so, I want to know. Thanks,Rich (talk) 10:01, 31 August 2010 (UTC)

My first guess is that it is unlikely. A more likely scenario is that something about those emails attracts the attention of the spam filter. For example, many spam filters are suspicious of mails where your address is just one of many in a mailing list. Astronaut (talk) 10:08, 31 August 2010 (UTC)

There are many ways an email can fall foul of a spam filter. Internet and email companies regular check several of the spam blocklists (e.g. http://mxtoolbox.com/SuperTool.aspx ) to see if a domain has been flagged on any of them, besides doing their own filtering on email received. How does a domain get flagged in the first place? Well, they could have sent an email to a honeypot trap, they could be using too many BCC contacts in their email which their own ISP would then flag as potential spam, they could be the victim of identity spoofing (very easy with emails) which would cause their domain to be wrongly identified as a spammer. There must be many more ways, no doubt someone more knowledgeable will enlighten us further. Zunaid 13:04, 31 August 2010 (UTC)

For mass mailings to pass a spam sniff-test, you need someone writing and sending it that really knows what they are doing. Most of the time political campaigns (especially small ones) are run very cheaply and (no offense intended) with minimal technical talent. Too many hyperlinks and too many spam keywords are a good way to end up in the junk bin immediately. Pushing too many messages at once can earn a spot on a blacklist, and if the email orchestrator for the campaign isn't technically savvy they may not even know what that means or how to remedy it. As far as your theory about content based filtering, in my experience Occam's razor should be invoked; how likely are you to find a company (especially a large one) where everyone has the same ideology to even form a consensus on which side to favor? OK, *except* Fox Broadcasting... --144.191.148.3 (talk) 14:49, 1 September 2010 (UTC)

Very informative. Thanks everyone! Rich (talk) 09:45, 2 September 2010 (UTC)

Processing-Core Price Disparity

Hi Everyone,

   I want to know why CPU cores cost twenty times as much as GPU cores. What is it that CPU cores can do that GPU cores can't, which makes soo much more expensive? Please refer to the table below for more information.

Product Name	Number of Cores	Unit Price	Effective Cost/Core
Intel Xeon Six-Core X5680 3.33GHz/12MB Processor	6	$2,400	$400.00
nVidia Quadro FX5800 Graphics Card	240	$5,475	$22.82
nVidia Tesla M2070 GPU Computing Module	448	$8,000	$17.86

Thanks to everyone, reader and respondent alike. Vickreman.Chettiar 12:34, 31 August 2010 (UTC)

GPUs can only process photos and videos, while CPUs can process everything. And it probably costs most to make CPUs than GPUs... Elspetheastman (talk) 12:37, 31 August 2010 (UTC)

Because you're paying for more than just a set of cores. Marnanel (talk) 16:51, 31 August 2010 (UTC)

A Xeon core and a Tesla core are designed for different needs. One of the most important requirements of the Xeon core is high single thread performance. This requires very sophisticated microarchitecture and circuit design in order to get things done as quickly as possible. And this is very expensive. There are only three companies that have the capability to design Xeon-class CPUs: Intel, AMD, and IBM.

A Tesla core, in comparision, it is very simple, consisting of some execution units, simple (two-way superscalar, Xeons are four- or six-way superscalar IIRC) dispatch logic, a register file, and misc. interface logic. Because of their simplicity, a Tesla core is not designed for performance, it is designed to be small and to consume little power so a few hundred of them can be placed on a single ~500 mm² chip to exploit massive parallelism, which the Xeon was not intended to do.

Another reason why Xeon cores are more expensive than Tesla cores are differences in circuit design and fabrication. The Xeon, as I mentioned before, requires sophisticated circuit design. That means they'll be designing the critical paths in full-custom logic, using circuit types such as dynamic logic. (Assuming that Intel doesn't synthesis a significant amount of logic first and then optimized it extensively over multiple generations as they did with the Pentium 4, instead opting for lots of full-custom design.) This requires lots of experienced engineers, lots of time and lots of tools. GPUs, I think, are still designed using commercially-available standard cell libraries. This is easier to design with: it requires less engineers, less tools, and less time. The fact that GPUs have more cores than the Xeon doesn't matter, for both designs, the cores are simply designed once and replicated multiple times.

Differences in fabrication also contribute to the price differences. Intel develops their own process to fabricate the Xeon with. And process development is done in conjunction with circuit design so both teams can optimize for each other. This is expensive. But less expensive than the fact that Intel fabricates the Xeon themselves. This is very expensive. Fabs cost billions of dollars. In contrast, Nvidia uses a commercially-available process from a foundary, in this case, TSMC. They have to buy capacity from TSMC, which shields them from the investment in process technology and the fabs. Another difference is that Intel's process is much more agressive than TSMC's, in regards to things such as features, ect. Process technology is not merely quantified in μm or nm. You can have two 32 nm processes, and one can be 40% slower than the other.

Finally, you ask what can a CPU do that a GPU can't. Many things from a microarchitectural view. For example, a GPU cannot execute instructions out-of-order to help mitigate long-latency events such as division or a load. A CPU can. Another thing is branching. CPUs have very sophisticated branch prediction, GPUs don't. I am not sure as to the exact reason why, but the way that each GPU core is connected has something to do with that.

This is as many differences as I can think of right now. There might be more, and some of my comments might not be 100% accurate (GPUs are foriegn to me, CPUs less so), but this should be a fair outline of why CPU cores cost more. Rilak (talk) 17:46, 31 August 2010 (UTC)

(ec) "What is it that CPU cores can do that GPU cores can't..." - well, to be blunt, a GPU "core" isn't a "core." It's only an ALU. So when you see a "240-core" count, that's number of EX stages, not number of true "cores." (Well, "core" is weakly defined in computer-architecture, and varies for each type of architecture). Each GPGPU core lacks its own Instruction Fetch, an Instruction Decode, and a proper memory writeback unit. A GPGPU is therefore typically programmed to operate in SIMD mode - so it shares one (or a few sets) of this hardware over the entire chip. In lieu of a per-element decoder, Teslas use a sort of "bitmask" to turn on operations chip-wide or per-thread block. This severely limits what you can program 240 Tesla GPGPU cores to do:: they must all operate in lockstep, because they share the computer hardware that determines what instructions to run. In other words, you can run six different programs on a 6-core Xeon; but you can only run one program on a 240-core Tesla. (In this loose usage, "program" means "unconditional set of instructions"). You basically have 240 copies of the same program. You can read about the basics of instruction pipelines - like the classic RISC pipeline - to become familiar with how this works. Modern CPUs like Intel use a extraordinarily sophisticated instruction pipeline - while a GPGPU like NVIDIA's Tesla series use the much more restrictive Parallel Thread Execution pipeline, defined by the PTX instruction set architecture. Nimur (talk) 17:54, 31 August 2010 (UTC)

Like a glorified array processor such as the Thinking Machines CM-1? Rilak (talk) 17:59, 31 August 2010 (UTC)

Yes. But, unlike predecessor array-processors like CM-1 and Cray, CUDA NVIDIA's hardware and software allows "SIMT" - "single instruction, multiple thread" - which does permit much more configurable and customizable programs. This essentially means you can perform different instructions on different subsets of the array. (Though, you really harness the full peak performance of a GPGPU when you treat it as if it were an "old-fashioned" SIMD array-processor with the same exact operation for every point, and carefully manage memory accesses). Nimur (talk) 18:29, 31 August 2010 (UTC)

Google Mail

Resolved

Does google mail delete accounts which have been inactive for a long time? I know Yahoo! Mail deletes stored messages if you don't log in every 6 months, but the account remains usable. Does the same apply to google? 82.44.55.25 (talk) 15:30, 31 August 2010 (UTC)

According to Gmail#Competition "The Gmail system flags as dormant every Gmail account which remains inactive for six months. After a further three months, for a total of nine months dormancy, the system may delete such accounts.". --jjron (talk) 15:49, 31 August 2010 (UTC)

By "delete such accounts", it means that they have the option to delete the email. They do not allow the username to be reused. It remains registered. -- kainaw™ 19:49, 31 August 2010 (UTC)

Could you clarify, by "delete the email" do you mean they delete all the messages in the account, but the account is still accessible? So even after say, 5 years of not logging in, you could still log in and send messages from that address? 82.44.55.25 (talk) 20:47, 31 August 2010 (UTC)

Jjron's referenced quote is pretty complete. They say "the system may delete such accounts", which means everything in the account is deleted, and since the account is deleted, you would no longer be able to log in. Comet Tuttle (talk) 21:07, 31 August 2010 (UTC)

Ok, thanks. 82.44.55.25 (talk) 21:19, 31 August 2010 (UTC)

SSD Roadmap

Is there any good way to predict the features / sizes / cost of solid state drives in the near future? Samsung and Toshiba have been making noises about toggle-mode NAND [5], but haven't shipped any actual products yet (AFAIK). I am wondering if this new technology is likely to have a significant impact on SSD sizes / cost during the next 4 months or so, or whether it is more likely to make only a small difference compared to the current technology. Dragons flight (talk) 19:29, 31 August 2010 (UTC)

I don't know much about toggle-mode, but an SSD with nearly identical features is available from Kingston - the Kingston SSDNow V+ 512GB 2.5” SATA II SSDs (part no. SNVP325-S2/512GB). IMHO, the toggle-mode drives will only make a small difference in regard of prices of SSDs. Rocketshiporion 08:01, 1 September 2010 (UTC)

Buying a kindle edition book from amazon

...without owning a kindle? I really don't know much about the kindle, but I'd rather download the kindle edition of Star Maker for five dollars instead of buying it for thirty. Would I be able to still read the book on my Macbook? Thanks : ]  ?EVAUNIT^{神になった人間} 21:18, 31 August 2010 (UTC)

Yes. What you need is the Kindle For Mac software. It's free. It's also available for PC, Android, iPhone, iPad, and Blackberry. APL (talk) 14:16, 1 September 2010 (UTC)