@Zylesea

Quote:

RAM access is faster on X1000, while cpu powerwise the PowerMac is leading, most cache things are close by each other. From those numbers it looks like both systems are probably more or less on par. The calculation intesive stuff will probably run faster on a PowerMac G4 while the throuput intensive stuff will be better off with an X1000.

Will see how RageMem numbers will do with full X1000 support and RadeonHD drives, as well as ... second core when possible

However, X1000 is still nice target machine for MorphOS too

_________________
Future Acube and MOS supporter, fi di good, nothing fi di unprofessionals. Learn it harder way!

@vox

Quote:

vox wrote: @Zylesea Quote: RAM access is faster on X1000, while cpu powerwise the PowerMac is leading, most cache things are close by each other. From those numbers it looks like both systems are probably more or less on par. The calculation intesive stuff will probably run faster on a PowerMac G4 while the throuput intensive stuff will be better off with an X1000. Will see how RageMem numbers will do with full X1000 support and RadeonHD drives, as well as ... second core when possible However, X1000 is still nice target machine for MorphOS too

Technically spoken it is a nice target machine, but MorphOS team decided to support more cost sensitive solutions. Both approaches are okay and have their pros and cons. I wouldn't mind if MorphOS ran on the X1000. but eventually neither MorphOS needs the X1000, nor does the X1000 need MorphOS. In sum I prefer if MorphOS team's development resources are spend elsewhere.

_________________
My programs: via.bckrs.de
MorphOS user since V0.4 (2001)

@Tuxedo

Quote:

Tuxedo wrote: Plz made lame test on that PowerMac to abotain valid test values... However from MacMini values on the above test I think G4@2GHz was about around 25 secs, not better...

Requires the audio source file and lame/mp3 settings.

_________________
My programs: via.bckrs.de
MorphOS user since V0.4 (2001)

@vox

Quote:

No luck for all Blendermaniacs, even that result would be "fake"

how would that be fake? does blender employ gfx hardware to render or what? ive never used this software but its news to me. care to enlighten?

what concerns lame test it looks like we are slowly getting some expectable numbers. 3xspeed of sam460 and 1,5-2xspeed of mac mini, that figures. out of interest was such test ever run already on a g5 machine?

@Zylesea

Quote:

Technically spoken it is a nice target machine, but MorphOS team decided to support more cost sensitive solutions. Both approaches are okay and have their pros and cons. I wouldn't mind if MorphOS ran on the X1000. but eventually neither MorphOS needs the X1000, nor does the X1000 need MorphOS. In sum I prefer if MorphOS team's development resources are spend elsewhere.

Seen this arguments, and they are valid until MOS 3.0 is out .. and then several MacMini models will be supported. Logic would be to go for some new and avail PPC HW even if its Playstation 3, SAMS, X1000 ....

In some 3.x update X1000 support would be great:
a) Because X1000 users might be interested to use it = more MOS licences
b) It would be about time for MOS to be avail for some new hardware, even its not cost effective as you say. This includes SAMs
c) OS4 users previously showed interest to have MOS on SAMs just as Peg2 users did get OS4 in some extent: this approach only hurts MOS, there is no better evidence then first hand May it be 100 licences, its 100 licences
d) Then get it on G5 for high end MOS machine
e) After MOS for X1000 is out we could moan on MOS team when there will be SMP

Last edited by vox on 02-Feb-2012 at 01:52 AM.

_________________
Future Acube and MOS supporter, fi di good, nothing fi di unprofessionals. Learn it harder way!

@wawa

Quote:

how would that be fake? does blender employ gfx hardware to render or what? ive never used this software but its news to me. care to enlighten?

Because it needs 3D gfx driver, which doesn`t exist for RadeonHDs.
He would have to use software emulation of Warp3D which is CPU only, so it will not show real results of X1000 potential. Works on SAM 460 + RadeonHD but is really slow.

_________________
Future Acube and MOS supporter, fi di good, nothing fi di unprofessionals. Learn it harder way!

@vox

Quote:

second core when possible

Please stop talking about dual core not going to happen for years if ever...

@vox

but it doesnt need it for rendering??? and the blender benchmark is rendering benchmark as far as i understand? am i wrong?

@vox

Quote:

e) After MOS for X1000 is out we could moan on MOS team when there will be SMP

And the hits keep on coming..

what you smoking there chief?

@realize

Quote:

@vox Quote: second core when possible Please stop talking about dual core not going to happen for years if ever...

Your starting to sound like kas1e. I suggest you lay back and enjoy your new os4 machine, no-one really knows when its become available. Atm its pure speculation. Remember its only February.

_________________

@realize

Quote:

And the hits keep on coming.. what you smoking there chief?

About 2nd core: hardware is there. It would be shame not to use it.

Well, I don`t smoke tobacco, only plain ganja and nothing less or more. But that is not a reason for last line of humor. Idea was to parody the MOS users moan on "second core will never be used" or "overclocked G4 is better".

Enjoy the good news, support it if you want, just don't spoil it.

Yes, we all do know drivers are missing and things can only improve from now.

_________________
Future Acube and MOS supporter, fi di good, nothing fi di unprofessionals. Learn it harder way!

@wawa

Quote:

but it doesnt need it for rendering??? and the blender benchmark is rendering benchmark as far as i understand? am i wrong?

You are wrong in testing possibility, not in thinking. Blender needs 3D to run but here Wazp3D is heavily burdening the CPU which will not be the case when RadeonHD is supported (as they say inWazp3D info = o Can be software only (=slow) : no need to have 3D hardware). That is currently the X1000 case.

That is why one of the Wazp3D features was
Bug fixes to make Blender/MiniGL/Warp3Dv5/OS4 works
... So Blender now works on Os4 machine without 3D-drivers
(Just remember that Wazp3D is limited to 1024x1024 screens)

Last edited by vox on 02-Feb-2012 at 01:51 AM.

_________________
Future Acube and MOS supporter, fi di good, nothing fi di unprofessionals. Learn it harder way!

@vox

you mean, gui costs unnecassary cpu time while it renders? okay, that might be, your point.

Another benchmark program, "Stream" on Depot, a memory bandwidth test.

http://os4depot.net/index.php?function=showfile&file=utility/benchmark/stream.lha

AmigaONE (512MB RAM, PowerPC 750FX @ 800Mhz
-------------------------------------------------------------
Function.....Rate (MB/s)----Avg time-----Min time-----Max time
Copy:.........160.0353-------0.2066-------0.2000-------0.2276
Scale:........145.7387-------0.2291-------0.2196-------0.2461
Add:..........151.5286-------0.3204-------0.3168-------0.3415
Triad:........154.9917-------0.3218-------0.3097-------0.3364

(Athlon XP 3000+ @ 2.2Ghz , 1GB RAM
-------------------------------------------------------------
Function....Rate (MB/s)---- Avg time----Min time-----Max time
Copy:........2133.3333-------0.0208-------0.0150-------0.0320
Scale:.......2133.3431-------0.0243-------0.0150-------0.0320
Add:.........1548.3923-------0.0313-------0.0310-------0.0320
Triad:.......1548.3923-------0.0312-------0.0310-------0.0320

PowerMac G5 (1500MB RAM, PowerPC G5 @ 1800Mhz, Max OS X 10.4.8)
-------------------------------------------------------------
Function....Rate (MB/s)-----Avg time-----Min time-----Max time
Copy:........1898.8954-------0.0173-------0.0169-------0.0199
Scale:.......1880.2741-------0.0176-------0.0170-------0.0210
Add:.........2149.7767-------0.0226-------0.0223-------0.0234
Triad:.......2147.3691-------0.0226-------0.0224-------0.0236

X1000 (4GB RAM, PA6T @ 1800Mhz, OS4 update 5)
---------------------------------------------
Function... Rate (MB/s)---- Avg time---- Min time---- Max time
Copy:....... 4177.5936-------0.0077------ 0.0077------ 0.0078
Scale:...... 3865.1614------ 0.0084------ 0.0083------ 0.0085
Add:........ 4037.9998------ 0.0120------ 0.0119------ 0.0122
Triad:...... 3905.6140------ 0.0124------ 0.0123------ 0.0125

_________________
Hate tends to make you look stupid...

@sundown

Nice numbers. Totally smokes a G5 - as they should.

I see you have 4GB on your machine.
- I assume only 2GB is seen by the OS at this point?
- I assume these results are using only one of the two memory controllers?

Also, I thought the PA6T was a 2GHz part, yet your clock is 1.8GHz? Am I wrong?

Thanks

@gregthecanuck

Only 2GB is seen by os4, the PA6T sees 4GB, both controllers are working & speed up the memory.
The PA6T mil part is rated at 2Ghz, which no mortal man could afford, so it's underclocked to be safe. Remember, the thermal tests were done with both core under max load.

_________________
Hate tends to make you look stupid...

@sundown

those numbers seems to be quite good indeed!

_________________
See my blog and collection website! . https://www.blog.amigaguru.com

@sundown

Nice!!

Athlon XP 3000+ 2.2Ghz and G5 1.8Ghz is 10+ times faster A1 800Mhz,
But 1.8Ghz PA6T is 2x faster than the 2.2Ghz Athlon or 1.8Ghz G5.

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

@KimmoK

Quote:

Athlon XP 3000+ 2.2Ghz and G5 1.8Ghz is 10+ times faster A1 800Mhz,

Yup, but the A1 is a micro, not my numbers, but very close to what I did get with the 750GX @ 800Mhz.

_________________
Hate tends to make you look stupid...

@rikky
Quote:

Your starting to sound like kas1e. I suggest you lay back and enjoy your new os4 machine, no-one really knows when its become available. Atm its pure speculation.

He sounds not like kas1e, but like any other sane person who follow to amiga developmnet for years, and know how everything going and how much time its take. There is no speculation, there is reality : its unpossible to do radical changes in little timeline because of current resources. Stop putting your mind in the happy place, and check reality.

I told you all before, that gallium + opengl will take 2 or more years ! And you all start to talk about "specualtion" , and all of you saying "you do not know how fast it will be done". Now, 1,5 year gone, still no gallium there or any news about. And that is _reality_. The same will happens and with SMP. I even not sure if work on it even starts.

So, better just not annoy sane persons with that fairy stuff about "soon SMP" and be more practical. Fanboism suck.

Quote:

Remember its only February.

I.e. 2 more weeks for SMP ? :)) Blah .. when you will understand that all that unreal hopes only make things worse in end ?

@mr2, tuxedo
Quote:

http://forum.amigaone.pl/topic65.html

Seems x1000 about twice faster than macmini g4/1.5ghz, but will be intersting to see powerpac results as well.

@sundown

In your bench x1000 seems beat powermac pretty well , not bad.

Last edited by kas1e on 02-Feb-2012 at 09:19 AM.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites