Audio latency problem with P35-DS3R

[estauber]

Hi,

I'm hoping I can get some help on a problem I'm having with my Gigabyte P35-DS3R. I use my computer for audio production, and consequently need excellent audio performance.

There is a program called DPC_latency checker, (http://www.thesycon.de/deu/latency_check.shtml), that checks how well one's computer can handle real-time streams, such as realtime audio or realtime video. When I run the program on my computer, I get periodic spikes every 5 seconds or so which spike up to about 600us. Even though this number is within the program's acceptable limit, it is no good when I'm trying to run my audio applications at extremely low latency, as the problem gets worse as the CPU gets loaded, and I start to hear little audio pops and clicks.

I've tried disabling every possible device (including the soundcard) and stopped every service from MSCONFIG, but the problem persists. I'm almost sure it is not caused by my soundcard, which is a pci RME9652.

Another interesting thing that happens is the problem gets worse if I change the motherboard BIOS from f9 to f10. My latencies go up to 750us with f10.

I had my friend run this little program on his Asus P5K and he gets latencies of ~30us, which is what I would expect from a modern motherboard/CPU. I've also tried using a Q6600 (I have a e6750), and the problem is worse - I get ~760us. This seems suspiciously proportional to the decrease in fsb speed.

I'm wondering if anyone has any insight to share, and if anyone can try running the program on their computer and posting the results. I'm in contact with Gigabyte, but so far they refuse to try to duplicate my problem and instead are trying to tell me that it's my setup.

Thanks!
Erik

 

[estauber]

Oh yes, my computer config is:

Gigabyte p35-ds3r
Intel e6750
2GB ddRII 800mhz
nvidia quadro fx550
dual monitors ( 1x dell 2560x1600, 1x samsung 1920x1200)
Seagate 500gb SATA
rme hdsp9652 sound
Windows XP home
1 ide DVD burner, 1 SATA DVD burner

 

[Onus]

What are your temps? Is the chipset getting toasty? Have you reset your BIOS to defaults, especially after flashing it to another version?

 

[estauber]

Temps are normal. Everything is running at stock frequencies. I've tried BIOS revisions F6, F7, F8, F9, and F10. All are the same, except for F10, which is worse than all the others.

 

[janwillem]

You could try the latest F11 bios. The changelog is:
1. Added new super I/O code
2. eliminated memory adress bug

maybe it helps.. the bios is released jan.9.2008

 

[janwillem]

hehe, nevermind. updated my bios and went from 880us to 1150us maximum... Not exactly an improvement I guess..

 

[estauber]

wow.... 1150us is really bad especially considering the Asus board did 30us. I think I'll stick with F6! (or maybe I should try F1 and it will be better???)

Gigabyte thus far has still not confirmed that they think this is an issue. I guess most users wouldn't notice this type of problem, so they are choosing to fix other bugs.

Has anyone else measured their system?

 

[janwillem]

I've doing some more searching on this subject and found a german gigabyte forum. it contains a thread with the exact same problem. a lot of people are experiencing this.

here is a google translation:
http://translate.google.com/translate?u=http%3A%2F%2F62.109.81.232%2Fcgi-bin%2Fsbb%2Fsbb.cgi%3F%26a%3Dshow%26forum%3D1%26show%3D5589%26start%3D&langpair=de%7Cen&hl=en&ie=UTF8

(I know it's hard reading, but it's easier that german )

I think these spikes are my main cause for my creative audio card to crackle. I'm having there spikes on my p35-ds3 and p35-ds3r.

Maybe trying F1 bios ins't a bad idea.. could you try that estauber?

 

[estauber]

Success!

I tried F2,F3,and F4, and the problem goes away with all of them. So whatever Gigabyte did between F4 and F6 (F5 isn't available), that was when they broke something.

Thanks for the link to the German post. I feel better knowing that I'm not the only one with this problem. I should have enough ammunition now to write Gigabyte and convince them that they've got a problem. I would like to be able to continue to use Gigabyte boards - I like them otherwise.

Thanks for the help, janwillem!

 

[janwillem]

YES!! SUCCES HERE AS WELL!!

I have the P35-DS3 (non-R) and was experiencing these kind of DPC spikes:

Above picture is with the F11 bios.

Changed the bios to F4 and got these results:


So, estauber! way to go for us!! mny thnx for your effort! I have a P35-DS3R with a lot of audio crackling and haven't tried updating a f4 version of the bios yet. Are your audio cracks gone now?

 

[estauber]

yes, much better! I hope that solves your problems

And yes, I can now run reliably at 128 samples (3ms) on my m-audio profire lightbridge - and I'm going to test my other computer once I go to the studio later today (same motherboard, but with a Q6600 and a RME HDSP9652 PCI sound card). I'm expecting I'll be able to go to even lower latencies with that since RME has written the driver to go all the way down to 32 samples (0.7ms) - that's really low! I'll post the result tonight.

Thanks again for the help!

 

[estauber]

OK, measured my system with the Q6600 and RME HDSP9652:

Wow - it works all the way down to 0.7ms without any popping. The BIOS made a huge difference. It barely worked at all before with F6

 

[rainer]

I have a P35c-DS3R rev 1 motherboard. I'm getting the DPC spikes which make Cubase unuseable. I reverted to the F4 BIOS and the spikes went down a bit but are still hitting 1000us and up regularly and Cubase is still popping. Next I tried BIOS F2 (F3 is not available) and the spikes went way down to where they normally stay at 30us but every minute or so I still get a smike around 500us.

I've disabled everything else I can. Anyone have succes eliminating the spikes with the P35C motherboard? On my old ASROCK motherboard I always ran at .6 latency without any pops and I did a major upgrade to a Q6600 and fast RAM but the DPC spikes are killing it.

best!
Paul

 

[Shadow703793]

Here are my results on a P35-DS3L:

 

[rainer]

I wonder if Gigabyte is working on a fix for the future BIOS revisions for this DPC spiking problem. I found the F1 BIOS works best for me with latencies usually below 12us which is great for Cubase. I did however notice that when I open CPUID that I get the spikes again which tells me whatever BIOS calls CPUID is making is triggering the problem.

It strange that later BIOS versions are getting progressively worse when Gigabyte only reports basic fixes for each version.

 

[Moba]

On my P35-DS3R Rev. 1, the problem was solved by going from F11 back to F5 (original BIOS that came with the board). F11 added a High Precision Event Timer in the system devices, so maybe that's the source of this latency issue...

Maybe someone could try disabling the High Precision Event Timer to see if it fixes the problem. I can upload F5 later if anyone would like to try it. BTW, I also had higher NB temps with F11 for some reason.

 

[estauber]

I've been trying for months now to get Gigabyte to fix this issue - I finally made it past the tech support, who said they'd forward my issue to the BIOS team, but still no response.

I've since started using Asus boards, and they don't show this bug.

 

[HoratioHB]

OK, so I have an EP35-DS3P and after reading your post, I decided to check my latency. When I first started DPC latency checker, I was regularly having spikes in the 4000-8000us range with an "Absolute Maximum" btwn 50000-60000us!!! Somehow, I noticed that it seemed to be related to the voltage/speedstepping done by the "Dynamic Energy Saver" software I had running. As soon as I stopped it, I got one last spike around 60000us, and then it came down closer to the ranges you're talking about.

Thinking there might be more "enhancements" like this in the BIOS, I checked through and found that the only thing that made any real difference was disabling "Enhanced Intel SpeedStep Technology." Disabling it finally got me to where you started - btwn 500-1000us.

Hoping I could further reduce my latency, I tried your solution of flashing older BIOS versions, but for my mobo, there are only three versions, not 11. The board shipped with F2, and neither F3 or F1 made any noticeable difference in latency.

Ultimately, I don't think latency will be a huge deal for me, but it sure would be nice to figure out the problem! For now, I'll just hope Gigabyte starts taking notice and be jealous that your mobo "goes to 11."

 

[maddog2k]

I have an EP35-DS3R revision 2.1 with this problem. It has audio dropouts/glitches/pops roughly every 5 seconds in all applications. After reading this thread I checked with DPC Latency Checker and sure enough there are latency spikes of > 26000us every 5 to 6 seconds. Every now and then the spikes will stop for 30 seconds or so but start up again.

Like HoratioHB says there are currently only two versions of BIOS for this motherboard, so I reverted to F1. At first, no difference, but then I unplugged my external eSATA backup drive. THAT FIXED IT!!! Zero spikes above 1000us and the audio became smooth.

I was happy until I noticed that I was having issues with my DTV tuner card and memtest started reporting errors on modules I know are good. So thinking the latency error was purely eSATA/SATA/RAID/AHCI related and not BIOS, I switched back to the F2 BIOS. Well that fixed the memtest errors, but the latency issue returned. @#*!

I am using the (yellow) Intel SATA ports for 2 Seagate 320GB hard drives in a RAID 0 configuration, 1 DVD reader, 1 DVD RW, and one eSATA port. The RAID setting automatically turns on AHCI which I highly suspect to be the culprit. The moment I disconnected the eSATA cable from my external Seagate Free Agent Pro backup drive, the latency problem disappeared, using the older BIOS. So, at least in my case, something about SATA/AHCI is part of the problem. The other part is related to the BIOS. The older version worked better than the new, even on the EP35 where there are only two versions.

I hope this info helps someone zero in on the problem and find a solution. I'll try to figure out how to communicate this to Gigabyte as well.

 

[estauber]

Well the gigabyte BIOS team actually sent me a beta BIOS (12n), which at first I was quite excited about. I asked them exactly what the problem was and how they fixed it, and I got a short reply "we just used the old code". Not exactly very descriptive, and they still haven't released it after about a month.

Maddog2k - that'd be great if you could press Gigabyte to fix this properly.

 

[squawkbert]

I just got a rev. 2.1 of this same mobo - it shipped w/ F2 - there's F3 on their web site, but they say it's a beta. I haven't done my XP reinstall yet, so no data from me, but I'll keep watching this thread as I've got pretty sensitive ears.

related(?) note - I built an older system w/ an old gigabyte mobo that did the same thing if you used a digital speaker output, the analog one worked tons better, so if you're using the coaxial output to feed a speaker system that takes analog or digital, try switching to analog.

 

[Timo79]

Another Gigabyte P35-DS4 revision 2.1 user with the IDENTICAL problem.

DPC latency monitor shows the same off-beat, 'staircase' type pattern:-



The mobo is paired up with a Q9450 2.66GHz CPU, which, because it's a new Yorkfield chip with 1333MHz FSB, cannot use BIOS revisions earlier than F9.

Full system specs:
* Intel Q9450 2.66GHz CPU
* Gigabyte GA-P35-DS4 revision 2.1
* 2 x 1GB Corsair DDR2 PC6400 CAS4
* 512MB MSI nVidia-based NX8800GT Zilent graphics
* 2x500GB WD Caviar RE2 SATA-II drives
* Samsung SH-203P DVD-/+ Rewriter
* Mitsumi FDD with multi card reader.
* Corsair 520W HX PSU
* Windows XP + SP2.

I've put mine through the mill. That same pattern is there right after a clean Windows XP install (absolutely nothing else installed), and remains like that at any time onwards (after all chipset/graphics/LAN/audio etc drivers are installed, and after any of the usual XP tweaks are carried out).

Oddly, if I start the DPC monitor ASAP after the Windows desktop has booted up, the latencies are nice and low for a good few seconds before the spikage starts (as shown in the following screengrab), which may be important (before a certain Service is loaded/executed at startup)?

However, I've already disabled every non-critical Service, and disabled every non-critical device from Device Manager (including the usual ethernet/LAN, DVD-RW, onboard audio etc.) to no avail. Have even gone as far as completely disabling USB and firewire, and using a PS/2 mouse and keyboard. Also disabled System.ini and Win.ini from Windows startup.

Already contacted Gigabyte today (waiting for a response). The motherboard is still currently within its short 7-day RMA period too, so I've gotta move quick.

 

[maddog2k]

Timo79,

I'd be VERY happy with the results you are getting. You should have smooth audio with that.

Here is what my Gigabyte EP35-DS3R rev 2.1 is currently doing....

Those big red spikes indicate delays long enough to cause the audio glitches - and most likely delay other processes as well, its just that the audio is easily noticed. It's not a subtle thing, you will hear the problem in any app that has sound. The audio will stutter every 5 to 6 seconds. It's extremely annoying and pretty much ruins games, movies, music, media center etc.

system specs:
* Intel E6750 2.66GHz
* Gigabyte EP35-DS3R rev 2.1
* BIOS F1 (also tried F2)
* 2 x 1GB Corsair DDR2 PC2-8500 C5, 5-5-5-15
* EVGA 8800GTX GPU
* 2 x 320GB Seagate SATA2 HDDs in RAID 0 config using the Intel ICH9R controller
* Asus DVD-E616A3T DVD-ROM, SATA
* Lite-on DVDRW LH-20A1L, SATA
* ATI TV Wonder 650 HD Tuner/PVR
* Seagate 600GB FreeAgent Pro external (tried USB2.0 and eSATA)
* Corsair 620W HX PSU
* Windows Vista Ultimate SP1

 

[Timo79]

>>It's not a subtle thing, you will hear the problem in any app that has sound. The audio will stutter every 5 to 6 seconds.

Yes, I've noticed even when playing certain games it locks up occasionally and creates a "BRRRR" loop in the sound, but worse the graphics jerk every second making it unplayable (I'm specifcally referring to the game Wolfenstein Enemy Territory). Even gaming PING times are increased. The game Trackmania Nations initially seems to work ok, though.

The music production I'll be doing will be merciless. 1.5 millisecond latency (a buffer of 64 samples) on all 16in/out channels at 24-bit/44.1KHz via firewire (Motu Traveler) to closely integrate it with my hardware mixer (S/Craft Ghost) and external synths, as well as sequencing software synths, and processing incoming audio with VST effects in realtime, so I'd be happier knowing everything else was up to scratch to avoid limitations elsewhere in the system. I wont be able to test it under full load until I start a decent project with the new setup, which will be too late to RMA it if it's found not worthy due to the DPC spikes.

But if it's not even good for gaming, let alone music...

(PS > You have disabled ethernet/LAN, wifi, etc. via Device Manager? They are often a separate/additional cause for the dreaded spikes. Yours are terrible! See this thread: http://www.soundonsound.com/forum/showflat.php?Cat=&Number=588140&page=&view=&sb=5&o=&fpart=all&vc=1)

 

[bapu]

On my P35-DS3R Rev. 1, the problem was solved by going from F11 back to F5 (original BIOS that came with the board). F11 added a High Precision Event Timer in the system devices, so maybe that's the source of this latency issue...

Maybe someone could try disabling the High Precision Event Timer to see if it fixes the problem. I can upload F5 later if anyone would like to try it. BTW, I also had higher NB temps with F11 for some reason.

I would like to have your opinion on whether the F5 BIOS would work on a P35-DS4? I get DPC spikes about every 3 seconds right around 1000u while the others are around 250u and an occassioanl 500-600u.

Can you post this BIOS somewhere?