Integer is King? - final thoughts about the EQ challenge
-
Music Engineer Music Engineer https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=15959 - KVRAF
- 4380 posts since 8 Mar, 2004 from Berlin, Germany
hi christian. thanx for doing this test. when you came up with it, i actually wanted to take part, but the very next day the thread was about ten pages longer than before and i didn't have the patience to go through all this. so, much thanx for summarizing the results as well.
-
Christian Budde Christian Budde https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=25572
- KVRAF
- Topic Starter
- 1538 posts since 14 May, 2004 from Europe
Yes, you were right. Sorry that I forgot you. Unfortunately I don't remember every detail, but I asked you to remove it since your guess was very close. Coincidence or not.mauseoleum wrote:I knew I was right all along.
Right, Christian?
-
- KVRAF
- 8389 posts since 11 Apr, 2003 from back on the hillside again - but now with a garden!
...hmmm...
Direct Form II...
Direct Form I has more bass...
DSP
Direct Form II...
Direct Form I has more bass...
DSP
-
AdmiralQuality AdmiralQuality https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=83902
- Banned
- 6657 posts since 10 Oct, 2005 from Toronto, Canada
-
- KVRist
- 86 posts since 14 Jan, 2007 from around
very interesting, extreamly interesting. 
Hi everybody. Oh well. King for a day. Not too bad.
I have a few questions here though.
The confusing part in this is that I rated the doubleprecision int the lowest and the single precision int the highest. ....mmh...
I didnt recognize Y as Int at all. To me it had worse high end than X( the float) but sucked out the fundament of the bass just as X(the float).
Seems that integer 64 really f**ked me up.
Now let me understand this. The int64 is just the doubleprecision of the int(int32). So same code, you just ask the processor to do int64 calculation instead of int 32 and keep everything in 64bit memoryspaces?
I have no explanation other than, that I noticed my characteristics too fast in the precision I am used to (which is 32int I would think) and after that the test was biased. But no... Y had the worst high end distortion and the same bass weakness..... mmh..
have to think about that a little. .. and listen again.
I would also like to hear what mauseoleum thinks, and how he got so close to the answer that he had to be removed?
thorKz
Hi everybody. Oh well. King for a day. Not too bad.
I have a few questions here though.
The confusing part in this is that I rated the doubleprecision int the lowest and the single precision int the highest. ....mmh...
I didnt recognize Y as Int at all. To me it had worse high end than X( the float) but sucked out the fundament of the bass just as X(the float).
Seems that integer 64 really f**ked me up.
Now let me understand this. The int64 is just the doubleprecision of the int(int32). So same code, you just ask the processor to do int64 calculation instead of int 32 and keep everything in 64bit memoryspaces?
I have no explanation other than, that I noticed my characteristics too fast in the precision I am used to (which is 32int I would think) and after that the test was biased. But no... Y had the worst high end distortion and the same bass weakness..... mmh..
have to think about that a little. .. and listen again.
I would also like to hear what mauseoleum thinks, and how he got so close to the answer that he had to be removed?
thorKz
-
- KVRist
- 86 posts since 14 Jan, 2007 from around
OK. here again in detail of what I remember how my test went.
I started on the AMD. default setting was X (I think). I remember turning to Y and thinking: thats int. But no I have to check Z first. After that I guess I was fiddeling around with several settings and found what I was looking for in Z.
Then I compared X and Y and found X had the better high end. Y had a lot of that distortion that I had heard numorous times. (on the AMD)
for me Y was like older floating point, X like modern floatingpoint and z fixed point.
Then I was curious how this all sounds on my new intel and changed machines.
On the Intel everything sounded different. I had my champion tho, and just checked if it still was my champion. it was. It sounded a little flatter than the others but was very tweakable without any of the unwanted sideeffects I was looking out for.
Then I did big bass boost and kinda checked Z against X, since Y had lost against X in the Highend dept.
Yes I found that bass weekness in X, while Z really rocked down there.
I remember doing some quick bass comparisms Z vs X, Z vs Y, Z vs X
OK, so Z was int, the other two float, X with a little better high end .
Thats it pretty much. If someone cares. If not dont read it.
thorKz
I started on the AMD. default setting was X (I think). I remember turning to Y and thinking: thats int. But no I have to check Z first. After that I guess I was fiddeling around with several settings and found what I was looking for in Z.
Then I compared X and Y and found X had the better high end. Y had a lot of that distortion that I had heard numorous times. (on the AMD)
for me Y was like older floating point, X like modern floatingpoint and z fixed point.
Then I was curious how this all sounds on my new intel and changed machines.
On the Intel everything sounded different. I had my champion tho, and just checked if it still was my champion. it was. It sounded a little flatter than the others but was very tweakable without any of the unwanted sideeffects I was looking out for.
Then I did big bass boost and kinda checked Z against X, since Y had lost against X in the Highend dept.
Yes I found that bass weekness in X, while Z really rocked down there.
I remember doing some quick bass comparisms Z vs X, Z vs Y, Z vs X
OK, so Z was int, the other two float, X with a little better high end .
Thats it pretty much. If someone cares. If not dont read it.
thorKz
-
- KVRian
- 770 posts since 2 Apr, 2003
The Intel sounded different to the AMD??
He's a troll, got to be, nobody could be that dumb, could they?
He's a troll, got to be, nobody could be that dumb, could they?
-
- KVRAF
- 6519 posts since 13 Mar, 2002 from UK
Congratulations to all those who guessed right
Now with improved MIDI jitter!
-
- KVRist
- 490 posts since 8 Mar, 2005 from Helsinki, Finland
Cue another boatload of worm-cans...JonHodgson wrote:The Intel sounded different to the AMD??
-
- KVRian
- 759 posts since 22 Mar, 2002 from fi
want another can? i remember some guy at gearslutz related that digital audio streamed from an external firewire drive sounded different to the exact same tracks streamed out of an internal SATA drive. or something like that anyway.
"..."
"..."
never stop loving music.
-
- KVRian
- 770 posts since 2 Apr, 2003
Isn't it amazing how some people's hard disks refuse to cough up identical data when they store audio on them, but all their word, excel and porn gifs read without errors?z15 wrote:want another can? i remember some guy at gearslutz related that digital audio streamed from an external firewire drive sounded different to the exact same tracks streamed out of an internal SATA drive. or something like that anyway.
"..."
I mean.. how does the hard disk know the difference so it can act differently?
-
- KVRAF
- 1907 posts since 29 Oct, 2003
How I got to the results?
Simple. Integer sounds "thinner" than float. You must "pump up the bass" more on integer, but when you do, the bass is "better" than on float. This way I recognized the float and int, the "unknown" was deductible from the two known variables.
Now onto the interesting topic:
What does this int/float thing remind me to?
On issues with impendance mismatch. I'm not joking.
Float (to me) sounds like (f.e.) 10k source going into 1Meg input, and the one with the likes of tl-071 in it (those in the know will know what 071 does to the sound )
Integer, on the other hand, sounds like 10k source going into input with similar impendance, f.e. an opamp with bipolars on inputs or a discrete, transistorized input stage.
Sound is less "watery", "cloudy", a little thinner per se, with more "presence", but if you turn up bass, it is "more stable", "in place ... " ... whatever you name it, while on mismatched example bass will be "booming".
My hearing is coming back after my player career - 5 years since my last gig.
Simple. Integer sounds "thinner" than float. You must "pump up the bass" more on integer, but when you do, the bass is "better" than on float. This way I recognized the float and int, the "unknown" was deductible from the two known variables.
Now onto the interesting topic:
What does this int/float thing remind me to?
On issues with impendance mismatch. I'm not joking.
Float (to me) sounds like (f.e.) 10k source going into 1Meg input, and the one with the likes of tl-071 in it (those in the know will know what 071 does to the sound
)Integer, on the other hand, sounds like 10k source going into input with similar impendance, f.e. an opamp with bipolars on inputs or a discrete, transistorized input stage.
Sound is less "watery", "cloudy", a little thinner per se, with more "presence", but if you turn up bass, it is "more stable", "in place ... " ... whatever you name it, while on mismatched example bass will be "booming".
My hearing is coming back after my player career - 5 years since my last gig.
-
- KVRian
- 770 posts since 2 Apr, 2003
That's impressive, mauseoleum just made an invalid connection between TWO phenomenons he obviously knows bugger all about at an engineering level.
-
- KVRist
- 86 posts since 14 Jan, 2007 from around
Awesome man!! I subscribe to all of that. I'm not that expereanced with impedances. From playing my bass without DI for a year or so I'm a little familiar with what your saying there though. Have never thought of that.
Thats cool.
Thats cool.
-
- KVRist
- 86 posts since 14 Jan, 2007 from around
JonHodgson wrote:That's impressive, mauseoleum just made an invalid connection between TWO phenomenons he obviously knows bugger all about at an engineering level.
Whats that Jon? You have nowhere else to go?