Poster | Thread |
olegil
| |
Re: Performance of the SAM440ep Posted on 21-Feb-2009 15:20:02
| | [ #61 ] |
|
|
|
Elite Member |
Joined: 22-Aug-2003 Posts: 5895
From: Work | | |
|
| @bernd_afa
And if you think a 233MHz e300 core with SIMM memory is faster than a 600MHz e300 core with DDR2 then you have one very active imagination. _________________ This weeks pet peeve: Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean. |
|
Status: Offline |
|
|
bernd_afa
| |
Re: Performance of the SAM440ep Posted on 21-Feb-2009 19:10:25
| | [ #62 ] |
|
|
|
Cult Member |
Joined: 14-Apr-2006 Posts: 829
From: Unknown | | |
|
| >And if you think a 233MHz e300 core with SIMM memory is faster than a 600MHz >e300 core with DDR2 then you have one very active imagination.
when you do some calculation compressing etc, then ram speed is maybe not so important and cache do the job.both have 32 kb cache.
only memspeed is important when have big c++ written programs that use many overload classes etc.
depend on the app you cant say without test, this CPU is faster.maybe somebody do some tests that show how fast SAM is.
http://en.wikipedia.org/wiki/PowerPC_400#PowerPC_440
""" The core lacks a floating point unit (FPU) but it has an associated 4 stage FPU that can be included using the APU (Auxiliary Processing Unit) interface. """
not sure what happen with performance when do this and if that work superscalar. programs that are too slow use most time FPU, and the 604 have a realy fast FPU.
I remember that the 68881 and 68882 too have a connectet FPU and it is not integrate.but FPU was really slow, in compare to integrate CPU as 68040 or 68060 have.
>That page doesn't show any SAM values. Also, the text explains how the test data >shows the bad performance of memory functions in glibc, not bad performance in >hardware.
Yes sam is not here, Of course there are methods show, that reach the Hardware limits and show real world speed. Last edited by bernd_afa on 21-Feb-2009 at 07:15 PM. Last edited by bernd_afa on 21-Feb-2009 at 07:13 PM. Last edited by bernd_afa on 21-Feb-2009 at 07:10 PM.
|
|
Status: Offline |
|
|
pavlor
| |
Re: Performance of the SAM440ep Posted on 21-Feb-2009 19:21:56
| | [ #63 ] |
|
|
|
Elite Member |
Joined: 10-Jul-2005 Posts: 9598
From: Unknown | | |
|
| @bernd_afa
Sure, 440EP is "embedded" CPU, but it is in almost every application three times faster than 604e in CyberstormPPC... |
|
Status: Offline |
|
|
bernd_afa
| |
Re: Performance of the SAM440ep Posted on 22-Feb-2009 11:29:56
| | [ #64 ] |
|
|
|
Cult Member |
Joined: 14-Apr-2006 Posts: 829
From: Unknown | | |
|
| maybe, but i like to see values.
only measuring show the Facts. If it is faster, wy not show that it is faster ? i think it is better show how fast it is, than show nothing.
or do you believe, when a car designer do not write how many seconds it take from 0-100 or the litres/100km it is very good, so he need this not write ?
|
|
Status: Offline |
|
|
ChrisH
| |
Re: Performance of the SAM440ep Posted on 22-Feb-2009 12:41:55
| | [ #65 ] |
|
|
|
Elite Member |
Joined: 30-Jan-2005 Posts: 6679
From: Unknown | | |
|
| @bernd_afa I don't see why you are being so pushy about this issue. It's pretty obvious Sam440's CPU is going to be faster than poor PPC accelerators for Classic Amigas. It is also obvious that it's CPU is going to be slower than a G3 or G4 of an AmigaOne. (Of course, CPU is not the only factor in speed, things like memory speed & disk speed (which Sam440 should be better at) also affect it - depending on what you are doing.)
But if no-one can be bothered to run some tests (that may tell us what we already think we know), just to please you, please don't get upset about it.... Last edited by ChrisH on 22-Feb-2009 at 12:42 PM.
_________________ Author of the PortablE programming language. It is pitch black. You are likely to be eaten by a grue... |
|
Status: Offline |
|
|
pavlor
| |
Re: Performance of the SAM440ep Posted on 22-Feb-2009 13:54:17
| | [ #66 ] |
|
|
|
Elite Member |
Joined: 10-Jul-2005 Posts: 9598
From: Unknown | | |
|
| @bernd_afa
If you like benchmarks...
AMCC PPC440EP - 667 MHz vs. Freescale MPC7455 - 1GHz vs. Freescale MPC755-400 EEMBC consumer benchmark EEMBC automotive benchmark
You can see that 440EP 666 MHz is faster than MPC755 400 MHz. I think that MPC755 is faster than your favourite 604e 233 MHz... |
|
Status: Offline |
|
|
Cool_amigaN
| |
Re: Performance of the SAM440ep Posted on 22-Feb-2009 14:27:26
| | [ #67 ] |
|
|
|
Super Member |
Joined: 6-Oct-2006 Posts: 1227
From: Athens/Greece | | |
|
| |
Status: Offline |
|
|
bernd_afa
| |
Re: Performance of the SAM440ep Posted on 22-Feb-2009 16:03:49
| | [ #68 ] |
|
|
|
Cult Member |
Joined: 14-Apr-2006 Posts: 829
From: Unknown | | |
|
| thanks for links, now its a little clearer it seem hard that my 604e is faster as SAM
but whats not clear, is if the compiler are optimize to 440ep.i dont know how much gcc PPC is optimized to use the additional 440ep instructions
https://www.amcc.com/MyAMCC/retrieveDocument/PowerPC/440EP/PPC440EP_PB2001.pdf
here stand this.I think when do digital signalprocessing benchmarks, the greenhills PPC compiler use them.sad that in the benchmarks not all must use same compiler.
""" and 24 digital signal processing (DSP) instructions, """"
http://tinyurl.com/d39j5b
sad that there are no results of 604e and it seem not possible to get the source to compile and test myself.
here seem see the slow fpu of 440ep.when i downscale the g4 1ghz Value to 233 MHZ (1000/233 =4.29) then in theory can reach 712. 440ep is then 82% faster than a G4 with 233 MHZ
but this show not real world values on app.
such code dont use altivec(see that the 755 with only 400 MHZ get good score.
""" Fast Fourier Transform (Auto/Indust. Version) 440ep 667 MHZ 1,321.34 G4 1 GHZ 3,058 MPC755-400 1,141 """
here a 440ep 667 is 65% faster than a G4 with 233 MHZ
""" Finite Impulse Response Filter (Auto/Indust. Vers) 440ep 667 MHZ 230,196.13 G4 1GHZ 600,500 MPC755-400 208,482 """
edited by wegster - please use tinyurl for URLs that scroll several lines.. Last edited by wegster on 22-Feb-2009 at 04:55 PM. Last edited by bernd_afa on 22-Feb-2009 at 04:13 PM. Last edited by bernd_afa on 22-Feb-2009 at 04:11 PM.
|
|
Status: Offline |
|
|
bernd_afa
| |
Re: Performance of the SAM440ep Posted on 23-Feb-2009 18:43:27
| | [ #69 ] |
|
|
|
Cult Member |
Joined: 14-Apr-2006 Posts: 829
From: Unknown | | |
|
| I look what the greenhills compiler are that are used for the benchmarks.
http://www.ghs.com/products/optimizingC++EC++Compilers.html
"""" Green Hills Software has been the leader in embedded optimizing compilers for over 25 years. Green Hills compilers consistently outperform competing compilers in certified results on industry standard benchmarks, such as those published by Embedded Microprocessor Benchmark Consortium (EEMBC). """""
so still not be sure what happen with performance, when use GCC.Only the 440ep use the greenhill compiler.
maybe Hyperion should port the GreenHill Compiler to OS4 and sam work much faster.
http://www.arc.com/software/development/metawarecompiler.html
Here stand not that this compiler are fastest and outperform other
""" For the past 20 years, ARC™ International's MetaWare® C/C++ Compilers have leveraged ARC's compiler optimization expertise and quality code generation into the successful development and deployment of complex embedded applications. MetaWare compilers continue to be recognized for their advanced technology. """
|
|
Status: Offline |
|
|