gogo-no-coda Ver.3.11 Options

[Nyanko]

gogo-no-coda Ver.3.11 was released.
http://member.nifty.ne.jp/~pen/free/gogo3/mct_gogo.htm

[DigitalDictator]

So GoGo is basically a tweaked Lame 3.88? Not much is said about the quality of this encoder, how good is it really? The quality of the new GoGo 3.11 is equivalent to what - Lame 3.88?

Also, my Japanese (or is it Korean?) is pretty rusty. Is it possible to make a simple homepage in English where you can read a bit about the history, progress and the developer(s) of GoGo..? Maybe that will draw some more attention to it.

[Jospoortvliet]

can't read the page either. and are there any good frontends?

[S_O]

The english page I found is still with the old version, I downloaded the version from this site and compiled it, it´s based on lame 3.88 and extremly fast:
-q 0 18,84x realtime
-q 2 24,33x realtime
-q 5 38,21x realtime
-q 9 77,89x realtime
all with enabled psymodel at a Athlon Thunderbird 1333MHz
I don´t know how fast xing is, but this sounds much better (-b 128 -m j).
If you don´t trust me test yourself: http://l.b.oltmanns.bei.t-online.de/gogo311.zip
dll and exe. Compiled with ICL6 (/O3 /QaxiMKW /Qip /Qsox-).

[harashin]

So GoGo is basically a tweaked Lame 3.88? Not much is said about the quality of this encoder, how good is it really? The quality of the new GoGo 3.11 is equivalent to what - Lame 3.88?

From readme_e.txt

CODE

       GOGO-no-coda ver. 3.11 for Windows, Linux and OS/2

                                               updated Nov. 25, 2002
* ABSTRACT

 This software is a mp3 encoder based on LAME3.88,
 which is optimized for Enhanced 3D Now!/SSE/SSE2 and dual-CPUs.

Also, my Japanese (or is it Korean?) is pretty rusty. Is it possible to make a simple homepage in English where you can read a bit about the history, progress and the developer(s) of GoGo..? Maybe that will draw some more attention to it

Their English page is not updated yet...
However you could read babelfished their site.

can't read the page either. and are there any good frontends?

Try WAV2GOGO.

This post has been edited by harashin: Nov 26 2002, 15:08

[john33]

Win32 binaries are available here: http://homepage.ntlworld.com/jfe1205/others.html

Edit: BTW, if anyone downloads the Windows '.tar' file, rename it to '.tgz', otherwise you will have trouble decompressing it.

This post has been edited by john33: Nov 26 2002, 15:25

[S_O]

Win32 binaries are available here: http://homepage.ntlworld.com/jfe1205/others.html

Which compiler-options do you use? My compile is 0,5x faster (but also twice as big).

Edit: BTW, if anyone downloads the Windows '.tar' file, rename it to '.tgz', otherwise you will have trouble decompressing it.

First I downloaded this lzh-archive, I could see the files, but I couldn´t uncompress it. The other archive could be decompressed by winrar without any problems.

[DigitalDictator]

Ah, thanx Harashin. I just couldn't seem to find the download link nor the English page! Must be tired today...

If I'm picky about the quality, can I use GoGo? (given that I wanna stick to mp3). I don't have time nor the equipment to perform any tests myself. I have a stack of CD's that I wanna convert to mp3 and I want to do it quickly. The file size should be around 190-200 kbps. I know Lame is pretty fast but I want Xing-speed and Lame quality. Possible? If so, which settings?

[john33]

Which compiler-options do you use? My compile is 0,5x faster (but also twice as big).

Intel 6.0 with '03 /QaxiMK /Qsox- /Qip'. What are you using?

They are compressed using UPX: upx --best .....

This post has been edited by john33: Nov 26 2002, 16:52

[FuRaL66]

Hi, I've tried to encode with this Command Line:

-v 2 -m j -vb 128 320 -q 2

Is this OK?

Can Gogo 3.11 reach neraly Lame APS Quality?

Greetz

[Benjamin Lebsanf...]

Can Gogo 3.11 reach neraly Lame APS Quality?

no way i think!

[john33]

Can Gogo 3.11 reach neraly Lame APS Quality?

no way i think!

But at the end of the day, if it is you who is going to listen to them and you're happy with the quality, it doesn't really matter what anyone else thinks! It won't be as good as LAME 3.90.2, etc, but it is probably better than the rest.

This post has been edited by john33: Nov 26 2002, 16:58

[S_O]

Intel 6.0 with '03 /QaxiMK /Qsox- /Qip'. What are you using?

They are compressed using UPX: upx --best .....

I´m using "/O3 /QaxiMKW /Qip /Qsox-" with ICL6, that´s nearly the same, I don´t have a Athlon-Thunderbird, so I cannot take advantage of the SSE2 optimization, so it must be randomly slower, because UPX decompresses the program at startup, so the start may take longer, but not the encoding.
I would like know if my compile is faster on Pentium 4 than yours because of the SSE2 optmizations.
Do you know if it´s better to use /QxiM for Athlon A/B/C & Pentium 2, /QxiMK for Pentium 3 & Athlon XP and /QxiMKW for Pentium 4? Because of Qx instead of Qax the optmizations has to be used, the programn won´t work without them. In HeadAC3he (ac3 transcoder) it is handled with the dlls like that.

[Chun-Yu]

I would like know if my compile is faster on Pentium 4 than yours because of the SSE2 optmizations.

I would guess no, because the speed critical sections of GoGo are pretty much all handwritten assembly. Plus, SSE2 isn't as cool as SSE because it only does 2 64-bit floats at a time (but it's still pretty cool).

Do you know if it´s better to use /QxiM for Athlon A/B/C & Pentium 2, /QxiMK for Pentium 3 & Athlon XP and /QxiMKW for Pentium 4? Because of Qx instead of Qax the optmizations has to be used, the programn won´t work without them. In HeadAC3he (ac3 transcoder) it is handled with the dlls like that.

Yes, using /Qx instead of /Qax produces slightly faster code, but it'll crash on older processors! /Qx is especially faster on small functions that could be inlined, since /Qax prevents the compiler from inlining the vectorized function. Using DLLs compiled with /Qx kind of defeats the advantage of using /Qx because /Qax uses a dispatcher kind of thing to select the right version of each function, which is pretty much what using a DLL does. Using a DLL might be helpful for targeting more than one platform (like P-Pro, P-III, P-IV), because /Qax produces only specialized code and generic code (e.g. /QaxiMKW produces code that only runs on P-IVs and generic code for all x86s).

ICL7 is out now (just came out 11/21/2002) - haven't tried it yet (but I sure will). Also, in spring 2003, Microsoft will be releasing version 7.1 of their C++ compiler, which finally can do vectorization with SSE/SSE2 like ICL can - it's currently in beta testing.

[S_O]

Using a DLL might be helpful for targeting more than one platform (like P-Pro, P-III, P-IV), because /Qax produces only specialized code and generic code (e.g. /QaxiMKW produces code that only runs on P-IVs and generic code for all x86s).

So it would be better to use /QaxiM for Athlon, because with /QaxiMKW it wouldn´t use any optimization, because SSE and SSE2 is not supported and it only generates optimized (with SSE / SSE2) and non-optimized code? And /QxiM would be even better, but won´t run on processors without that optimizations? What would happen with /QxiM /QaxiMKW? Does it generate optmizated code (with MMX) and as alternative if avaible also SSE and SSE2?
I´ll try that.

ICL7 is out now (just came out 11/21/2002) - haven't tried it yet (but I sure will).

Do I need to a new license for that or will the old still work?

Now I finally found a bug in GoGo: It supports now Lame-Tags (-lametag on):
-MP3s with tags have terrible sync-errors and lots of other errors (mad-winamp-plug-in)
-It writes in the tag "LAME3.92"!

[DSPguru]

if you'll open the .dll in an hexa-editor and look at address 197835 (decimal) you'll see 'LAME3.92'.

[Guest_SK1_*]

I don't get this. So it's fast sure, but
1. it's actually LAME version 3.92 (THANKS DSPguru!), LAME 3.90.2 is a better, recommended version, not 3.92.
2. the newer "unofficial" release with tons of changes and improvements made by Takehiro produces files with MUCH better quality!! Why on earth would you want to encode files in lower quality when there is a much better alternative?
3. so WHAT if it's sooOO00oo fast, if Dibrom's builds' speeds aren't enough for you, i don't know what your problem is, i can understand a need for speed, but this is just useless..

I just don't understand all this GoGo thing.. It's inferior, period. It's like the Audi TT of encoders!! (if you know what i mean..)

[S_O]

1. it's actually LAME version 3.92 (THANKS DSPguru!), LAME 3.90.2 is a better, recommended version, not 3.92.

see the readme file: This software is a mp3 encoder based on LAME3.88,
which is optimized for Enhanced 3D Now!/SSE/SSE2 and dual-CPUs.

And in the history:

The following bug is fixed.

gogo happens to make mp3-data which can't be played by Windows Media
Player version 6.

supported LameTag output (-lametag on)

lame 3.92 is nowhere mentioned, I guess they just took the lametag-writer-source from lame3.92 and forgot to change the version-string. Look into the source-file vbrtag.c at the function CreateLameVBR, you´ll see:
'const char *szVersion = (const char *)"LAME3.92";'
I thought I could simply change that to "GOGO3.11", but encspot etc. finds the lametag by searching for "LAME", so that tag won´t be found anymore.
The tag has nothing to do with the real version, I can also write LAME3.96 in it, or even LAME5.27.

[Guest_SK1_*]

OK, my mistake, thanks for clearing that out.. Anyway, this means that it's even worse then..

[sven_Bent]

3. so WHAT if it's sooOO00oo fast, if Dibrom's builds' speeds aren't enough for you, i don't know what your problem is, i can understand a need for speed, but this is just useless..

exactly my philosophy
Why do it fast when you can do it right...

[cd-rw.org]

For quite some time I have wondered if any part of GoGo could be ported to LAME for speed benefits. Speed boost would be a good way to market LAME and high fidelity MP3s, and for my PIII-500mhz - I'd like a speed boost for a change. Usually new revisions have slowed things down (for quality of course, but still..)

[SometimesWarrior]

3. so WHAT if it's sooOO00oo fast, if Dibrom's builds' speeds aren't enough for you, i don't know what your problem is, i can understand a need for speed, but this is just useless..

I just don't understand all this GoGo thing.. It's inferior, period. It's like the Audi TT of encoders!! (if you know what i mean..)

What's wrong with having a faster encoder available? Maybe I'm in a rush and I just want to get my MPC album transcoded for a portable as quickly as possible... GoGo to the rescue!

GoGo's also ideal for people running music streaming servers. If you want to run multiple MP3 streams along with your Ogg streams (and whatever else), you might need a fast encoder like GoGo.

I'd never use GoGo for MP3 archiving (that's what the alt-presets are for!), but it's still nice to have it around and to see that it's being updated.

[LIF]

GoGo is my 2nd choice encoder, mainly when using my old machine(PII 233MMX/192 RAM).
To my ears it does a decent job.
I've found these simple ABR settings and both are fast like the "plain" 192 ~ 256 CBR:

-b 200 -a -q3
or
-b 250 -a -q2

I use plain stereo, because I believe GoGo doesnt carry the same JS improvements as the latest Lame compiles.
ABR settings around 180 kbits/s can also be used.

Last: Is it possible to create an GoGo filter for Cool Edit?(.flt)

LIF

This post has been edited by LIF: Nov 28 2002, 03:46

[twostar]

What's wrong with having a faster encoder available? Maybe I'm in a rush and I just want to get my MPC album transcoded for a portable as quickly as possible... GoGo to the rescue!

How do I do that exactly? I've tried gogo with MPC2MP3 and it won't work (probably because MPC2MP3 is only for lame).

Is there an easier way to transcode MPC to MP3 using gogo?

[Chun-Yu]

So it would be better to use /QaxiM for Athlon, because with /QaxiMKW it wouldn´t use any optimization, because SSE and SSE2 is not supported and it only generates optimized (with SSE / SSE2) and non-optimized code? And /QxiM would be even better, but won´t run on processors without that optimizations? What would happen with /QxiM /QaxiMKW? Does it generate optmizated code (with MMX) and as alternative if avaible also SSE and SSE2?
I´ll try that.

Yep. /Qx switches mean that the binary will require those instructions, and /Qax means that it will use them if available. You can combine /Qx and /Qax exactly like you mentioned. /Qx is usually faster, but it will only run on newer processors that have those instructions. Athlon XPs have SSE, so /QxiMK would be good for them (actually /QxK implies /QxiMK so you can save yourself 2 bytes ).

ICL7 is out now (just came out 11/21/2002) - haven't tried it yet (but I sure will).

Do I need to a new license for that or will the old still work?

I think the old one will still work, but I'm not sure (well, my old evaluation one works - ICL7 is so good that I'm going to buy it for sure). I finally tried ICL7 and it pretty much blows VC++ .NET away when it comes to floating-point stuff. ICL7 seems to be a bit faster than ICL6.