XXHighEnd Model 0.9z-4-1

[Flecko]

Hmm... The right things for me, turned out to be the last things that I found. I think I found every version since z2 better than z2 at their release and in the end I always returned to z2. So my statistic with first impressions is not so good unfortunatly.

[PeterSt]

All, please put problems in their own topic, ok ? Thanks !

Eric,

However, playing  SC;SFS=10, I am getting a click every 4 seconds or so.

And this wasn't with 0.9z-4-0 ?

Thanks,
Peter

[Eric]

Peter, for sure it was 0.9z-4-1.
After reboot the problem was gone, so I cannot recreate it anymore.

BTW I did read the Release Notes (I always do). It says "When you're really ready to play maybe reboot for one time". So in my case the maybe in fact turned out to be needed. I wonder why the reboot is needed ot what is does.

Cheers,
Eric.

[PeterSt]

Haha Eric, that is not what I meant; I was referring to "and you didn't encounter those clicks in 0.9z-4-0 ?".
But if it's solved ... never mind.

But ...

There is nothing I can think of that may need a reboot to avoid clicks there. So for sure I wasn't referring to that (and reading release notes). This was just about the possible SFS size.
And to be complete : what I actually referred to was to quite explicitly not using a low SFS. Also not 100. But just the largest size possible in your system. So I just didn't understand (or wasn't waiting for) remarks about anything else than advised.
That you, or anyone - in the mean time, try other settings just the same is very OK of course, but then I expect results from that largest size at least. Not "12 sounds better than 100". If you'd tell me "12 sounds better than 350" (or whatever your maximum is), *then* I know more. Now I know nothing. So ... I just claim that the largest size is the best, and if that is not the case I really like to know !

That's all.
Peter

[PeterSt]

Dang Eric, you're right. I just saw it (by accident) on the analyser. So now there's ticks in between track parts again (also with your SFS of 100, but you won't notice it, unless by coincidence).
It looks like we can't have it all ?

Oh well ...

[Eric]

Peter,
I have determined the Max SFS for the 3 different memory options in my setup.
Other parameter settings are: CR=0.5ms;Wallpaper=OFF;KeepLAN=ON.

m:MaxSFS=509
mc:MaxSFS=68
sc:MaxSFS=12

So far I did not hear abnomalies, just superb high quality music reproduction.
Michaela defines the sound as "like velvet", which I interpret as very positive.

Next step would be lowering the SFS and compare.
After that I like to determine the effect of changing Other parameter settings.

Cheers,
Eric.

[PeterSt]

Thank you Eric.

Stupid "question" :
Can't you get yourself a total of 8GB of memory ? With a not preferred trick I could "get you" some SFS of 170 SC, but getting more memory in your system really would be a better solution. And as I said in the release notes : 12GB would be the max utilizeable at this moment (actually SFS = 500 would be), so maybe make it 12, or 16 if 12 is not possible.
Or just 8, which seems enough to me. Well, at this moment.


PS: But something seems wrong;
Your sig is old because you already have more than 4GB, *or* your SFS of 500 was used with an output of 16 bits or something. Things don't combine. Especially not because your SC-SFS can be 12 only. What's wrong ?

[Eric]

Peter,
mobo is Gigabyte 965P-DS4 having 4 memory slots occupied by 4 x 1GB DDR2 (Kingston).
Any suggestions?

I use OctoFX-PeakExtnd-ArcPredict 352.8 playing a simple WAV track with the parameter settings as above. Volume=-6. 
#4 Engine; Adaptive 1024; No Appointment-Noting-RealTime; Q1=1; Q2=Q3=Q4=Q5=0.
That's it. The rest is a mystery (for me).

My sig is old because I DID remove RAMDisk completely. I will update it today.

Cheers,
Eric.

[PeterSt]

Haha, my sig is old too.

But I think I know what is happening ... (not good !)

With Mixed and the way it all works now, you will be using Virtual memory (in the paging file).
Never thoght of that ...

With Mixed Contiguous and Straight Contiguous that can't happen. But with Mixed it can, and this wasn't the case before.
At this moment I'm not sure how to tackle that (sense it, etc.), but it's REALLY not good ...

I must tell you (or all) honestly, that I tested it all, but as it turns out now, wrongly;
My W7 64 bit just recognizes the 8GB of memory, and there it goes 100% okay because of further "constructions";
My 32 bit Vista sees 4GB of memory, but I tweaked some stuff there. There too it goes 100% ok.
So, I should have tested without tweaks, and *then* I would have seen it was not okay at all.

Summarized, with 4GB of memory, and 500SFS you *will* be using 4GB - but maybe less when your tracks aren't that long.
Since you only have 4GB of memory, where's the OS ? well, it's just in memory, but your track will definitely be in the paging file.
This creates the exact oppositie of the intentions ...

Must edit some release notes ...

Thanks for helping out.
Peter

[Flecko]

First thing is, that since the asio driver I found out, that scheme 2 sounds now incredible good. The asio driver increased the dynamics and control and with scheme 2 it gets even better. Never before it sounded that good. I played around and think z2 is still better. It has more space between the instruments and every instrument has a teriffic dynamic and clearness. A single sound of a guitar string is an acoustical event. z4.1 MC is very close to z2 but in the end it doesn't gets there. The sound of z2 is in a way better than z4.1, like scheme 2 is better then scheme 3. So if you like "air" between the instruments and a flat sound, use scheme 3 and if you like "space" between the instruments and a dynamic sound use scheme 2.
A question btw: I optimised windows for background services, is that recomended? Using this setup and scheme 2, the computer gets slow like a snail. With scheme 3 the computer is a lot faster.

[PeterSt]

Thank you Adrian.

No, the "Background" setting is plain wrong, and you won't be able to reset it to default by normal means.
It can be reset though through the Registry. Tell me you want that and I'll tell you how (must dig it up).

[Eric]

Peter,
still experimenting, I disabled Paging (by settting the size of Virtual memory to =0) just to see what happens.
To my surprise .......I got the following results.

m:MaxSFS=511   (was 509)
mc:MaxSFS=107 ( was 68)
sc:MaxSFS=12    (was 12)

I am lost, but you will probable be able to understand and explain.
Cheers,
Eric.

[Flecko]

Thanks Peter but I have made a system copy, so I can restore my windows. I realy liked z4.1 to be better, so sorry that I can not confirm my yesterdays opinion. But today I could hear a lot louder because it was not so late. Plus I was relaxed and not so exited by the new release. I have this "background" setting since today, maybe I have to check again.

[Flecko]

After having a new windows again I rechecked but found it like before. The main reference with which I check is "Friday Night in San Francisco". It is nost the best record but I am sure most people have it. I listen to how the guitar strings sound, especially when there are single notes by one single guitar. I know the sound very well and that sound I know is reproduced best by z2 with scheme 2.

[PeterSt]

Peter,
still experimenting, I disabled Paging (by settting the size of Virtual memory to =0) just to see what happens.
To my surprise .......I got the following results.

m:MaxSFS=511   (was 509)
mc:MaxSFS=107 ( was 68)
sc:MaxSFS=12    (was 12)

I am lost, but you will probable be able to understand and explain.
Cheers,
Eric.

Maybe (to myself), maybe not ...

The whole problem is, *all* memory is virtual (read : not real). So that's where it starts.
Next, paging can't be shut off. You think you do, but you really can't. Not till some degree.
Then, the OS wants the whole of your memory to be in the paging file. Well, in the normal situation (you not shutting it off). This has reasons (never mind those for now).
Furthermore, you'll be forcing things with unknown results (well, for me at this moment) by actually allocating memory which physically is there, while the virtual "mapping" will always allow it. This will be your "M" situation.
What is more difficult to test (or lengthy) is that the "M" situation will only use the around second half of the original set SFS when the first half has been used. This means that you need to "run out" this first half just by consuming as many minutes of track-play which SFS=511 implies, and by heart this will be something like over 6 minutes for your 32/352.8 situation. But then it needs another 6 minutes to be able to load, meaning that your track must be 12 minutes long at least (but take 15 to be sure). At half of this track time things may start to go really wrong (crash, memory exception (error). Btw, the cpu graph will show you when track parts are loaded (long 100% spike for the Core concerned). BUT, if the memory is still available in the paging file, it will just keep in running ...

The "MC" situation is the one I don't get. However, it can be something I just don't think about, because just similar tricks are being used to get as far in the first place (where before this clearly wasn't possible by far).

Your SC situation only shows / prooves that you will have 4GB of memory only (and not more by accident without you knowing). However, it also seems to show that you "mangled" standard OS settings, because it's really too low (now). So, it's my guess that you may have turned off DEP (Data Execution Prevention) which really is related. It should be ON. What this can do to your reported now larger MC possibility is beyond me at this time, and will need extensive testing and research (knowing already that MS herself is far from really clear about this, which by itself maybe because they don't know themselves (bugs).

Remember, I can get an SC of 170 myself, and this is Vista, which really is far more difficult to get than for W7. This too depends on a dirty trick on knowing how the OS "behaves" (and what you won't find anywhere, not even here ). But it may be in a next version, *if* important at all for the few who really need it, because what it really needs is just more memory. Done (cheaper than my attempts to not mangle your OS with negative results for normal useage).


All assumed that you rebooted before attempting a next memory type try (and don't hop over to e.g. MC right after trying M).

In any case, I'm glad that these things happen with explicit coorporation on finding out what is going on, because in the end I can only learn from it. If only -and this is important- people wouldn't just start copying suggestions from others, which really destroys and makes all unreliable - hence nothing I (thus XX) can depend on. The almost most clear example is Adrian's change, which is even a completely legit one. Still the as legit means to set it back won't end up in the default behaviour. It needs manually adjusting the Registry to get back.
Or a reinstall if you just don't know what happened (which is common of course).

Don't know anything better to say at this moment. But at some time I hope i can.
Peter