@Kronos

Logan RULEZ!

I'm not too sure if the CPU is the main concern when it comes to the Amiga nowadays. I'd say it's more of a sound and visual issue. I got my AmigaOne 500 the other day and noticed it was underpowered when it came to graphics. I also noticed sound was kind of iffy. Not as good as the original Amiga's build in sound. Maybe something I have to change, but haven't noticed where. And sorry for saying this, but the one guy has to be seriously kidding when he says the X1000 is only a little less powerful than a intel i7. Those CPUs are wicked fast and process WAY more information than any CPUs have ever done in the past. I think the only way we, as Amiga users, will ever see our former glory again is going to be through software. Just my 2 cents though.

Ed

_________________
Amiga 1200 - ACA 1233 68030 128MB Ram
8GB CF With tons of Classics

AmigaOne X5000

Raspberry PI 400 - PiMiga 1.5
"That which doesn't kill you, only makes you stronger" - Someone important, but I forgot who

Spotted that MOS ppl have found pretty nicely some hidden bits and pieces of info about NG PPCs:
http://www.morphzone.org/modules/newbb_plus/viewtopic.php?forum=3&topic_id=7001&post_id=91497&viewmode=flat&sortorder=1&showonepost=1

(interesting to find information that confirms that MIPS (CAVIUM chips) became a threat to freescale PPC line, thanks to that, PPC SoC development has accelerated)
UPDATE: Link to CAVIUM MIPS chips info.

Last edited by KimmoK on 09-May-2012 at 09:23 AM.
Last edited by KimmoK on 09-May-2012 at 07:39 AM.
Last edited by KimmoK on 09-May-2012 at 07:38 AM.

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

@edponpon

"kidding when he says the X1000 is only a little less powerful than a intel i7."

I think PA6T is something like 0.05% of i7 in integer performance when multiprocessing is in use.
(OpenCL could give us best way to get more FP performance, I think.)

But at the same time PA6T is as powerfull as some (Atom) 1.6Ghz (also 1...1.4Ghz x86 chips are being sold today) of the x86 world. And everyone can see that 1.6Ghz dualcore can do a lot of productivity stuff and play all iPad caliber games (plus xbox1...PS2 level of games, even xbox360 caliber of games with high end GPU), even when there is inefficient operating system on the HW. With Amigatized OS and SW 1.6Ghz dualcore can do little bit more even. IMO, that is.

When using LinuxMintLXDE on 1.7Ghz AMD, I'm surpriced how little it is faster than SAM440. And the AMD box feels/seems slower ... untill I tried AROS, then it's pretty nice.

UPDATE: with chip like T4240 we could have perhaps 50% of i7 performance in integer... with just 10% of i7 performance it would be superb for our needs. But it all needs a lot of SW (drivers+new apps).
UPDATE: and as long as we have non-multiprocessing OS+SW we get only 1/2 of the speed of PA6T, 1/12 of the speed of T4240 etc.. Also some i7 would "only" be less than 10x faster than PA6T.

Last edited by KimmoK on 09-May-2012 at 08:40 AM.
Last edited by KimmoK on 09-May-2012 at 07:36 AM.
Last edited by KimmoK on 09-May-2012 at 07:02 AM.
Last edited by KimmoK on 09-May-2012 at 07:00 AM.
Last edited by KimmoK on 09-May-2012 at 06:59 AM.
Last edited by KimmoK on 09-May-2012 at 06:56 AM.

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

@KimmoK

Let's just say that most CPUs are underutilized.

Modern games are able to wring out every last drop from one or two cores of the CPU and the whole GPU, anything else is just wasting loads of cycles.

Programming efficient multithreaded software is dang hard. Tasks that easily break up into discrete elements that can be processed more than one at a time without affecting the other elements exist, but games and office work just isn't like that.

Yet (hopefully ). It's tedious work discovering parallelizable tasks and making sure you don't have race conditions and wait states.

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@olegil

Actually, I have a question.
What sort of application is it we need more oomph for now. We know the PA6T isn't an i7, so pointing this out any more times now isn't actually helping anyone. But quite honestly, where is our bottleneck now?

I can't actually think of anything, so I would suggest that now we have hardware and can focus on software for a bit.

So what is it we cannot do right now due not to lack of software but lack of processing power?

Also, it would be nice to hear some more suggestions for problems which can be parallelized. I mentioned compiling earlier but many didn't agree with me. Funny, because it's not that hard to break up a large file into several smaller ones and do multiple stages of compiling and linking. I always test a parallel build of my code as a principle. If it doesn't parallel build, the Makefile is not correctly written.

it might not be obvious why it helps to parallelize something like compiling, but remember that when you compile, you read in a ton of files, then you process them, then you write back something. So it's bottlenecked first by IO, then by CPU, then to a lesser extent by IO yet again. If another thread is able to do IO while the first is using CPU, then it can use CPU while the first one is back on IO for the next file. It really works, and you actually get a speedup even from a uniprocessor with parallel jobs. Besides, it's fairly easy to get a RAID setup these days that is actually faster than the memory we had just 10 years ago

Last edited by olegil on 09-May-2012 at 08:04 AM.

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@olegil

"I can't actually think of anything, so I would suggest that now we have hardware and can focus on software for a bit."

Yes the lack of SW and drivers is the main issue now.
Another thing is that the (powerfull) Amiga HW is not really affordable for a lot of people. (I have SAM667Mhz and it's a bit slow for some things I would like to use it for)
I know SAM460 should do my needs pretty nicely, but I would need to gather extra 1keur to get it.
(I should get some productive stuff out so that people would then donate me SAM460 motherboard. )

About parallelly running things:
Gcc, Blender, video editing(mpeg encoding+decoding) can use parallel processing easily. If we had two cores we (people who tend to run a lot of apps at the same time in multitasking) would get immediate general boost of performance. And heavier tasks, like SW running in UAE, would have even less affect on system responsiveness. Also we could start to use some sane background stuff (automated backups etc.) without big affect on the application running in front.

UPDATE: About high performance computing.
If we want to be around the top performance of computing power, we should get drivers for latest GPUs + OpenCL + SW to use.

LUXMark Top 20 Complex Benchmark Results:
i7(12cores3.2ghz) CPU scores 509 in luxmark
Radeon HD7970 (1 GPU only) scores 1206 in luxmark

LUXMark Top 20 Simple Benchmark Results:
i7(12cores3.2ghz) CPU scores 6294 in luxmark
Radeon HD7970 (1 GPU only) scores 18561 in luxmark

So, depending on what you do, the fastest possible CPU might not be needed. Sometimes a fast GPU can substitute fast CPU. And then you can have several GPUs.
Source (LuxMark results).

Some low end CPU result from Medium complexity:
CPU only score 27 - AMD Athlon(tm) 64 Processor 3000+ [1 units @ 2502MHz]

I imagine x1000 could get around 25 in that CPU only test, but with just one HD5850 card one might get score of 700+.

UPDATE:
For some power Amiga user a PPC board with 10Gb ethernet port connected to a cluster of T4240/i7 units might be interesting (for multithreaded & cluster aware apps).

Last edited by KimmoK on 09-May-2012 at 10:13 AM.
Last edited by KimmoK on 09-May-2012 at 10:01 AM.
Last edited by KimmoK on 09-May-2012 at 09:46 AM.

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

@amigadave

Quote:

The only way this could happen would be if the whole Amiga community accepted something that is cutting edge and modern, state of the art, as an Amiga and left all of our current legacy support behind, only to be visited through emulation on what ever that new "cutting edge" platform of the future might be.

You just described the Commodre USA Vision macihnes. LOL.

I am not at all a technical guy but the first thing
i think of is, if we have AmigaOS and Linux on the
same machien would it be possible to run them
on each core at the same time?

That way we could have for example 2 screens,
1 with AmigaOS and Amiga-software and an other
screen with Linux and the open-software.

Is this doable or is it SF?

_________________
A1G3-SE + OS4.1 u1 iso (x2)

@Hypex

Quote:

You just described the Commodre USA Vision macihnes. LOL.

Somethimg even smaller C=USA cult would love us to believe.

He said "innovative" and "cutting edge". Overpriced existing boards with integrated gfx just isn`t that.

Last edited by vox on 09-May-2012 at 04:35 PM.

_________________
Future Acube and MOS supporter, fi di good, nothing fi di unprofessionals. Learn it harder way!

@Dirk-B

Quote:

That way we could have for example 2 screens, 1 with AmigaOS and Amiga-software and an other screen with Linux and the open-software.

This was kind of promised for X1000, but will see when OS 4.2 is out. Could it be "hardware running virtual box" or similar, just to launch PPC Linux apps (OpenOffice, Iceweasel etc.) rendering dual boot obsolete, that would be just great

_________________
Future Acube and MOS supporter, fi di good, nothing fi di unprofessionals. Learn it harder way!

@amigadave

Quote:

The only way this could happen would be if the whole Amiga community accepted something that is cutting edge and modern, state of the art, as an Amiga and left all of our current legacy support behind, only to be visited through emulation on what ever that new "cutting edge" platform of the future might be.

Kind of portable device would do the magic.

Why that could not be by inner evolution that OS evolves being close to state of art, just lacking some modern features and apps? OS is the weakness, but is also a strenght and difference as it becomes more updated and modern.

Small form factor machines like tablets and smartphones have proven a lot can be done on something that would not be perceived as "cutting edge" when compared to desktop machines, but was cutting edge as new introduced technology seems like today people would be pleased by using some of the recent technologies.

_________________
Future Acube and MOS supporter, fi di good, nothing fi di unprofessionals. Learn it harder way!

@Drewlio77
Quote:

I know many people will disagree with this but the only thing that would make the Amiga become ahead of the curve that it was back in 1985, is custom chips. That is what made the Amiga what it was. A revolutionary pre-emptive mutlitasking OS built around a custom chipset. It was pure genious, ahead of it's time, built from the ground up, not from existing mainstream technology. Unfortunately for a customized Amiga to be developed would require an investor with a company with a sizeable capitol to invest in a technology that will have to face stiff, well entrenched competition. Risk Management anyone?

So find someone like Trevor to do it out of love or for some other non-business-sensical reason. Maybe a huge education project.

And I am one that disagrees that custom chips will make Amiga better today. I doubt even Trevor and however many friends he might come up with could afford to compete with Nvidia, AMD, whoever makes good audio chips these days, etc. all at the same time, when each of those alone spends several $Billion per year improving their stuff. If custom chips are a requirement, then it's flat-out absolutely impossible to do IMHO. Besides, why is CPU on your list? Amiga always used an off-the-shelf CPU. Hombre was to change to PA-Risc instead of 68k if I remember correctly, but that was not unique either.

_________________
All glory to the Hypnotoad!

"if we have AmigaOS and Linux on the same machien would it be possible to run them on each core at the same time?"

PPC chips have had that option for a long time.
Recently I saw P1022 demo where Android was run on the other core and RTOS on the other.
But it requiress HW resource virtualization SW to be built. And I think our niche is too small for that to happen.
I rather see SMP done. And then perhaps if linux PPC is really wanted that much, sandbox/virtual machine for it.

Not heard what is the way how SMP is being done in AOS4.2.
One way could be that AOS4.2SMP would not be compatible with current SW and sandbox might be needed to be built to run AOS4.1 and AOS3.x SW.
(IMHO, it's annoying that user input is not wanted on how users would like to have the SMP done/working. With more open communication, I doubt there is any financial risk of some other AOSlike system stealing the ideas. As I see it. MOS is saying bye bye to PPC while AROS is already mainly elsewhere.)

@billt

"Hombre was to change to PA-Risc instead of 68k if I remember correctly"

I think the plan was to have one PA-Risc to do the CPU stuff and another helping GFX to do 3D.

Last edited by KimmoK on 09-May-2012 at 05:08 PM.
Last edited by KimmoK on 09-May-2012 at 05:01 PM.

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

@Kronos

Quote:

In 2012 we are using G4 class CPUs in the topmodels (old Macs,Pegs,A1s and X1000)

The PA Semi in X1000 is a G5 class chip, not G4.

_________________
All glory to the Hypnotoad!

@billt

Quote:

The PA Semi in X1000 is a G5 class chip, not G4.

PA Semi wishes

_________________

@olegil

Quote:

What sort of application is it we need more oomph for now. We know the PA6T isn't an i7, so pointing this out any more times now isn't actually helping anyone. But quite honestly, where is our bottleneck now?

Considering how infrequently we get a major performance update these days, I think it's better to consider tomorrow's needs, not today's. If anyone is working on a design, I'd hope they are thinking as far down the calendar as existing chips can possibly take us. That can of course be frustrating when a new chip poofs into existence that does so much more than what someone started a design with, but if they haven't got too far done yet then changing is something to consider.

Even if we can't use 24 cores today (or 24 virtual cores from 12 real ones), maybe some years from now we will be able to. And by then perhaps there's a good use for them. Maybe there's good uses for them now and we're just not sure of them. If we get a Mac PPC emulator someday, some cores could be running OSX-PPC. (which I'd still like to be able to do) Perhaps 3d rendering. Or if we ever see some simulation tools like Icarus, GHDL, Verilator though I'm not sure these do much threading, I know the big-gun commercial ones do. But remember multitasking, we could have all sorts of stuff going on in background. Transcoding videos such as Tivo to divx or something like that. If we'd be running MythTV with a bunch of tuners, there could be lots of transcoding going on behind the scenes while compiling software while being a Linux virtual machine server while rendering our 3d movie while playing Doom7 all at the same time.

Perhaps no one thing will strain more than a couple cores, but add them all together...

I took an embedded programming class last fall, and the instructor talked about some company working on a 300core ARM chip. Wiggy... Though I've heard that the quad-core Kalel chip got beat out by a dual-core competitor, possibly due to weakness in memory system design. I think AMD's Bulldozer/FX originally suffered some performance due to software scheduling issues, and has since improved to some degree. So we do need to remember that more cores doesn't automatically mean faster, even when software is written to use multi cores.

_________________
All glory to the Hypnotoad!

@KimmoK

Quote:

IMHO, it's annoying that user input is not wanted on how users would like to have the SMP done/working

Try to think many cooks making one soup. When every one cook favours one type of spice over all the rest. You would get a mess instead of anything tasteful in the end. And that kind of comments makes one think you don't trust the developers at all. That might be bad keeping their motivation up.

@billt
I think it was that risc CPU was targeted to be used for graphics stuff and the main CPU being 68k or PPC still.

Custom chips. Nvidia and ATI chipsets are still using the same old ideas. Maybe if somebody came with new ideas. (Complex real time ray tracing for example.) Or change the whole computing to something completely different like GUI was compared to text based systems back then. Controlling a computer with your mind...

Last edited by Tomppeli on 09-May-2012 at 07:10 PM.

_________________
Rock lobster bit me. My Workbench has always preferences. X1000 + AmigaOS4.1 FE
"Anyone can build a fast CPU. The trick is to build a fast system." -Seymour Cray

@WolfToTheMoon

"PA Semi wishes"

It seems that without any PA6T optimizing, PA6T is in between G4 and dualcore G5. PA6T is faster than PPC970FX but slower than PPC970MP.
So it is a G5 class chip. (with 1/4 of the electricity bill (and heat) of G5)

My Blender results summary post.

(still wonder what they did in Bryce render benchmark to make G5 1.6Ghz 2x faster than G4 1,6Ghz ... to me it indicates that Blender could be optimized to make it run a lot faster on G5 and PA6T)

UPDATE & btw. Checked G5 power usage ... they had up to 1400W PSU in those beasts. And the cooling noice is referred as "wind tunnel". (G5 wind tunnel etc...)

Last edited by KimmoK on 09-May-2012 at 09:01 PM.
Last edited by KimmoK on 09-May-2012 at 08:10 PM.
Last edited by KimmoK on 09-May-2012 at 08:08 PM.

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

@olegil

Quote:

olegil wrote: @olegil Actually, I have a question. What sort of application is it we need more oomph for now.

I think a PA6T or G4 1.67 GHz is close to be convenient. From my experience with the G4 I still desire more power for:
- Full HD video replay (the PA6T may do some full hd codecs, but probably not everything an dwill be on its limit, the G4 is definietly not able to handle full hd in a convenient way)
- webbrowsing with heavy computing stuff and/or embedded video in high resolutions
- I recognized that much of my Hollywood stuff is rather cpu demanding.
- complex pdfs are also not scrolling smoothly.
- high resolution compressed pics (read: digi cams with high resolutions) take a while to display (the 8 MP jpgs from my digicam take about 0.35-0.8s to load on my 1.5 GHz G4 with Showgirls. That's of course still quick, but not instant. My older pics ( 2 MP) take about 0.06-0.12s to load, that's about the time of the double click,. That's instant, it feels real time)

With CPU power it is easy: the more the better. But all in all I think reaching i7 level is not the most pressing issue, but a bit more juice than the G4 or PA6T provides would be very handy for rather basic everyday stuff. A G5 2.7 GHz would proabably be fine cpu power wise - unfortunately it's a bit energy hungry and neither OS4 nor MorphOS supports it anyway...

_________________
My programs: via.bckrs.de
MorphOS user since V0.4 (2001)