@KimmoK

however i think hardinfo give strange result.

in FPU Raytracing is E8500 only double fast than AthlonXP 2500?...
i should be much much much faster..

you should try Athlon and Sempron with Blender 2.58 (it's in linux mint repository)

Last edited by Seiya on 01-May-2012 at 11:11 AM.

_________________

however X1000 with 1.8 Ghz cpu and AthlonXP 2500+ (1,8 ghz) will be very interesting to see with linux in other more complex test like Lame, Blender and mplayer.
Com'on boys :)
i think to test linux on athlon64 3500 or pentium 4 or athlon 900, to make slax distro on cd with all software that i need and usb key like hd :)

Last edited by Seiya on 01-May-2012 at 11:16 AM.

_________________

Slax 6.1.2 benchmark

Intel Core2 Duo E8500 via vmware with only 1 core enabled because don't work on real hardware (however the result are +/- the same.

Blender 2.48a (the same AmigaOS 4 version)
02:24.18

Lame 3.98.4
7s

Mplayer
24s

----------
Athlon 64 3500+ (not vmware, but native)

Blender 2.48a
04:24.22

Lame 3.98.4
14s

Mplayer
38s

-------------------------------------------
With Vmware on E8500, Linux Mint 12 with Blender 2.58:
00:44.81

E8500 native with linux mint 12 and blender 2.58:
00:40.70

Last edited by Seiya on 02-May-2012 at 11:13 AM.
Last edited by Seiya on 02-May-2012 at 11:13 AM.
Last edited by Seiya on 02-May-2012 at 11:12 AM.
Last edited by Seiya on 01-May-2012 at 09:48 PM.

_________________

G4 1.67 GHz results here:
http://www.amiga.org/forums/showthread.php?t=61493

@aorg thread (moobunny-NG?)

Because x1000 is so expensive, a lot of people can not afford it, that's why it gets so negative comments (people are bitter).

In reality, it is the fastest AOS4 HW by big margin, even when it's still in BETATESTER phase without full HW support.

Next we need the affordable HW + SW.

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

@KimmoK

Quote:

(people are bitter)

Wishful thinking.

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

@itix

I am bitter. My favorite Sam440ep speedster is getting trashed by the other Machines .
.

But this is a good CPU benchmark that can be used on all PPC hardware to get an idea of how the CPUs perform. I am looking forward to some results from Pegs,Sam460eps,A1's,more PowerPC Macs and even Yellowdog Powerstations etc.

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

@Channel_Z

Quote:

G4 1.67 GHz results here:

Interesting, I expected better values from G4 in comparison to PA6T (or even older x86 CPUs). HardInfo seems to give some strange results.

However, FPU-FFT result of X1000 looks really good.

@pavlor

Quote:

However, FPU-FFT result of X1000 looks really good.

Interesting, I expected better values from PA6T in comparison to G4 (or even older x86 CPUs).

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

To summ it up:

HardInfo benchmarks PA6T 1.8 GHz to G4 comparison:

CPU Blowfish: 1.08 GHz (PA6T is as fast as 1.08 GHz G4)
CPU Fibonacci: 1.33 GHz
CPU N-Queens: 1.76 GHz
CPU CryptoHash: 1.62 GHz

FPU FFT: 3.67 GHz
FPU raytracing: 1.51 GHz

Arithmetic mean: 1.83 GHz
Geometric mean: 1.68 GHz

@itix

Quote:

in comparison to G4 (or even older x86 CPUs).

It seems PA6T was intended for FFT demanding applications, but such great result is really surprising.

@Spectre660

Quote:

I am bitter. My favorite Sam440ep speedster is getting trashed by the other Machines .

Actually it isn't. On some benchmarks Sam440ep has better bang for the buck ratio.

Last edited by itix on 02-May-2012 at 04:50 PM.

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

@itix

Quote:

Actually it isn't. On some benchmarks Sam440ep has better bang for the buck ratio.

That would be quite rare cases (even assuming single core is used).

@pavlor

CPU Blowfish, CPU Fibonacci and FPU Raytracing. There X1000 is less than twice as fast than Sam440ep but costs 2x more.

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

@itix

Quote:

FPU Raytracing

Better look at Blender benchmark.

As Seyia wrote, HardInfo results are sometimes a little bit strange (eg. FFT or CPU Blowfish in case of PA6T). Application benchmarks show much better real speed of respective computers.

@itix

are you sure that both X1000 cores are used in these benchmarks

Last edited by klx300r on 02-May-2012 at 05:33 PM.

_________________
____________________________
c64-2sids, A1000, A1200T-060@50(finally working!),A4000-CSMKIII
! My Master Miggies- Amiga 1000 & AmigaOne X1000 !
mancave-ramblings
X1000 I BELIEVE

Another benchmark for Linux systems: http://www.netlib.org/benchmark/linpackjava/
Gave me 367Mflop/s for AMD64x2-3800+ (not sure if it's sane, and I do not know if it uses only one core or what)
Here one can compare with other chips: http://www.roylongbottom.org.uk/linpack%20results.htm
It seems that with SIMD optimizations etc the results multiply.

Last edited by KimmoK on 02-May-2012 at 05:38 PM.
Last edited by KimmoK on 02-May-2012 at 05:36 PM.

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

@klx300r

Quote:

are you sure that both X1000 cores are used in these benchmarks

Single core benchmarks of course.

Quote:

pavlor wrote: @klx300r Quote: are you sure that both X1000 cores are used in these benchmarks Single core benchmarks of course.

why of course so I'm sure someone will now adjust all those nice stats to reflect this simple fact

hmm considering the results are only using HALF of the PA6T in the X1000 I would say that's pretty good for a machine DESIGNED TO RUN AMIGAOS

_________________
____________________________
c64-2sids, A1000, A1200T-060@50(finally working!),A4000-CSMKIII
! My Master Miggies- Amiga 1000 & AmigaOne X1000 !
mancave-ramblings
X1000 I BELIEVE

@klx300r

Quote:

hmm considering the results are only using HALF of the PA6T in the X1000 I would say that's pretty good for a machine DESIGNED TO RUN AMIGAOS

Results aren´t bad. However PASemi expected performance around 900 SPECInt2000 for 1.8 GHz PA6T core, published benchmarks show score in the range of 500-600 SPECInt2000.