Your support is needed and is appreciated as Amigaworld.net is primarily dependent upon the support of its users.
[ home ][ about us ][ privacy ]
[ forums ][ classifieds ]
[ links ][ news archive ]
[ link to us ][ user account ]
Re: Back when Ben was the White Knight!|
Posted on 5-Jul-2021 4:30:56
|[ #121 ]|
|Joined: 9-Mar-2003 |
As poor as the PPC 603 design is, their data bus is 32/64 bits. That could have been an option for a 68060 successor where transistors were cheaper although more pins would still increase the cost.
Are you claiming 68060 FPU is superior when compared to Pentium FPU?
Pentium's 64-bit bus is better for increased FPU workloads.
PPC 603 used a 0.5 μm process node with around ~80 Mhz clock speed. PPC 603's cache is small for 68K emulation.
Quake (timedemo demo3) benchmark
Pentium 150 delivered 33.9 fps (easy OC with 166 Mhz via a simple 60 Mhz to 66 Mhz FSB jumper)
Pentium 166 delivered 37.30 fps
Pentium 200 delivered 41.4 fps
Pentium MMX 200 delivered 44.3 fps
Good luck with 68060 at 95 Mhz matching that.
For Warp 1260 at 100 Mhz, https://www.youtube.com/watch?v=0_dW-21gdkw
At 100 MHz it seems to perform like a Pentium 75. Yup, that's better than the 486, almost twice as better.
Clock by clock it's probably on par with the Pentium overdrive, which is to be expected, seeing the data bus is 32 bits wide on both
Intel recognized the need for higher clock speeds quickly unlike Motorola who stuck with shallow pipeline PPC designs as customers complained about the low clock speeds. One of those customers at Apple even went to IBM to have them design a higher performance G5 CPU with a deeper pipeline.
PowerPC 970 (Apple's G5) failed to reach 3 Ghz, hence Apple's move towards Intel Core 2.
IBM tried convince Apple to use weak IPC PPE 3.2 Ghz.
The 68060 should have an advantage over the Pentium and most PPC cores due to the deeper pipeline for higher clock speeds. Perhaps it was rushed to market before the stages could be rebalanced for maximum clock speed. We were fortunate to get the 68060 at all as Motorola had already switched the high performance focus away from the 68060. The 68060 was a balanced design where the Pentium was a higher performance design (trades power and area for performance) so it was doing good to be as competitive as it was.
Clock speed is dictated by the slowest part of the CPU design.
During the Ghz race, Pentium III incurred errors at 1.13 GHz
For 68060, https://www.nxp.com/docs/en/data-sheet/MC68060UM.pdf
The six stages in the dual integer execution pipelines
68060 FPU is not pipelined, hence the clock speed bottleneck can be the FPU.
At 0.6 μm process node, Motorola has released XC68LC060RC75 and XC68EC060RC75 at 75 Mhz while XC68060RC66 reached 66 Mhz.
My XC68060RC50 Rev 1 reached 63 Mhz on TF1260 and it passed the Quake test. 75 Mhz attempt has resulted in a freeze. 3rd party PSU's 5V rail can deliver ~26 watts, and PSU is rated for about 50 watts.
At 0.6 μm process node, classic Pentium reached 120 Mhz.
PS; I have 4.5 KW solar cells on my roof.
For Quake era, Classic Pentium FPU.
Motorola had a 5 stage 68040V which ran at 3.3V and was 1.5W@33MHz without an FPU. There is just one stage less than the original Pentium design and the lower voltage may have allowed it to be clocked up more than a normal hot 68040. Despite the confined space of an Amiga 1200, CBM would never have considered it though.
68040V has 6 stages integer (ref 1).
68040V doesn't include an FPU.
Unlike Apple's cost-competitive Quadra 605 (with 68LC040 at 25 Mhz and entry price of $1000 during October 1993), C= didn't release A1200 with either 68LC040 or 68040V + 4MB fast ram official SKU or official reference design. My point: out-of-the-box experience and target ID Software's Doom game.
CPU by itself is not a computer.
Last edited by Hammer on 05-Jul-2021 at 06:09 AM.
Last edited by Hammer on 05-Jul-2021 at 06:08 AM.
Last edited by Hammer on 05-Jul-2021 at 06:07 AM.
Last edited by Hammer on 05-Jul-2021 at 06:01 AM.
Last edited by Hammer on 05-Jul-2021 at 05:59 AM.
Last edited by Hammer on 05-Jul-2021 at 05:41 AM.
Last edited by Hammer on 05-Jul-2021 at 05:33 AM.
Last edited by Hammer on 05-Jul-2021 at 05:31 AM.
Last edited by Hammer on 05-Jul-2021 at 05:15 AM.
Last edited by Hammer on 05-Jul-2021 at 05:14 AM.
Last edited by Hammer on 05-Jul-2021 at 05:13 AM.
Last edited by Hammer on 05-Jul-2021 at 05:02 AM.
Last edited by Hammer on 05-Jul-2021 at 04:59 AM.
Last edited by Hammer on 05-Jul-2021 at 04:57 AM.
Last edited by Hammer on 05-Jul-2021 at 04:41 AM.
Core i9-9900K, DDR4-3800 32 GB RAM, GeForce RTX 3080 Ti
Ryzen 9 3900X, DDR4-3200 32 GB RAM, GeForce RTX 2080 Ti
Amiga 1200 (rev 1D1, KS 3.2, TF1260, 68060 @ 63 Mhz, 128 MB)
Amiga 500 (rev 6A, KS 3.2, 68K 50Mhz, 12 MB RAM)
| Status: Offline
Re: Back when Ben was the White Knight!|
Posted on 5-Jul-2021 8:07:58
|[ #122 ]|
|Joined: 20-Aug-2003 |
From: Trondheim, Norway
“ethernet” isn’t Internet, that’s not where the obstacle is.
The Amiga was slow to adopt networking (like TCP/IP over ethernet) and where there was not a connection there was an obstacle. Many Amiga users were still on serial dialup back then but the lack of ethernet support usually meant the loss of an Amiga user, at least daily user, when ethernet became available at the location.
That would be long after CBM was gone.
|Lack of networking support was a problem for educational markets as well.|
Nevertheless I have a good handfull of A2065 cards (and used to have a few A2060 cards as well) that I got from a local school that had a lab with 12-16 A2000, A3000 and A4000s used for art classes. When a small school in rural Norway can manage to get hold of such stuff...
|Apple did well in the education market with the Macintosh and the Amiga did not. Coincidence?|
No, not at all any coincident, but it had little to do with "ethernet". And as you know, Amiga had Envoy for such simple local non-IP networking, file sharing and printer.
|There are porting issues with converting Unix originating networking software to the Amiga but there is no big obstacle because of AmigaOS.|
Yes there are, even simple things like the lack of fork() creates headaches already, never mind more complex "modern" stuff that we take for granted, like process isolation, thread isolation etc that is widely used in IP software for the sake of security.
|There are more primitive OSs than AmigaOS on mobile and embedded devices which also have no problem supporting networking software.|
Heh, then please name them. iOS/XNU, Android/Linux, VxWorks?
|The MacOS with no multitasking and then only cooperative multitasking added later was inferior to the AmigaOS and it had no problem supporting networking.|
Yes, it did have problems supporting networking well too, and it was blatantly obvious at the time - really, you cannot remember? You could halt the entire OS and MacTCP by sending a stream of ctrl+b (bell) to mac terminal users - imagine that, DoS by "quack quack quack" or "dong dong dong", it was great fun! Not to mention what happened if resolvers were unreachable - full halt, awaiting timeouts! Really, MacTCP could only take you so far. If we wanted "proper" networking Apple had A/UX for us. Or we just installed NetBSD on 68k, or mkLinux on PowerPC macs. Then the rescue for Apple came in the form of NeXTStep, aka Rhapsody, aka OSX, aka BSD, aka UNIX.
For Amiga, the best we have is currently Roadshow, and it too has its limits and compromises, and quite a few of those are also due to the nature of AmigaOS - you essentially have to build an entire OS (BSD) stack that runs in parallelle with AmigaOS to make things fly. Why do you think we have things like ixemul.library etc? Do you remember AmiNIX? Just ask Olsen if you don't believe me.
|You are smarter than this. Don't be a troll.|
Classy. I am a networking engineer by profession, and have been for more than 25 years, the last 20 working for an NREN. But what is that against anecdotes about CBM Pets from the 80s...
Last edited by kolla on 05-Jul-2021 at 08:08 AM.
| Status: Offline