Talk:Motorola 68060

(untitled)
Since Freescale does not, and as far as I can figure never did, produce an mc68060, I'm changing the mod that gives them credit for it.

--User:kjs3

I don't understand ... http://www.freescale.com/webapp/sps/site/taxonomy.jsp?nodeId=0162468rH3YTLC61654622 shows that FSL is selling the 040, the 060, etc. Motorola does not sell those devices any more, and while they were developed in the era when SPS was a part of Motorola, it is no longer a Motorola property. Neier 05:14, 1 February 2006 (UTC)

The pentium FPU was definitly pipelined, this was the core of some of the programming tricks used in Quake and is the reason why games based on the quake engine are a lot slower on 68060. --Qdr 22:44, 15 Jul 2004 (UTC)

Random links from google:  

I'm going to differ there and so are Intel. Please revert this page to remove the factual inaccuracy. P5's FPU was a low-latency single issue design very much like 68060 and K6. It was not what we'd recognise today as a pipelined design, Intel simply did not have the transistor budget to do that. You can read more on this link. The significant FP advantage of the PentiumPro (P6) came from the pipelined dual-issue FPU design. Quake's performance came from it being coded with Pentium FPU latencies in mind.


 * I know the Ars technica article. It is very inaccurate, especially when it comes to the Pentium I. But even then it does not even state anything about the Pentium I FPU not being pipelined. How about checking out the actual Intel manuals? Intel Architecture Optimizations Manual


 * For reference, FPU latency (ADD,MUL) and throughput for the cpus you mentioned:


 * 68060: 3/3 - P5:   3/1 - K6:    2/2


 * (Wayne replies) I have read your link, and it clearly supports me that P5 did not have a pipelined FPU. Like every other source, like Intel's documentation, like the very people who designed the processor. P5's FPU is not pipelined, it is a single issue design. Any layman examining the processor can see this.
 * You linked something to support you, but it supports me instead. I think I and established authorities in the field are right here.


 * Sorry, you are starting to get obnoxious. I quote page 5-2 of the referenced document: Pentium, Pentium Pro and Pentium II processors have a pipelined floating-point unit. By scheduling the floating-point instructions maximum throughput from the Pentium processor floating-point unit can be achieved. --Qdr 14:57, 13 Nov 2004 (UTC)


 * A thought just crossed my mind: Maybe you are mixing up "pipelining" and "dual issue"? --Qdr 15:59, 13 Nov 2004 (UTC)

Motorola 68070 merged here
Its AFD debate agreed to do so; I redirected the article here to preserve its page history. Johnleemk | Talk 10:46, 11 December 2005 (UTC)


 * There was no Motorola 68070 - but there *was* a 68070 - it was used in the CD-i, made by Philips, and I believe it was a clone of the 68000, so it's a bit misleading to redirect 68070 to here. Mdwh 19:17, 18 December 2005 (UTC)


 * The AFD debate had nothing to do with 68070 being a redirect - feel free to turn it into a disambiguation page. Johnleemk | Talk 19:18, 18 December 2005 (UTC)

Usage in desktops
I added a paragraph in usage referencing the CPUs use in the Amiga 4000T and DraCo. As far as I know, the use in the 4000T was the only use in a desktop computer (apart from upgrades) shipping with a 68060. Interestingly, I believe that the CPU card used in the 4000T was actually an accelerator from a third party. 60.240.207.146 (talk) 22:59, 23 March 2011 (UTC)

68060 variants and 75MHz clock, and fakes
The "Architecture" and "Variants" ("Feature Table") sections both indicate that there is a 75MHz version of the (full) 68060 model, but I was unable to find information to support this. On Amiga-related websites/forums/... there is occasional talk about fake 68060s, often claiming there is no 75Mhz version (only 68EC060 and 68LC060). For example, this page does not list a 68060@75Mhz: http://www.cpu-world.com/CPUs/68060/index.html Can someone clarify, esp. the "Variants" section, add links, sources, pictures, ... - verified, proven information? RoMancer (talk) 11:30, 4 January 2020 (UTC)

There was supposedly PC68060 versions that were early engineering samples equivalent to a Rev 0, with 'D11W' printing. The 'XC'68060 revision of 68060 CPUs covers mask revision 1 (xyF43G) and 5 (xyG65V), and these are the valid program identifiers in the on-chip PC register. Versions were marked 50, 50 and 66MHz The 'MC'68060 revision of 68060 CPUs designates the chip as 'Qualified' and covers the two mask revisions 4 (xyG59Y) and 6 (xyE41J). Both main Rev values are available to read off the internal PC register by software. The Rev 4 was the official EC/LC mask, and was made in variations of 50, 60, 66, and 75MHz (RCxx) with related markings. The Rev 6 was the official full CPU rev, and was only ever marked as RC50 (50MHz), but has been successfully clocked in some cases over 100MHz with external cooling.

My distillation of full Motorola mask values, errata and related speeds documents can be found on this thread/post:

https://www.amibay.com/showthread.php?100206-68060-PCR-Information-(-amp-Request)

Including copies of the Motorola/Freescale/NXP source documents that they came from. Note that mask Rev 2 and Rev 3 are hinted at, but no evidence of their public release occurred.

The documents at the bottom of the main post are:

DSAFRAZ0013418.pdf MC68060DE.pdf PCN2378-LC060 G59Y Mask Set Introduction.pdf PCN6850-MC-Qual-EC&LC.pdf PCN2401-68060 G65V Mask Set Introduction.pdf PCN3448-042umTo035um.pdf PCN4061-More035um.pdf PCN4460-945048375992collateral.pdf

Most of the docuemnts can be searched for on the web using the PCNxxxx number. The words after the - are my references to the detail they contained. — Preceding unsigned comment added by 108.34.193.9 (talk) 22:15, 3 March 2021 (UTC)

Presence of overclocking data unnecessary?
There's a lot of talk in this article about how much the CPU could be overclocked, but that's generally not present on similar era CPU pages. Shouldn't we ditch the overclocking mentions? Particularly since there's at least one dodgy source cited and marked as an issue already. Fishmech (talk) 02:53, 19 April 2023 (UTC)