Klanghelm MJUC and MJUC jr. released
-
- KVRian
- 677 posts since 22 Sep, 2007 from Caracas, Venezuela
Quality comes with a price and that's CPU load in our world. HQ mode uses like two times the amount of CPU but it's not just an oversampling switch, it takes the whole modeling to a higher level. It's been set to ON here since the beta and it won't change any time soon. The difference is night and day, and it's worth it, in my opinion.
-
- KVRAF
- 14739 posts since 19 Oct, 2003 from Berlin, Germany
I can confirm the high CPU usage on my Core i7 920, which is a 1st Gen i7.
In Cubase 7.5 x64, I get the following readouts on 128 samples buffer (RME HDSPe), single instance:
MJUCjr: roughly 10-12% ASIO (depending on the signal applied)
MJUC Mk 1: roughly 18-20% ASIO (HQ on), else 10-12% ASIO (HQ off)
MJUC Mk 2: roughly 15-18% ASIO (HQ on), else 8-10% ASIO (HQ off)
MJUC Mk 3: roughly 15-16% ASIO (HQ on), else 8-10% ASIO (HQ off)
Further instances eat like 5-10% with HQ off, and 15-20% in HQ on.
Cubase 7.x's ASIO Guard v1 (128 samples buffer) can eat 20 instances of MJUC Mk3 (HQ on, IStage on, Density on, Drive on
before it starts to crap out on me (dropouts, pops and clicks).
You can influence the numbers even more if you pull back the Drive to 0 (default is 2), which somewhat turns off the saturation stage.
If you're interested, a comparison (single instance):
IKM Vintage Comp 670 (Fairchild): 4-6% ASIO usage - default (the eldest of the bunch)
SKnote GTS-39 (from mid 2011): 14-16% ASIO Guard usage, around 22% ASIO realtime (fixed 8x OS!)
Slate VBC FG-MU: 5-6% ASIO usage - default preset
TDL Kotelnikov GE: 10% ASIO usage (idle), 13% ASIO usage (in use) - default preset
Sorry, no Gates STA-Level or Abbey Road RS-124 clone on my HDD.
Maybe Tony can further optimize this with a future update.
But personally I am happy that it's working without severe issues. And if I'm after a "vintage mix"; I won't go into the 24+ channel ranges anyway.
And as Christian Budde once said to me (which adds further weight to Mercado's post):
"You can only have two out of three: High Quality (i.e. through OS), Low Latency, Low CPU usage"
In Cubase 7.5 x64, I get the following readouts on 128 samples buffer (RME HDSPe), single instance:
MJUCjr: roughly 10-12% ASIO (depending on the signal applied)
MJUC Mk 1: roughly 18-20% ASIO (HQ on), else 10-12% ASIO (HQ off)
MJUC Mk 2: roughly 15-18% ASIO (HQ on), else 8-10% ASIO (HQ off)
MJUC Mk 3: roughly 15-16% ASIO (HQ on), else 8-10% ASIO (HQ off)
Further instances eat like 5-10% with HQ off, and 15-20% in HQ on.
Cubase 7.x's ASIO Guard v1 (128 samples buffer) can eat 20 instances of MJUC Mk3 (HQ on, IStage on, Density on, Drive on
You can influence the numbers even more if you pull back the Drive to 0 (default is 2), which somewhat turns off the saturation stage.
If you're interested, a comparison (single instance):
IKM Vintage Comp 670 (Fairchild): 4-6% ASIO usage - default (the eldest of the bunch)
SKnote GTS-39 (from mid 2011): 14-16% ASIO Guard usage, around 22% ASIO realtime (fixed 8x OS!)
Slate VBC FG-MU: 5-6% ASIO usage - default preset
TDL Kotelnikov GE: 10% ASIO usage (idle), 13% ASIO usage (in use) - default preset
Sorry, no Gates STA-Level or Abbey Road RS-124 clone on my HDD.
Maybe Tony can further optimize this with a future update.
But personally I am happy that it's working without severe issues. And if I'm after a "vintage mix"; I won't go into the 24+ channel ranges anyway.
And as Christian Budde once said to me (which adds further weight to Mercado's post):
"You can only have two out of three: High Quality (i.e. through OS), Low Latency, Low CPU usage"
Last edited by Compyfox on Tue Jul 14, 2015 3:54 pm, edited 1 time in total.
-
- KVRAF
- 2065 posts since 14 Sep, 2004 from $HOME
Congrats to the release! Always impressed by the quality of your GUI, great design, excellent execution and great usability.
Not sure I really need the big version, judging from the junior i am frankly not sure that it is that so much better than all the other compressors i have
but it seems to work well on problematic (harsh and sibilant) vocals and acoustic guitars. Also with SDRR i didn't really fancy IVGI, but i loved the full product. And i like that it is simple and has only eight knobs. And of course i need to keep my Klanghelm collection komplete, sorry, complete, so I'll eventually buy it anyway
Just one thing I noticed: Especially on the fast and auto setting it seems that even though the meter doesn't react at all there is already compression happening. Also I have to dial in 1-2dB more of make up than the highest (lowest?) GR meter value to match the uncompressed loudness. This is not the case with DC8C or other compressors i have. Might that be because VU ballistics? I tried switching from slow to fast but no changes.
Here's an example, the left waveform is MJUC disabled, the right is enabled, the needle didn't move at all during playback, but clearly, there's GR going on. Well, not tragic of course, i am just curious
Not sure I really need the big version, judging from the junior i am frankly not sure that it is that so much better than all the other compressors i have
Just one thing I noticed: Especially on the fast and auto setting it seems that even though the meter doesn't react at all there is already compression happening. Also I have to dial in 1-2dB more of make up than the highest (lowest?) GR meter value to match the uncompressed loudness. This is not the case with DC8C or other compressors i have. Might that be because VU ballistics? I tried switching from slow to fast but no changes.
Here's an example, the left waveform is MJUC disabled, the right is enabled, the needle didn't move at all during playback, but clearly, there's GR going on. Well, not tragic of course, i am just curious
You do not have the required permissions to view the files attached to this post.
-
- KVRAF
- 2065 posts since 14 Sep, 2004 from $HOME
Let this be known from now on as the "Budde theorem" or the "QLC theorem" (analog to the CAP theorem)Compyfox wrote: And as Christian Budde once said to me (which adds further weight to Mercado's post):
"You can only have two out of three: High Quality (i.e. through OS), Low Latency, Low CPU usage"
-
- KVRist
- Topic Starter
- 225 posts since 14 Dec, 2008
The GR meter only shows the gain reduction caused by the actual GR-cell. But the in and out stages plus transformer have a slight compressing effect too. I guess, that's what you're experiencing.fese wrote:Congrats to the release! Always impressed by the quality of your GUI, great design, excellent execution and great usability.
Not sure I really need the big version, judging from the junior i am frankly not sure that it is that so much better than all the other compressors i havebut it seems to work well on problematic (harsh and sibilant) vocals and acoustic guitars. Also with SDRR i didn't really fancy IVGI, but i loved the full product. And i like that it is simple and has only eight knobs. And of course i need to keep my Klanghelm collection komplete, sorry, complete, so I'll eventually buy it anyway
![]()
Just one thing I noticed: Especially on the fast and auto setting it seems that even though the meter doesn't react at all there is already compression happening. Also I have to dial in 1-2dB more of make up than the highest (lowest?) GR meter value to match the uncompressed loudness. This is not the case with DC8C or other compressors i have. Might that be because VU ballistics? I tried switching from slow to fast but no changes.
Here's an example, the left waveform is MJUC disabled, the right is enabled, the needle didn't move at all during playback, but clearly, there's GR going on. the GR meter only shows the gain reduction by the actual GR cell (the actual variable-mu stages). The other parts (xformers and in-out stages) are causing some compression too. I think, that is what you're experiencing.
Tony
Well, not tragic of course, i am just curious
Last edited by tonhelm on Tue Jul 14, 2015 4:04 pm, edited 1 time in total.
-
- Banned
- 454 posts since 30 Apr, 2013
Softube CL1B, Plugin-Alliance VSC-2 and Waves CLA-2A are the "high quality" compressors that are least CPU intensive (that I know of). If MJUC uses, say, 5+ times more processing power, does it sound 5+ times better? I'm not really asking.
-
- KVRAF
- 14739 posts since 19 Oct, 2003 from Berlin, Germany
Softube CL1B - Opto Comp
Plugin Alliance Vertigo VSC-2 - VCA comp
Waves CLA-2A - Opto Comp (the new UAD Version is clearly blowing this one out of the water btw!)
MJUC is a Vari MU design.
It's like comparing Apples with Oranges in this case.
Plugin Alliance Vertigo VSC-2 - VCA comp
Waves CLA-2A - Opto Comp (the new UAD Version is clearly blowing this one out of the water btw!)
MJUC is a Vari MU design.
It's like comparing Apples with Oranges in this case.
-
thecontrolcentre thecontrolcentre https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=76240
- KVRAF
- 37262 posts since 27 Jul, 2005 from Scottish Borders
Bought. Can't wait for my license to arrive ... 
Last edited by thecontrolcentre on Tue Jul 14, 2015 4:57 pm, edited 2 times in total.
-
thecontrolcentre thecontrolcentre https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=76240
- KVRAF
- 37262 posts since 27 Jul, 2005 from Scottish Borders
-
- KVRist
- 410 posts since 9 Oct, 2013 from uk
VST3 not mentioned anywhere...
Will definitely be trying this out very soon; sounds/looks like another quality release from you Tony..!
thebutler
Will definitely be trying this out very soon; sounds/looks like another quality release from you Tony..!
thebutler
Last edited by thebutler on Tue Jul 14, 2015 5:26 pm, edited 1 time in total.
System 1 - Win11; i9 13900HK miniPC; 64Gb; Iris XE graphics; Cubase 15.0.30; Studio Pro v8.1;UR44 i/o
System 2 - Win10; i7 4790; 16Gb; GTX750Ti; Cubase v14.0.41; WLab Pro v12.0.51; StudioOne v6.6.4
System 2 - Win10; i7 4790; 16Gb; GTX750Ti; Cubase v14.0.41; WLab Pro v12.0.51; StudioOne v6.6.4
-
- KVRAF
- 1866 posts since 18 Feb, 2012
Downloaded free version and cpu usage on i7 4790 is 3-4%, thats nothing...added on each channel one of these and cpu is so low, its totally not worth mentioning. Great job! And sound is top notch...I am trying to restrain myself from buying full version, but its not easy haha. GAS lvl 99
-
- Banned
- 454 posts since 30 Apr, 2013
Not really... It's a very valid comparison of the processor load of several compressor plugins. These compressors don't have to be exactly the same in every respect in order to be comparable, you know. And doing such a comparison is especially relevant FOR ME, because I will use them (or not) on MY MACHINE. So my judgement is the only thing that matters for me. Do you get that, professor?Compyfox wrote:Softube CL1B - Opto Comp
Plugin Alliance Vertigo VSC-2 - VCA comp
Waves CLA-2A - Opto Comp (the new UAD Version is clearly blowing this one out of the water btw!)
MJUC is a Vari MU design.
It's like comparing Apples with Oranges in this case.
