today i just tried to rip a song with three diff versions of lame encoder.

Artist: Opeth
Song: When
Style: Extreme Progressive Metal
Length: 9:14
Source: When.wav (uncompressed wav ripped with EAC in secure mode, no error)
All settings: --alt-preset extreme
System: P4 3.0c with 800 FSB Hyper Thread (Enabled; Non Overclock), 1GB DDR pc3200, Intel 865PE Motherboard

- And the speed results were very weird:

lame 3.903 : ~6x; 98 seconds; 17,328 kb
lame 3.95 : ~8x; 66 seconds; 17,390 kb
lame 4(a6) : ~12x; 50 seconds; 16,980 kb

- I only can judge the quality with my own ears and this is personal opinion:

lame 3x sound the same..., lame4 sound pretty much metallic...maybe cause the new vbr mode..

- When I tried to encode in alt-preset cbr 192; lame 3.90.3 and lame 4.0 are same slow ( around 2x) and lame 395 was fast, it was about 10x


any commend welcome

You're gonna need abx results for that.

what about the cbr issue...why lame 3.90.3 cbr mode is much slower than vbr

thanks

Qoute
lame 3.903 : ~6x; 98 seconds; 17,328 kb
lame 3.95 : ~8x; 66 seconds; 17,390 kb
lame 4(a6) : ~12x; 50 seconds; 16,980 kb



Something is wrong with your PC. I encode most of my music on my brother's PC (just an Intel Celeron 1200 MHz). I get about 5x with APS 3.90.3 and even faster with APE. Check your BIOS settings.

I don't think anything looks unusual with those values (although it depends on your kit). I get 8x with APS on 3.95 and 3.90.3 gave me about 5x (from memory).

These values look fine. Personally I'm comparing APS with Nero AAC Normal at the moment and finding it very difficult to tell the two apart. Whenever I think I can here a difference I can't consistently detect it so I guess it isn't there!

Do some ABXs and post the results like the man said. Yeah I know documenting things is a drag and it's a bore hearing people go on about ABXing all the time.

Love,
Fairy

What is your system?
I am sure I get about 5x with lame.exe 3.90.3 and LAMEDropXPd on Intel Celeron 1,2 GHz 256 Kb cash. So Opeth have to get about 12-14x

Agreed with de Mon about the speed issue - on my P4 overclocked to 3.0GHz (no HT) I get results like this:

LAME 3.90.3:
--aps = 7.1x
--ape = 7.9x
--apfs = 11.1x
--apfe = 11.4x
--cbr 192 = 9.9x

LAME 3.95:
--aps = 11.3x (59% faster)
--ape = 12.3x (56% faster)
--apfs = 14.2x (28% faster)
--apfe = 14.2x (25% faster)
--cbr 192 = 16.4x (66% faster)

maybe this is hyper threading issue...

i forgot to note that i was encoding divx511 @ slowest mode (but i put the priority to slowest)....

why lame 3.09.3 is SO SLOW with my system...

I get about 2.4x speed with --preset standard, Lame 3.95 on my Duron 600 processor PC .
But I ripped with EAC today with --preset standard -m s (stereo) and the speed was only about 1.2x and less. Anyone know why that is?
And using Lame 3.94 beta 1 with EAC was even a greater pain in the *ss. It went like 0.4x with 192 CBR setting . I don't understand why?
Lame encoding was always about 2.5x with VBR and about 4x using CBR on my machine. Up to now....

APS:

LAME 3.90.3: 0.7x
LAME 3.95.0: 1.2x
LAME 4.00.0: 3.4x

Tested on my Pentium II with 333 MHz.

Wow, fast machine. On my C3 600MHz, and preset standard, 3.95 is arround x0.7, 4.00 is arround x2.5.

Note: 4.00 is daily chainging and discussing sound quality and speed is quite meaningless...

Beeing fanless has its price...

Well, I get ~1.3x on my fanless iMac with it's blazingly fast 400MHz G3 I used to get 0.6-0.8 with 3.90.3 - quite an improvement.

Celeron 1.7
3.90.3 aps ~ 2.8-3.2x realtime
3.95 aps ~ 4.2x

very nice improvement, if there are no major dropouts we should have an excellent successor
How do I turn off replaygain calculation please?

Hell! I can't understand speeds you post here. Are you playing Quake 3 while encoding? I speak about 5x on my brother's Celeron 1.2 GHz and when I encode I do it from .WAV (not from CD) and I don't run any other applications. The only application is RazorLame + 3.90.3 (or LAMDropXPd).

This is off-topic, but, were you using the -m s switch just for testing?

linux kernel 2.6.1-love1, glibc with nptl enabled, lame compiled with gcc 3.3.2 (CFLAGS="-march=athlon-xp -mfpmath=sse -O2 -pipe -frename-registers -fomit-frame-pointer -s") on athlon-xp @ 2.1 GHz, 1GB DDR-RAM DC PC-400:

lame 3.93.1 (I never encode to mp3...)
ape: 7.3x realtime with my sample
aps: 5.7x

lame 3.95.1
ape: 10.3x
aps: 9.0x