13 May 2024, 12:53 | #4281 | |
Registered User
Join Date: May 2017
Location: Munich/Bavaria
Posts: 2,361
|
Quote:
"Please keep in mind, the parallel threads do not mean smp here!" It is still using just one core! There is no exposure of more than one core in Emu68 (yet). That said: A multicore variant of AROSx86 does exist, so with a modified scheduler this could work under AROS-68k, if Emu68 would support it. AFAIK this method under AROS was more a proof of concept and adds quite some instability to the system. More likely Emu68 will some day provide access to other cores via a an ARM.library or use the other cores for gfx-accelertion. |
|
13 May 2024, 13:02 | #4282 | |
Registered User
Join Date: Jun 2010
Location: PL?
Posts: 2,810
|
Quote:
|
|
13 May 2024, 13:40 | #4283 | |
Registered User
Join Date: Aug 2020
Location: Australia
Posts: 898
|
Quote:
SMP (symmetric multiprocessing) treats all CPU cores as equal. Asymmetric multithreading has a master thread with slave threads i.e. treating the other CPUs as unequal co-processors. Saturn did not have SMP despite the two SuperH2 CPUs. The two CPUs were not interchangeable. Each CPU is assigned to different VPx (visual processor) chips. This makes programming complicated. I prefer uber singe core 120 Mhz PA-RISC / PowerPC / Pentium. SMP needs OS change. There's "A-EonOS" System 54+ with SMP. Legacy Amiga apps are not SMP friendly. Last edited by hammer; 13 May 2024 at 14:09. |
|
13 May 2024, 14:08 | #4284 | |||
Registered User
Join Date: May 2017
Location: Munich/Bavaria
Posts: 2,361
|
Quote:
Quote:
Quote:
It is not there yet (and may never come). |
|||
13 May 2024, 14:12 | #4285 |
Registered User
Join Date: Aug 2020
Location: Australia
Posts: 898
|
|
13 May 2024, 14:23 | #4286 | |
Registered User
Join Date: May 2017
Location: Munich/Bavaria
Posts: 2,361
|
Quote:
Michal Schulz: Now, I wanted to have a look at old AROS code which was written there by me and Kalamatee when we were testing AROS SMP. I have ported both SmallPT path tracer and Buddhabrot fractal generator to AmigaOS using P96, removed SMP code from there ... https://www.patreon.com/posts/61450003 SmallPT is just starting 4 ordinary Exec tasks in this video. Nothing more. This is running under unmodified AmigaOS. One Core. |
|
13 May 2024, 14:24 | #4287 |
Registered User
Join Date: Aug 2006
Location: Scunthorpe/United Kingdom
Posts: 2,041
|
Like I said, Trippin'
Edit: As I think I may have pointed out previously, this is the guy that thinks Commodore was done in by yours truly for enjoying Frontier on a stock 1200. I mean, we all had a good laugh about that on the discords and the facebooks and such but I'm beginning to think he's posting here earnestly and believes this stuff. |
13 May 2024, 17:15 | #4288 |
Registered User
Join Date: Nov 2021
Location: Utrecht/Netherlands
Posts: 331
|
Can we say Commodore did everything wrong after the A500? A600 was a big flop, not wanted at that time by the consumers, and managers thought it will sell like hot-cakes and created a large stock. CDTV same story, they thought big hit and it sold a few hundred in US. What was the reason of this large discrepancy between market realization and expectation? CD32 was last ditch effort what they coould bring together and it did not save the sinking ship.
|
13 May 2024, 17:28 | #4289 | |
Registered User
Join Date: Sep 2013
Location: Poland
Posts: 847
|
Quote:
Now you might think it's fine because at some point you can just emulate 68k and DSP all together. Sure, it's possible. And that is exactly why decent Nintendo64 emulation requires PC at level of late 2000s, PS2 requires PC with at least 2cores, sse4.1 and few gigs of ram which translates to at least Wolfdale iirc, and that's bare minimum. Recommended is 4 cores and AVX. And of course fairly good Passmark single core performance. That's for ~300MHz MIPS processor from late 90s and 2 vector co-processors. Now the first PPC (601e) was hardly fast enough to emulate properly 68k with actual speedup over things like 030 ... why it's so hard to imagine that even G3 would've struggled to emulate faster 68k AND DSP on top of that? It's all fine if you create some middleware between application and dsp hardware. Some libs which have standardized subroutines etc. Then you can recreate that basically with any other architecture and it should work. But if you do binaries for specific DSP ... well you are screwed. Especially because DSP requires - by design - specific processing to get proper results in proper amount of time; which means to accurately emulate DSP behavior you cannot just JIT it and hope for the best. |
|
13 May 2024, 19:16 | #4290 | |
Registered User
Join Date: Jun 2017
Location: France
Posts: 65
|
Quote:
Thanks for these informations. I reread what had been said to me and I was the one who didn't remember quite well enough. In fact they explained to me that as you say, the most common DSPs at the time were those with fixed point. On the other hand, they said that the DSP 56001 was one of the best sellers. They didn't understand why Commodore had somehow isolated itself with the choice of the DSP 3210. |
|
13 May 2024, 20:11 | #4291 | ||
Registered User
Join Date: Mar 2018
Location: Hastings, New Zealand
Posts: 2,649
|
Quote:
Quote:
|
||
13 May 2024, 20:14 | #4292 | |
Inviyya Dude!
Join Date: Sep 2016
Location: Amiga Island
Posts: 2,784
|
Quote:
But then, some people here do not either . |
|
13 May 2024, 20:27 | #4293 | |
Registered User
Join Date: Jun 2017
Location: France
Posts: 65
|
Quote:
Maybe. The Amiga 500 and 2000 were the ideas of Thomas Rattigan. Irving Gould fired him right after. But Thomas Rattigan protested and won over millions in compensation against Commodore. If Thomas Rattigan had remained in office, perhaps the management would have been more competent. Medhi Ali hired in his place clearly showed that he was totally incompetent. So combined with Irving Gould, we probably couldn't make a worse combination. |
|
13 May 2024, 21:47 | #4294 | |||||
Registered User
Join Date: Mar 2018
Location: Hastings, New Zealand
Posts: 2,649
|
Quote:
Amiga World October 1993Happy now? Quote:
Quote:
But since you insist... There's more than a slight difference between $399 and $1495. Sure, if you added all the stuff needed to give each machine equivalent hardware specs the PC would probably turn out a little cheaper. That's totally logical - and what I have always said (anyone who expected otherwise was an idiot). But you couldn't get any PC for $399, let alone one that would run the latest PC games - while a base model A1200 would run the latest Amiga games. Even with a hard drive and accelerator card it was cheaper, with the advantage that you could add those things as your budget allowed (by which time they would probably be significantly cheaper). Not that I'm advocating one way or the other. Buy both an Amiga and a PC if you can afford it, and enjoy using both (I did). But if you couldn't afford both the A1200 was a whole lot of fun for a lot less money. Quote:
|
|||||
13 May 2024, 21:48 | #4295 |
Inviyya Dude!
Join Date: Sep 2016
Location: Amiga Island
Posts: 2,784
|
|
13 May 2024, 21:50 | #4296 | |
Registered User
Join Date: Aug 2013
Location: Marseille / France
Posts: 1,480
|
Quote:
|
|
13 May 2024, 22:10 | #4297 | |
Registered User
Join Date: Jan 2010
Location: >
Posts: 2,913
|
Quote:
I am of course talking from a UK perspective *other perspectives may vary |
|
13 May 2024, 22:17 | #4298 | |
Registered User
Join Date: May 2017
Location: Munich/Bavaria
Posts: 2,361
|
Quote:
The last round of lay offs was way too harsh on the engineering departments. Development came to an halt. Spitting the development of the A500 and A2000 into two different teams was questionable - at least the way it was done. The German group did not communicate with the US group and vice versa. Leading to such absurdities as not letting Germany know about the integration of a many discrete parts into Garry, which was only fixed with the B2000. Also the integrating the bus switch into Agnus but not making the small changes to be able to address the full 1MB as ChipRAM, despite of the package providing enough pins to do so, was a grave lack of oversight ... the SlowRAM stupidity was haunting the A500 for all its life. Rattigan was a good overall manager, but coming from a softdrink company he had of course no clue about computers or electronics whatsoever. Last edited by Gorf; 13 May 2024 at 22:22. |
|
13 May 2024, 22:21 | #4299 | |
Registered User
Join Date: May 2017
Location: Munich/Bavaria
Posts: 2,361
|
Quote:
https://archive.org/details/1984-amiga-presentation/ The original sales pitch document from Amiga Inc. emerged from the deep. This document was used in the early negotiations between the companies. (I still think the stackable approach would have been a much better idea compared to the Zorro slot on the side on the A1000.) |
|
13 May 2024, 22:39 | #4300 |
Registered User
Join Date: Nov 2021
Location: Utrecht/Netherlands
Posts: 331
|
|
Currently Active Users Viewing This Thread: 2 (1 members and 1 guests) | |
jamonyorke |
Thread Tools | |
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
A1200 RF module removal pics + A1200 chips overview | eXeler0 | Hardware pics | 2 | 08 March 2017 00:09 |
Sale - 2 auctions: A1200 mobo + flickerfixer & A1200 tower case w/ kit | blakespot | MarketPlace | 0 | 27 August 2015 18:50 |
For Sale - A1200/A1000/IndiAGA MkII/A1200 Trapdoor Ram & Other Goodies! | fitzsteve | MarketPlace | 1 | 11 December 2012 10:32 |
Trading A1200 030 acc and A1200 indivision for Amiga stuff | 8bitbubsy | MarketPlace | 17 | 14 December 2009 21:50 |
Trade Mac g3 300/400 or A1200 for an A1200 accellerator | BiL0 | MarketPlace | 0 | 07 June 2006 17:41 |
|
|