Poster | Thread |
Zylesea
| |
emulated 68k speed vs. emulated ppc-speed Posted on 20-Dec-2014 22:02:16
| | [ #1 ] |
|
|
|
Elite Member |
Joined: 16-Mar-2004 Posts: 2263
From: Ostwestfalen, FRG | | |
|
| As I lack a decent x64 or x86 hardware I cannot test it myself. Hence my question on the community: how fast is an emulated 68k processer and an emulated ppc processor on the very same hardware using WinUAE 3.0?
Maybe a test with lame for both processors would be a good benchmark.
lame 68k on OS 3.x. lame ppc on OS 4.1 FE.
Or some comparable task. I guess 68k is faster, but how much faster is it actually? _________________ My programs: via.bckrs.de MorphOS user since V0.4 (2001) |
|
Status: Offline |
|
|
pavlor
| |
Re: emulated 68k speed vs. emulated ppc-speed Posted on 20-Dec-2014 22:30:04
| | [ #2 ] |
|
|
|
Elite Member |
Joined: 10-Jul-2005 Posts: 9578
From: Unknown | | |
|
| @Zylesea
Quote:
Maybe a test with lame for both processors would be a good benchmark. |
PowerPC FPU is not properly emulated (sloooow), lame is highly FPU demanding benchmark.
In my experience 68k JIT and QEMU PowerPC deliver nearly the same performance on my hardware (Core i5 2500K 3.3 GHz).
eg. dnetc benchmark (note: versions aren´t the same, but results still should be comparable)
dnetc v2.9101-507-CTR-08102711 for AmigaOS (OS 3.9, 68K). OGR-NG: [5,031,979 nodes/sec] RC5-72: [4,801,249 keys/sec]
dnetc v2.9109-518-CTR-10121712 for AmigaOS (OS 4.1, PowerPC). OGR-NG: [5,451,861 nodes/sec] RC5-72: [3,693,154 keys/sec]
OGR-NG scales good with performance, so it is better for comparison.
Edit: MPlayer benchmark shows greater CPU performance than SAM460: 238.031s (MUI-MPlayer with WinUAE 2.9.0/3.0.0 beta 26)
In most cases, emulated CPU is in range of 500 to 1000 MHz G3 CPU on my hardware (except FPU of course).
I posted numerous benchmarks in WinUAE with Cyberstorm emulation thread.Last edited by pavlor on 20-Dec-2014 at 10:39 PM. Last edited by pavlor on 20-Dec-2014 at 10:38 PM.
|
|
Status: Offline |
|
|
itix
| |
Re: emulated 68k speed vs. emulated ppc-speed Posted on 20-Dec-2014 23:45:12
| | [ #3 ] |
|
|
|
Elite Member |
Joined: 22-Dec-2004 Posts: 3398
From: Freedom world | | |
|
| @pavlor
Some cores perform better in emulation than other so results may change. But PPC emulation is surprisingly good. _________________ Amiga Developer Amiga 500, Efika, Mac Mini and PowerBook |
|
Status: Offline |
|
|
Zylesea
| |
Re: emulated 68k speed vs. emulated ppc-speed Posted on 21-Dec-2014 13:42:46
| | [ #4 ] |
|
|
|
Elite Member |
Joined: 16-Mar-2004 Posts: 2263
From: Ostwestfalen, FRG | | |
|
| Indeed it's quite good. Far better than I thought (which was that 68k emu was much faster than the ppc emulation). Pretty promising.
_________________ My programs: via.bckrs.de MorphOS user since V0.4 (2001) |
|
Status: Offline |
|
|
Hypex
| |
Re: emulated 68k speed vs. emulated ppc-speed Posted on 22-Dec-2014 14:28:12
| | [ #5 ] |
|
|
|
Elite Member |
Joined: 6-May-2007 Posts: 11180
From: Greensborough, Australia | | |
|
| @itix
Quote:
But PPC emulation is surprisingly good. |
I've pointed it out before but I expect perfomance would be good as PPC is a simple CPU to emulate in the instruction structure as each one is a long word and usually split into high word and low word. It's a simplified RISC design. So interpreting each code is simple enough, just emulating it is the next step. With enough CPU cycles anytihng is possible,
OTOH, 68K is CISC, and has its own complications there. But so is x86 and a lot of things can be remapped over with a bit of endian comnversion. As the UAE core emulation would demonstrate. |
|
Status: Offline |
|
|