Poster | Thread |
pavlor
|  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 17:02:31
| | [ #201 ] |
|
|
 |
Elite Member  |
Joined: 10-Jul-2005 Posts: 9673
From: Unknown | | |
|
| @number6
I know, but I don´t hold my hope too high.  |
|
Status: Offline |
|
|
number6
|  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 17:07:32
| | [ #202 ] |
|
|
 |
Elite Member  |
Joined: 25-Mar-2005 Posts: 11689
From: In the village | | |
|
| @pavlor
Fair enough.
I just wanted anyone wishing current information from Mike regarding his work with Toni and OS4.x application developers to be part of the discussion if you wished. And I know most people are still unaware of the question and answer podcast from this prestigious list of people that were interviewed 3 days ago.
#6
_________________ This posting, in its entirety, represents solely the perspective of the author. *Secrecy has served us so well* |
|
Status: Offline |
|
|
Spectre660
 |  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 17:27:01
| | [ #203 ] |
|
|
 |
Elite Member  |
Joined: 4-Jun-2005 Posts: 3918
From: Unknown | | |
|
| @pavlor
Thanks. Do you have one for the AKsack.wav ?
Quote:
pavlor wrote: @Spectre660
Of my WinUAE?
My last lame benchmark is with WinUAE 3.2 beta 3, OS4.1FE (I think I used debug kernel during test, so real result may be cca 10 % higher) on Core i5-2500K 3.3 GHz:
0.2770x with WhereDoAll....vaw
For comparison more lame results in Benchmarks time! thread. |
_________________ Sam460ex : Radeon Rx550 Single slot Video Card : SIL3112 SATA card |
|
Status: Offline |
|
|
cdimauro
|  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 18:12:08
| | [ #204 ] |
|
|
 |
Elite Member  |
Joined: 29-Oct-2012 Posts: 4155
From: Germany | | |
|
| @pavlor
Quote:
pavlor wrote: @cdimauro
Quote:
the 128MB limit can be lifted with a little patch |
Of course.
Quote:
if some OS4 dev works on it. |
And that is the problem... |
OK, but... why don't you open a bounty for it? Quote:
FPU performance in WinUAE is not an issue. Only 3D games (Quake) or lame require faster FPU. I was surprised how non-essential is FPU in most applications. |
I know it. Since the FPU was a rare bird for the Amiga o.s., coders usually used integer or fixed-point (using integers) math, instead of the floating point one.
Of course, some applications strictly required FP math, but they are the minority.
That's why usually FPU isn't so much important in the Amiga scene. Quote:
Quote:
and I think also QEMU performance should increase, because now it can run in 64-bit mode using the extended set of registers. |
In theory, we don´t have new qemu.dll yet. 
|
Let's wait for Toni. 
@number6
Quote:
number6 wrote: @cdimauro
Quote:
the 128MB limit can be lifted with a little patch, if some OS4 dev works on it. |
and @pavlor
Quote:
Quote: if some OS4 dev works on it.
And that is the problem... |
Last speaker on the podcast:
the podcast
#6
|
I was there, so I listened it (I've also made some pictures of the slides ). After his talk, I tried to talk with Michele Battilana of his plan regarding improving the PowerPC emulation, but unfortunately there were too many persons around talking with him, so I wasn't able to ask him...  |
|
Status: Offline |
|
|
pavlor
|  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 18:45:54
| | [ #205 ] |
|
|
 |
Elite Member  |
Joined: 10-Jul-2005 Posts: 9673
From: Unknown | | |
|
| @Spectre660
Now, I´m not able to download needed wav from Piru´s site. Strange. |
|
Status: Offline |
|
|
Spectre660
 |  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 19:23:43
| | [ #206 ] |
|
|
 |
Elite Member  |
Joined: 4-Jun-2005 Posts: 3918
From: Unknown | | |
|
| |
Status: Offline |
|
|
Everblue
|  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 19:36:30
| | [ #207 ] |
|
|
 |
Cult Member  |
Joined: 24-Sep-2006 Posts: 679
From: Amigaland | | |
|
| I am in the market for a new board.... and X5000 is beyond my budget. At first I was excited with the "Tabor" board, but following some folks reactions here and other forums, it seems that it is not that great any more due to issues related to compatibility and slow FPU through emulation. Will there be another board coming form ACube? or THIS is what's coming from them next with collaboration with A-EON? What are the chances of last minute changes to fix the problems with this board?
Thank you!
_________________
|
|
Status: Offline |
|
|
bennymee
|  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 19:38:50
| | [ #208 ] |
|
|
 |
Cult Member  |
Joined: 19-Aug-2003 Posts: 700
From: Netherlands | | |
|
| @Everblue
Are there any Linux benchmarks anywhere to show if the claims are right?
|
|
Status: Offline |
|
|
Everblue
|  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 19:45:41
| | [ #209 ] |
|
|
 |
Cult Member  |
Joined: 24-Sep-2006 Posts: 679
From: Amigaland | | |
|
| @bennymee
I have no idea, I am just the average Joe end user trying to understand the technical stuff some people here are discussing :) _________________
|
|
Status: Offline |
|
|
cdimauro
|  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 20:30:17
| | [ #210 ] |
|
|
 |
Elite Member  |
Joined: 29-Oct-2012 Posts: 4155
From: Germany | | |
|
| Quote:
bennymee wrote: @Everblue
Are there any Linux benchmarks anywhere to show if the claims are right?
|
Linux benchmarks are useless if they are directly compiled with SPE unit support. I mean: you're not using the infamous sloooooow trap emulation, but proper SPE code is generated for binaries, in order to avoid it.
Whereas on AmigaOS4 it's unlikely (correct term here, I think ) that you'll see o.s. and applications recompiled with SPE support, so the trap emulation is used instead.
Last but not least, Linux (and apps) supports 64-bit (thanks to the 64-bit general purpose registers), so double precision can be supported (with DPESFP) as well.
OS4, instead, runs in 32-bit mode, so it's not possible to correctly save the high 32-bit part of the integer registers. To support it, the OS4 kernel should be changed, but it means that the context-switch time for tasks increases a lot (doubling the time, circa). Alternatively, every application has to disable and enable the interrupts for each "critical region" (when the DPESFP instructions are used), which compromise the performance.
In short: wait for some OS4 benchmarks. 
EDIT: corrected Linux & 64-bit support.Last edited by cdimauro on 13-Oct-2015 at 08:46 PM.
|
|
Status: Offline |
|
|
pavlor
|  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 20:41:51
| | [ #211 ] |
|
|
 |
Elite Member  |
Joined: 10-Jul-2005 Posts: 9673
From: Unknown | | |
|
| @Spectre660
0.3642x with AKsaw.wav |
|
Status: Offline |
|
|
Spectre660
 |  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 21:33:57
| | [ #212 ] |
|
|
 |
Elite Member  |
Joined: 4-Jun-2005 Posts: 3918
From: Unknown | | |
|
| @pavlor
Thanks. Of course you can still do interesting things with the overall emulation right ?
My results with the AKsaw.wav
Sam440ep is 3.6575x (1:15S) Sam460ex is 5.8033x (0.47S) Quote:
pavlor wrote: @Spectre660
0.3642x with AKsaw.wav |
Last edited by Spectre660 on 13-Oct-2015 at 09:34 PM.
_________________ Sam460ex : Radeon Rx550 Single slot Video Card : SIL3112 SATA card |
|
Status: Offline |
|
|
Tuxedo
 |  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 21:37:10
| | [ #213 ] |
|
|
 |
Elite Member  |
Joined: 28-Nov-2003 Posts: 2350
From: Perugia, ITALY | | |
|
| @cdimauro
was so difficult gets OS4 exes compiled for P1022 SPE? Last edited by Tuxedo on 13-Oct-2015 at 09:37 PM.
_________________ Simone"Tuxedo"Monsignori, Perugia, ITALY. |
|
Status: Offline |
|
|
broadblues
 |  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 21:47:43
| | [ #214 ] |
|
|
 |
Amiga Developer Team  |
Joined: 20-Jul-2004 Posts: 4449
From: Portsmouth England | | |
|
| @cdimauro
Quote:
Whereas on AmigaOS4 it's unlikely (correct term here, I think ) that you'll see o.s. and applications recompiled with SPE support, so the trap emulation is used instead.
|
yes correct terminology , but I'm not I agree that it's unlikely, certainly every componemt of the OS won't be recompiled, but only certain components might require it, and as I pointed out earlier (or was it on another site?) each machine type already has thier own kernel, so build a dedicated kernel is 100% likely.
As to apps, surely it'll just be a matter of a compiler flag, so a fraction nmore complexity in a makefile. Wouldn't put me off doing it for sure. I already build a altivec special ffmpeg lib for blender, why not an SPE one?
_________________ BroadBlues On Blues BroadBlues On Amiga Walker Broad |
|
Status: Offline |
|
|
cdimauro
|  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 21:52:26
| | [ #215 ] |
|
|
 |
Elite Member  |
Joined: 29-Oct-2012 Posts: 4155
From: Germany | | |
|
| @Tuxedo
Quote:
Tuxedo wrote: @cdimauro
was so difficult gets OS4 exes compiled for P1022 SPE? |
No, but the same should happen with the applications, because you cannot run the same binaries if they use some instructions, because the e500v2's SPESFP and DPESFP instructions are mapped to some existing instructions (seems to be "string" and Altivec; don't know if some instructions from the FPU opcode space is used, but from what I've read 'til now it doesn't seems to be the case).
So, there's some incompatibility with the existing binaries, that might not run as they are.
You need to fork both o.s. and applications, generating confusion to the users, and with the possibility that some software cannot be run because the author doesn't release a properly compiled version for the e500v2. |
|
Status: Offline |
|
|
cdimauro
|  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 21:58:23
| | [ #216 ] |
|
|
 |
Elite Member  |
Joined: 29-Oct-2012 Posts: 4155
From: Germany | | |
|
| @broadblues
Quote:
broadblues wrote: @cdimauro
Quote:
Whereas on AmigaOS4 it's unlikely (correct term here, I think ) that you'll see o.s. and applications recompiled with SPE support, so the trap emulation is used instead.
|
yes correct terminology , but I'm not I agree that it's unlikely, certainly every componemt of the OS won't be recompiled, but only certain components might require it, and as I pointed out earlier (or was it on another site?) |
Maybe on another site. I don't remember to have read something like that here. Quote:
each machine type already has thier own kernel, so build a dedicated kernel is 100% likely. |
Fine. But it means that the task scheduler can handle the 64-bit registers, with the cost of a longer task switch time.
Plus, you have to remove or rewrite tricks like using the FPU to fast memory transfer, as it was reported. Quote:
As to apps, surely it'll just be a matter of a compiler flag, so a fraction nmore complexity in a makefile. Wouldn't put me off doing it for sure. I already build a altivec special ffmpeg lib for blender, why not an SPE one?
|
Yes, you'll for sure , but would it happen for all OS4 applications (see my message to Tuxedo)?  |
|
Status: Offline |
|
|
klx300r
|  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 13-Oct-2015 23:07:58
| | [ #217 ] |
|
|
 |
Elite Member  |
Joined: 4-Mar-2008 Posts: 3857
From: Toronto, Canada | | |
|
| Quote:
Tuxedo wrote: @all
But why they presented hw but not specs??????
Suspance or they know that the hw was weak and wait for something....? |
well that seems to be the Amiga way of doing things lately eh
create excitement for the people actually interested and Cannon fodder for the trolls & nay sayers _________________ ____________________________ c64-2sids, A1000, A1200T-060@50(finally working!),A4000-CSMKIII ! My Master Miggies- Amiga 1000 & AmigaOne X1000 ! mancave-ramblings X1000 I BELIEVE  |
|
Status: Offline |
|
|
wawa
|  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 14-Oct-2015 0:01:23
| | [ #218 ] |
|
|
 |
Elite Member  |
Joined: 21-Jan-2008 Posts: 6259
From: Unknown | | |
|
| seems os4 prefers to run straight into cpu dependant binary versions hell which was already so annoying on amiga. good luck. at least the number of files on os4 depot will be growing more significantly.
|
|
Status: Offline |
|
|
iggy
|  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 14-Oct-2015 1:17:11
| | [ #219 ] |
|
|
 |
Super Member  |
Joined: 20-Oct-2010 Posts: 1175
From: Bear, Delaware USA | | |
|
| @klx300r
Quote:
well that seems to be the Amiga way of doing things lately eh create excitement for the people actually interested and Cannon fodder for the trolls & nay sayers |
Actually, at the undoubted cost of offending even more friends, I have a ...slightly different take on that.
When Eyetech introduced its boards, the flaws inherent in them were not known quantities. Many of you bought this hardware and soldiered on with it, even to the point of getting Acube to repair it for you when it broke.
Now A-eon announces a board with flaws it knows exist at the outset and apparently wants to connect Acube's name with this product.
What none of you seem to understand is that this is NOT the opinion of a 'naysayer'.
I have steadfastly supported A-eon , Varisys, and Acube in their continued attempt to assure that we have a range of PPC based hardware to choose from.
But I find in hard to support a move that can be clearly seen as primarily marketing oriented.
And haven't we, of all people, suffered enough at the hands of decisions made by marketers and not engineers? |
|
Status: Offline |
|
|
bison
 |  |
Re: New hardware: PPC-Motherboard A1222 "Tabor" by Acube/A-eon Posted on 14-Oct-2015 1:28:09
| | [ #220 ] |
|
|
 |
Elite Member  |
Joined: 18-Dec-2007 Posts: 2112
From: N-Space | | |
|
| @pavlor
Quote:
Who would buy out-dated, over-priced desktop oriented motherboard for Linux? |
Yes, indeed. If I were in the market for a Mini ITX board for Linux I'd get something like this and save a bundle of money. The Tabor is only interesting to me because it will run AmigaOS 4, and only then because of the limited choices available.
_________________ "Unix is supposed to fix that." -- Jay Miner |
|
Status: Offline |
|
|