G5 port seems to progress nicely:

http://morphos.pl/423


Piru even posted first benchmarks on morphzone.org:






I hoped for better lame results (less than 40 % faster than G4 1.42 GHz ), MPlayer result is nice.

Last edited by pavlor on 25-Dec-2012 at 09:32 AM.

@pavlor: Interesting!

_________________
AmigaOne X5000/40, 2.2 Ghz, 4 GB RAM, Radeon R9 280X, M-Audio Revolution 5.1, 240 GB SSD

@pavlor

Can I update the SAM460ex 1155mhz results.
WIth the latest kernel and Update 6.

lame Work:Downloads/AKsack.wav
LAME 3.99.5 32bits (http://lame.sf.net)
Using polyphase lowpass filter, transition band: 16538 Hz - 17071 Hz
Encoding Work:Downloads/AKsack.wav to Work:Downloads/AKsack.mp3
Encoding as 44.1 kHz j-stereo MPEG-1 Layer III (11x) 128 kbps qval=3
Frame | CPU time/estim | REAL time/estim | play/CPU | ETA
10529/10529 (100%)| 0:47/ 0:47| 0:47/ 0:47| 5.8033x| 0:00
-------------------------------------------------------------------------------
kbps MS % long switch short %
128.0 100.0 74.4 13.4 12.2
Writing LAME Tag...done
ReplayGain: +0.5dB
===============================================================
WorkBench:> mplayer -benchmark -nosound -ao null -vo null -lavdopts skiploopfilter=none "Media:Test/Prometheus-Trailer.mp4"
MPlayer UNKNOWN-4.4.3 (C) 2000-2010 MPlayer Team
160 audio & 349 video codecs

Playing Media:Test/Prometheus-Trailer.mp4.
libavformat file format detected.
[lavf] stream 0: video (h264), -vid 0
[lavf] stream 1: audio (aac), -aid 0, -alang eng
VIDEO: [H264] 1920x816 24bpp 90000.000 fps 6828.3 kbps (833.5 kbyte/s)
Clip info:
major_brand: mp42
minor_version: 0
compatible_brands: mp42isomavc1
artist: 20th Fox
title: Prometheus - Trailer
encoder: HandBrake 4344svn 2011111001
date: 2012
==========================================================================
Opening video decoder: [ffmpeg] FFmpeg's libavcodec codec family
Selected video codec: [ffh264] vfm: ffmpeg (FFmpeg H.264)
==========================================================================
Audio: no sound
Starting playback...
Movie-Aspect is 2.35:1 - prescaling to correct movie aspect.
VO: [null] 1920x816 => 1920x816 Planar YV12


BENCHMARKs: VC: 398.120s VO: 0.013s A: 0.000s Sys: 1.658s = 399.791s
BENCHMARK%: VC: 99.5821% VO: 0.0032% A: 0.0000% Sys: 0.4147% = 100.0000%

Exiting... (End of file)

Last edited by Spectre660 on 25-Dec-2012 at 04:01 PM.
Last edited by Spectre660 on 25-Dec-2012 at 03:57 PM.
Last edited by Spectre660 on 25-Dec-2012 at 03:57 PM.

_________________
Sam460ex : Radeon Rx550 Single slot Video Card : SIL3112 SATA card

I think the most interesting will be benchmark between G4 2Ghz and G5 2Ghz on MOS.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@KimmoK

Well we all knew the G5 would be faster. The point I think is that the PASemi can do what it does probably without a heatsink, the G5 needs stoopid cooling - look at the probs Mark is having until he implements fan control!

_________________
Fraggle

- insert profound text here -

@Frags

Quote:

Well we all knew the G5 would be faster.

Current lame benchmark shows 2 GHz G5 would be as fast as 2 GHz G4. 32 bit mode in G5 is probably slower than 64 bit mode.

Now I´m realy interested in Blender benchmark - G4 can´t "cheat" there with its superior AltiVec implementation.
Note: Blender benchmark under Linux shows 2.3 GHz G5 is as fast as both cores of PA6T 1.8 GHz.

@pavlor
Quote:

32 bit mode in G5 is probably slower than 64 bit mode.

That's a common misconception, which I believe partly there thanks to Apple's "64bit reality distortion field" when the G5s appeared and because how other architectures perform when comparing their 32bit and 64bit mode.

On x86-64, 64bit mode offers real advantages over plain x86's 32bit mode - more registers, better ISA at places, which can be easily translated to better code by the compiler, so in one word: performance - but PowerPC is different.

PowerPC has a good ISA also in 32bit and 64bit ISA is practially the same on user level. 64bit on PPC is really offers advantages only when addressing huge areas of data or 64bit integer math is involved. The drawback is a somewhat worse utiliztation of cache lines on average, because sometimes 64bits of data is moved or used in calculations, where less would be enough. Which means an average code might even work a bit slower in 64bit mode than in 32bit mode, on a "G5" class CPU.

So I think for an Amiga-like (lightweight) OS as we know it nowadays, there's really little-to-no advantage of running the 970 series in 64bit mode. However, please feel free to fix me, if I got this wrong.

_________________
MorphOS, classic Amiga, demoscene, and stuff
"When a bridge is not enough, build a Viaduct!"
"Strip the Amiga community of speculation and we can fit every forum on a 720k floppy" (by resle)

I think for most people G4 2.0 Ghz in´t really interesting. There is no stock-Powermac with this speed and CPU-Cards with 2.0 Ghz G4 are rare and expensive. Sonnet CPU-cards with G4 1,6 and 1,8 Dual-G4 may be easier to find (but still expensive).

Just for the performance it is easier and cheaper to buy a Powermac G5 instead.

_________________
AmigaOne X5000/40, 2.2 Ghz, 4 GB RAM, Radeon R9 280X, M-Audio Revolution 5.1, 240 GB SSD

@Chain-Q

Quote:

However, please feel free to fix me, if I got this wrong.

I admit I´m no expert.


I understood there is some performance penalty when using current MorphOS applications on 970. Am´I right?

@Chain-Q

Also on x86-64 architecture 32-bit applications may run slightly faster than their 64-bit versions.

On G5 some instructions are emulated (mcrcx and few more) and its cache line is longer than G4's (128 bytes vs 32 bytes). For some performance critical applications it may be good idea have alternate code paths for G5 and maybe also tune cache alignment...

@pavlor

Yes there can be performance penalty. Due to missing instructions, due to different cache line size, different cache alignment and microcoded instructions.

Last edited by itix on 25-Dec-2012 at 05:00 PM.

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@AP
I have a dual 1.8Ghz PowerLogix G4 accelerator in my QuickSilver with 7448 CPUs, and (using a single core, obviously) it performs roughly as my 1.67Ghz PowerBook G4. The PowerBook has a faster memory bus, while the QS is slightly better at number crunching, but the differences are nowhere significant in real world.

In fact the PowerBook feels a bit faster, due to its R300 Radeon outperforming the (old) R100 Radeon I have in the QuickSilver.

_________________
MorphOS, classic Amiga, demoscene, and stuff
"When a bridge is not enough, build a Viaduct!"
"Strip the Amiga community of speculation and we can fit every forum on a 720k floppy" (by resle)

@pavlor

Quote:

Now I´m realy interested in Blender benchmark - G4 can´t "cheat" there with its superior AltiVec implementation. Note: Blender benchmark under Linux shows 2.3 GHz G5 is as fast as both cores of PA6T 1.8 GHz.

Mmmmmmmmmm

Back when I ran that Blender bench on my 2x1.8GHz QuickSilver I scored results slighly better than the X1000, something I've allways contributed to the G4 having superior Altivec compared to P6T or G5.

The benches shown here sofar are a bit disapointing as the G5 only seems slightly faster clock by clock than a G4.

_________________
- We don't need good ideas, we haven't run out on bad ones yet
- blame Canada

@pavlor
There might be, because as itix said, it needs differently optimalized code at places, due to the different architekture of the CPU (longer pipelines, different cache lines, some instructions missing). But these possible optimizations also available in 32bit mode for user space apps, and the kernel and other system components will try to hide the differences to applications where possible anyway. So it's still likely to outperform anything we've got so far, with the possibility to improve further a bit in the future. (I don't have a G5 myself, so this is just a prediction/estimate.)

_________________
MorphOS, classic Amiga, demoscene, and stuff
"When a bridge is not enough, build a Viaduct!"
"Strip the Amiga community of speculation and we can fit every forum on a 720k floppy" (by resle)

Test Lame on my Efika:

Ram Disk:> lame_ppc AKsack.wav
LAME 3.98.4 32bits (http://www.mp3dev.org/)
Using polyphase lowpass filter, transition band: 16538 Hz - 17071 Hz
Encoding AKsack.wav to AKsack.wav.mp3
Encoding as 44.1 kHz j-stereo MPEG-1 Layer III (11x) 128 kbps qval=3
Frame | CPU time/estim | REAL time/estim | play/CPU | ETA
10529/10529 (100%)| 2:20/ 2:20| 2:20/ 2:20| 1.9536x| 0:00
-------------------------------------------------------------------------------
kbps MS % long switch short %
128.0 100.0 76.1 13.4 10.5
Writing LAME Tag...done
ReplayGain: +0.5dB

SYSTEM: bplan,EFIKA5K2 (revision 2B3)
PROCESSOR: 5200 (G2-LE) (V1.4) 396MHz (FSB 132MHz)

_________________
A1200 BlizzardPPC BVision FastATA Eth 3Com, EFIKA5K2

@SKOLMAN_MWS

Sam440ep-Flex 800mhz. AmigaOS 4.1 Update 6 and new kernel .

lame Work:Downloads/AKsack.wav
LAME 3.99.5 32bits (http://lame.sf.net)
Using polyphase lowpass filter, transition band: 16538 Hz - 17071 Hz
Encoding Work:Downloads/AKsack.wav to Work:Downloads/AKsack.mp3
Encoding as 44.1 kHz j-stereo MPEG-1 Layer III (11x) 128 kbps qval=3
Frame | CPU time/estim | REAL time/estim | play/CPU | ETA
10529/10529 (100%)| 1:15/ 1:15| 1:15/ 1:15| 3.6575x| 0:00
-------------------------------------------------------------------------------
kbps MS % long switch short %
128.0 100.0 74.4 13.4 12.2
Writing LAME Tag...done
ReplayGain: +0.5dB
==================================================================
mplayer -benchmark -nosound -ao null -vo null -lavdopts skiploopfilter=none "Media:Test/Prometheus-Trailer.mp4"
MPlayer UNKNOWN-4.4.3 (C) 2000-2010 MPlayer Team
160 audio & 349 video codecs

Playing Media:Test/Prometheus-Trailer.mp4.
libavformat file format detected.
[lavf] stream 0: video (h264), -vid 0
[lavf] stream 1: audio (aac), -aid 0, -alang eng
VIDEO: [H264] 1920x816 24bpp 90000.000 fps 6828.3 kbps (833.5 kbyte/s)
Clip info:
major_brand: mp42
minor_version: 0
compatible_brands: mp42isomavc1
artist: 20th Fox
title: Prometheus - Trailer
encoder: HandBrake 4344svn 2011111001
date: 2012
==========================================================================
Opening video decoder: [ffmpeg] FFmpeg's libavcodec codec family
Selected video codec: [ffh264] vfm: ffmpeg (FFmpeg H.264)
==========================================================================
Audio: no sound
Starting playback...
Movie-Aspect is 2.35:1 - prescaling to correct movie aspect.
VO: [null] 1920x816 => 1920x816 Planar YV12


BENCHMARKs: VC: 531.155s VO: 0.015s A: 0.000s Sys: 1.762s = 532.932s
BENCHMARK%: VC: 99.6666% VO: 0.0028% A: 0.0000% Sys: 0.3306% = 100.0000%

Exiting... (End of file)
==================================================================

Last edited by Spectre660 on 25-Dec-2012 at 06:29 PM.
Last edited by Spectre660 on 25-Dec-2012 at 06:28 PM.

_________________
Sam460ex : Radeon Rx550 Single slot Video Card : SIL3112 SATA card

@

The lame results for Sam440ep and Sam460ex are about 11.75% faster with update 6 and the new kernels .

_________________
Sam460ex : Radeon Rx550 Single slot Video Card : SIL3112 SATA card

@Spectre660

Quote:

The lame results for Sam440ep and Sam460ex are about 11.75% faster with update 6 and the new kernels

That I call speed improvement!

@Spectre660

Test new Lame 3.99.5 on Mac mini G4 1.42 GHz

For MorphOS there is only old version LAME 3.98.4 (slower)

Last edited by SKOLMAN_MWS on 25-Dec-2012 at 06:25 PM.

_________________
A1200 BlizzardPPC BVision FastATA Eth 3Com, EFIKA5K2

@pavlor

I knew that both of my machines feel faster.
Good stuff.


Quote:

pavlor wrote: @Spectre660 Quote: The lame results for Sam440ep and Sam460ex are about 11.75% faster with update 6 and the new kernels That I call speed improvement!

_________________
Sam460ex : Radeon Rx550 Single slot Video Card : SIL3112 SATA card

@SKOLMAN_MWS

Quote:

Test new Lame 3.99.5 on Mac mini G4 1.42 GHz

I hope older version was used for G5 benchmarks. However, same speed as G5 2 GHz looks impressive.