@cdimauro

no only application ...(real?0_0) because if you need benchmark is time for buy something for clear your curiosity buy an x5000 when it is done and not wait some one do this for you ;)


Now is time to post another time the video of Aros running in qemu for pavlor i think he loose the thread because many post ot especially from some one who like do this
@pavlor i know you like aros i did this for you :P

https://www.youtube.com/watch?v=CEQ4MEOpdjY

Last edited by tlosm on 10-Oct-2014 at 12:06 AM.
Last edited by tlosm on 10-Oct-2014 at 12:05 AM.
Last edited by tlosm on 09-Oct-2014 at 11:58 PM.

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

Application = software. So games are included.
Real = no synthetic benchmarks like Ragemem, Sysinfo, et similia.
only CPU = stress only the CPU, minimizing the impact of other components (graphic card, I/O) = do as much as possible running on the CPU (no hardware acceleration).

Quote:

tlosm wrote: but in any way .. the G5 have a really good performances for a 2005 machine with a 2005 hacked video card :P

true

_________________

@cdimauro

e5500 core has much higher DMIPS/MHz than PA6T (3.0 vs 2.2). I don´t think AltiVec will speed up tasks like emulation.

Quote:

Clock for clock, an e5500 core has a lower performance on average on single core/thread compared to the X1000's CPU

Source for such claim? I wonder because I didn´t find any e5500 core benchmarks.

@pavlor

Faster Integer better emulation ... right words .
Im pretty sure on X5000 on LinuxPPC i will have much faster Qemu machine compared my G5 probably i will have a qemu faster like VirtualPc
... and im more sure Mol will much faster on one core than my G5 on single core

i found only this for the happyness of cdimauro
Bho

Last edited by tlosm on 10-Oct-2014 at 04:50 PM.
Last edited by tlosm on 10-Oct-2014 at 04:25 PM.
Last edited by tlosm on 10-Oct-2014 at 04:23 PM.

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

@tlosm

I think a quad 2.7 GHz G5 is still faster than a P5040.

_________________

@WolfToTheMoon

Quote:

I think a quad 2.7 GHz G5 is still faster than a P5040.

Over-clocked monster?

@tlosm

Your link contnains only some benchmarks of e6500 core. There are better (English ) presentations with more results for the same core. Still nothing about e5500.

@WolfToTheMoon

Quad 2.5 ghz is from beginning a overclocked 2.3 ghz G5 this why it need a water cooling system and not air system only
i know there are 2.7 and 3.0 ghz dual G5 but i dont think are much faster than the same cpu of G5 quad. 2.5 ghz
the 2.7 and 3.0ghz are different from the chip of the quad, like you comparing a P4 3.0ghz with a I5 2.3 ghz (is a just an example)

you can check my words here .

http://www.primatelabs.com/blog/2008/06/mac-performance-june-2008/

note my G5 Quad on Geekbench did 3600points in 64 bit mode. 3400 in 32bit

@pavlor

I post that in jap. for cdimauro ;)

Last edited by tlosm on 10-Oct-2014 at 07:11 PM.
Last edited by tlosm on 10-Oct-2014 at 07:07 PM.
Last edited by tlosm on 10-Oct-2014 at 07:06 PM.
Last edited by tlosm on 10-Oct-2014 at 07:06 PM.
Last edited by tlosm on 10-Oct-2014 at 07:05 PM.
Last edited by tlosm on 10-Oct-2014 at 07:03 PM.

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

however here there are some spec about e5500

_________________

@pavlor
a good benchmark for linux platform (hope there is for ppc) im too courious to check because debian kernel is in 64 bit (macos x is 32) and i can say for sure there everything is much much more faster compared macosx

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

Quote:

Seiya wrote: Quote: tlosm wrote: but in any way .. the G5 have a really good performances for a 2005 machine with a 2005 hacked video card :P true

False. Take a look at the benchmarks that I reported.

Quote:

pavlor wrote: @cdimauro e5500 core has much higher DMIPS/MHz than PA6T (3.0 vs 2.2).

That's what Freescale states, but doesn't explain how she can got it, since an e5500 core can decode 2 instructions per cycle, and can retire also 2 per cycle. See below.
Quote:

I don´t think AltiVec will speed up tasks like emulation.

Not entirely, but can be used to speedup MOVE16 instructions, for example.
Quote:

Quote: Clock for clock, an e5500 core has a lower performance on average on single core/thread compared to the X1000's CPU Source for such claim?

Me. See below.
Quote:

I wonder because I didn´t find any e5500 core benchmarks.

Neither I did. Freescale generically claims a 20% speed improvement over a e500mc core, but it's highly questionable how it got such data.

However if you take a look at the e5500 core and the PA6T one, you can make a guess like the one that I did.

http://cache.freescale.com/files/64bit/doc/white_paper/64BTTCHNLGYWP.pdf
http://cache.freescale.com/files/64bit/doc/fact_sheet/QP5020FS.pdf
http://www.realworldtech.com/pa-semi/3/

The e5500 can only decode 2 instructions and retire 2, has 32K+32K L1 caches, 512KB L2 cache per core, and 2MB L3 cache shared. It has 1 unit for branches, 1 for complex instructions, 2 for simple integer ones, and 1 for the FPU. There's no data about the branch prediction unit, but it has the advantage of a shorter pipeline, so branch mispredictions aren't that bad.
The PA6T can decode/issue 3 and retire 4, has 64K+64K L1 caches, and 2MB shared L2 cache. It has 1 branch/integer instructions unit, 1 integer, 1 FPU, and 2 Altivec. It has also a strong branch prediction unit, due to its longer pipeline.

Summing it up, the conclusion is that the latter is much better in terms of performance for single core/thread (on average, of course), as I previously stated. And it also clear that 3MIPS per Mhz isn't achievable, with a core which can only decode and retire 2 instructions per cycle...

Quote:

tlosm wrote: @pavlor Faster Integer better emulation ... right words .

Only if you have a core which is able to provide good performance on "integer" calculations. But see above: the PA6T is MUCH better for this.
Quote:

Im pretty sure on X5000 on LinuxPPC i will have much faster Qemu machine compared my G5 probably i will have a qemu faster like VirtualPc ... and im more sure Mol will much faster on one core than my G5 on single core

The e5500 has less performance than the PA6T, and the latter isn't better than the G5, so you can continue to use your PowerMac G5 and have better performance with Qmu.
Quote:

i found only this for the happyness of cdimauro Bho

I don't see why I could be happy: it says nothing...

Quote:

WolfToTheMoon wrote: @tlosm I think a quad 2.7 GHz G5 is still faster than a P5040.

Right.

Quote:

pavlor wrote: @tlosm Your link contnains only some benchmarks of e6500 core. There are better (English ) presentations with more results for the same core. Still nothing about e5500.

I don't expect a different picture for the e6500, except that it finally has an Altivec unit, so it'll be "on-par" with PA6T (but performance should be tested: Altivec implementations aren't the same).

@cdimauro

your benchmark reported are like you ...
i reported better

http://www.primatelabs.com/blog/2008/06/mac-performance-june-2008/

if you know about powerpc cpu axpecially about G5 you will know for sure that benchmark reported by some one are made with some program for sure did for the G4
and think this "can be possible apple did something wrong with first mac with G5?" check the benchmark i had been post ...

If that bench was did with progams compiled for G5 you will have totally different results..
NOW STOP SPOIL THREADS and OPEN A NEW ONE .. with your considerations.

Last edited by tlosm on 10-Oct-2014 at 07:53 PM.

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

@cdimauro

Quote:

Me.

Quote:

I don't expect a different picture for the e6500, except that it finally has an Altivec unit, so it'll be "on-par" with PA6T (but performance should be tested: Altivec implementations aren't the same).

There are Coremark scores for 970, PA6T and e6500

970 1800 MHz: 3988.83 CM/thread, 2.22 CM/MHz
PA6T-1682M 1.7GHz: 2748.9 CM/thread, 1.61 CM/MHz (result was on Coremark website, removed since then (??) )
T4240 1800 MHz: 7490.13 CM/thread, 4.16 CM/MHz.

Sure, Coremark is not ideal benchmark (some scores are "strange"), but there aren´t many performance tests of e6500.

Quote:

tlosm wrote: @cdimauro your benchmark reported are like you ...

Please, better elaborate: how are my benchmarks? And what's your opinion about me? Be clear.
Quote:

i reported better http://www.primatelabs.com/blog/2008/06/mac-performance-june-2008/

So what?
Quote:

if you know about powerpc cpu axpecially about G5 you will know for sure that benchmark reported by some one are made with some program for sure did for the G4

I don't understand your not so good English. Please, can you clarify? Thanks.
Quote:

and think this "can be possible apple did something wrong with first mac with G5?" check the benchmark i had been post ...

I already saw them. So what?
Quote:

If that bench was did with progams compiled for G5 you will have totally different results..

It depends on the application. Emulation is something which hurts the G5, due to some instructions that can be useful and are used. But I really doubt that the tested applications use them, at least not on the hot spots.

But if you have more precise information about this, you can report it.
Quote:

NOW STOP SPOIL THREADS and OPEN A NEW ONE .. with your considerations.

Writing in upper case is against the netiquette, and you should know it. So, be quite, and be polite please: that's not your forum.

Second, and more important, if you don't want to continue discussing, you should be the first one to do NOT reply providing information and reporting link.

Third, other people seems to be interested on the performance discussion, since it also apply to emulation. Especially what I reported in my previous message.

So, you have to stop complaining. Again: be quite and polite.

@cdimauro

This thread is Emulation Passion ... can you made a new theread ?
About Netiquette

Quote:

MUCH better for this.

this is you



I repeat you are usually spoil tread all around the world with you convinctions ...
OPEN (ops) a new Thread ...

Quote:

don't understand your not so good English. Please, can you clarify? Thanks.

Sorry this is my english and when you will understand something about PPC cpus and code optimizations probably i will clarify
Now again Open a New Tread
"Cdimauro convintions"

..

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

Quote:

pavlor wrote: @cdimauro Quote: I don't expect a different picture for the e6500, except that it finally has an Altivec unit, so it'll be "on-par" with PA6T (but performance should be tested: Altivec implementations aren't the same). There are Coremark scores for 970, PA6T and e6500 970 1800 MHz: 3988.83 CM/thread, 2.22 CM/MHz PA6T-1682M 1.7GHz: 2748.9 CM/thread, 1.61 CM/MHz (result was on Coremark website, removed since then (??) ) T4240 1800 MHz: 7490.13 CM/thread, 4.16 CM/MHz. Sure, Coremark is not ideal benchmark (some scores are "strange"),

Exactly.
Quote:

but there aren´t many performance tests of e6500.

Let's wait for real applications benchmarks.

Quote:

cdimauro wrote: False. Take a look at the benchmarks that I reported.

mmh..he said another thing.
G5 in 2014 is yet a good cpu to run some software and emulate some systems with good satisfaction.

Last edited by Seiya on 10-Oct-2014 at 08:40 PM.
Last edited by Seiya on 10-Oct-2014 at 08:39 PM.

_________________

Quote:

tlosm wrote: @cdimauro This thread is Emulation Passion ... can you made a new theread ?

For sure, but some people already talked about performance. This is you: http://amigaworld.net/modules/newbb/viewtopic.php?mode=viewtopic&topic_id=39020&forum=8&start=0&viewmode=flat&order=0#734081

Is "nbench" an emulator? Please, clarify.
Quote:

About Netiquette Quote: MUCH better for this. this is you

Yes, but do you know the difference between emphasizing words and yelling?
Quote:

I repeat you are usually spoil tread all around the world with you convinctions ...

I don't know anyone that have no personal ideas.

BTW, "convinctions" is a false friend.
Quote:

OPEN (ops) a new Thread ...

First remove YOUR (notice: I'm not yelling. Just emphasizing) OT messages when you talked of performance. Is it the "Emulation: A passion" thread, right?
Quote:

Quote: don't understand your not so good English. Please, can you clarify? Thanks. Sorry this is my english and when you will understand something about PPC cpus and code optimizations probably i will clarify

I've a good understanding, so you can continue if you want or are able to do it.

Quote:

Now again Open a New Tread "Cdimauro convintions" ..

Another false friend. And another sentence from you just for flaming.

I repeat myself: be quite and polite. This isn't your forum, where you can flame and offend the people which you don't like.

Last edited by cdimauro on 10-Oct-2014 at 08:47 PM.

Quote:

Seiya wrote: Quote: cdimauro wrote: False. Take a look at the benchmarks that I reported. mmh..he said another thing. G5 in 2014 is yet a good cpu to run some software and emulate some systems with good satisfaction.

He made a different, more generic, sentence: "the G5 have a really good performances for a 2005 machine".

But numbers speak, as I stated, and shown a different picture.