Talk:Motorola 68040

"80486 had the ability to be clocked significantly faster"

This is simply not true! http://wiki.riteme.site/wiki/80486#Models Fastest 80486 is 33MHz (while 68040 is 40MHz)!! Note that intel used internal clock when talk about CPU speed while Motorola used external clock!! —Preceding unsigned comment added by Calimero (talk • contribs) 13:32, 26 July 2008 (UTC)

There were 486 CPUs with 50MHz external clock. However if we are to go by external clock speeds then my current Core 2 Duo has a clock speed of 266 MHz.--Anss123 (talk) 13:54, 26 July 2008 (UTC)

"There were 486 CPUs with 50MHz external clock" same goes to 68040 if I understand comment bellow on this page: "There was a 50Mhz 68040 variant" so my objection that "80486 had the ability to be clocked significantly faster" IS NOT TRUE still stand!

What exactly are you referring to when you say external clock speed? The FSB or the clock input. In the latter case a 50MHz 68040 has 100 MHz external clock speed, according to Apple anyhow.

The fastest 486 is in any case a lot faster than the fastest 040 (even if the FSB is clocked slower), which is what the article tries to communicate with "the 486 can be clocked faster", and by conventional wisdom the CPU's clock speed is measured by the internal clock speed - not the external.--Anss123 (talk) 00:02, 27 July 2008 (UTC)

External clock speed is "FSB". Please give me reference to 486 clocked faster than 68040 (greater than 100MHz internal clock)! There was AMD "486" CPUs that worked faster than 100MHz but not intel i486. Highest speed for 68040 and i486 is 100MHz (if we measured internal clock speed). Do you agree?

No. The fastest i486 was the DX4, with 99.99 MHz clock speed and 33.33 MHz FSB. There was also Intel CPUs with slower clock but faster FSB (Like the 486 DX-50 with 50 MHz FSB). The fastest 040 out of Motorola had a 40 MHz FSB and 40 MHz internal clock speed. A faster variant was planned, but unless Freescale built them the 040 tops out at 40 MHz.--Anss123 (talk) 14:13, 27 July 2008 (UTC)

Ok, let me get this straight: MC68040 manual (from www.freescale.com) says: "The M68040 uses two clocks to generate timing: a processor clock (PCLK) and a bus clock (BCLK). The PCLK signal is twice the frequency of the BCLK signal and is internally phase-locked to BCLK. PCLK is also distributed throughout the device to generate additional timing for additional edges for internal logic blocks and has no bearing on bus timing. The use of dual clock inputs allows the bus interface to operate at half the speed of the internal logic of the processor, requiring less stringent memory interface requirements. Since the rising edge of BCLK is used as the reference point for the phase-locked loop (PLL), all timing specifications are referenced to this edge." So we have PCLK and BCLK (as you call it FSB). Further in Manual you can read that PCLK Cycle Time for 40MHz version of 68040 is 25 ns which corespondent to 40MHz; and for BCLK manuel says that Cycle Time is 50 ns which corespondent to 20MHz. Bottom line is that 40MHz Motorola does not operate at 80MHz internal but only on 40MHz. You could simple told me that 68040 does not support doubling rate internally ;) although at http://www.cpu-world.com/CPUs/68040/ you could read: "Some internal logic of the Motorola 68040 uses 2x clock frequency, but the processor cannot be considered double-clocked."

While on http://www.bbs.ingedigit.com.ve/TechInfo/68040.Microprocessor.html you could read: "The main internal units work at twice the clock speed of the bus interface unit. For instance, when processing most instructions internally, a 68040 clocked at 33 MHz effectively runs at 66 MHz." which is incorrect?!

I also find this http://www.skepticfiles.org/cowtext/comput~1/486vs040.htm text where you can find explanation why 68040 is way faster than 80486 at same clock. Only DX2 version of 80486 could compare to 68040. --Calimero (talk) 11:57, 28 July 2008 (UTC)

I don't know much about how the 040 works internally, but from what I recall (from benchmarks back then) was that the 486 had a clock for clock performance advantage on common integer workloads while the 040 beat the 486 at select integer benchmarks and beat the pants of the 486 at floating point benchmarks that was not written for older 68K FPUs.

However if you wish to compare clock for clock performance you should also keep in mind the manufacturing technology and die size. The DX/2, for instance, was manufactured on a 0.8 micron process. Whereas the Pentium 66MHz was built on a 0.6 micron process, giving the lather an "unfair" advantage as it had more transistors to devote to enchanting performance.--Anss123 (talk) 15:11, 28 July 2008 (UTC)

Mystery is solved! ;) http://milan.kovac.cc/atari/040CLOCK.TXT "About the PCLK (Processor CLocK), the 040 needs it to cadence its internal logic, especially its pipeline. This clock could be 200 or 500 MHz, which wouldn't change the performance of the 040 !!" —Preceding unsigned comment added by Calimero (talk • contribs) 16:28, 28 July 2008 (UTC)

Interesting read. Guess even back then marketing material was less than trustworthy.--Anss123 (talk) 18:45, 28 July 2008 (UTC)

So now we are sure that 80486 clocked better (DX2 - double clock rate) and we know that at same clock 80486 was two times slower than 68040 and that 68040 does not support double clock ability (at least not one that will affect performance!). Missing reference (in wikipedia text) for this two statements could be Rodolphe text (link above) and other link that I also mention above. Thank you Anss for discussion on this topic! :) --Calimero (talk) 18:54, 28 July 2008 (UTC)

Additionally to this, I would simply point towards the latter generation of "overdrive" and "5x86" type processors, which were essentially just 486DXes with better tolerances and clock-tripling or quadrupling to reach 120 or 133mhz officially (and 150, 160 is achievable with enough effort). Trying to split hairs on these fronts is silly, because it's a simple fact that the *processor* is running at a higher speed. If that's your argument, then you've lost it. Ignoring matters of internal pipeline synchronisation (and who's to say that Intel don't do exactly the same thing but didn't see fit to crow about it?), the 486 core is capable of at LEAST twice that of the 040, and arguably actually 3 to 4 times more (...indeed, I'm fairly certain some budget Pentium-motherboard-compatible chips were actually just 486s with different pinouts, and ran at upto 200mhz if not higher before their brief day in the sun faded).

If it's a matter of how much bus I/O they peak at - which is why the clock doubling entered the argument - then things get very messy very fast. Both manufacturers had half-width buses for their 32-bit (and 16-bit!) parts in the past, after all, which if we're to use that as a metric means they were actually half the speed written on the label. And indeed they were, for some types of processing that was heavily I/O bound; for others which chewed up the ALUs and other internal parts for a good many clock cycles, they were only fractionally slower than the full-width models. And I seem to recall that the 040 still has the ability to operate on 16 or even 8-bit buses if needs be, though no-one ever seems to have been crazy enough to try it.

However, neither chip to my knowledge has ever used anything other than 32 bit, so let's forget that. So, their bus width is the same. For the sake of argument, let's say their I/O cycles take the same number of clocks (I don't know if that's true off the top of my head and I sure as hell can't be bothered finding out right now), and their memory chips and controllers (whether internal or external) are the same speed, bit-width, and number of wait-states/latency rating. So all that matters is the frequency they're clocked at.

...well, golly gee would you look at that. They're the same. Both chips were offered on buses running at 25, 33 and 40mhz, and indeed even a few at 20 and at 50. So if we take clock multiplication out of the picture, and ignore per-cycle calculation efficiency, THEY'RE EXACTLY ALIKE. Though it might be worth pointing out that I've never seen, or heard tell of a production machine offering a raw 50mhz as an option for hosting an 040, whereas I *ACTUALLY OWN* a 486 system (ISA based, with an SX25 installed at the moment, but I do fancy installing a DX50 if I can get one off ebay at a sensible price) that can be jumpered for a 50mhz FSB. Think I have another motherboard that might offer it too, but that'll be a walk through its manual should I still have it...

Therefore, advantage 486, until someone turns up with an official (rather than dodgily overclocked) system to show off with a 50mhz 68040 in it.

And, well, if we bring the clock doubling into play, then it's all over. We'll start with a board running at 40mhz, and an Intel Overdrive "83mhz" chip (meant for older boards that couldn't actually run a 40mhz FSB, so it runs some arcane multiplier from 25 or 33mhz), removed from its adapter and plugged in directly, with suitable motherboard jumper settings for voltage, multiplier, frequency etc. Voila, a DX2/80. This WILL STILL RUN ITS EXTERNAL I/O AT 40MHZ, the same as the Motorola. BUT, the internal logic runs at 80mhz! So, things that would have been IO bound run at the same speed, and match the 040 performance, with the CPU core basically idling half the time. Those which AREN'T IO bound, e.g. anything that really abuses the FPU and need a great many clock cycles, go twice as fast (and still don't challenge the IO lines). It's a bit like having a 64-bit chip on a 32-bit bus, you could say. And if you then take an actual 120mhz overdrive (or an overclocked 160mhz one), then the advantage for complex operations gets even greater, still with no detriment to IO.

Though really we'd probably have been fine with the gloss-coat primary-coloured Ford Fiesta of the mid 90s CPU world, the DX2/66, on a 33mhz board. There's a slight slowdown with IO ops, but actually with PCI, accelerated graphics, etc, it didn't turn out to matter so much. The 33mhz systems of the time had bus bandwidth to spare - it was the CPU cores that needed turbo boosting. The DX2/66 was successful because it had a good mix of integer and FPU processing speed and external bandwidth without having to break the bank making the motherboards super-fast... (and indeed, what turned out to be the common base speed for PC interface slots, and even SDRAM when it came along? 33mhz, then 66mhz and 133mhz in the Pentium days. having a faster main bus would have been kind of a waste.)

And, of course, there's that internal cache, which ALSO speeds up with the rest of the core. Forgot about that, didn't you? :D 193.63.174.254 (talk) 19:44, 10 May 2017 (UTC)

Renaming from Motorola to Freescale

Neier can't be bothered to figure out Freescale didn't produce a 68040, and I've apparently screwed up backing out his st00pidity.

KJS3 - Please see http://www.freescale.com/webapp/sps/site/taxonomy.jsp?nodeId=0162468rH3YTLC61654622 . The 68k, CF, i.MX, etc are all divested from Motorola with the Freescale spinoff. Neier 05:29, 1 February 2006 (UTC)

Neier - By your rationale, we should call my old 1991 Volvo 240 a Ford 240 because Ford purchased Volvo. Please stop making edits were one is neither accurate nor desirable.

No, Ford has not discontinued the Volvo brand name. Please point to me a spot on Motorola's web site where you can order the 68040, or for that matter, any device formerly produced by their Semiconductor Products Sector (now, Freescale).

Nor has Ford ever marketed it as the "Ford 240". They are content to let it remain named Volvo. Freescale markets the 68040, etc. and has no intention of leaving Motorola in their product names at all. Wikipedia should follow the companies' decisions. Neier 22:29, 1 February 2006 (UTC)

PS - You can sign your posts with four ~~~~ characters.

The problem is that Wikipedia isn't a brochure of currently available products, it is an encylopedia, and how things are known in a historical context is important. Whilst the CPUs are now being sold by Freescale, they were originally developed and sold by Motorola, and they are surely far more well known as "Motorola 68K CPUs" than "Freescale 68K CPUs" (eg, 36,700 Google hits for "Motorola 68040" versus 2 for "Freescale 68040"!) Mdwh 00:58, 2 February 2006 (UTC)

Thanks Mdwh. Articulated much better than I could. But I suspect Neier won't get it. Kjs3 02:09, 2 February 2006 (UTC)

Thanks, Mdwh. The differences between a product brochure and an encyclopedia are duly noted. The MoS seems quite clear as well, at least as far as "common names" goes. This st00pid guy gets it. So, I guess it is better to focus on making the articles more up-to-date in a historical context rather than enforcing a company POV on the article names.

Kjs3, why the hostility? Mdwh's explanation makes much more sense to me than your claims that "Freescale didn't produce a 68040", "Freescale does not, and as far as I can figure never did, produce an mc68060", and "A trivial check would show Freescale never produced the 68060". In fact, it was a trivial check that showed me that Freescale does still produce the chip. A well-reasoned and researched response is better than bad analogies and blanket statements that are just false.

As for the name calling and such, if it makes you feel better, fine. Myself, I've never been cool (er, k3wl) enough to keep up with the new spelling fads. Neier 02:47, 2 February 2006 (UTC)

I agree that the article should still be named 'Motorola'. Everyone knows it as a Motorola part. The chips themselves are now mainly of historical interest. Parts that are produced in the present day using the 680x0 architecture can have their own articles. - Richard Cavell (talk) 11:21, 4 July 2008 (UTC)

3.3V Version

There was at least one 3.3V variant, which I would expect (but have not tested) to run cooler than the 5V versions. I remember this as being a 68EC040, sadly I don't think there was ever a 3.3V full 68040. Anyone have data to back up my (non-ECC ;-) memory? AndrewBall 01:55, 18 January 2007 (UTC)

ok, we have text from Rodolphe Czuba (http://www.czuba-tech.com), maker of accelerators for Atari Falcon computer: http://milan.kovac.cc/atari/040CLOCK.TXT as you can read: "A remark : the 040 3.3V model only exists in two versions: - 1. The 68040V which doesn't have an FPU ! - 2. The 68EC040V which has neither an FPU, nor a PMMU !" —Preceding unsigned comment added by Calimero (talk • contribs) 16:17, 28 July 2008 (UTC)

50 MHz Variants

There was a 50Mhz 68040 variant: http://www.micromac.com/products/speeddoubler040.html http://www.sonnettech.com/product/quaddoubler.html

Note that the '100/50mhz' noted in the pages refers to the internal clock of the chip, which ran at twice the speed of the external clock

Or was the 50Mhz part just an overclocked 40Mhz part?

If I recall correctly, it was the other way around - the internal clock was half the external, i.e. a "25Mhz" '040 had a 50Mhz clock input. Apple took to mentioning the external clock rate in advertising, in order to keep up appearences relative to the 486. Mirror Vax 23:45, 24 Apr 2005 (UTC)

Apple always advertised 68040/LC040 desktop machines using the lower clock speed, but indeed the '040-based PowerBooks were advertised using both, as in "Featuring an advanced 66/33-megahertz Motorola 68LC040 processor, the PowerBook 540c…" (from the 540c sales brochure, cf http://www.saldanah.com/apple-collection/images/datashet-540c-2.jpg ). I don't consider that especially wrong, since the CPU core used both speeds internally in different parts of the CPU (almost everything uses the lower clock, but some parts of the arithmetic units use the higher clock). For sure I can understand why Apple did that, since a 25MHz 68040 could run circles around a 25MHz 486DX in actual performance. tooki (talk) 11:05, 10 April 2016 (UTC)

Three pictures

Does this article really require three almost indistinguishable photos? Convince me otherwise or I will delete two of them Mtpaley (talk) 23:01, 16 August 2009 (UTC)

Delete away.--Anss123 (talk) 11:39, 17 August 2009 (UTC)

External links modified

Hello fellow Wikipedians,

I have just added archive links to one external link on Motorola 68040. Please take a moment to review my edit. If necessary, add {{cbignore}} after the link to keep me from modifying it. Alternatively, you can add {{nobots|deny=InternetArchiveBot}} to keep me off the page altogether. I made the following changes:

• Added archive https://web.archive.org/20120216013510/http://www.bbs.ingedigit.com.ve/TechInfo/68040.Microprocessor.html to http://www.bbs.ingedigit.com.ve/TechInfo/68040.Microprocessor.html

When you have finished reviewing my changes, please set the checked parameter below to true to let others know.

This message was posted before February 2018. After February 2018, "External links modified" talk page sections are no longer generated or monitored by InternetArchiveBot. No special action is required regarding these talk page notices, other than regular verification using the archive tool instructions below. Editors have permission to delete these "External links modified" talk page sections if they want to de-clutter talk pages, but see the RfC before doing mass systematic removals. This message is updated dynamically through the template {{source check}} (last update: 5 June 2024).

• If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
• If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers. —^{cyberbot II}_{Talk to my owner:Online} 05:54, 27 August 2015 (UTC)

"MC status"?

In the 68LC040 section, it says: "This new mask was introduced in mid-1995 and converted the 68LC040 chip to MC status." What does "MC status" mean? It isn't mentioned anywhere else in the article. The link at the end of that sentence refers to "MC-qualified", but it doesn't explain what that means either. -- Hux (talk) 17:04, 18 January 2023 (UTC)