Debunking the "perf. loss using onboard" myth

Status
Not open for further replies.

criccio

Fully Equipped
2FA
Joined
Mar 26, 2008
Messages
14,688
Now, I would ideally like this thread to be a resource for those looking to get information on the topic, but I know threads like this have a tendency to go downhill. We'll see.

It has come to my attention that there a still quite a few "gamers" out there still believing that unless they buy a ridiculously expensive "gaming" soundcard, their performance (FPS) will be lower while gaming. The argument is that the CPU will have to take over sound "processing" duties, therefore lowering possibly FPS.

Now I'm aware that this might have been true "way back when" to an extent, but how much of it do you really think was marketing fluff by Creative and the like to get you to buy their products?

And the main question being, is it really an issue at all today? With ~3GHz Quad Cores and 8GB of RAM?

A recent thread was what actually spurred me to write this post. The OP claimed he was experiencing a "50 FPS" loss when using onboard audio. I cannot and will not believe that under any circumstances. There must have been some other factors at work (crappy software... etc..)

Discuss... :)
 
onboard audio still has popping, just installed the latest onboard drivers for SoundMAX and the onboard still blows. Oblivion has popping. Playing MP3s has popping. So even if the performance thing didn't exist the fact of the matter is that most onboard audio just sucks still.
 
onboard audio still has popping, just installed the latest onboard drivers for SoundMAX and the onboard still blows. Oblivion has popping. Playing MP3s has popping. So even if the performance thing didn't exist the fact of the matter is that most onboard audio just sucks still.

Not all, my Realtek ALC888 sounds great. And I play FLAC's through active monitors. Well I did, until I got a nice external DAC.

Anyway, I don't want this thread to focus on quality, that we all know the answer to.
 
I really think it depends on what you're comparing it to.

There's probably not going to be much difference when comparing onboard to a card based on AV200 or Cmedia 8788, because both of those chips require lots of software, specially because they do EAX in software (at least the Asus solutions do).

A card that does EAX via hardware will probably outperform the onboard solution.

The onboard loss of performance is neither a myth nor the truth. It falls somewhere in between. But, I think that CPUs are fast enough today that we can tolerate using extra cycles to process audio. How many people out there have quad cores, but are only using apps that can use 2 threads? And how about core i7 with HT?

The additional work can be offloaded to those cores not being used, and you won't see a performance decrease. However, onboard will still use more CPU. It's still less efficient, but it doesn't really matter anymore.
 
I thought that the only hardware acceleration any sound cards ever offered was in cards offered by Creative prior to the release of Windows Vista, and that since Vista there is no sound hardware acceleration of any kind.

My impression was that add-on sound cards in general were more about quality and extra features ("can it do this at all", rather than "can it do this faster").
 
onboard audio still has popping, just installed the latest onboard drivers for SoundMAX and the onboard still blows. Oblivion has popping. Playing MP3s has popping. So even if the performance thing didn't exist the fact of the matter is that most onboard audio just sucks still.

My X-fi has poooping too :(. I for one love CMSS 3D for headphones. The soundmax/dolby headphone equivalent doesn't even come close too the level of immersion creative tech provides. Sure it changed the nature of the sound, but If it helps me in a game, I could not care less.
 
And the main question being, is it really an issue at all today? With ~3GHz Quad Cores and 8GB of RAM?

Not everyone has a ~3GHz Quad and 8GB of ram. The question should really be "is it still an issue for someone even with a 2-3 generation old K8 with 1GB of ram?"

Honestly, I have not put in a sound card yet, but I have been contemplating it. Not for performance reasons, but for sound quality reasons. On board sounds good, for the most part, however the extreme highs (cymbals) are complete mush and have no clear definition.
 
Not everyone has a ~3GHz Quad and 8GB of ram. The question should really be "is it still an issue for someone even with a 2-3 generation old K8 with 1GB of ram?"

Honestly, I have not put in a sound card yet, but I have been contemplating it. Not for performance reasons, but for sound quality reasons. On board sounds good, for the most part, however the extreme highs (cymbals) are complete mush and have no clear definition.

Ok yea, but again, I don't want to talk about quality.

What I really want to focus on is are we really going to see large (~10FPS) variations during gaming switching between on board and an add-in card.
 
Ok yea, but again, I don't want to talk about quality.

What I really want to focus on is are we really going to see large (~10FPS) variations during gaming switching between on board and an add-in card.

It depends on a lot of factors as I stated in my post (O/S, the add-in card type, your hardware configuration, etc), but overall I would say that it more than likely won't affect it.

It was a big deal when we all had 1.7Ghz single core P4s and AMD 2200s with 512mb of ram, but those days are long, long gone.
 
Thats my general assumption aswell. Its just the thread I originally mention where the OP was claiming a "50FPS" drop when using onboard audio. I called BS and he told me not to post in the thread again. :D

I would post the thread here, but I don't really want to piss him off anymore.
 
onboard audio still has popping, just installed the latest onboard drivers for SoundMAX and the onboard still blows. Oblivion has popping. Playing MP3s has popping. So even if the performance thing didn't exist the fact of the matter is that most onboard audio just sucks still.

No,your specific mb onboard audio blows but that is not always the case if you do more research the next time you buy a mb. I use discrete audio cards but I think there are mbs now with decent onboard solutions. For music I use external EMU 0404 USB for even better isolation from RFI.

On Vista I highly doubt there is hardly any performance difference with discrete compared to onboard or even USB external so I am with the OP.
 
how is this a myth? it has been empirically proven, demonstrated and documented on numerous occasions

the fact that it matters less now because of more powerful processors and less use of EAX and the like doesn't really make it a myth

the real question is why build a gaming system with 8GB of ram with billion gagillion hertz proc and big booming speakers or quality headphones but not want to buy a $30 chaintech sound card :D

(let it be said that I am relatively happy with the Analog Devices onboard HD sound chip in my laptop)
 
Thats my general assumption aswell. Its just the thread I originally mention where the OP was claiming a "50FPS" drop when using onboard audio. I called BS and he told me not to post in the thread again. :D

I would post the thread here, but I don't really want to piss him off anymore.

Maybe he meant 5.0 fps? That I could believe but not 50 unless the game had a major issue with the sound driver and that would be the game's fault and not the soundcard.
 
how is this a myth? it has been empirically proven, demonstrated and documented on numerous occasions

the fact that it matters less now because of more powerful processors and less use of EAX and the like doesn't really make it a myth

the real question is why build a gaming system with 8GB of ram with billion gagillion hertz proc and big booming speakers or quality headphones but not want to buy a $30 chaintech sound card :D

(let it be said that I am relatively happy with the Analog Devices onboard HD sound chip in my laptop)

Chaintech no longer makes soundcards and if you are referring to the AV-710 it never did hardware acceleration anyway. I have one right now in my XP PC.
 

So that pretty much agrees with my point, any performance hit is marginal at best. And definitely not noticeable. Unless of course you think you can really notice ~1-2FPS.

Interesting quote from that article:

Guru3D said:
There are a few notes to be aware of. First is that we did not test any of these games with Vista, and as such we did not test the controversial DS3D GX technology Asus developed to enable EAX extensions with Vista.
 

Those benchmarks are done on XP and not Vista. Vista uses a completely different sound API with no hardware acceleration so if they benchmarked on Vista the scores would be quite different. My main OS is Vista64 so I don't pay any attention to XP benchamrks any more. Win7 will be my main OS by then end of this year and Vista64 will get rlegated to the XP machine and XP will go to my WinME machine. I know others still use XP so for them the benchmarks have some relevance.
 
I thought that the only hardware acceleration any sound cards ever offered was in cards offered by Creative prior to the release of Windows Vista, and that since Vista there is no sound hardware acceleration of any kind.

Not quite. There is no DirectSound hardware acceleration, but OpenAL works just fine, and many developers are switching to it for that very reason. I believe Microsoft's plan was to switch Vista's sound acceleration over to the XNA framework, but they didn't finish. Creative just likes to pretend that there is no hardware acceleration in Vista, because they're too lazy to port EAX.
 
Those benchmarks are done on XP and not Vista. Vista uses a completely different sound API with no hardware acceleration so if they benchmarked on Vista the scores would be quite different. My main OS is Vista64 so I don't pay any attention to XP benchamrks any more. Win7 will be my main OS by then end of this year and Vista64 will get rlegated to the XP machine and XP will go to my WinME machine. I know others still use XP so for them the benchmarks have some relevance.

The masses still use XP.
 
Since this is sparked from my thread and this is the topic of this thread I will go into more depth.

No where did I ever state that I got 50 FPS less accross the board in all games/apps. For the most part there is no noticeable difference between onboard and discreet solution (from a strictly performance stand point). However in certain games that use hardware 3D positional sound heavily including EAX effects then switching to a card that requires complete software emulation and useing the CPU then you can have a significant drop in performance. If a game is not coded to use hardware paths in the first place then there wont be this significant drop or any drop at all as it is still being done by the CPU.

The game I was specifically referring to was World of Warcraft. It was brought to my attention when they removed hardware sound to accommodate vista's new sound system. However the new sound system they moved to also had hardware channels that they implemented with help of creative and the makers of the new system to appease those who were complaining of dropped performance (which in some cases for me was as high as 50fps, at the time I was running on average 200 fps). For more info on this see creatives and wow forums.

As for the quad core dual core argument and what not.... if the game in question is coded for a single thread then that means its sound drivers are still running within that thread. Other cores can't just magically pick up the slack, it has to be coded for it.

The problem is and the reason there is so much confusion around this is because its not based around standardized technology. Creative had a monopoly with there sound cards and EAX and newer games and vista dont take advantage of it anymore thanks to the ditching of directsound. However it has been proven in many games that you can still take advantage of this hardware acceleration under vista useing the new methods. They are there and support is availible to developers to use it. I can't remember the name of the company that developed the API that is used in WoW and Bioshock but it works.

Does buying a gameing card make a huge difference and that every gamer should run out and spend $300+ on a card for such a thing? no. But for the hardcore gamers like me looking to squeeze every bit of performance out of supported games an X-Fi is well worth the money. Its just like useing a dedicated NIC for online gaming. A $30 nic can game you more performance (less lags and stutters) then buying a new video card (depending on system of course) when online gaming.

What we need is all developers and manufacturers to pick a technology like openal and stick to it. Then we could see some real benefits to discreet sound solutions and it would be more common and probably cheaper too. Or even better yet just use DDS/DTS and make all sound cards able to decode it... thats the way it should be then I wouldn't even have had to make the other thread in the first place.

And thank you to the OP for keeping this out of my thread.
 
The game I was specifically referring to was World of Warcraft. It was brought to my attention when they removed hardware sound to accommodate vista's new sound system. However the new sound system they moved to also had hardware channels that they implemented with help of creative and the makers of the new system to appease those who were complaining of dropped performance (which in some cases for me was as high as 50fps, at the time I was running on average 200 fps). For more info on this see creatives and wow forums.

I'm not trying to be argumentative here, but WoW is a difficult game to benchmark. Are you sure it was a sound issue that caused this drop in performance? Even if it was, does a 50 fps drop in performance really matter if you still have 150 fps left over?

Or even better yet just use DDS/DTS and make all sound cards able to decode it... thats the way it should be then I wouldn't even have had to make the other thread in the first place.

That's not really feasible. Games cannot be pre-encoded into surround formats like movies can, simply because of their dynamic nature. You don't really know which direction the next sound is coming from until it happens. Thus, making a game output DD/DTS independent of the sound card would require a built-in software encoder, which is pretty processor-intensive and tends to introduce sound lag, not to mention the additional licensing costs for the technology.

Of course, some games do bear the Dolby logo, but it's nothing more than a certification, much like THX.
 
I think what Bun-Bun wanted to say (or maybe not..) was that sound chips that support DDL/DTSC should become more mainstream. That would allow games to be encoded in those formats in real-time, allowing the wider use real audio equipment (receivers, HT speakers), and end "gamers" reliance on sub-par speakers from companies like Logitech, Klipsch, and Creative...

We all know how terrible those sound.
 
Keep in mind that when looking for an FPS loss of any sort, that benchmarking games will show several discrepancies due to the sound system that's used. For one, usually for any software based DSP, the game will automatically use a software based audio stack, like Miles2D or software emulated DirectSound3D. Miles is a very high performance system, but still doesn't support the same features and 3D features that EAX or OpenAL provide.

But with hardware based sound also comes a usually more complex audio stack. So in BF2 for example, you're limited to 32 streams and a reduced audio model for software mode, but under OpenAL/X-Fi mode, you can do 128 with full positional audio, occlusions, etc. It's not just the echo effects and reverb, it's also positional accuracy and calculations.

Other games like HL2 don't actually use hardware audio no matter what (well you can force it, but that's a console variable). It has it's own hybrid audio renderer. In this case you wouldn't be able to measure a difference in the first place.

For pure CPU usage, Rightmark is probably best. Using an equally high number of streams would give the best comparison. Even then, the test is fairly simple and usually you only see 3-10% CPU under DS3D. But also note that 10% of a 2.66ghz Core 2 is a helluva lot more CPU needed than 10% of an Athlon XP 1700. So normalizing this to MHZ values might be keen.

In the real world, the best reason to use discreet audio has always been component quality and features. So would you rather play with muddy sounding software positional audio in a game, or a more immersive OpenAL hardware environment, even if they are the same FPS?

P.S. Just noticed that this was originally about WOW audio. You do know that Blizzard completely gutted their audio subsystem around the time of the Burning Crusade patch right? It's all software now and surround sound for many people turned to shit.
 
I think what Bun-Bun wanted to say (or maybe not..) was that sound chips that support DDL/DTSC should become more mainstream. That would allow games to be encoded in those formats in real-time, allowing the wider use real audio equipment (receivers, HT speakers), and end "gamers" reliance on sub-par speakers from companies like Logitech, Klipsch, and Creative...

We all know how terrible those sound.

That's exactly what I was talking about. Give sound cards a REAL purpose usable in mass. To get other things such as video/physics acceleration you need to buy a dedicated card. I don't see why this shouldn't be the same with true 3D Audio.

Keep in mind that when looking for an FPS loss of any sort, that benchmarking games will show several discrepancies due to the sound system that's used. For one, usually for any software based DSP, the game will automatically use a software based audio stack, like Miles2D or software emulated DirectSound3D. Miles is a very high performance system, but still doesn't support the same features and 3D features that EAX or OpenAL provide.

But with hardware based sound also comes a usually more complex audio stack. So in BF2 for example, you're limited to 32 streams and a reduced audio model for software mode, but under OpenAL/X-Fi mode, you can do 128 with full positional audio, occlusions, etc. It's not just the echo effects and reverb, it's also positional accuracy and calculations.

Other games like HL2 don't actually use hardware audio no matter what (well you can force it, but that's a console variable). It has it's own hybrid audio renderer. In this case you wouldn't be able to measure a difference in the first place.

For pure CPU usage, Rightmark is probably best. Using an equally high number of streams would give the best comparison. Even then, the test is fairly simple and usually you only see 3-10% CPU under DS3D. But also note that 10% of a 2.66ghz Core 2 is a helluva lot more CPU needed than 10% of an Athlon XP 1700. So normalizing this to MHZ values might be keen.

In the real world, the best reason to use discreet audio has always been component quality and features. So would you rather play with muddy sounding software positional audio in a game, or a more immersive OpenAL hardware environment, even if they are the same FPS?

P.S. Just noticed that this was originally about WOW audio. You do know that Blizzard completely gutted their audio subsystem around the time of the Burning Crusade patch right? It's all software now and surround sound for many people turned to shit.

You hit it right on the nose. Except the last part.

The makers of the new system that Blizzard is using (I can not remember the name for the life of me) have hardware paths included in it that work with creative cards. Allowing for more streams and better accuracy. At much better performance to boot. When they first gutted the system everyone was complaining due to crap audio quality and sounds missing because the software can't handle it. So after the wow forums and creative forums got flooded with hate the mod over at creative contacted blizz directly and blizz started working with the software vendor to implement the hardware channels in place. And I still notice a difference by turning off the hardware support though its not as bad as when the software system was first brought out as they have optimized it a bit. I can understand why they did the move. So that people without sound cards can still enjoy surround sound and reverb. However I would rather see a hardware solution become standard and cheaper (due to being standard) instead of crap software support.

People always think that the bottle neck is the CPU when its not. Most of the time (esspecially in games like wow) its the memory sub system (ram, nb, bus ect). So every little bit you can pull out of there the better.

That's when people tell me they are buying a new video card or processor to play wow I first ask them what mobo and ram they are useing. Usually they have some cheapo mobo and only 1gb of ram. Get 2-4gb of ram and then a good mobo next and its amazing how much improvement wow gets.
 
However in certain games that use hardware 3D positional sound heavily including EAX effects then switching to a card that requires complete software emulation and useing the CPU then you can have a significant drop in performance. If a game is not coded to use hardware paths in the first place then there wont be this significant drop or any drop at all as it is still being done by the CPU.


On what kind of hardware are we talking?


Its just like useing a dedicated NIC for online gaming. A $30 nic can game you more performance (less lags and stutters) then buying a new video card (depending on system of course) when online gaming.

I don't mean to threadjack but do you have links to any benchmarks? Or perhaps a link to a thread which is already discussing this? I'm very curious about this as well, again on new-ish hardware.
 
Not all, my Realtek ALC888 sounds great. And I play FLAC's through active monitors. Well I did, until I got a nice external DAC.

Anyway, I don't want this thread to focus on quality, that we all know the answer to.

I have that same onboard, at least I think I do.. I remember downloading drivers for ALC888.

Anyways, how big was the jump going from external dac? Would it have to be expensive to notice much?
 
I have that same onboard, at least I think I do.. I remember downloading drivers for ALC888.

Anyways, how big was the jump going from external dac? Would it have to be expensive to notice much?

I think you have to have quality speakers to get alot out of an upgrade like that.

That being said, I am very please with my Devilsound DAC and it was completely worth it for me.
 
I think you have to have quality speakers to get alot out of an upgrade like that.

That being said, I am very please with my Devilsound DAC and it was completely worth it for me.

I have the M-Audio AV40's. I plan to upgrade to the Audioengine A5's in a year or so.
 
Those are some nice monitors, I think a nice DAC will suit them well.

Are you reffering to the m-audios or a5's or both? lol.

I mostly just want to get more out of my current monitors now. Not sure if the A5's will end up going on the TV.
 
Well I was originally talking about the A5's but both are very nice.
 
On what kind of hardware are we talking?




I don't mean to threadjack but do you have links to any benchmarks? Or perhaps a link to a thread which is already discussing this? I'm very curious about this as well, again on new-ish hardware.

Computer in my sig. I've done testing myself on both discreet audio and discreet NIC. With NIC my ping dropped by average of 50ms in wow and things i use to call server side lag are no longer present. Also my FPS went up on average of 5-10fps and was noticeably smoother.

http://www.nordichardware.com/Reviews/?page=10&skrivelse=524 here is a review of the Killer dedicated NIC. Its not the review I was looking for. there is another one (and some testing done by linus at NCIX i beleive) that a Intel NIC or any other good NIC will get you 95% of of the performance of the Killer NIC. You will see in that review that there is a statistical significant improvement in both pings and fps when using a discreet NIC.

There was also another review that did blind testing with a group of people where they had x amount of machines set up with onboard NIC and x amount of machines with Intel NIC's (i think) and x amount of machiens with killer nic's and had online gamers of various skill levels play games like battfield, wow, ect and rotate the machines and statistically there was a improvement with machines with dedicated NIC's with the players saying things like "Gameplay was smoother on x machine" or "There was noticeably less lag". I found this review on the NCIX forums if your interested to look for it yourself. I just dont have the time atm.

Sorry I don't know of any third party testing down with sound cards. Just know my own personal experience with wow and a couple other games that support hardware sound.
 
So that pretty much agrees with my point, any performance hit is marginal at best. And definitely not noticeable. Unless of course you think you can really notice ~1-2FPS.

Interesting quote from that article:

Guru3D likes to cripple creatives cards.
Eg. use the lowest form of EAX so the other cards won't look like a pile of turd when gaming.

And if you think EAX is dead, you should really google OpenAL...and good look getting any decent soundpicture of your onboard sound...thingy.

Gaming with no hardware support for EAX/Open AL is like gaming with no AA/AF...it's plain, it's ugly...it's lacking.
 
Guru3D likes to cripple creatives cards.
Eg. use the lowest form of EAX so the other cards won't look like a pile of turd when gaming.

And if you think EAX is dead, you should really google OpenAL...and good look getting any decent soundpicture of your onboard sound...thingy.

Gaming with no hardware support for EAX/Open AL is like gaming with no AA/AF...it's plain, it's ugly...it's lacking.

Depends. I built my computer audio setup with absolute music quality in mind. It now consists of a Devilsound DAC connected to Swan M200MKII's. As far as gaming goes it still sounds great.

Actually, about 5 minutes ago I bought my first internal soundcard. :eek: Audiotrak Prodigy HD2 :D
 
Depends. I built my computer audio setup with absolute music quality in mind. It now consists of a Devilsound DAC connected to Swan M200MKII's. As far as gaming goes it still sounds great.

Actually, about 5 minutes ago I bought my first internal soundcard. :eek: Audiotrak Prodigy HD2 :D

I take it you have never heard and EAX 5.0 game on a 7.1 setup?
If onboard sound is "great", then that is SUPER-DUPER-TRILLI-FANTASTICO!!!!111!!

Don't call onboard "great sounding"..that is like raping the language.
 
I take it you have never heard and EAX 5.0 game on a 7.1 setup?
If onboard sound is "great", then that is SUPER-DUPER-TRILLI-FANTASTICO!!!!111!!

Don't call onboard "great sounding"..that is like raping the language.

Let me guess, you are a fan of the "Z-5500" types of computer speakers?

I'm just not all that enthusiastic about crazy positional audio in PC games...
You don't have to lecture me about quality audio... lol. For me its all about quality. A card that supports EAX5.0 is just not geared towards quality. Not to mention to get an EAX5.0 card connected to a worthwhile 7.1 setup I would require the card have full 7.1 analog preouts so I could use my own amplifier and speakers.

I would eaisly take a modern "on-board" audio chip and quality stereo speakers + amp over a EAX5 card and "computer" speakers.. even if they are 7.1.

Now HT is a different story. I am all for full 7.1 and HD audio. My Integra DTR-9.8 takes care of that. :)
 
Let me guess, you are a fan of the "Z-5500" types of computer speakers?

No?
But for gaming you get nothing better than 7.1 speakers.....and you started this thread about gaming...don't try and move the goalposts now.

I'm just not all that enthusiastic about crazy positional audio in PC games...
The stop posting about gaming hardware...

You don't have to lecture me about quality audio... lol. For me its all about quality. A card that supports EAX5.0 is just not geared towards quality. Not to mention to get an EAX5.0 card connected to a worthwhile 7.1 setup I would require the card have full 7.1 analog preouts so I could use my own amplifier and speakers.

I would eaisly take a modern "on-board" audio chip and quality stereo speakers + amp over a EAX5 card and "computer" speakers.. even if they are 7.1.

Now HT is a different story. I am all for full 7.1 and HD audio. My Integra DTR-9.8 takes care of that. :)
I take it you havn't heard the 3D soundpicture in an EAX 5.0 game on 7.1...so why are you wasting my time?

Only thing I have seen debunked here is your "debuunking".
 
Status
Not open for further replies.
Back
Top