Overview Features Coding ApolloOS Performance Forum Downloads Products Order Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.
Please visit our Apollo-Discord Server for support.



All TopicsNewsPerformanceGamesDemosApolloVampireAROSWorkbenchATARIReleases
Performance and Benchmark Results!

Vampire2 80MHz Outclasses the AMIGAONE 800MHzpage  1 2 3 4 

Daytona 675x

Posts 32
11 Sep 2017 12:24


@Gunnar
 
 

 
And still no answer why the 1k-to-8k results of sortbench didn't make it to your performance comparison charts...

 
  This is not a clever question.
 
  Sortbench can provide thousands of different results from 1 to size of your memory.
  People can easily read a chart with 1 line.
  Drawing a chart with tens, or hundreds, or thousand lines is unreadable. So it makes sense to "pick" one value and choose it.
 

 
Actually, this is not a clever answer...
 
- the other test results of sortbench are at least as interesting in real-world as the 32k variant. Why only pick the one the Vampire "wins" against the select competiton?
 
- and why ignore valid 440ep results you definitely were aware of that change the bar order of that 32k chart?
 
- you can easily post a chart of that very same kind for 1, 4, 8, 16 and 32k, would be no problem at all.
Readability problems? Are you kidding?
 
@Hbarra 2Pi
Ah, well, not worth an answer. Oh boy...


Obetto Sannala

Posts 61
11 Sep 2017 12:33


@Gunnar @Daytona 675x

Why don't you just make a call. You're both german, right?

Or better: Just go to drink a Beer together. You both are very talented guys - some of the very best and rare developers in Amiga-Land.

Try to put the forces together. A lot of people really love what both of you do for the Amiga community.

To me, both of you are real heros in a small Amiga World.



Gunnar von Boehn
(Apollo Team Member)
Posts 6207
11 Sep 2017 12:35


Daytona 675x wrote:

  - the other test results of sortbench are at least as interesting in real-world as the 32k variant. Why only pick the one the Vampire "wins" against the select competiton?
 

Lets be clear here.
The 32K value is not the best value for the VAMPIRE.
 
If we wanted to bash all PowerPC, then we would have chosen a higher  value than 32K.
With higher than 32k values - the VAMPIRE will outrun many more PPC machines even 1.4 GHz G4-PowerPC will be beaten by the VAMPIRE.




Javier R. Santurde

Posts 16
11 Sep 2017 12:37


Obetto: I agree with all although Gunnar is not from Germany.
  Come on guys!


Daytona 675x

Posts 32
11 Sep 2017 12:39


@Gunnar

Yes, and it loses in the lower value tests, which as being said more than once, are at least as important (or even more) in real-world apps.

If you want to outrule any concerns that you beautify your stats, then you should also present values that are not so good on current Vampire.

@Gunnar
@Obetto Sannala

That's actually a nice idea.
You can get my phone number from Kymon :)



Gunnar von Boehn
(Apollo Team Member)
Posts 6207
11 Sep 2017 12:41


Daytona 675x wrote:

That's actually a nice idea.

My phone number is visible on this website click on CONTACT


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
11 Sep 2017 12:46


Daytona 675x wrote:

@Gunnar
Yes, and it loses in the lower value tests, which as being said more than once, are at least as important (or even more) in real-world apps.

It goes without saying that in every chart you have to focus on something.

We have posted results of all values - nothing is hidden or kept secret.
I agree with you that more results are allways better.
MINIBENCH is very insightful as it does hundreds of tests and clearly shows all results. But for many people having "hundreds" of scores to look at is not clearer at all, than having 1 score.


Javier R. Santurde

Posts 16
11 Sep 2017 13:13


"Obetto: I agree with all although Gunnar is not from Germany."
Oh man... I was not right. I dont know what I was thinking.
Anyway, looks like you both, gunnar and daniel, can do great things for the amiga. And a beer is always a good start ;)


Daytona 675x

Posts 32
11 Sep 2017 14:05


We just had a pretty long, nice and prosper phone-session :)
Yes, it's true: it's easy to get wrong impressions through forums.
The beer will have to wait until the Amiga32 event in Neuss though ;)



Thellier Alain

Posts 141
11 Sep 2017 14:21


Time to know better who is Gunnar :

EXTERNAL LINK


Daytona 675x

Posts 32
11 Sep 2017 14:33


@Alain Thellier
 
  ??
  I guess everybody knows who Gunnar is. Wonder what this has to do with the discussion here. And just in case you want to imply anything:
  nobody questioned the quality of the Vampire or whatever, only the quality of some benchmarks respectively the results presented...


David Wright

Posts 373
11 Sep 2017 14:41


Everything posted lately is all about speed. It's nice to see for development purposes, but many of us are interested in the other features.

Once this goes rgb through hdmi and aga is added it will be a much more complete system. I know those are priority but it doesn't get much mention with all the speed demons here.

At that point I really want to install my Vampire in the A1000.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
11 Sep 2017 14:43


Daytona 675x wrote:

I guess everybody knows who Gunnar is.

I'm just an average guy who likes AMIGA and tries to do his part to revive the dream.
And I think WE ALL share this motivation.


Daytona 675x

Posts 32
11 Sep 2017 14:52


Absolutely, a big Amen to that ;)


Olaf Schoenweiss

Posts 690
11 Sep 2017 15:10


good that it is sorted out ;)


Thierry Atheist

Posts 644
11 Sep 2017 15:54


Daytona 675x wrote:
Very nice results for the Vampire, great work!
But...

I'm a Vampire fan. I got one myself and I love it.
But I am no fan of "beautified", imcomplete or simply wrong benchmarks.

Let's check out this one here and the claim you made.
I have no 800 MHz AONE, but I have a 667 MHz sam440ep.


Hi Daytona 675x,

Are you perfectly aware of the AmigaOne SE and XE models?

The AmigaOne XE was sold with a CPU that was rated for 933 MHz and yet we were always told to leave it running at the stock 800MHz speed that was set on the motherboard right from when it arrived in the mail!

Also, the RAM was sdram which you could use in speeds of 100 or 133 MHz.

The Articia S was rather problematic.


Daytona 675x

Posts 32
11 Sep 2017 16:12


Yes, I am, I know perfectly well that the memory controller of those AONE models sucks hard.

But as I said I am also perfectly aware of the fact that the original stream build executable for ppc delivers wrong values on the machines I got (also the pretty crappy sam440ep), due to absolutely sub-optimal code produced by an antique compiler.

Therefore I provided a fresh (and otherwise unmodified) build that produces much more correct data and I suggest to re-measure using this one.

So if anybody with such a machine would be so kind?
EXTERNAL LINK 
And to make it clear: I don't care who "wins", in contrast to others I am no militant Amiga-flavour-fanboy (as you might know I always try to support AOS / MOS and AROS with my progs; and if possible I'll also support the Vampire, after all that's the reason I bought one :) ).

If it turns out that the AONE doesn't deliver better results with the new build, fine with me. I just want to know and suggest others to be equally curious.


Thomas Blatt

Posts 200
11 Sep 2017 16:15


Hey Daytona,see you in Neuss, i will come with my Amiga600+Vampire2.


Thellier Alain

Posts 141
11 Sep 2017 20:03


>I guess everybody knows who Gunnar is
Please dont overeact :-)
That wasnt supposed to be offending in any way
I just found this interview that I found interesting because Me I wasnt knowing especially who is Gunnar

Alain Thellier


Vojin Vidanovic

Posts 770
11 Sep 2017 20:44


Daytona 675x wrote:

Ah, I actually got an G4, a PowerMac 3,5 G4 733 MHz with MorphOS to be exakt.
  Here's the results:
 
  -------------------------------------------------------------
  Function      Rate (MB/s)  Avg time    Min time    Max time
  Copy:        407.5924      0.0790      0.0785      0.0798
  Scale:        413.6547      0.0780      0.0774      0.0801
  Add:          472.6132      0.1021      0.1016      0.1031
  Triad:        473.2230      0.1020      0.1014      0.1024
  -------------------------------------------------------------
  Solution Validates
  -------------------------------------------------------------

Here are PA Semi results
CPU: P.A. Semi PWRficient PA6T-1682M B1 @ 1800 Mhz
Caches Sizes: L1: 64 KB - L2: 2048 KB - L3: none
Cache Line: 64

3.RAM Disk:> bustest CHIP/S,FAST/S ROM/S
BusSpeedTest 0.19 (mlelstv) Buffer: 262144 Bytes, Alignment: 32768
========================================================================
memtype addr op cycle calib bandwidth
chip $62368000 readw 13.3 ns normal 149.9 * 10^6 byte/s
chip $62368000 readl 14.6 ns normal 274.1 * 10^6 byte/s
chip $62368000 readm 17.4 ns normal 229.6 * 10^6 byte/s
chip $62368000 writew 13.3 ns normal 150.3 * 10^6 byte/s
chip $62368000 writel 13.9 ns normal 286.8 * 10^6 byte/s
chip $62368000 writem 15.7 ns normal 254.8 * 10^6 byte/s
rom $00F80000 readw 757.7 ns normal 2.6 * 10^6 byte/s
rom $00F80000 readl 786.8 ns normal 5.1 * 10^6 byte/s
rom $00F80000 readm 744.8 ns normal 5.4 * 10^6 byte/s

Stream OS4 New
-------------------------------------------------------------
Function Rate (MB/s) Avg time Min time Max time
Copy: 2672.4355 0.0121 0.0120 0.0124
Scale: 3976.1147 0.0082 0.0080 0.0083
Add: 3650.9909 0.0145 0.0131 0.0187
Triad: 3775.0388 0.0128 0.0127 0.0130

Stream OS4 Old
-------------------------------------------------------------
Function Rate (MB/s) Avg time Min time Max time
Copy: 4144.5692 0.0079 0.0077 0.0080
Scale: 3957.9407 0.0083 0.0081 0.0084
Add: 3602.7736 0.0140 0.0133 0.0182
Triad: 3743.2432 0.0130 0.0128 0.0132

RAGEMEM v0.37 - compiled 11/06/2010

---> CPU <---
MAX MIPS: 3082

---> L1 <---
READ32: 6845 MB/Sec
READ64: 13603 MB/Sec
WRITE32: 6844 MB/Sec
WRITE64: 13668 MB/Sec

---> L2 <---
READ32: 3269 MB/Sec
READ64: 4884 MB/Sec
WRITE32: 2543 MB/Sec
WRITE64: 4086 MB/Sec

---> RAM <---
READ32: 2857 MB/Sec
READ64: 3994 MB/Sec
WRITE32: 2728 MB/Sec
WRITE64: 3385 MB/Sec
WRITE: 351 MB/Sec (Tricky)

---> VIDEO BUS <---
READ: 115 MB/Sec
WRITE: 161 MB/Sec


posts 70page  1 2 3 4