Welcome, Guest. Please login or register.

Author Topic: dnetc benchmarks  (Read 13037 times)

Description:

0 Members and 1 Guest are viewing this topic.

Offline PiruTopic starter

  • \' union select name,pwd--
  • Hero Member
  • *****
  • Join Date: Aug 2002
  • Posts: 6946
    • Show only replies by Piru
    • http://www.iki.fi/sintonen/
dnetc benchmarks
« on: February 04, 2012, 02:14:09 PM »
Even more benchmarks: This time distributed.net client benchmark result for OGR-NG and RC5-72.


sources:
Power Mac G4 1.8GHz values from Károly "Chain|Q" Balogh
PowerBook G4 1.67GHz and Mac mini G4 1.5GHz values by me
AmigaOne X1000 PA6T 1.8GHz values
Sam 460ex 1.0GHz values

AltiVec is used if present.


*Update* The original picture had Sam460 OGR-NG and RC5-72 values swapped. It's fixed now.
« Last Edit: February 04, 2012, 02:34:01 PM by Piru »
 

Offline yakumo9275

  • Sr. Member
  • ****
  • Join Date: Jun 2008
  • Posts: 301
    • Show only replies by yakumo9275
    • http://mega-tokyo.com/blog
Re: dnetc benchmarks
« Reply #1 on: February 04, 2012, 02:42:13 PM »
I'm starting to wonder if the compilers for the pa6t are just crap and its hitting cache misses/branch misses etc and nuking its pipelines. I'd expect similar results of x1000 to the other g4's when it comes to raw cpu tasks...

and it may just be the pa6t really isnt all that fantastic and meeting the hype....
--/\\-[ Stu ]-/\\--
Commodore 128DCR, JiffyDOS, Ultimate 1541 II, uIEC/SD, CBM 1902A  Monitor
 

Offline krashan

  • Full Member
  • ***
  • Join Date: Jan 2003
  • Posts: 247
  • Country: pl
  • Thanked: 1 times
  • Gender: Male
  • Hardware designer and programmer
    • Show only replies by krashan
    • Personal homepage
Re: dnetc benchmarks
« Reply #2 on: February 04, 2012, 03:52:52 PM »
Quote from: yakumo9275;679145
I'd expect similar results of x1000 to the other g4's when it comes to raw cpu tasks...

PA6T is not a G4. Its core is an independent design. And for sure it is not the 8-th wonder of the world.

Offline bbond007

  • Hero Member
  • *****
  • Join Date: Mar 2009
  • Posts: 1517
    • Show only replies by bbond007
Re: dnetc benchmarks
« Reply #3 on: February 04, 2012, 04:26:28 PM »
Quote from: Piru;679140
Even more benchmarks:
sources:
Power Mac G4 1.8GHz values from Károly "Chain|Q" Balogh
PowerBook G4 1.67GHz and Mac mini G4 1.5GHz values by me
AmigaOne X1000 PA6T 1.8GHz values
Sam 460ex 1.0GHz values

AltiVec is used if present.


*Update* The original picture had Sam460 OGR-NG and RC5-72 values swapped. It's fixed now.


Piru

I think it would be really beneficial to test the various amiga compatible platforms at 68K emulation because there is a lot of apps that will never be ported to PPC so running legacy code is a large part of what these systems will do...

Would it be possible to rerun the 68K dnetc benchmark on those machines?

Also, what is the procedure?

dnetc-68K -bench

it does several runs 68000/68030/68060 .... just take the fastest?

if so

UAE i740qm (1.7ghz) ORG-NG : 4,674,671 RC5-72 : 4,116,455
A1200/060(50mhz) ORG-NG : 251,437 RC5-72 : 141,118
Minimig 1.1 (50mhz) - unable to initialize core. (maybe not enough ram?)
 

Offline takemehomegrandma

  • Hero Member
  • *****
  • Join Date: Oct 2002
  • Posts: 2990
    • Show only replies by takemehomegrandma
Re: dnetc benchmarks
« Reply #4 on: February 04, 2012, 04:39:39 PM »
Quote from: Krashan;679156
And for sure it is not the 8-th wonder of the world.


That's for sure! :lol: At least it's faster than the Sam... ;)
MorphOS is Amiga done right! :)
 

Offline yakumo9275

  • Sr. Member
  • ****
  • Join Date: Jun 2008
  • Posts: 301
    • Show only replies by yakumo9275
    • http://mega-tokyo.com/blog
Re: dnetc benchmarks
« Reply #5 on: February 04, 2012, 05:45:32 PM »
Quote from: Krashan;679156
PA6T is not a G4. Its core is an independent design. And for sure it is not the 8-th wonder of the world.


did I say it was? I said It expected its raw compute power to be on par with the other g4s, meaning the 3 g4's cpu's in the benchmark.

my only thought is the g4 compiler pipeline is mature, the pa6t is not, and I just wonder if poor compiler support is more to blame for the initial showing of low benchmarks in areas I'd expect it to be on par with g4 class cpu's.
--/\\-[ Stu ]-/\\--
Commodore 128DCR, JiffyDOS, Ultimate 1541 II, uIEC/SD, CBM 1902A  Monitor
 

Offline freqmax

  • Hero Member
  • *****
  • Join Date: Mar 2006
  • Posts: 2179
    • Show only replies by freqmax
Re: dnetc benchmarks
« Reply #6 on: February 04, 2012, 06:22:35 PM »
Seems the military PowerPC processor used in X1000 is approximatly 30%  faster compared to an ordinary PowerPC using the same clock frequency.
 

Offline krashan

  • Full Member
  • ***
  • Join Date: Jan 2003
  • Posts: 247
  • Country: pl
  • Thanked: 1 times
  • Gender: Male
  • Hardware designer and programmer
    • Show only replies by krashan
    • Personal homepage
Re: dnetc benchmarks
« Reply #7 on: February 04, 2012, 06:29:15 PM »
Quote from: freqmax;679188
Seems the military PowerPC processor used in X1000 is approximatly 30%  faster compared to an ordinary PowerPC using the same clock frequency.
It has 30% worse benchmark results and is 30% faster? Interesting.

Offline Iggy

  • Hero Member
  • *****
  • Join Date: Aug 2009
  • Posts: 5348
    • Show only replies by Iggy
Re: dnetc benchmarks
« Reply #8 on: February 04, 2012, 07:16:12 PM »
What I'd like to see, is how well Freescale's T4 and T5 AMP processor fair in a similar comparison.

Also, to be fair, the PA6T is a very nice Soc with great PCIe support.
No other PPC based processor offers as many lanes as this processor does.
"Not making any hard and fast rules means that the moderators can use their good judgment in moderation, and we think the results speak for themselves." - Amiga.org, terms of service

"You, got to stem the evil tide, and keep it on the the inside" - Rogers Waters

"God was never on your side" - Lemmy

Amiga! "Our appeal has become more selective"
 

Offline jorkany

  • Hero Member
  • *****
  • Join Date: Sep 2006
  • Posts: 1009
    • Show only replies by jorkany
    • http://www.amigaos4.com
Re: dnetc benchmarks
« Reply #9 on: February 04, 2012, 07:38:58 PM »
Quote from: freqmax;679188
Seems the military PowerPC processor used in X1000 is approximatly 30%  faster compared to an ordinary PowerPC using the same clock frequency.


Look again once you've sobered up.
 

Offline runequester

  • It\'s Amiga time!
  • Hero Member
  • *****
  • Join Date: Oct 2009
  • Posts: 3695
    • Show only replies by runequester
Re: dnetc benchmarks
« Reply #10 on: February 04, 2012, 08:09:04 PM »
Is it possible to buy G4 processors new from anywhere these days?

Not talking about old mac's or whatever, but are there sources where factory new chips for a new computer design could be sourced, or is that a completely closed off option?
 

Offline TheDaddy

  • Hero Member
  • *****
  • Join Date: Feb 2005
  • Posts: 1154
    • Show only replies by TheDaddy
    • http://www.loriano.pwp.blueyonder.co.uk
Re: dnetc benchmarks
« Reply #11 on: February 04, 2012, 08:13:58 PM »
MOS and old Apple machines are better than AmigaOS 4 and any SAM/X1000.
 

Offline zylesea

  • Hero Member
  • *****
  • Join Date: Feb 2006
  • Posts: 638
    • Show only replies by zylesea
    • http://www.via-altera.de
Re: dnetc benchmarks
« Reply #12 on: February 04, 2012, 08:15:28 PM »
The poor performance of the PA6T could shed new light on the whole ISA shift affair by Apple. I mean, the PA6T was (inofficially) the scheduled successor of the 744x in Apple's laptops. But it performs rather worse. Not well suited for a 7447A successor laptop. ibm hat no fitting offerings for Apple and Freescale's 86xx were also not the huge step forward. No wonder why Apple looked elsewhere eventually.
But okay, Apple would have had one huge advantage over AOS4 and the X1000, tOS X has SMP support, so a Powerbook PA6T would have benefited quite a lot from the 2nd core.

Anyway, from those historical what-if games I would like to see an benchmark between a PA6T and 8641D based system. It seems to me as if Freescale's chip would have been the better choice eventually. A shame it was not used in a GP computer back when it was current.

Offline runequester

  • It\'s Amiga time!
  • Hero Member
  • *****
  • Join Date: Oct 2009
  • Posts: 3695
    • Show only replies by runequester
Re: dnetc benchmarks
« Reply #13 on: February 04, 2012, 08:19:43 PM »
Quote from: TheDaddy;679216
MOS and old Apple machines are better than AmigaOS 4 and any SAM/X1000.


Linux on my dual core is better than MOS and old Apple machines.

But that's not really the point, is it?
 

Offline zylesea

  • Hero Member
  • *****
  • Join Date: Feb 2006
  • Posts: 638
    • Show only replies by zylesea
    • http://www.via-altera.de
Re: dnetc benchmarks
« Reply #14 on: February 04, 2012, 08:20:44 PM »
Quote from: runequester;679215
Is it possible to buy G4 processors new from anywhere these days?

Not talking about old mac's or whatever, but are there sources where factory new chips for a new computer design could be sourced, or is that a completely closed off option?


Freescale still sells 744x processors, e.g 7447A:
http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=MPC7447A&tab=Buy_Parametric_Tab&nodeId=018rH38653&pspll=1&fromSearch=false

or 7448

http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=MPC7448&tab=Buy_Parametric_Tab&nodeId=018rH38653&pspll=1&fromSearch=false

Prices are high, but not completely nuts.

But G4 series rather shifted to the 86xx chip line - which unfortunately never made it into a GP computer..