Engine#3 Vs Engine#4 comparison

[Calibrator]

I've been flipping between Engine#3 and Engine#4 all day on my music server here, in both a Vista partition and Windows 7 partition, and I'm having a hard time trying to distinguish any noticeable difference(s). They all sound great --- certainly no backwards step from 0.9y-4 from what I recall from yesterday. !!

I suppose the proof of the pudding, so to speak, is, after living with 0.9y-5 for a few days, to jump back to 0.9y-4 and see if anything screams out at me as being poorer.

Now, in order to get Engine#4 running on my Juli@ card, I need to run analogue out, yet under Engine#3 I can use analogue or SPDIF to my processor, so the data paths aren't exactly the same. Even so, these old ears are still content

Anyone else have thoughts on differences, if any ?

Cheers,

Russ

[boleary]

Hey Russ, do you notice much of a difference between Engines 3 and 4 in 9y--5? Just wondering.

[AUDIODIDAKT]

Russ,

I'am Playing 1 Blue-Note album after another, (with Engine3)(have Juli@ too, as you know)
Latest version is way better, less loud.
But bass is more firm and overall very smooth.
It wasn't my interntion to test xx, but I know I ain't wrong, it does a better job.
Little surpriced by this..............were does this ends............wha !!!
So I was drawn to the speakers, without me wanting it.....such
Just let it play for some weeks will see.
But I know 1 thing for sure previous version had some bad elements (loudisch, cold, less analog)

I don't know what Peter did, but it is a workaround of W7 problems I guess.
So.........

[Calibrator]

I took a different approach today on my music selection for testing. Yesterdays sessions were more vocal orientated; midrange and percussive in nature. Today, following Roy's thoughts that the bass was improved, I first put on the XRCD24 mastering of Sheffield Labs "Drum and Track Record". Queueing up the first track and flipping back and forth between Engine#3 and #4 under Windows7, it was becoming apparent there was some slight improvements in the dynamics and control in the lower registers. Moving onto Stanley Clarke's "Lords of the low frequencies" from his "East River Drive" album, helped reinforce this notion.

As I can only use analogue out under Engine#4 with the Juli@ card, it dawned on me that I could now use quad sampling, whereas I was limited to double previously when using the digital out and feeding it to my processor.

Hmmmm ... that made a nice improvement, and I kicked myself for not thinking of trying the analogue approach in the earlier versions.

I'll stick to this combination under Windows7 for a week and then jump back to Vista for a session to see if that influences matters. At this point though, I'm pleased with the incremental increase in overall quality. Not as massive as the improvements that Arc Predictor brought to the table, but very welcome nevertheless.

Thanks again for your efforts Peter 

Cheers,

Russ

[Fidelio]

Latest version is way better, less loud.

y5 is much quieter than y4 for some reason. I don't like it, why is this so? My heart almost stopped when I switched back to y4 from y5 and some drums kicked in.

[PeterSt]

It shouldn't be ... ? (except for 1.5dB when Arc Prediction is used, as denoted in the Release Notes).

What does the Wallpaper show for volume in either case ?

[thermalbat]

Hmmm...everybody seems pretty quite on this one, so I'll throw my  two cents in.

I'm listening on a pair of AVI ADM9.1's, a Veleodyne MicroVee sub with a DSPeaker anti-mode 8033 to tame the bass a bit(and it does a very good job), and a Prodigy HD2 audio card with the Toslink feeding the ADM9.1's digital input. This resides on my desk(sub is under desk), so my listening is definitely near-field.

 With that out of the way, I've been enjoying  9y-4 with arc prediction and 2x upsampling. I know, the Adm9.1's already upsample-but it sounded very good.

 Now, last night I tried 9y-5 with Engine 4 and arc prediction and 2x upsampling. Guess what, it just didn't sound right. Actually bothered my ears a bit.

 This morning I tried Engine 4 without arc prediction and 2x upsampling. WOW! Sounds REALLY good!

 Right now I'm listening to Chick Corea Akoustic Band and it has never sounded so good! My ears are smiling.

  Excellent job Peter!

     

               Rich W

 whoops..almost forgot-using Windows 7 RC.

[PeterSt]

Hi Rich - Thank you. And, interesting !

Everybody is quiet because nobody gets it really running I guess. And those who expressed so far have to deal with apples and oranges.

About your observations ... it won't be so that Engine#3 now also sounds better without Arc Prediction ? This *is* possible, because it has changed ...
Thanks,
Peter

[boleary]

Cause its early morning here am listening at a low volume; however,engine 3 with arc prediction is definitely softer or maybe not quite as loud as engine 3 with arc prediction turned off. Need more volume to know which is preferred.  9y-5

[boleary]

Not just because I can't get kernel streaming to work, the topic of this thread should be 9y-5 arc prediction versus everything else. Don't know about quad but 2x is the most natural sound I've heard in any kind of playback format. sh*t, I am bottoming out the volume.....Thanks Peter.

[boleary]

One more post on this and i'll stop. 9y-4 with 2x arc prediction is very similar to 9-y5 without arc prediction: they have a digital feel with what feels like more of an emphasis on the top end. 9y-5 is much more natural sounding, the music is smoother and sweeter if not as enhanced as the others; it certainly is not as loud either. It seems balanced so that the the high and mid ranges get lost in each other until they are not. Very natural sound. I love it.

Now all of this comes from one whose delta sigma shouldn't sound better with arc prediction. But I'm just sayin.......

[PeterSt]

[GerardA]

Just returned from Vista to 7.
Here I don't get #4 to work (Phase 24FW-problem) so only engine 3 double arc.
Only one apple, no orange, but this apple tastes very good!
I'm gonna finish it with a glass of Martini....

I felt like it is better on 7 now then on Vista (and 7 before) but all this strange kinds of fruit make A-Bing very difficult.
Hope to get #4 working soon,please! Because Peter's description was right to the point for me, more emotion, like analog.
But for now I don't miss it.

[PeterSt]

You guys can't be more vague, right ? fruit cakes ...

Just finished the last glass of whiskey here. Hopefully I'll understand more of it tomorrow.
The only thing I know is that L.A. Woman (which I must have played a 200 times since I was a kid) sounded the best I ever heard. And that was before the whiskey. IIRC my OS is something with a 7 in it.

Ok, I'll try to be more serious after some sleep.

Cars hiss by my windowS ...

[thermalbat]

“Perfection is attained by slow degrees; it requires the hand of time”  -   Voltaire

"and a little whiskey"  -  RW

[PeterSt]

Haha ... now look at your email address for the forum ...

[boleary]

Good God I hate fucking fruitcake! Rotting fruit, yuck.

But, thanks, the music here is best yet. (I 'm also experiencing the huge difference V-Caps have made in my tube amp.......blissmush not fruitcake. )
 

[Fidelio]

Just been listening to Engine #4 today, for me this is the perfect gift. I always enjoyed the clarity, sense of space and "realism" of #3, but it could get a bit analytical. However, I got used to this.

Engine #4 is pure old-school flat earth sounding. Great PRAT (better than #3), really gets the foot tapping, but not as crystal as #3. #4 is also more bassy than #3, no question, but the bass isn'y muddy. To be able to switch between these two would be fantastic, I get a round earth and a flat earth source in one

Edit: Been listening some more. Music sounds fucking good now.

[Telstar]

Quick comparison with y5-01. I think I prefer engine #4, more "lively".
Settings Q1=4, 4x arc prediction, attended.

[GerardA]

So, got the phase 24FW working in 7.
Comparing it to the Gigaworks-Dac, both engine 3, the latter wins on extra information/space etc.
But comparing #3 and #4 on the phase makes the 4 a clear winner in the sense that going back and forth between the two you want to stay with the 4 and don't go back to 3 just to enjoy the music inmensely!
One song with bass, cello and guitar I can follow the line of the Cello, making a very 'emotional' performance while with engine 3 you still here all the parts but it's a bit more messy making it impossible to hear a fluid line and connect to that.

So we need a Kernel streaming driver for SPDIF!!
Boleary get your hiface working man!!
And Peter thanks for the good work!!!

Gerard.

[PeterSt]

Quick comparison with y5-01. I think I prefer engine #4, more "lively".
Settings Q1=4, 4x arc prediction, attended.

Hey Telstar. You might try Q1 at 30 or so !
But it isn't operable ...

I am still thinking on how to let it work similarly as with Engine#3, and maybe I can't. Anyway, it should be disabled currently. Sorry.
(Q2/3/4/5 still work and they work the same as before)

[Calibrator]

And those who expressed so far have to deal with apples and oranges.

Well here we are at another day, with a new set of alpha waves coursing through the head, and a few more pieces of fruit to mix in with the apples and oranges !

I'm having a hard time coming to any definite conclusion now as to what sounds better in my setup. While remaining within the Windows7 environment, I've been flipping between Engine#3 (both analogue and digital output from the Juli@ card) and Engine#4 (restricted to analogue only), and trying different combinations of None/Double/Quad sampling with/without Arc Predictor, together with Q2/Q3 set to 30/30 or 0/0.

When using Engine#4 I'm tending towards settings of Q2/Q3 at 0/0 and having NO sampling engaged.

When using Engine#3 I'm tending to remain with the Double/AP together with Q2/Q3 at 30/30, and may have a subconscious factor at play here as that is what I've been using for some months now under Vista.

Now, this inability to get a definitive preference may be due to the processor I use having a global influence, limiting the effects on anything connected upstream. Certainly connecting either analogue or digital would bring into being different internal routes within the processor.

Regardless of my ability to come to a conclusion, it all sounds great !

I'll play the devils advocate here for a moment and pose the question whether or not subconscious thoughts are influencing our preferences for one engine over another, or that one set of parameters is better than the other. One sure way to come to an answer to that would be to have someone else make the parameter changes without your knowledge of what has been changed ( if anything ) and evaluate each set of changes on their merits. In essence, doing some blind testing.

Cheers all,

Russ

[PeterSt]

Hehe, nice Russ. Although you seem to work with a kind of old technology on the driver's side. But ok, you wanted to do it all over again anyway, I guess.

Well, I'm sure you have seen it by now : So, a Juli@ doesn't dig KS, right ? and I can only say I'm sorry I didn't find the time earlier to try it out (getting the idea last Thursday).

[PeterSt]

Just been listening to Engine #4 today, for me this is the perfect gift. I always enjoyed the clarity, sense of space and "realism" of #3, but it could get a bit analytical. However, I got used to this.

Engine #4 is pure old-school flat earth sounding. Great PRAT (better than #3), really gets the foot tapping, but not as crystal as #3. #4 is also more bassy than #3, no question, but the bass isn'y muddy. To be able to switch between these two would be fantastic, I get a round earth and a flat earth source in one

Edit: Been listening some more. Music sounds fucking good now.

Well, that seems to be a description I would make 100% the same. Each aspect of it.
Must be a victory for both of us, haha.

To be able to switch between these two would be fantastic

... but I don't get what you mean by this. Or wasn't it a question ?

Peter

[Fidelio]

To be able to switch between these two would be fantastic

... but I don't get what you mean by this. Or wasn't it a question ?

Aha, I can switch, never mind

[Gerard]

Peter,

Well what can i say..... Sounds real great  ... love the softness that comes with it. But i really needed some time to get use too it was not love at first sight i have to admit  

The naming stuff is working without any problems ....  (even after a reboot hahaha)

  

[bernharddiner]

Placebo or not and I have to laugh about myself, but playing the #4 in sequences of 6 sec let me think: this is the best sound I ever got from my system 

Very analog, extremly clear, with a lot of sense for the space, sounds really live 
I hope this does not change once I can play a whole song 

Cannot await that you fix it!


Will post in the support thread my log file

Thank you

[Telstar]

I decided to give engine #4 another try. This time without upsampling.
And I noticed that it does sound more involving!

My thought is that with poor sigma-delta upsampling DACS engine4 sounds best without any upsampling.
I have much more to test on the real system.

[pedal]

Just installed 0.9y-5-02. The new "KS" alternatives poped up in the Output Device window, as expected. Engine #4 working just fine. When switching between Output Devices (with/without KS) XX automatically switches between Engine #3 and #4. Thats nice!

Listening: (Jazz at the Pawnshop, from original CD issue)
Engine #3: Slightly more "distinct" in the treble region. Sometimes I think the microdynamics definition is better, but sometimes it sound just "thinner". Engine #4: Sounds "fuller". Vibes sounds more rich, more "kick" on the kickdrum. Drumsticks are more "woody".

Anyway, quite small diferences in my system. (You can get bigger diferences if swapping interconnects). Hard to choose an overall favourite for all music tracks. Maybe long term listening will decide. When I started writing this post I prefered #3, but right now I think I'll leave it on #4, he-he.

[AUDIODIDAKT]

Up and running, #Engine4 with Juli@ on W7 x64,
just minutes ago, it deffinitly sounds different,

Latency on 2048 is a must also, otherwise I get gliching in sound.
Btw latency on 2048 is also better on 3#Engine.

Tnx Peter for making the Juli@ update, give me some days of testing here.........

 

[PeterSt]

Hey Roy, good.

But what happened that you got it working ? or was it just a matter of enough  ?
Or the other way around of course hehe

[Telstar]

In my current situation, I think I now prefer 5-02 engine #3 without upsampling. Q1=4 is a fixed thing for me.
HF comparing engine #3 and #4 sound less distorted with #3. I know with a NOS DAC things will be different, but i suspect the cpu usage has something to do. For now i'll stick this way.

Funny thing is, the Quartet drivers do auto sample rate in W7 BUT the VST host doesnt, so it's a moot point
Anyway, i'm sure ESI will fix the drivers soon.

[AUDIODIDAKT]

#4Engine is definitely more distorted,
I already said this to Calibrator today.
But if you want massive try Enigma 4 - The Screen Behind the Mirror
Then I kind of liked it.
But I heard some more albums today, sometimes its just to distorded it gets on my nerves.
Especially when there are lots of samples or instruments at the same time, it becomes to difficult
Then really I prefer the latest #3Engine.

Will try some more settings, like upsampling on/off, latency, different drivers, 0,0,0,0 vs 30,30,0,0

 

[PeterSt]

Maybe don't get fooled by that higher buffer setting in the driver. I imagine not to hear a real difference with the lower setting, and remember, it was just a "strategy".

I am quite sure nothing disturbs here, and I like it very much.

But either way, keep it coming. The more the better, and negative doesn't harm. Something to work on on this side.

[PeterSt]

And Roy, would you care to describe "distortion" in this case ?
(just Roy, not Telstar please )

[AUDIODIDAKT]

I do understand why you like #4Engine, I like it (that) too, don't understand me wrong.
I will try to pinpoint this more exactly, with some tracks somewhere tomorrow.

It differs with music styles, I don't get it myself 100%. (but what to expect after just 1 day)
When parts become more complex, it can get into the direction of distortion (or preceived as)(have to compare)
This has nothing to do with wrong setup or something like that.
All sounds very nice, both engines for sure are better than the previous version of xx (0.9y-4)
This was just my first impression, but it does sounds very interesting.

Good to hear, it does great for you. ( I can only imagine)

I'am very happy with you bringing us back on the right track with the 0.9y-5 version.
4 was to clean, to analitycal.

For now more enjoyment ! more lively on both engines anyway

[PeterSt]

Ok, I'll wait for that. In the mean time, do you have an example of such a more complex piece ?

[AUDIODIDAKT]

Try Al Di Meola - Casino\02 - Chasin' the Voodoo for eg.

I think the guitar is cleaner on #3Engine but the percussions are better on #4Engine.
I mean some parts are better on #3 than #4 and v. v.

[AUDIODIDAKT]

Earl Klugh - Naked Guitar (acoustic guitar) for eg. sounds better with #3Engine, IMO
 

#3Engine gives it just that little warmth to the instrument while #4Engine makes it more .........?? distorded

[PeterSt]

Roy, I think you are right.
I'll try to explain after some .

[PeterSt]

Ok, Al Di Meola ... difficult to analyse with some sense ...

I played the album from track 1 (Engine#4), and after track 2 I didn't have the idea of more - or too complex at all. So, it came to me that I didn't know what you exactly were talking about, btw explicitly without the idea "it will be your system". Then I thought maybe you were using the remaster of it, and with the placebo "that must be it" I tried that for track 2. I kind of made the mistake to play it less loud for safety (remasters usually are way more loud), but the track came the same to me. Next I used the Volume Normalization which would show me the difference between the two versions, but apart from the track lengths being slightly different, they are the same. So, here is another hoax. Anyway, both are 9 dB softer than my "reference", and because of using the Normalized Volume they now played 9dB louder. Right ...

A useless story so far, although there was one thing : I just didn't like it much.
Engine#4 generally comes to me as more bright than Engine#3, and here the more bright disturbed. There's too much of cymbals, and they aren't right at the same time. It wouldn't fit in my brains.

After hearing the track quite some times now, I thought to hop over to Engine#3. Now a strange thing happened, or at least quite unexpected : NOW it showed as complex and "crowded" music. But, now it was interesting at the same time, and now I liked it very much ...

Next I "put up" Scenario (also from him) which seemed even more suitable for this kind of judgement, and is somewhat better to my likings to begin with. Still Engine#3 and nothing seemed to be wrong with it. Then, at the second track I "heard" my ever so often made mistake : forgot to set the latency to what it should be for Engine#3, which is 48 while it still was at 2048. Changed it, continued the track where it stopped, but immedietaly I noticed the life has gone out of it. Hmm ...
After one minute I set the latency back to 2048 and enjoyed the remainder with great pleasure.

In the end, Roy, this has all been noticed by you, and to what extend this depends on the type of music, I don't know (but I never like such a dependency so I'd rather like to think this "setting" is suitable for all. And to keep in mind : this is Windows7.

Right now I can't know what the higher latency will do to Vista, and all I have in mind is that with Vista I wasn't disturbed at all (Engine#3), while with W7 it was very apparent it needs the higher latency. What to make of the "fact" that for Engine#4 the latency doesn't do much to my ears, I don't know (yet).


Allright. Apart from these more technical issues, I think it is good to recognize what I actually perceived from Engine#4 and the "not complex at all" while Engine#3 showed the opposite. I think this is fairly easy to see : Engine#3 shows much more separation, and while Engine#4 exhibits a kind of "all together", Engine#3 shows dozens of rototoms, cymbals and everyting, and it just shows crowded where a crowd should be. It shows the crowd. Engine#4 shows the (coherent ?) mass.
It may be important that both these albums are not much compressed while using the full dynamic range at the same time. This means that high transient instruments (as used all over here) get the room to exhibit the transients, which with normal "audiophile" recordings are used sparsely (like one hit on a cymbal, which then shows nicely natural).
Engine#4 smears more and while smearing shouldn't create "emotion", maybe it does to some extend while first something else it not completely right. Changing the latency from 48 to 2048 seems to do similar to Engine#3, although I couldn't recognize something from "no emotion" to "emotion" because I just don't like Engine#3 (with the latency of 48 as it now turns out) on Windows7 to begin with. With 2048 I do (or at least after this short time I think I do), and it might just be that compensation needed for W7 using some crazy CPU, as explained by me in the "Windows7 maybe not" topic (and notice that a latency of 48 exhibits even way more cpu usage than what I encounter with Engine#4 by W7 itself), easiliy shown by TaskManager and which by itself is logic.


So, what's next ?
Maybe it would be good that people not having problems with Engine#3 on W7 recognize that they use a high latency (or can't set it, which implies the same), and that those who do have problems recognize this solves it.

Lastly, I must honestly say that I loved the "good old sound" from Engine#3 after so many weeks not listening to it. It came to me that not preferring this over Engine#4 is a waste of time. But as honest, these albums are not the most to get involved in, and they are "technical" albums again (suitable for Engine#3 to begin with). But, since I heard how the life got out of Engine#3 with this low latency, it well can be that the high latency is just the solution to everything (and not Engine#4 ! -> which I will continue for XP anyway).

So far,
Peter

[boleary]

I believe there was once a wise man who said:

[boleary]

Knowing that it's really difficult to say one sound is preferable to another given that everyone's system is so different, here's another observation on 9y-5 engine #3 with AP: Last week I went to a house concert at a local McMansion (one of the kinds of huge houses that sprung up all over the US earlier this decade, many of which now sit empty). The acoustics in the livingroom were quite good; thirty seven people in attendence. The artist used very mellow amplification cause it just didn't need to be blasting in a room like that. The sound was very natural and sweet; it was almost identical (slightly better) to 9y-5, engine #3 with AP. One way of describing the similarity is what Dan Hageland described earlier in this thread (though he said he didn't like the sound) regarding the vocals sounding diffuse. To me, on my system, this diffuse sound provides a kind of softness that is just VERY natural sounding, much better than 9y-04.

The fact that the artist at the concert was my step-daughter has nothing to do with my overall judgement that it was a great show, though it was only her with a guitar and no band. 

Oh, she will be touring in Europe beginning February 17th. Check out: http://www.carsieblanton.com/

[PeterSt]

[AUDIODIDAKT]

Hi, (feel I little like Youp van 't Hek with his Buckler beer, only he enjoyed it !) hehe

Thnx Peter for your great explanation and exploration, somewhat the same I had.
Your communicative skills good as ever, you saved me.
To be honest I've been quite lost for a while (kinda lost interest, not in the subject at all, but in the general SQ)
But this discovery only came to me with listening to the 0.9y-5 series.
Now I know what I've missed, because "its" back.

Shulman --->> #3Engine3 DAP (Lat. 2048) --->> Ears--->>Brains--->> WOW WOW Stunning.


PS: Gerard I think NOS1 will do very good right now, make me jealous, haha (Do grab that change in this state, must be best ever !)

[earflappin]

I can report that the SQ on the latest 0.9y-5 on Win7 64 into my Berkeley Alpha DAC is a nice improvement over 0.9y-4.  As to the engines I find a slight preference for engine 4.  Engine 3 to my ears is a tad more smooth, but less live sounding. 

[AUDIODIDAKT]

If latency does nothing but changing SQ
OK,

When you want a TRUE compare of for eg. upsampling vs no upsampling you should adjust latency accordingly.
Why ?

When you double the samplerate you half the latency as your using the buffer up twice as fast

So should there be bigger buffers when using QAP or oct-AP,
to keep smooth sound ??
 

So, QAP can only be better then DAP (@ 2048) if Latency goes above 2048 to lets say 4096
or WOW how low can we go right now, without SQ changes.
if TRUE compare upsampling: DAP set @1024 vs QAP set @2048
Right ?

I can imagine this is not a nice thing for the volume control and overall speed (it will take longer again)

[PeterSt]

Yes, true. But so many more things are involved that I wouldn't bother about it. Theoretically true though.

The volume change taking longer will hardly be noticeable because it takes "relatively way long" already. Notice that if 48 samples incurs for 1ms, 2048 will be 42ms. You won't notice that (you will when playing a keyboard though).