EMU 1212m/1616m on Windows 7 - The Thread to End All Other EMU Threads!

[Diego_R]

Today I removed the 2GB ram stick, and recording worked again.
For testing, I recorded Winamp (WaveOut output) into Cool Edit (Wave input).
I waited several minutes, opened a few various softwares, and it still worked.
Then I put a fourth 1GB stick instead, totally identical to the 3 other ones I have. It worked... but only for a few minutes, then recording became totally silent, not even just choppy but totally silent.
I restarted and it worked again.
I manually fired CleanMem and the choppy recording issue appeared right away. I disabled CleanMem and closed it, and the problem disappeared ! But a few minutes later, the choppy recording problem came back... and a few minutes later it was working again.. I feel like running in circle.
(I left the fourth 1GB plugged in place of the 2GB.)

I now remember that since my last combination of driver+patchmix versions, it often works after restart, not just after reinstall. I had forgotten about this detail, but it is written in my diary.

It's a Windows image downloaded directly from Microsoft website. My system is in french : "Windows 7 Edition Familiale Premium" and I translated without really thinking, but I guess the english name is rather "Home".

1. At startup, there are 33 processes, 17% of the RAM is used . Swap file is fixed 1024MB, on the SSD.

diegor task list at startup .jpg

2. I did follow an online guide about which services to disable, but I do not remember which one. Here is a screenshot of the services list, but I was not able to change the language to english, so names are in french.

diegor services .jpg

3. The problem only happens with Wave input. I use Cubase only in ASIO and only for midi and vsts, I never record audio with it. I just checked, recording works in it (in asio), while in Cool Edit it does not (in wave input).
I seem to remember trying to record in ASIO back in 2005 or so with another software, but there was a problem because Cubase was also using ASIO at same time for vsts, and I was not able to sort it out, so I decided to stick to Cool Edit and its Wave input to avoid this. In Patchmix, in the main inserts I put a send to Wave, so Cool Edit can record everything coming out.

4. I unplugged all usb : drawing tablet, wifi dongle, and the display hub, on which the mouse was too. Oddly, Windows could not boot anymore, I tried four times in a row and each time it stayed stuck on the logo, until I plugged back the mouse and it got to boot.
But as in all my other tries today, it worked for a few minutes after restart then recording became choppy, without opening or closing anything.

I hate having to spend hours learning new tools, which is why I stick to old stuff that I know like the back of my hand, but I will at least try these softwares.
Interesting. In Audacity, using MME or DirectSound, the bug happens just like in Cool Edit. With "Windows WASAPI" however it often works, but sometimes it becomes choppy exactly like the other options.

( I am interested in either 4x 2GB DDR2 or 2x 4GB DDR3. Let's talk in private messages how much you would sell me these, and how much would the shipping to France cost. )

Thanks for the advice. I did not know that most new interfaces could not anymore record windows sound.

[planetearth]

Today I removed the 2GB ram stick, and recording worked again.
For testing, I recorded Winamp (WaveOut output) into Cool Edit (Wave input).
I waited several minutes, opened a few various softwares, and it still worked.
Then I put a fourth 1GB stick instead, totally identical to the 3 other ones I have. It worked... but only for a few minutes, then recording became totally silent, not even just choppy but totally silent.
I restarted and it worked again.
I manually fired CleanMem and the choppy recording issue appeared right away. I disabled CleanMem and closed it, and the problem disappeared ! But a few minutes later, the choppy recording problem came back... and a few minutes later it was working again.. I feel like running in circle.
(I left the fourth 1GB plugged in place of the 2GB.)

I now remember that since my last combination of driver+patchmix versions, it often works after restart, not just after reinstall. I had forgotten about this detail, but it is written in my diary.

It's a Windows image downloaded directly from Microsoft website. My system is in french : "Windows 7 Edition Familiale Premium" and I translated without really thinking, but I guess the english name is rather "Home".

1. At startup, there are 33 processes, 17% of the RAM is used . Swap file is fixed 1024MB, on the SSD.
diegor task list at startup .jpg

2. I did follow an online guide about which services to disable, but I do not remember which one. Here is a screenshot of the services list, but I was not able to change the language to english, so names are in french.
diegor services .jpg

3. The problem only happens with Wave input. I use Cubase only in ASIO and only for midi and vsts, I never record audio with it. I just checked, recording works in it (in asio), while in Cool Edit it does not (in wave input).
I seem to remember trying to record in ASIO back in 2005 or so with another software, but there was a problem because Cubase was also using ASIO at same time for vsts, and I was not able to sort it out, so I decided to stick to Cool Edit and its Wave input to avoid this. In Patchmix, in the main inserts I put a send to Wave, so Cool Edit can record everything coming out.

4. I unplugged all usb : drawing tablet, wifi dongle, and the display hub, on which the mouse was too. Oddly, Windows could not boot anymore, I tried four times in a row and each time it stayed stuck on the logo, until I plugged back the mouse and it got to boot.
But as in all my other tries today, it worked for a few minutes after restart then recording became choppy, without opening or closing anything.

I hate having to spend hours learning new tools, which is why I stick to old stuff that I know like the back of my hand, but I will at least try these softwares.
Interesting. In Audacity, using MME or DirectSound, the bug happens just like in Cool Edit. With "Windows WASAPI" however it often works, but sometimes it becomes choppy exactly like the other options.

( I am interested in either 4x 2GB DDR2 or 2x 4GB DDR3. Let's talk in private messages how much you would sell me these, and how much would the shipping to France cost. )

Thanks for the advice. I did not know that most new interfaces could not anymore record windows sound.

Thanks for the detailed information, @Diego_R! I know this was quite a bit to test, but it appears the problem is your RAM configuration.

I'm going to look at this a bit more, but since I won't be able to translate the Services easily, I'll do that last. For comparison however, on my Win7 (x64) system with PatchMix and my 1820m card, Windows has 27 processes running before I load Cakewalk by Bandlab. The percentage of memory in use on my machine won't compare to yours, since I have 20 GB of RAM. But I'm using 20% fewer processes than you are, on basically the same system (though I have Win7 Professional), with basically the same audio interface and drivers. So we would need to see what's going on there, and why you have so many services running. But increasing your RAM -- with good, reliable sticks -- would definitely help you.

Win7_1.jpg

While I look into your Services, what is the recommended size for your swap file with 4 GB of RAM in the system? The 1024 kb you say the file is sounds a bit low for a 4-GB system. (There are websites that could tell us the "recommended" or "ideal" swap file size for a 4-GB system, but it would be just as fast for you to see what Windows tells you. It's usually roughly the same amount as your physical RAM, though in low-memory systems, it can be 1.5x the amount of physical RAM.)

If I understand you correctly, you have three "matched" 1-GB sticks of RAM, and 4 slots. If it's not too difficult to determine the model number of the sticks, would you mind letting me know what they are? If you can't figure it out, just post a photo of the front and back of all the sticks, please. I'll see if I have something that matches them. It sounds like the 2-GB stick is what's causing the problems and that could be because the speed is very different (which forces all the RAM to run at the lowest common speed, basically), or there's something actually wrong with the stick...or both.

You mention you'd like either DDR2 or DDR3 RAM for the system. Unless your motherboard accepts both types, you can only use one or the other. And if your mobo does accept both types, you shouldn't mix them. Do you know if you have DDR2 and DDR3 RAM on your system?

CleanMem is not doing you any favors, so you really should consider disabling that completely. As you saw when you enabled it, you started having problems again and you're not gaining any memory by using it. On the contrary, it takes up memory and processor cycles.

Without unplugging everything again, could you at least unplug the Wi-Fi adapter and turn off wireless networking in Windows? That's another CPU hog and it causes problems with recording and playback and creates latency issues. Is there a reason you're using Wi-Fi and not wired Ethernet?

Have you disabled anything under "Performance" in the "Advanced system settings"? Most of what's in there makes Windows look nice, but it takes up RAM and processor cycles. I used to run Sonar just fine on a 4-GB Windows XP (and then Win7) by disabling many of the "display enhancements" which simply weren't necessary.

Win7_2.jpg

(Note: Nothing above "Enable Aero Peek" is enabled; there just isn't room to show all the options in the box, so I chose to show the ones that are enabled. This works fine for me on my system; your mileage may vary, of course.)

As @rawl747 suggested, the Focusrite interfaces aren't very expensive, and you get a decent suite of software with them. PreSonus also makes inexpensive interfaces and they're usually cheaper than Focusrite, though the Focusrite ones are generally better.

Cakewalk by Bandlab is basically Cakewalk Sonar. It's been around as long as Cubase, and you'll probably find it familiar and fairly easy to jump right in and start using. Sonar and Cubase have pretty much "grown up together", and they handle recording audio and MIDI in similar ways. They're not "loop-based" or "pattern-based" like FL Studio or Ableton and others. Audacity has come a long way since the first version, and can easily do as much as Cool Edit now. The main goal is to keep you working within the ASIO protocol, either until you have more stable RAM on your system or you speed up your processor. (I'm not suggesting you overclock your processor, but since ASIO is much more efficient at handling real-time, multi-channel, high-definition audio, it's the best choice for what you have now.)

Drop me a PM when you can and include the photos of the RAM sticks and I'll see if I have something that matches.

Steve

[Diego_R]

I did not explain myself well. Yesterday I removed the 2GB stick at the beginning of my tests.
Then I tested with only my 3 identical sticks, and recording kept working for as long as I tried (maybe fifteen minutes or so)
Then I borrowed a fourth stick from another computer that I built at same time as mine with the same ram sticks, in order to check about the >3GB thing. I put it in place of the 2GB. For all the other tests there were now 4 identical sticks, the 2GB stick was out.
During all my tests, it was always the same : recording worked for a few minutes after restart, then the bug was happening, then it worked again for a random few minutes, then the bug was happening again, just like it did with the 2GB stick. (In Audicity in ASIO it works though. I think the bug never happens in ASIO, only in waveout/windows/directsound, because Win7 handles this differently than WinXP)

I am adding a photo of one of the 1GB sticks along with the 2GB stick as attachement. But as I said, I removed this 2GB stick even before I tried other things and the bug is still happening.

Yes, my motherboard can use either 4 DDR2 slots or 2 DDR3 slots, I know they can't be used at same time, that is why I said that I was interested in either four 2GB DDR2 sticks or two 4GB DDR3 sticks, not 2GB DDR3 sticks.

Without the wifi dongle, as in all my other tries, recording worked for a few minutes after restart, then it got choppy. Then it was working, then it was not working again, and so on. I plugged it back, and there was no apparent difference.

I changed the swap file size to 4000 following your advice, and restarted. I see no difference.

The only visual effects checked are :
- desktop composition
- transparent
- smooth fonts
- use visual styles on windows and buttons
because without these it looks unbearably ugly

I am pretty sure that the processor is not the problem, it is very rarely in high load for music stuff, even in my biggest projects.
I did a little test. In one of these moments when recording was working again for a random few minutes, I had the computer compress a big video, I gave the process high priority, and while it was working I tried to record again in Cool Edit. And to my surprise it still worked.... The display was a bit laggy, but the record was ironically perfect.
Also, the holes in the faulty records are very regular, if the cpu was at fault wouldn't the holes be random ? Ironically, I wanted to take a screenshot of the holes to show how regular they are, but for over twenty minutes recording kept working... I tried to load the computer on purpose, I launched photoshop, firefox and stuff, but it kept working, as if it was mocking me... Then at last, it started being choppy again, while I did not open or close anything else.
By the way, maybe it tells something, holes are 21ms.
Most of the time it is like on the first screenshot, the pattern is : 21ms hole, 39ms sound. Exactly the same in 32bit and 96KHz than in 16bit 48KHz by the way.
Sometimes they are strange though. On the second screenshot the pattern is : 21ms hole, 0.9ms of sound, 16ms hole, 21ms sound, 21ms hole and so on.
And just after I noted this, I tried again, and now it was working again...

Thanks for spending time on my problem.

Later today I will remove the fourth stick and try again with only my 3 original ones, to confirm if it really keeps working forever or if I was just lucky when I tried shortly yesterday.

[rawl747]

Diego_R

I submit that your real problem is using any flavor of Windows audio drivers rather than the ASIO drivers. And it may have to do with the fact that when you add that last stick of RAM you are pushing RAM into areas that the audio drivers that your recording software is also trying to use for data transfer. Many drivers use upper RAM addresses for data transfer expecting that actual RAM would never populate that high. This would also explain why it doesn't start out as a problem but soon appears.

BTW, Cubase has a setting that allows it to "Release" the ASIO driver being used when Cubase is not in the Windows focus foreground. This presents other logistical issues when recording WAVE sources because Cubase may then "lose" the audio source when you click on the app generating the WAVE audio. You have to then click back to Cubase to give it back control of the ASIO driver. You can also allow Cubase to NOT release the ASIO audio driver when Cubase is in the "background" but then as long as Cubase is running, no other app may make use of that same ASIO driver.

My recollection was that using my Emu interface I still could record WAVE audio with Cubase "owning" the ASIO driver (not releasing it) as long as I didn't make the Emu card the default Windows "Speakers" device. PatchMixDSP still could "see" the WAVE audio and route it to Cubase.

If you have Cubase, I would try using it to record your WAVE audio (with Cubase using the Emu ASIO driver) and see of your problem goes away even with 4 or 5 GB of RAM in the system.

If so, I would then consider the conclusion that your real problem is using Windows audio drivers and not timing issues with RAM or with too much RAM.

Rawl

[Diego_R]

With now only my original 3x 1GB, I checked from time to time and it kept working all the afternoon. The bug never happened, not a single time.
So this confirms that the bug happens only when having over 3GB and using windows/wavein/directsound. (and does not happen with ASIO for Windows is bypassed then)

I did not know that about drivers. To me it's weird for a program, especially when written in c or asm like drivers are, not to reserve its own memory area so the system keeps other programs from using it, but they must have their reasons and this is off topic here.
This made me think of something though. My SSD is DRAMless, but somehow it "writes" way faster than it is supposed to, so maybe that part of the ram is used as a cache for it and this conflicts with the driver. For Cool Edit's temp folder is on the SSD, when it records data first go into that cache in my ram waiting for the ssd to actually write them. I never thought about that, but this is how things work, don't they? But then why would the conflict only happen with over 3GB? I will put the fourth stick back and move the temp folder to the hard disk instead and try anyway.

I never record audio with Cubase because it is multitrack purposed and I do not record for multitrack. Cool Edit (in editor mode that I use) or audacity are way more suitable for just quickly recording and editing something like I do.

Well, if there are no other solution, I will have to get used to Audacity instead of Cool Edit from now on. Shame, I don't like to change my habits but, well, if I must. Thank you guys anyway for your time.

[rawl747]

Diegao_R
I never record audio with Cubase because it is multitrack purposed and I do not record for multitrack. Cool Edit (in editor mode that I use) or audacity are way more suitable for just quickly recording and editing something like I do.

I use Cubase for just mono or stereo recordings all the time. It works just fine for this purpose. If you choose an "Empty" project template when you create a new project, there are no tracks of any kind and you can then add a single mono or audio track.

So, there is nothing that inherently means you are starting with a 16 track audio project for example. I actually saved an empty single stereo track project as WAVE recording template myself.

Something to think about.

Rawl

[planetearth]

With now only my original 3x 1GB, I checked from time to time and it kept working all the afternoon. The bug never happened, not a single time.
So this confirms that the bug happens only when having over 3GB and using windows/wavein/directsound. (and does not happen with ASIO for Windows is bypassed then)

I did not know that about drivers. To me it's weird for a program, especially when written in c or asm like drivers are, not to reserve its own memory area so the system keeps other programs from using it, but they must have their reasons and this is off topic here.
This made me think of something though. My SSD is DRAMless, but somehow it "writes" way faster than it is supposed to, so maybe that part of the ram is used as a cache for it and this conflicts with the driver. For Cool Edit's temp folder is on the SSD, when it records data first go into that cache in my ram waiting for the ssd to actually write them. I never thought about that, but this is how things work, don't they? But then why would the conflict only happen with over 3GB? I will put the fourth stick back and move the temp folder to the hard disk instead and try anyway.

I never record audio with Cubase because it is multitrack purposed and I do not record for multitrack. Cool Edit (in editor mode that I use) or audacity are way more suitable for just quickly recording and editing something like I do.

Well, if there are no other solution, I will have to get used to Audacity instead of Cool Edit from now on. Shame, I don't like to change my habits but, well, if I must. Thank you guys anyway for your time.

Now that we've sorted out what kind of RAM you have (and how much), the real issue appears to be the audio protocols, not the RAM. You'll just need to get enough RAM for Windows to do what it needs to do. And then you'll need to use ASIO.

I've had moderately powerful machines for the last 14 years that I've been doing this, and I've never had any luck using anything other than ASIO. Any of the core Windows XP or Windows 7 audio protocols would always result in pops, clicks, and glitches, and some wouldn't even let me record above 44.1 kHz. (They would say the audio interface didn't support it, when it clearly did, because it worked fine with ASIO.)

One limitation of ASIO that you may have already seen is that only one app can access it at a time. So you can't leave Cubase (or Cakewalk by Bandlab) running and then launch something else that also needs ASIO. As @rawl noted, Cubase has an option to release the ASIO driver when Cubase is in the background. (Cakewalk by Bandlab has this option too.) This is supposed to allow a foreground app to use ASIO temporarily, but I've never really had much luck with it. I just close CbB (or Sonar before it) when I need to load SoundForge to master anything.

But the performance boost in using ASIO far outweighs that minor inconvenience.

Steve

[rawl747]

One limitation of ASIO that you may have already seen is that only one app can access it at a time. So you can't leave Cubase (or Cakewalk by Bandlab) running and then launch something else that also needs ASIO. As @rawl noted, Cubase has an option to release the ASIO driver when Cubase is in the background. (Cakewalk by Bandlab has this option too.) This is supposed to allow a foreground app to use ASIO temporarily, but I've never really had much luck with it. I just close CbB (or Sonar before it) when I need to load SoundForge to master anything.

I have been able to run both Cubase and other applications playing audio. You must set Cubase to release the ASIO drivers when Cubase is in the background for this to work.

Rawl

[planetearth]

One limitation of ASIO that you may have already seen is that only one app can access it at a time. So you can't leave Cubase (or Cakewalk by Bandlab) running and then launch something else that also needs ASIO. As @rawl noted, Cubase has an option to release the ASIO driver when Cubase is in the background. (Cakewalk by Bandlab has this option too.) This is supposed to allow a foreground app to use ASIO temporarily, but I've never really had much luck with it. I just close CbB (or Sonar before it) when I need to load SoundForge to master anything.

I have been able to run both Cubase and other applications playing audio. You must set Cubase to release the ASIO drivers when Cubase is in the background for this to work.

Rawl

That's why I noted you said that in my comment, () but I guess I left the "747" off accidentally. Sorry about that.

Steve

[rawl747]

Steve

So you can't leave Cubase (or Cakewalk by Bandlab) running and then launch something else that also needs ASIO.

My response was not about my "handle" - rawl is fine without the 747.

I responded because I disagree with your statement that "you can't leave Cubase running and then launch something else."

That is NOT my experience. Sure you may have to work around the fact that whichever program using ASIO that has the Win foreground will "grab" it. But it is still possible to leave Cubase running in the background and launch another audio generating program IF you set Cubase to release the ASIO drivers.

Now, I am running Cubase Pro v11 so mileage may vary with older versions but I have been using Cubase since v 3SX and have always been able to record WAVE "system" audio into Cubase with Cubase using ASIO and my default Windows "Speakers" device using the same hardware as default.

It is actually trickier to do so with my 1st Gen Focusrite Scarlett 18i20 interface. Newer versions of the 18i20 have a WAVE "loopback" capability built-in. On my 1st Gen 18i20 however, I have to use a SPDIF physical loopback jumper on the back to get WAVE playback from default speaker audio back into the interface where Cubase can "see" it.

This was never a problem to my recollection with my Emu 1820m.

Rawl

[planetearth]

Steve

So you can't leave Cubase (or Cakewalk by Bandlab) running and then launch something else that also needs ASIO.

My response was not about my "handle" - rawl is fine without the 747.

I responded because I disagree with your statement that "you can't leave Cubase running and then launch something else."

That is NOT my experience. Sure you may have to work around the fact that whichever program using ASIO that has the Win foreground will "grab" it. But it is still possible to leave Cubase running in the background and launch another audio generating program IF you set Cubase to release the ASIO drivers.

Now, I am running Cubase Pro v11 so mileage may vary with older versions but I have been using Cubase since v 3SX and have always been able to record WAVE "system" audio into Cubase with Cubase using ASIO and my default Windows "Speakers" device using the same hardware as default.

It is actually trickier to do so with my 1st Gen Focusrite Scarlett 18i20 interface. Newer versions of the 18i20 have a WAVE "loopback" capability built-in. On my 1st Gen 18i20 however, I have to use a SPDIF physical loopback jumper on the back to get WAVE playback from default speaker audio back into the interface where Cubase can "see" it.

This was never a problem to my recollection with my Emu 1820m.

Rawl

Yes, I said "you can't leave Cubase (or Cakewalk by Bandlab) running and then launch something else that also needs ASIO". But in the same paragraph, I also noted that you mentioned the option to release ASIO but that "I've never really had much luck with it".

It wasn't your experience that you couldn't share ASIO with Cubase and another app, and I understand that. However, it was my experience. But we aren't talking about the same versions of Cubase, either. The last time I tried it was the very old Cubase 4 LE, which came with the E-MU 0404 PCIe card in 2006. Maybe it was a limitation of the "LE" version of Cubase 4, or a problem with the fairly early version of ASIO and the E-MU drivers on my (then) 4-GB WinXP system. All I remember is that it wasn't reliable in Cubase 4 LE or any version of Sonar I had on that system (even after I'd updated the E-MU drivers and PatchMix). And I'm just so in the habit now of either FL Studio's ASIO or the MME protocol with SoundForge that I haven't really tried it with Cakewalk by Bandlab. I have seen it mentioned on another forum, though.

To the OP's original issue however, running two audio apps with only 4 or 5 GB of RAM can cause problems because, in general, both will want (preferably) exclusive access to the audio protocol (ASIO, MME, etc.), the sound card driver, and the MIDI interface; and both will usually take as much RAM as they can. And a system running Win7 has virtually no available RAM as it is. Two audio apps releasing and then grabbing again the same audio and MIDI drivers back and forth over and over can make a system unstable as it relies heavily on the swap file while the apps trade off "ownership" of the drivers. So until he gets at least 6 GB of RAM in the system (and I'm working with him on that), I wouldn't even recommend trying to go back and forth between two old audio apps that usually want exclusive access to the audio/MIDI engines if they are to remain stable.

I do appreciate your input and experience with Cubase and the E-MU drivers, @rawl747.

Steve

EDIT: For "fun", I decided to check Cakewalk by Bandlab's handling of ASIO. Using E-MU's ASIO engine with CbB set to "Suspend Audio Engine When Cakewalk Is Not In Focus", I loaded SoundForge and set it to also use E-MU's ASIO engine. I didn't get any errors about "cannot open audio interface" (or whatever the error used to say), and both would play audio. But SoundForge's audio wouldn't make it out of the computer with CbB paused and SoundForge in the foreground. CbB's audio could still play in the background (if I chose to do that), and I could hear it over the speakers. But SoundForge's audio would not make it out of the computer unless I changed it to the FL Studio ASIO driver or MME (or something else).

So it appears that Bandlab hasn't fixed this issue, unfortunately.

Steve

[Smasha]

New Windows updates has killed my EMU 1212m sound card drivers loading (patchmix).

Dam you Gates.

[Smasha]

Phew,reinstalled drivers and patchmix,working again!

[RiP2]

Thanks for reply! Mine is the newer revision without firewire port.
There's no IRQ conflict. I removed all the cards and disabled the onboard stuffs.
It can happen to any of left/right/both channels. Here's a sample of the noise I get every few seconds:

I couldn't still get rid of the noise coming from 1212m. I also tried it on G41 chipset but no luck.
I even disconnected the analog board and connected the SPDIF Coaxial output to another device but the noise still comes from it

[rawl747]

Smasha
KVRist
Contact Smasha
57 posts since 4 Jun, 2010
Post Tue Jan 26, 2021 8:38 am

Phew,reinstalled drivers and patchmix,working again!

It is possible to create a Policy that will stop Windows Updates from updating your Emu drivers. There is a previous post discussing this process. Since Gates & Co. will surely strike again, it is probably worth the time and effort to work through setting up a policy to prevent this from happening in the future...

Rawl