Click Here
home features news forums classifieds faqs links search
5618 members 
Amiga Q&A /  Free for All /  Emulation /  Gaming / (Latest Posts)
Login

Nickname

Password

Lost Password?

Don't have an account yet?
Register now!

Support Amigaworld.net
Your support is needed and is appreciated as Amigaworld.net is primarily dependent upon the support of its users.
Donate

Menu
Main sections
» Home
» Features
» News
» Forums
» Classifieds
» Links
» Downloads
Extras
» OS4 Zone
» IRC Network
» AmigaWorld Radio
» Newsfeed
» Top Members
» Amiga Dealers
Information
» About Us
» FAQs
» Advertise
» Polls
» Terms of Service
» Search

IRC Channel
Server: irc.amigaworld.net
Channel: #Amigaworld
Channel Policy and Guidelines

(Uses JAVA Applet and Port 1024)
Visit the Chatroom Website

Who's Online
 55 guest(s) on-line.
 0 member(s) on-line.



You are an anonymous user.
Register Now!
 OneTimer1:  7 mins ago
 Vidar:  14 mins ago
 K-L:  18 mins ago
 RetroManCave:  19 mins ago
 g01df1sh:  28 mins ago
 Frank:  30 mins ago
 outrun1978:  36 mins ago
 tonyw:  36 mins ago
 kamelit0:  46 mins ago
 freak:  48 mins ago

/  Forum Index
   /  Amiga OS4 Hardware
      /  A1222 and CPU multiplier
Register To Post

Goto page ( Previous Page 1 | 2 | 3 | 4 Next Page )
PosterThread
iggy 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 11:50:36
#41 ]
Super Member
Joined: 20-Oct-2010
Posts: 1175
From: Bear, Delaware USA

@AP

Quote:
...significant faster on Tabor


Well, what no one is factoring in is that even though the A1222 uses an fpu that is relatively 'weak', the rest of the Freescale core in that cpu performs better than the Applied Micro cpus used by Acube (at similar clock speeds) and its clocked slightly faster.

So in most application I expect to see things balance out.
However....'significantly faster'? Unlikely.

That kind of hyperbole has backfired when it was applied to the X5000, which regardless of tlosm's posts does NOT perform better than a PowerMac G5.
BUT, there are still a LOT of features that make the X5000 a worthwhile purchase (as I've pointed out in previous posts).

However, when you oversell something and apply a distortion field to your information, it can lead those who take your advice to future disappointment.

So a caution here, you don't need to be a fanboy. Aeon hardware has features that sell itself.
There's no need to BS it.

Last edited by iggy on 20-Jul-2017 at 11:52 AM.

 Status: Offline
Profile     Report this post  
AP 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 12:14:23
#42 ]
Regular Member
Joined: 31-Jul-2003
Posts: 374
From: Vienna/Austria

@iggy

Quote:

iggy wrote:
@AP

Quote:
...significant faster on Tabor


So in most application I expect to see things balance out.
However....'significantly faster'? Unlikely.

So a caution here, you don't need to be a fanboy. Aeon hardware has features that sell itself.
There's no need to BS it.


I am neither a fanboy nor want to create a distortion field, I am speaking of first-hand information and facts from a developer (source: my interview in AF 126).

Tower57 on SAM460 (using FPU): 30 fps
Tower57 on Tabor (optimised): 50 fps
Tower57 on Tabor (using FPU-Emulation): 20 fps

So at least in this case I would say Tabor is significant faster.

Last edited by AP on 20-Jul-2017 at 12:18 PM.
Last edited by AP on 20-Jul-2017 at 12:16 PM.

_________________

 Status: Offline
Profile     Report this post  
KimmoK 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 12:24:02
#43 ]
Elite Member
Joined: 14-Mar-2003
Posts: 5045
From: Ylikiiminki, Finland

From tlosm...
timedemo demo2, 640x480, quake warpos (software rendering)
- 138 fps on x5000/020
- 119fps on G5 Quad

Interesting. But I assume G5 was not running quake warpos?

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

 Status: Offline
Profile     Report this post  
TrevorDick 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 12:24:55
#44 ]
Elite Member
Joined: 30-Dec-2004
Posts: 2635
From: Wellington

@KimmoK

Thanks, will check this out

TrevorD

_________________
No, I don't need no reason, I'm just breezin'

 Status: Offline
Profile     Report this post  
iggy 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 12:26:51
#45 ]
Super Member
Joined: 20-Oct-2010
Posts: 1175
From: Bear, Delaware USA

@AP

Quote:
So at least in this case I would say Tabor is significant faster.


No, what you CAN say is that the developers version for the A1222 is faster.
THAT can be a issue related to the code itself, depending on what specific instructions the author is implementing.

It doesn't make a more generalized statement like yours valid.

Nor would I credit tlosm's statement about Blender being the sole valid benchmark as being anymore credible as it not optimized for that fpu, and performance will depend on what instructions the app relies on.

So yes, you are both fanboys, and the distortions you both want to present to support your causes only serve to weaken us in the long run.

Last edited by iggy on 20-Jul-2017 at 12:27 PM.

 Status: Offline
Profile     Report this post  
iggy 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 12:41:02
#46 ]
Super Member
Joined: 20-Oct-2010
Posts: 1175
From: Bear, Delaware USA

@TrevorDick

Quote:
Thanks, will check this out


Trevor, I'm not sure this constant fixation on comparisons to the G5 will really serve the market.
As long as the G5 is not supported by OS4, its a apples and oranges comparison anyway, since the closest comparison that can be made is to a G5 running MorphOS (and MorphOS doesn't run on a PCIe G5).

Comparing Linux or OSX generated benchmark figures of PCIe G5s may, in some cases, help to bolster a particular argument (they also may backfire and do the reverse), BUT the salient point is that it is NOT a relevant comparison.

AND, I wouldn't necessarily want to see an OS4, X5000 vs G5 comparison, because each platform would have some advantages, but the fact that the X5000 is the only one of the two that is available new makes it the obvious clear winner.
And the distinct advantages the X5000 would hold, other than being available new? DDR3 vs DDR2, faster SATA, more modern PCIe slots, etc...

Any further exploration of the comparison between these two relatively comparable platforms only serves to refocus people on the pricing disparity.
And that is a unfair comparison because you are comparing used hardware over ten years old with new hardware that is readily available.

Last edited by iggy on 20-Jul-2017 at 12:42 PM.

 Status: Offline
Profile     Report this post  
AP 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 12:45:22
#47 ]
Regular Member
Joined: 31-Jul-2003
Posts: 374
From: Vienna/Austria

@iggy: It would be nice, If you read my comments carefully before you answer.

I never made a general statement. I quote myself:

„at least Tower57 (which makes heavy use of FPU) is significant faster on Tabor (in comparison with SAM 460 and with optimized code for Tabor).“ ("at least Tower57“ is the key)

and

"So at least in this case I would say Tabor is significant faster. (again: I wrote „at least in this case“)

BTW: I answered to pavlor, who wrote, that it is impossible that Tabors „FPU“ is on par with SAM460ex. FPU. At least im my example Tabor is faster, so I don´t think Tabor will be slower than a SAM460ex in general (when you use code written for Tabor´s FPU-unit).

I suggest to leave your own distortion-field.

_________________

 Status: Offline
Profile     Report this post  
iggy 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 13:01:34
#48 ]
Super Member
Joined: 20-Oct-2010
Posts: 1175
From: Bear, Delaware USA

@AP

Quote:
"So at least in this case I would say Tabor is significant faster."


Quote:
I suggest to leave your own distortion-field.


Yes, you have a point, you included a qualifier.
My apologies.

And its important to note that this will be the issue with each package that will be made available for the A1222.

How well is it optimized for the processor?

That processor does have a fpu with less functionality, but in the long run, will that make a difference in packages written for it?
I'd assume the authors would do the best they can with what is available.

Then again, it WILL make comparisons difficult, as performance difference between platforms will relate to differences in code.

SO, if I take the liberty to abuse your statement by taking part of it out of context and only considering "...Tabor is significantly faster"...
Its primarily because your focus would appear (at least to me) to be the use a selected case to make a generalized argument (which isn't valid).
And the addition of an easily overlooked qualifier only serves to make you look like a lawyer.

Last edited by iggy on 20-Jul-2017 at 01:13 PM.
Last edited by iggy on 20-Jul-2017 at 01:12 PM.

 Status: Offline
Profile     Report this post  
AP 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 13:13:06
#49 ]
Regular Member
Joined: 31-Jul-2003
Posts: 374
From: Vienna/Austria

I was to fast with my answer, answer is following...

Last edited by AP on 20-Jul-2017 at 01:14 PM.

_________________

 Status: Offline
Profile     Report this post  
iggy 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 13:17:33
#50 ]
Super Member
Joined: 20-Oct-2010
Posts: 1175
From: Bear, Delaware USA

@AP

Quote:
Don´t quote it, do it


I like that.

Hey...as to quotes, how about Three Dog Night's "Mama told me...not to come".
Nice double entendre.

"In this case..."

Ah, God! We all all beginning to sound like spin doctors.

 Status: Offline
Profile     Report this post  
iggy 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 13:22:37
#51 ]
Super Member
Joined: 20-Oct-2010
Posts: 1175
From: Bear, Delaware USA

@AP

Don't sweat it, I get your point.
The qualifier is valid.
But it will be interesting to see how people compensate for this 'feature'.

Potentially it could work out well in packages where the author works to optimize code for the platform.
But for packages that aren't optimized?
They will be at a disadvantage or they may not run at all (unless Hyperion does produce an OS variant that traps unsupported calls).

In any case, we are beating to death an issue that isn't going to be clearly understood until this hardware is in the hands of more users.

 Status: Offline
Profile     Report this post  
AP 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 13:33:12
#52 ]
Regular Member
Joined: 31-Jul-2003
Posts: 374
From: Vienna/Austria

@AP
Quote:
Yes, you have a point, you included a qualifier.
My apologies.


No problem.

Quote:
And its important to note that this will be the issue with each package that will be made available for the A1222.


Agree. There are even cases, when Tabor is slower than a SAM460ex (every time when software uses a FPU and Tabor has to emulate this FPU). But you can´t say (like pavlor) that Tabor isn´t on par with SAM460ex in general.


Quote:
And the addition of an easily overlooked qualifier only serves to make you look like a lawyer.


That was clearly not my intention. As English isn´t my fist language I tent to build "German style"-sentences...

_________________

 Status: Offline
Profile     Report this post  
iggy 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 13:48:48
#53 ]
Super Member
Joined: 20-Oct-2010
Posts: 1175
From: Bear, Delaware USA

@AP

German tends to be so much more compact and concise.
I can't imagine the challenges that must be presented trying to translate it into something as sloppy as a 'mutt language' like English.

The primary problem with English is when they don't have a precise way to word something, they invent totally new phrases to convey the idea.

Definitely the lawyers choice of languages.

In any case, I don't really feel comfortable with all this SAM460 vs. A1222 comparison, because the SAM460 isn't available.
Its the same problem with the G5 vs. X5000 arguments.

One you can buy, the other you have to find used (and good luck finding a SAM460...just because used APPLES can be found everywhere for peanuts... ).

But again, you have a valid point, in some application this will be a fast solution.
That will depend largely on the talent the coder.

 Status: Offline
Profile     Report this post  
Rob 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 14:46:23
#54 ]
Elite Member
Joined: 20-Mar-2003
Posts: 5564
From: S.Wales

@iggy

Quote:
In any case, I don't really feel comfortable with all this SAM460 vs. A1222 comparison, because the SAM460 isn't available. Its the same problem with the G5 vs. X5000 arguments.


Sam460 owners might want to know if it could be considered as an upgrade or not.

 Status: Offline
Profile     Report this post  
tlosm 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 14:50:25
#55 ]
Elite Member
Joined: 28-Jul-2012
Posts: 2302
From: Amiga land

@KimmoK

quake software rendering on osx leopard

Last edited by tlosm on 20-Jul-2017 at 02:50 PM.

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32; A4k.
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
AmigaOne X5000/40 16GB

 Status: Offline
Profile     Report this post  
tlosm 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 14:54:40
#56 ]
Elite Member
Joined: 28-Jul-2012
Posts: 2302
From: Amiga land

@iggy

X5000/20 is faster than the G5 single thread in integer, but is slower in fpu.
but the kernel on x5000 is not optimized many things need to be been fixed on x5000 linux kernel.
only 2 persons work on x5000 linux kernel and in free time (im sure and expecting better performances on x5000 compared today in future) . On G5 was working many ppl and this performances on G5 are after 12 years of kernel fixing.
X5000 linux kernel is not mainstream ... G5 yes

Last edited by tlosm on 20-Jul-2017 at 02:58 PM.
Last edited by tlosm on 20-Jul-2017 at 02:56 PM.
Last edited by tlosm on 20-Jul-2017 at 02:54 PM.

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32; A4k.
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
AmigaOne X5000/40 16GB

 Status: Offline
Profile     Report this post  
iggy 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 15:31:31
#57 ]
Super Member
Joined: 20-Oct-2010
Posts: 1175
From: Bear, Delaware USA

@tlosm

Quote:
X5000/20 is faster than the G5 single thread in integer, but is slower in fpu.


Yes, I get that, and frankly even though the P5020 is at a clock speed disadvantage when compared to the higher speed G5s, I still see a fair level of parity there.
Plus, you can buy a NEW X5000. I point I can't repeat often enough.

The fpu in the PPC970 performs pretty well as it was the last PowerPC developed by IBM (for desktop use), and their intent was to make it perform better than X86 cpus of that era.
I'm not sure the difference in performance between that and the P50xx is all that significant.
And you're right, the integer performance of the e5500 cored cpus is better (and the e6500 cored cpus are better still - at the same clock speeds).

So while it draws the ire of some of my friends in the MorphOS community, I'm all in favor of supporting NEW hardware. Especially when it is designed for the benefit of our market.

It will get even better with the P5040 powered X5000/40 as it could be powered by an even faster cpu (I'm not aware of what Aeon's intentions are in that regard, but they are available).

Plus, as you've pointed out, our Linux distributions will only get better with time.
And this new hardware is helping drive support for big endian PowerPC Linux distributions.
Freescale cpus can support little endian mode, but they lack the Vector and Scalar Unit (VSU) that Power 8 and 9 cpus have, making them incompatible with future releases of little endian Linux.

So what do you do?
Well...with Linux, you gather a group together interested in continuing support for the ISA and you continue on.
Its how the 68K still gets Linux support.
Apple PPC hardware, Aeon PPC hardware, older AmigaOnes, the proposed T2080 based laptop, all communities/projects that will benefit from that focus.

Hey guys, we're helping Apple diehards with this (not sure how I feel about that ).

Last edited by iggy on 20-Jul-2017 at 03:33 PM.
Last edited by iggy on 20-Jul-2017 at 03:32 PM.
Last edited by iggy on 20-Jul-2017 at 03:32 PM.

 Status: Offline
Profile     Report this post  
pavlor 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 15:48:41
#58 ]
Elite Member
Joined: 10-Jul-2005
Posts: 8415
From: Unknown

@AP

Quote:
at least Tower57 (which makes heavy use of FPU)


Well, pure FPU benchmarks like lame and blender show performance in the G3 600 MHz league. I don´t know ratio of FPU/integer code in Tower 57, but higher integer performance of e500 may play role here. Once I get A1222, I will make many benchmarks myself, I don´t expect my estimate will be wrong...

I think useful benchmark for potential Tabor buyers would be Quake 3 performance with regular PPC binary. However, I don´t think we will know answer to this one befere the boards are on sale.

 Status: Offline
Profile     Report this post  
bennymee 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 17:16:39
#59 ]
Regular Member
Joined: 19-Aug-2003
Posts: 343
From: Netherlands

@pavlor

Regular PPC binairy, you mean not a new compiled SPP-version ?

 Status: Offline
Profile     Report this post  
AP 
Re: A1222 and CPU multiplier
Posted on 20-Jul-2017 17:49:49
#60 ]
Regular Member
Joined: 31-Jul-2003
Posts: 374
From: Vienna/Austria

@pavlor

Quote:

Well, pure FPU benchmarks like lame and blender show performance in the G3 600 MHz league. I don´t know ratio of FPU/integer code in Tower 57, but higher integer performance of e500 may play role here. Once I get A1222, I will make many benchmarks myself, I don´t expect my estimate will be wrong...


According to the coder of Tower57 the game uses FPU very much (KI), but I agree that there could be other factors, too. I don´t expect, that Tabor is a powerhorse, but I expect that its overall-performance is at least on par with SAM460ex. And than there is the second CPU-core, which will hopefully be used in AmigaOS4 sometime in the future. If the prize is right, It could be my return to AmigaOS4. And benchmarks from your side are very welcome!

Quote:

by pavlor on 20-Jul-2017 15:48:41

@AP

Quote:
at least Tower57 (which makes heavy use of FPU)


Well, pure FPU benchmarks like lame and blender show performance in the G3 600 MHz league. I don´t know ratio of FPU/integer code in Tower 57, but higher integer performance of e500 may play role here. Once I get A1222, I will make many benchmarks myself, I don´t expect my estimate will be wrong...

I think useful benchmark for potential Tabor buyers would be Quake 3 performance with regular PPC binary. However, I don´t think we will know answer to this one befere the boards are on sale.


You mean a benchmark without using FPU? That would be indeed interesting. FPU-related software (which isn´t optimised for SPE on Tabor) is of course slower, because of the FPU-emulation. Regarding Tower57 20 fps against 30 fps on SAM460ex (on AOS4Beta).





_________________

 Status: Offline
Profile     Report this post  
Goto page ( Previous Page 1 | 2 | 3 | 4 Next Page )

[ home ][ about us ] [ forums ][ classifieds ] [ links ][ news archive ] [ link to us ][ user account ]
Copyright © 2000 - 2017 Amigaworld.net.

Amigaworld.net was originally founded by David Doyle