Talk:Sound Blaster X-Fi

From Wikipedia, the free encyclopedia

Contents

[edit] Added line

"Two additional models were added in October 2006; the X-Fi 'Xtreme Gamer' and X-Fi 'Xtreme Audio'. Initial comments (October 2006) suggest that the former is similar in capability to the XtremeMusic model, while the latter, intended more for music and movie audio than computer gaming, does not support hardware EAX processing."

Just some initial info gleaned from recent hardware posts (first learned of them on Anandtech forums)

Not sure of any more detail than that; I might be interested in picking one up if it's like the SB Live! Value (Essentially feature-comparable with a significantly more expensive card) Chiaroscuros 3:38 24 October 2006 (UTC)

[edit] Removed line

I removed the following line:

While the card is incredibly versitle, keep in mind that this card's features are made for headphones. The card's market is power gamers and most power gamers use headphones beacuse they are cheap and don't bother others (For example they are essential at a LAN party).

It was unsourced and appears totally untrue. There is strong evidence to suggest Creative has also targeted those who use speakers as well and that many of the features also matter when you use speakers, especially multispeaker configs. Furthermore, it also isn't true that the cards market is power gamers. Sure it is one of the key markets, especially given the price of the more expensive cards. But Creative's adverts (such as the claim it will improve your MP3s so they're better then CD quality) strongly and other factors strongly suggest they are also targetting non power gamers (casual gamers and non gamers). Nil Einne 14:17, 16 June 2006 (UTC)

I also removed another line:

For this feature, whatever sounds good is good, but in general these are nothing more than novelty.

Firstly I have no idea what it means whatever sounds good is good. Secondly, the idea it's simply a novelty is POV and many would not agree. Indeed it doesn't even really agree with the intro. The article is still in need of serious copy-editing but I think I've removed all clearly untrue or misleading lines Nil Einne 14:21, 16 June 2006 (UTC)

Thanks for cleaning it up. If I owned one I'd try to work more on this article, but I don't. We've had some very POV additions of late with no sourcing. Especially the "proper setup" section. --Swaaye 18:16, 16 June 2006 (UTC)

Removed the following line:

However, Creative did not mention any Linux OS support in the minimum system requirements that are printed on X-Fi packaging.

Lets be logical here, does NVIDIA state on their boxes that they have the best Linux driver support in the minimum requirements? Hell no, they don't even mention the word Linux even. The list of examples from various hardware vendors can go on and on, so this line was nothing more than a pointless jab at Linux users wanting drivers for this respective card. 142.163.160.196 15:28, 16 January 2007 (UTC) Spike

[edit] Configuration differences? (Fatality, Xtrememusic et al)

The article mentions XtremeMusic, Platinum, Fatal1ty FPS and Elite Pro configurations of the X-fi cards. I have noticed that the prices of those models vary widely.

What are the differences, if any, between them? Peripherals? Driver-enabled characteristics? Other? Porfyrios 15:57, 4 July 2006 (UTC)

XtremeMusic is just the card. Platinum has a breakout box with it, that fits in a 5.25" drive bay. Fatal1ty has the breakout box and 64MB of, uh... some sort of dedicated buffer to increase the amount of channels / sound quality. (Games need to be programmed with that in mind; Battlefield 2 and some others support it.) The Elite Pro lacks a breakout box, but has a very large external unit instead with more features. It also has the 64MB of RAM. AthlonBoy 15:23, 12 September 2006 (UTC)

Exactly. They've since added three more models: the X-Fi XTreme Gamer Fatal1ty Professional Series is the X-Fi Fatal1ty without the I/O front panel and remote, the X-Fi XTreme Gamer is a budget model with the gaming mode only, and the X-Fi XTreme Audio is a budget model with the audio and movie modes only. --JaceCady 01:57, 26 November 2006 (UTC)

X-Fi Xtreme Audio (and perhaps some other models) seem to use a sort of crippled X-Fi chip or even no X-Fi chip at all. "Xtreme Audio" doesn't even use the X-Fi drivers but a variation of the Audigy 2 drivers (some sources say it lacks hardware acceleration, which was instead supposed to be a trademark of the X-Fi chipset). It only has some polishing effects supposedly done in hardware, but no hardware polyphony. EpiVictor 16:48, 3 February 2007 (UTC)
Perhaps I'm wrong but it's my understanding the XtremeGamer supports the audio/movie mode and the gaming mode but not the recording mode Nil Einne 10:57, 6 March 2007 (UTC)

[edit] "at the time of its release"

This article is peppered with qualifiers on positive comments that strongly imply something better is on the market. If this is the case, could someone tell me what it is so we can at the very least put it in 'See Also'? --Ted 01:13, 4 October 2006 (UTC)

Yes, please - I agree with Ted, what cards are being referred to? I haven't much knowledge about sound cards. Graphics cards, i'll tell you the name, shader count, cost, and time of release - but I have no such information for sound cards. It would be very encyclopedic to at the very least mention a competitor in passing. Someone who knows needs to put the info down. -Skorpus McGee 22:19, 4 October 2006 (UTC)
I agree but one of the problems is there isn't really a competitor in the gaming market (well other then Creatives older cards)

[edit] Doesn't make sense

The first function of the card is to modify the frequencies, e.g. raising a frequency of 25 Hz to 26-27 Hz. The second function alters the quantization (bit-depth) of all the input audio (which is usually set at 16-bits) to 24-bit "audio" (to enable higher calculation precision for the Crystallizer effect). For instance, the Crystallizer helps headphones and speakers that have substantial performance penalties to output audio at a better quality of sound. However, because it shifts lows lower, it will likely reduce some of the bass but this will reduce strain on a lower-end speaker and by shifting higher frequencies to even a higher frequency, the audio will appear to have a greater quality of sound.

The whole tuning sense needs clean up but this one is particularly bad... First it gives an example of raising low frequencies. Then it says it shifts lows lower. Also, won't shifting low's lower increase the strain on low end speakers? High end speakers of course should be able to handle this lower bass! Nil Einne 21:19, 29 October 2006 (UTC)

[edit] 10,000 MIPS?

I find this difficult to believe. Given 10,000 million instructions per second for a chip running at 400 megahertz, this would come down to 25 instructions per cycle. 10*10^9 operations per second would sound more reasonable if it were an OMG-wide superscalar SIMD-style DSP, possibly with multiple instruction streams (which wouldn't be that far-fetched given that it's the approach used by contemporary graphics processors). Still, 25 operations per cycle sounds more like the theoretical upper bound of some part of the execution pipeline rather than real-world performance. In other words, marketing material which, in the article, is neither tagged as such nor elaborated. 88.112.3.21 14:48, 2 November 2006 (UTC)

You are undoubtedly right that the rating is way off. Creative rated the old EMU10K1 on Live! as a 1000 MIPS DSP. A rather miraculous number there too. --Swaaye 22:27, 2 November 2006 (UTC)

Thanks for doing the maths! Of course sound cards tend to have a lot of small tasks (or is it a few big ones? I mix that up with graphics.) Eoghanzer 05:20, 2 July 2007 (UTC)

[edit] Criticisms

The 20K1 chip is able to utilize a significant amount of RAM to store sound effects for faster and improved processing, just like the previous E-mu 10K-series and E-mu 8000. This feature, now dubbed X-RAM by Creative, is claimed to offer quality improvement through audio processing capability enhancement, in addition to further reduction in host system CPU overhead.

How, exactly, is this a criticism?

Moved it. --JaceCady 22:38, 28 November 2006 (UTC)

I have added quite a bit of criticisms regarding the "X-Fi Xtreme Audio" model, based from what I could find from forums and Creative Labs themselves. It's really weird, since that model costs 3 times as much as an Audigy 2 SE (which is only marginally better than an integrated codec), performs like one (no hardware acceleration in games) and costs almost as much as a bulk Audigy 4 (which was almost Creative's top-of-the line before X-Fi, and isn't crippled in any way). So really, what's the deal here? Is the X-Fi "Xtreme Audio" just an Audigy 2 SE with a crippled X-Fi chip and the only addition of some audio polishing post-processing ? Is it really worth paying as much as an Audigy 4 to get it? EpiVictor 13:43, 3 February 2007 (UTC)

I've left most of it in but I wonder if it is really valid criticism. If I'm not mistaken, the Xtreme Audio is not supposed to support hardware acceleration for games by design. I'm pretty sure when it was launched I did the research and it only supports the audio/movie mode not gaming mode. One possibility is they have purposely ensured they used special drivers to make it more difficult for people to softmod the drivers to enable features they disabled by design. So it isn't really surprising IMHO then that the card results in slow downs in games etc. Really this seems a little silly to me. I'm all for blaming Creative when there is a valid reason but if Creative have released a card which they have made resonably explicit does not support a certain feature then it seems silly to blame Creative for not supporting this feature. If the card really lacked the EMU20k1 then there might be valid criticism but then again, did Creative ever advertise it as having the processor? One thing is for sure IMHO, talking about problems in games is silly. In any case, if I'm not mistaken the XtremeGamer supports gaming mode and audio/movie mode. What are the drivers like for this one? Nil Einne 11:16, 6 March 2007 (UTC)
I did BTW remove this:
The PCI-E version of the card [1], while different in profile and PCB layout than the PCI, makes the difference in processor size compared to true X-Fi cards more evident, as well as clearly lacking any external memory.
It's unsourced speculation (it's a forum post and no one has even really discussed the processor size issue from what I can tell). As this is the only PCI-express X-Fi card so far, there is no way we can know what the 'real' processor size is. It's easily possible I presume that Creative had to redesign the EMU20k to support PCI-express and so ended up making it for a different process or whatever. Or put it another way, if the PCI-express version of the XtremeAudio has a smaller processor then the PCI XtremeAudio, how do you come to the conclusion the XtremeAudio uses a crappier processor (maybe it does, but comparing the PCI-express to PCi seems silly)? Until and unless other X-Fi PCI-expresses are released we can't really compare the processor sizes IMHO and regardless if we do want to talk about this processor size issue, we at least need a reliable source.. Nil Einne 11:16, 6 March 2007 (UTC)
Addendum: Actually, the X-Fi XtremeAudio looks awfully like an Audigy SE. This is fairly damning evidence IMHO, much more so then the gaming thing which as mentioned, I find a little silly. And I still feel the PCI-express thing is not valid since your comparing apples to oranges Nil Einne 11:21, 6 March 2007 (UTC)
Uhh...well, Creative Labs have been (and are) subject to heavy criticism regarding their products, and the XtremeAudio is just a typical example of why this happens: it's marketed as X-Fi but it's not an X-Fi (drivers are different, lacks the hardware acceleration which should be the trademark of every X-Fi card, so what's left to it?).
Regarding the PCI-E bus, using it is not a great advantage for a sound card (the data that has to be exchanged is amply within the standard PCI bus capabilities, save for Creative's well-known issues with it). What made the PCI-E model "interesting" was that the photo showed the "processor" for what it was, without any covers, so it should be possible to identify it, if someone can provide a better source with the codes readable. It's smaller size just hinted to the fact that it's probably not a true X-Fi, something which is impossible to tell directly on "normal" PCI XtremeAudios. I'm not saying the PCI-E version is crappier than the PCI, I'm saying they are (probably) equally crappy, only the bus changes, and that pic showed the chip "naked".
Even assuming that it uses an EMU20K is far too optimistic, in fact the XtremeAudio probably doesn't even use a "fully working" EMU10K processor, just like the Audigy SE doesn't. Anyway...I'm not in position to judge Creative's marketing moves, but it wouldn't surprise me if the XtremeAudio will someday be marked as the "only non X-Fi product of the X-Fi line", and is nothing more than a beautified Audigy SE. Too bad that the tag price and Creative's reputation for buggy hardware doesn't help it in that aspect. EpiVictor 12:17, 6 March 2007 (UTC)
While I may disagree with Creative's marketing ultimately Creative is the one who define what an X-Fi is. Therefore, you cannot say a card is not an X-Fi just because you don't feel it should be one. Additionally, as I pointed out, Creative have never marketed the X-Fi XtremeAudio as supporting gaming or EAX. Also my point was not that the PCI-E would be better but rather as it is a completely different card it was easily possible that Creative redesigned the EMU20K e.g. with a different process. Although this does not appear to have been the case, it was not and is not possibly to conclude anything about the XtremeAudio PCI from the PCI-ex version. Ultimately all this is OR unless good sources can be found. Nil Einne 11:13, 6 April 2007 (UTC)


Other criticisms, at least judging from Creative Forums, is that none of their PCI products, including the X-Fi series, have solved the SCP (Snap, Crackle & Pop) problems that plague the Soundblaster line since the Live! series, and their PCI bus mastering remains poor and off-specs, making the cards dangerous to install. EpiVictor 12:17, 6 March 2007 (UTC)
I have yet to see any real analysis of the reasons for Creative cards problems from someone who really knows what they're talking about. Creative have their own theories and claims, others make their own claims. None of the stuff I've seen approach a reliable source. Some people suggest Creative are doing things in spec but in a somewhat abnormal way which doesn't work with certain mobos that don't follow the specs properly. Others suggest as you that they're not on spec. Regardless tho, all this is OT. It's also not 100% clear from what I can tell why Creative's latest cards have SCP problems. It may be the same as the previous problem, it may be different. From what I can tell, it occurs mostly in insane systems like SLI and the like and although as with all such matters, it's impossible to know for sure it does seem to only effect an extreme minority of users (albeit a minority large and vocal enough to merit attention). IIRC they only have a measly 7 ms buffer so the cards are extremely latency sensitive. Obviously as sound cards, bandwidth is not such an issue, it's latency that matters the most. If the card doesn't receive any data for more then 7 ms, you get SCP. Whether the small buffer comes from skimping on cost or whether they didn't want to make it much larger because it will reduce the 'realtimeness' of sound is anyone's guess. At best, what we can perhaps report the problems, assuming a reliable source can be found but I doubt there is any call to expand on the problems much. Nil Einne 11:37, 6 April 2007 (UTC)

BTW, in terms of criticisms, there are a lot of valid criticisms that could probably be sourced. Not directly related to the X-Fi but criticisms of Creative in general like the lack of support for anything non-Windows (which we already mention to some degree), their refusal to provide tech specs to enable OSS developers to support their cards, their extremely bloated drivers and software and the way they like to think they own your computer, their much lambested tech support etc. This would probably merit an article in itself or a mention either in Creative or the more general Sound Blaster article and we could link to it if someone works on it. These seem to me to be more valid criticisms then complaining about the X-Fi XtremeAudio which may be an overprice POS which doesn't support gaming or hardware acceleration but was never marketed as supporting gaming or hardware acceleration anyway Nil Einne 11:49, 6 April 2007 (UTC)

Rewrote the Xtreme Audio criticisms a bit better. They actually don't market it saying "Watch out! It doesn't have gaming mode!" or "Beware! It doesn't have hardware acceleration!" (however they do say it "enhances gaming experience"), at least on the box/early versions. And on their site it doesn't say ofc "Guys, this is a rebranded Audigy SE, so no hardware acceleration here, even if we sell it 2x-3x as much!". These things are basically all for the user to discover himself. I just tried to write them down in the most painless way possible. EpiVictor 18:48, 6 April 2007 (UTC)

The comment about the graphic eq is unwarranted. All graphic eqs will introduce audible changes, which is the point. They will sometimes introduce audible group delay, but it is possible to design a filter with constant group delay. Furthermore, when the sliders are set to 0 on any eq, it is entirely possible to pass audio without any change, provided that the underlying DSP transformation is capable of perfect reconstruction and the precision is great enough to eliminate roundoff error. Typical errors in the Matlab implementation of x = ifft(fft(x)) is on the order of 10^-14 if I remember correctly, which are neglible with 24-bit precision.

Actually, 32-bit floating points (at least the IEEE 754 standard) are not enough for that kind of precision: 32 bit floating point gives about 6-7 significant digits, while 64-bit give 12-14, which is near the error you mentioned. 24-bit precision is nowhere near that (would be 1 part per 16.7 million). EpiVictor 09:51, 7 June 2007 (UTC)
I am wrong about not being able to represent but 10^-14 in 24-bits, but that doesn't change the situation. A graphic eq set to 0 will likely not have any harmful effect because mathematically it will be transformed into multiplication by 1.0 (specifically a FIR digital filter with a coefficient 1.0 at some delay), or at the very worst something it might drop the least significant bit. (Even if it's a 24-bit DSP core, the Motorola 56000 family and other fixed-point DSPs use 56-bit accumulators internally to minimize round-off.) Have you done any tests which suggest that the graphic eq introduces significant artifacts or is misbehaved in the all-0 case?
Not personally, but a lot of people on the the Creative Forums have complained about very noticeable distortion (especially on high frequencies) when enabling the DSP, even with minimal or "zero" settings, although those were mostly directed versus the old Audigy series, which have a quite nasty DSP with just a 16-bit/48 KHz internal precision and a poor resampling engine. It's unknown what the X-Fi DSP inner precision is, in terms of bit depth and sampling rate (perhaps again 16-bit, 48 KHz but with a better resampling engine than the Audigy series). So, the very least, enabling the DSP also enables resampling, unless a special "mode" exists to bypass it or minimize its effects. EpiVictor 10:41, 14 June 2007 (UTC)
I think the comment had general value, and should be reintroduced, perhaps written less strongly. However we need some confirmation that the EMU20K1 chip has internal 24-bit precision for its DSP, and not still 48 KHz/16 bit like the older EMU10K1/10K2. The resampling engine is better though. EpiVictor 09:14, 23 May 2007 (UTC)
It would be helpful to find out the internal precision of the chip. But don't all graphic eqs process audio to some degree, even when set to 0? Or at least, shouldn't one expect them that they might? It seems a bit moot. If one doesn't like it, one can turn it off. And no, it doesn't change the fact that Creative should be punished for how they marketed this thing....

[edit] Why are there so many citations?

Most of them are unneccesary.... Article wont be readable soon if every sentance needs backing up with a citation

The thing is that there are a lot of unreferenced claims (specifically criticisms) here. Most of these are likely true, hence they have not been removed but they still need to be referenced and failing that, should be removed. You might want to read wikipedian policies and take a look at a featured article to get an understanding of what's expected of a good article. The trouble is, a lot of people don't seem to understand that this is an encylopaedia and not in fact a place to try and correct Creative's misleading marketing Nil Einne 08:52, 16 June 2007 (UTC)
The bit about "Xtreme Audio" not being part of the X-Fi line was written since a very similar statement is used in the Sound Blaster Audigy and Sound Blaster Live! article about the low-end Live! 24-bit and Audigy/Audigy2 SE: they too, do NOT truly belong to their product lines, feature and hardware wise. So perhaps you want to "correct" those ones too? EpiVictor 12:19, 16 June 2007 (UTC)

[edit] 'Squeal of death'

I moved that particular criticism here, because it's written in a horribly POV way, openly endorsing Creative's competitors, and citing only one post on CL forums as a reference. It is true there are some hardware problems with the X-Fi lines (some, like the infamous SCP, were actually carried on from the Live! and Audigy series), others are more critical and unique to X-Fi (e.g. outright defective firmware on some batches, resulting in totally unusable cards) but independent backing up for each of these facts is needed. And surely openly endorsing another manufacturer isn't in Wikipedia's style. Here's the removed excerpt:

====The 'Squeal of Death' problem====
A great many Creative customers have suffered from an intermittent fault known colloquially as the Squeal of Death'. This manifests as a high pitched whine or squeal which plays constantly until the computer is rebooted.[2] It manifests on many high performance PCs. The X-FI card appears to be incompatible with NVidia SLI based systems and with ASUS motherboards.
Creative have failed to fix the problem, directing blame at the motherboard manufacturers. Users report that replacing the Creative card with a rival such as the Razer Barracuda card cures the problem. The 'Squeal of Death' only affects Creative Labs products.

If it's fixed, backed up properly and expanded...then it may go in the article again. EpiVictor 20:52, 19 June 2007 (UTC)

[edit] Crystallizer and mathematics

Criticism of Crystallizer in this article is somewhat unwarranted. Smart processing can restore information that was lost during compression/encoding if it's known where data originally came from.

"Aoccdrnig to a rscheearch at an Elingsh uinervtisy, it deosn't mttaer in waht oredr the ltteers in a wrod are, the olny iprmoetnt tihng is taht frist and lsat ltteer is at the rghit pclae. The rset can be a toatl mses and you can sitll raed it wouthit porbelm."

Information about correct order of letters in individual words has been lost, but you can restore it because you know it's English.

X-Fi can in principle do the same with sound.--Itinerant1 06:33, 12 July 2007 (UTC)

You can only "restore" information if it's known to be comprised withing certain limited frequency and bandwidth boundaries, and in any cause, only if it consists of frequencies lower and values "intermediate" to the ones you already have, which is also know as "conveying no extra information", exactly like what happens with the letter order example above. The example phrase above contains all the information needed to be "reconstructed" only from the first, final and intermediate letters of each word, without ambiguities. But that means that we actually have all the information the sentence contains even without "reconstructing" it. If the sentence contained ambiguous scrambled words that could be reconstructed in more than one ways, then it would not be possible to reconstruct the original sentence without using non-informational methods e.g. intended context, grammar, etc.

How does that apply to the Crystallizer and other similar "sound enhancers"? If a certain frequency isn't present in a signal after recording and encoding it with a lossy or loseless technique, you can't rebuild it. It could be frequency A or frequency B, there's no telling. If you *can* rebuild it without recording it, then it means you didn't need to record it in the first place, and it wasn't essential information anyway. The words example just proves that words have redundancy and could be written in shorted forms, and also that natural languages use finite sets of symbols and meaningful words/grammatical structures which make "recovering" easier. EpiVictor 17:42, 12 July 2007 (UTC)

Perhaps a better analogy is the "instant fix" buttons in Picasa and other imaging software. Mathematically, no, you simply cannot correct an over/underexposed image perfectly because the image is inherently lacking resolution. But you can adjust the highlights/shadows and attempt to improve image's appearance. Just as there are ways of optimizing an image for the eye, there are ways to psychoacoustically optimize a given sample of audio so that it sounds better to the human ear, even if resolution is actually sacrificed to make the transformation. (Easiest way: increase the volume.) One could also create an algorithm to detect the sort of artifacts introduced by compression and attempt to remove them (e.g. pre-echo could be detected using autocorrelation).
What Creative attempted to do -- improve audio quality of compressed audio according to a subjective measurement, and to do so by attempting to estimate the effects of a dynamic range compressor and reverse it -- is entirely possible. The marketing is, I suppose, misleading. I believe that the original comment (not mine) is correct in stating that some criticism is unwarranted. The article seems quite biased on the whole. —Preceding unsigned comment added by 24.68.157.104 (talk) 09:08, August 28, 2007 (UTC)
Sorry, but the effects of audio compression cannot be reversed. They can be surely masked or be replaced with something else that's more pleasing to ear, but cannot be even partially reversed. If that was true, then lossy compression wouldn't actually be lossy. Compare snake oil. EpiVictor 16:47, 10 October 2007 (UTC)

[edit] Vista issues

Wouldn't it be worth mentioning the initial problems with vista and that directSound still does not support EAX-features of X-fi?--213.183.10.41 07:27, 9 October 2007 (UTC)

[edit] The Mixer

The article says that we have to keep all the sound sources at 50% to avoid distortion but some time ago it used to say that the correct setting is 100%. While I happen to prefer it at 50% I'd like to know if it's proved that it's the 0db amplification setting or not. I'd also like to know if "50% for all sources" includes the master volume or if it doesn't have anything to do with the sound sources mentioned above. 89.131.127.92 (talk) 20:27, 14 December 2007 (UTC)


AFAIK it has never said 100%, I added the mixer category in initially a year ago and I wrote 50% then. You can quite easily tell that setting the volume at 100% with any channel is bad if you play bassy content. Play something bassy then put the master volume to 100%, it will sound distorted. Then put it back down to 50% then put wave upto 100%, it sounds equally distorted. Move them both up to 100% and it sounds dreadfully distorted. Check the creative forums for more information. —Preceding unsigned comment added by 81.178.112.251 (talk) 14:06, 15 January 2008 (UTC)

[edit] Added a line too

Xtreme Audio series also has CMSS 3D and Crystalizer Mindlesslysiao (talk) 04:52, 22 December 2007 (UTC)mindlesslysiao