• From: Houston, TX
  • Registered: 2007-05-11
  • Posts: 1,191

Topic: DAW-Bench website suggests 3.24 is less efficient than 3.085?

So, the fairly well-respected and knowledgeable www.dawbench.com website has updated its low-latency interface comparisons, and posted an interesting comment over at Cubendo.com about HDSP(e) 3.24 vs 3.085 performance:

A few side notes - the baseline reference is the RME HDSPe AIO card using the 3.08.5 driver. You will notice that even the new version 3.24 driver is also tested and gauged against that baseline result as the  performance has dropped. If a later RME HDSPe driver levels or betters the 3.08.5 driver, then it will become the new baseline, unless of course another card has better overall performance and becomes the new baseline.

http://www.cubendo.com/showthread.php?1 … to=newpost
http://www.dawbench.com/images/dawbench … -06-11.jpg

Seems like 3.085 can run 593 RXC's, while 3.24 only ran 561 RXC's before pops and clicks began (a 6%-7% loss).  The other 2 tests also showed worse performance with the 3.24 driver. 

What does RME say to this?  I thought I recall RME stating that 3.24 was supposedly more efficient if anything, but these non-partial tests (if anything they are biased TOWARDS RME in general wink ) clearly show otherwise.  Granted this is a small decrease, but I'd figure a driver should become more efficient as time goes on (as 2.94 to 3.085 clearly showed with more efficiency AND lower latencies!), but this seems like a small step backwards for RME's well-known driver development practices.  I believe at least a few have posted that projects with 3.085 were no longer "clean" with the same settings using 3.24 (a buffer increase was needed).

Also - any news on an update for the reported multiclient issues in 3.24?  Any chance that fix will get the new "updated" driver back in-line with 3.085's performance (or hopefully surpass it)?


Thanks for clearing this up! :-)

cool

MADIface-XT+ARC / 3x HDSP MADI / ADI648
2x SSL Alphalink MADI AX
2x Multiface / 2x Digiface /2x ADI8

2 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 24,913

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

Interesting. Not sure if they want to test again, but if so they could uncheck the 'Optimize Multi-client Mixing' option in the driver's About dialog ;-)

Regarding multiclient - which issues exactly?

Regards
Matthias Carstens
RME

MC's Website

3 Reply by Randyman... (edited by Randyman... 2011-06-29 19:31:41)

  • From: Houston, TX
  • Registered: 2007-05-11
  • Posts: 1,191

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

So it is your supposition that the DAW-Bench testers simply didn't try the driver with this setting un-checked?  And you are implying that 3.24 should indeed be more efficient (or on par with) 3.085 with this setting un-checked?

On the Multiclient thing - I thought I saw a few threads where people were having Multiclient issues after updating, and I could have sworn I saw you or Knut mention that an update was on the way?  Maybe it was the USB/FW drivers, but I thought I recalled HDSP in particular...

Anyways, I'll try to see if I can get any clarification on the performance benchmarks (what setting they used) and get back to the board later in the week...


Enjoy!

cool

MADIface-XT+ARC / 3x HDSP MADI / ADI648
2x SSL Alphalink MADI AX
2x Multiface / 2x Digiface /2x ADI8

4 Reply by TAFKAT (edited by TAFKAT 2011-06-30 01:49:00)

  • From: Melbourne : Australia
  • Registered: 2007-06-17
  • Posts: 46

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

MC wrote:

Interesting. Not sure if they want to test again, but if so they could uncheck the 'Optimize Multi-client Mixing' option in the driver's About dialog ;-)

Hey MC,

I actually left the driver install as default, so I'll test again with the option unticked and report back... :-)

Just a quick question re firmware , I have the latest respective firmware for each driver build which I swap between when needed, is there any known performance issues if I leave say the version 11 firmware ( 3.085) on the AIO and use the 3.24 driver instead of updating to the latest Version 12.

I tested 3.24 with both and got identical results so I'd be interested on your take.

Peace

Vin Curgiliano
AAVIM Technology
DAWbench.com

TAFKAT's Website

5 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 24,913

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

Version 12 adresses only TCO functionality. Nothing that has influence on audio data transfer has changed, so performance should be the same.

Regards
Matthias Carstens
RME

MC's Website

  • From: Melbourne : Australia
  • Registered: 2007-06-17
  • Posts: 46

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

Thanks M.C,

Saves me swapping firmware when switching between drivers.

I'll retest and report back shortly

Vin Curgiliano
AAVIM Technology
DAWbench.com

TAFKAT's Website

  • From: Houston, TX
  • Registered: 2007-05-11
  • Posts: 1,191

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

...insert eating popcorn smiley here...

:-)

MADIface-XT+ARC / 3x HDSP MADI / ADI648
2x SSL Alphalink MADI AX
2x Multiface / 2x Digiface /2x ADI8
  • From: Melbourne : Australia
  • Registered: 2007-06-17
  • Posts: 46

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

Quick heads up re the 3.24 driver,

M.C was right on the money re the Multi-Client option, with it disabled the performance is identical to the previous 3.085 driver.

I'll update the tables and make a note.

Peace

Vin Curgiliano
AAVIM Technology
DAWbench.com

TAFKAT's Website

  • From: Houston, TX
  • Registered: 2007-05-11
  • Posts: 1,191

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

Will kill this setting on my dedicated DAW then.  A blurb mentioning this somewhere might be a good idea IMO since the default setting (Optimize Multiclient Mixing = Enabled) will result in worse performance comapred to v3.085...

Good stuff to know - Especially for those of us who tend to teeter "on the edge" as-is :-)

cool

MADIface-XT+ARC / 3x HDSP MADI / ADI648
2x SSL Alphalink MADI AX
2x Multiface / 2x Digiface /2x ADI8
  • Registered: 2008-01-01
  • Posts: 3,025

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

:-° "Living on the edge..." :-°

  • From: Melbourne : Australia
  • Registered: 2007-06-17
  • Posts: 46

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

Randyman... wrote:

Good stuff to know - Especially for those of us who tend to teeter "on the edge" as-is :-)

It will give some extra headroom, so for those not needing the Multiclient capability its better to have it off.

Its great to have the option, and also refreshing that the actual devs know where the bodies are buried within their own drivers.

That reminds me, I need to go back and retest the latest FW and USB drivers with the setting disabled as well.

Vin Curgiliano
AAVIM Technology
DAWbench.com

TAFKAT's Website

12 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 24,913

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

Sorry Vin, that option is not available in those. Not yet. Will come after the summer holidays.

Regards
Matthias Carstens
RME

MC's Website

  • Registered: 2011-06-02
  • Posts: 3

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

Thanks for posting this, I am getting much better sound with "...Multi-client..." unchecked. I am currently working with some old piano performances and the difference in fidelity is clear. Who would have thunk it? Ahh, the many mysteries of digital audio...

14 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 24,913

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

There is no difference in 'fidelity'. Your system isn't working correctly as you posted in a different thread. So are we talking about 'fidelity' or real errors like scratching sound, pops, clicks or distortion? Did your TV problems go away or are still the same?

Regards
Matthias Carstens
RME

MC's Website

15 Reply by TAFKAT (edited by TAFKAT 2011-07-09 00:38:48)

  • From: Melbourne : Australia
  • Registered: 2007-06-17
  • Posts: 46

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

MC wrote:

Sorry Vin, that option is not available in those. Not yet. Will come after the summer holidays.

Hey MC,

You just beat me to the punch.. :-)

No problem Mate , I'll retest when the new drivers hit and post the results..

Peace

Vin Curgiliano
AAVIM Technology
DAWbench.com

TAFKAT's Website

  • Registered: 2011-07-01
  • Posts: 77

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

Beware switching off "optimize" when using hardware loopback and multiclient, though. I got strange feedback delays and searched hours and hours why, only to accidentaly switched optimize on and every problem dissapeared...

Can rme explain what it does exactly, how it works technically?

17 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 24,913

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

It's an erase routine for the buffers, which prevents clicks/noise that occurs on start/stop of a multiclient usage. So if you don't use multiclient you don't need it. If you don't care about an up to 0.5 s noise when starting/stopping in mc mode you don't need it. If you just want the system to work without thinking what to do you will want it.

Unfortunately such a routine can not be switched off in-between, so has to run all the time (although it would be needed only for the beginning and end). It causes a small CPU load, which - these days - either nobody notices, or doesn't care. Only in an extreme situation like the DAW bench this load becomes visible.

We did not disclose this just to show the users that the effect is indeed not noticable and nothing to worry about. 99.9 % of the users will never use their DAW to the absolute max. They use this new driver and did not notice any load or performance problem.

Note that the load increases with the number of channels (more to erase). Using 2 MADI cards or up the effect becomes visible in the Task Manager, especially if you have multi-core CPUs. This also explains why we did not add the switch (yet) to the limited FW and USB drivers. It's not that important there.

Regards
Matthias Carstens
RME

MC's Website

  • Registered: 2011-07-01
  • Posts: 77

Re: DAW-Bench website suggests 3.24 is less efficient than 3.085?

Thanks for the insights! smile this explains the repeating/accumulation effect in my situation. I checked hours and hours, cables, software, routing... *lol*