XC / MC: correct, MC are the newer ones.

RC has nothing to do with LC or EC. LC or EC would be written between "68" and "040". RC says that it's a PGA ceramic package.

Not fully sure what the last letter is. All my 040 CPUs have an "M" there. Probably not important.

ps: be careful with overclocking. Most Macs change their Gestalt ID if you overclock them. If you have a MacOS version installed that uses an enabler to run on that system, it may no longer boot after that. And Apple A/UX may tell you it can't run on that system if overclocked due to the different ID.

In regard to counterfeits, sorry if this sounds like a platitude, but if the deal is too good to be true, it most likely is. It's getting incredibly hard to find genuine, non-LC 040s. A lot of advice coming from the 68kMLA forum is along the lines of, "unless you're doing 'work' that seriously benefits from an FPU, an LC040 is perfectly acceptable." but to each their own.

In regard to chipping, AppleFool's collection of OC and clock chip guides will be pretty decent: http://www.applefool.com/clockchipping/q630.html

Also, you've probably thought of this, but the 040s run HOT when you get to and above their ratings. A 40 MHz chip will definitely require a heatsink, and possibly a fan (I haven't looked into that style of case since my family owned a Performa back in '96, so I don't know what the airflow is like. The heatsink on my DayStar 040 (running at 40 MHz) was always above a comfortable-to-touch temp threshold, though that's subjective, I guess.

Good luck.

XC68040RC40M

That is an '040 CPU with an FPU; I'm not sure why you'd think otherwise? The LC630 has the FPU less '040 while all Quadra 630 variants had full.

- RC is the full chip
- LC lacks FPU
- EC lacks FPU and MMU

Run a benchmark such as Norton System Info under your Q630 to confirm

Given my intent with Apple A/UX, an FPU should make a difference, and an extra 7mhz will be greatly appreciated.

How is this for a heatsink

It measures at 45mm by 45mm, while the NXP claims the chip is 47.2mm by 47.2mm.

Concerning derSammler's warning... Will doing a clean install after overclocking circumvent the 'enabler' issue? Current OS on the system is kanjitalk 7.5.1, which I believe is just the Japanese version of System 7.5.1.

OP has a typo. LC is stamped on the chip's title. I forget the exact model, I can check the physical system tomorrow. I wrote Quadra just out of my poor memory. It is some type of Performa or LC. Maybe someone has messed with the interior. After all, it is running kanjitalk, and this was a unit intended for the USA market.

smevans526 wrote on 2020-02-27, 22:01:

Concerning derSammler's warning... Will doing a clean install after overclocking circumvent the 'enabler' issue? Current OS on the system is kanjitalk 7.5.1, which I believe is just the Japanese version of System 7.5.1.

While a valid concern when chipping, according to AppleFool, on this model (series):

Unlike the Q605 modification, the gestalt does not change at the different speeds, so there are no software install problems, and there will never be any software install problems on a modified 63x machine in the future.

re: the model name, I wondered 😀 I've never seen a Quadra with a "CD" appendage on the model name/plate, which was more prevalent on the LC and Performa series. Makes sense now.

In Amiga world many people upgraded their "turbo" cards from 68040 to 68060. You may ask on their forums. For example Apollo 1240 in standard configuration had MC68040 40MHz.

Byrd wrote on 2020-02-27, 21:46:

- RC is the full chip

*facepalm*

No, it's not. Read my post above, RC has nothing to do with FPU/MMU. LC and EC have RC in the name as well.

derSammler wrote on 2020-02-28, 09:14:

Byrd wrote on 2020-02-27, 21:46:

- RC is the full chip

*facepalm*

No, it's not. Read my post above, RC has nothing to do with FPU/MMU. LC and EC have RC in the name as well.

I'm with derSammler on this. It is a Performa 631CD. Isn't that nice, I get an extra memory slot compared to the standard 630 - 48mb can be added (52mb total, accounting for the soldered-in 4mb).

Anyway, the chip is XC68LC040RC33B.

jessenator wrote on 2020-02-27, 22:53:

While a valid concern when chipping, according to AppleFool, on this model (series): […]
Show full quote

smevans526 wrote on 2020-02-27, 22:01:

Concerning derSammler's warning... Will doing a clean install after overclocking circumvent the 'enabler' issue? Current OS on the system is kanjitalk 7.5.1, which I believe is just the Japanese version of System 7.5.1.

While a valid concern when chipping, according to AppleFool, on this model (series):

Unlike the Q605 modification, the gestalt does not change at the different speeds, so there are no software install problems, and there will never be any software install problems on a modified 63x machine in the future.

re: the model name, I wondered 😀 I've never seen a Quadra with a "CD" appendage on the model name/plate, which was more prevalent on the LC and Performa series. Makes sense now.

Awesome. Thank you sir. Who needs to fork over the cash for an 840 when you can just overclock a 630?

smevans526 wrote on 2020-02-28, 17:05:

Awesome. Thank you sir. Who needs to fork over the cash for an 840 when you can just overclock a 630?

Good luck! And definitely source a heatsink. Hopefully you're a steadier hand with a soldering iron than I am. I tried something similar on a StarMax 5500 board (all for a 250 to 300 MHz bump) and …didn't succeed. I might send it away to get redone eventually. Having a magnifier and a quality iron/hot air station, tweezers, and lots of flux is key.

If you've done stuff like this before, feel free to disregard this advice 😉 , but I'd watch a few recapping videos by Bruce over at Branchus Creations. It's his business to do a lot of Mac re-caps and the concepts are the same, especially with this SMD components. A magnifying lens is also quite helpful, or a microscope in his case.

Edit: I think the confusion surrounding the "RC" designation comes form the Wikipedia article on the Motorola 68040. Whoever authored that Wikipedia article tried to simplify it for us mere mortals who don't dive deep into the CS/EE world, perhaps? ¯\_(ツ)_/¯

I ordered the processor along with the heat-sink I linked to in an above post. There is a nearby electronics shop that should be able to move the two resisters, further, I will ask them to mount the heat-sink. Given the board's lack of screw sockets, I will ask them to use hot glue (along with a thermal paste).

My first step is to install the new chip and run Apple System Profiler to see whether or not the FPU registers (or, perhaps, whether an 'LC' registers). If there is an FPU, I will be convinced enough that the chip is legit and take it to the shop. I had a set of 16mb and 32mb FPM SIMMs sitting around, and already have the thing running at 52mb RAM.

Before putting on A/UX, I think I will experiment a bit with gaming on System 7. I am curious how the likes of Marathon and Wolfenstein will operate on a 40mhz Quadra. I have seen the videos of them running on a Quadra 700, and performance is atrocious. Will the extra 15mhz make a difference? We shall see.

A good test will be King's Quest VII. I am not a fan, but nevertheless, I remember it slowing a friends 486-DX2 (66mhz) system to crawl. Perhaps memory was the issue, but I have found similar complaints on reviews. I guess I am getting at 'How does the 68040 compare relative to 486?'. Looking at a few games is a laughable and anecdotal comparison, but nevertheless what interests me.

It may have nothing to do with '040 vs 486, and simply be down to if Sierra programmed more efficiently for the '040 or 486 version.

Two other gaming issues of interest will be how a 40mhz handles 'relatively' vintage games. In my memory Space Quest IV was unplayable on a DX2, even after slowing-down in the options menu. The speed would skyrocket at certain points regardless. But on this 33mhz kanjitalk 7.5.1 system, the speed meter still seems responsive and proportional. Monkey Island will also make a good test. Again, it probably has nothing to do with CPU, and just how the game was programmed.

My ranting to myself also makes me wonder if there are games with both a PPC and 68k binary, and the 68k version runs better. I could see this happening if the studio emphasized development first on the 68k version, and then rushed the PPC version. Would not surprise me if some of Sierra games behaved like this; Their x86 games always have some bugs concerning the processors - clock issues, if memory serves correct.

Not sure why you think an FPU is important for compiling c code...

kolderman wrote on 2020-02-28, 21:29:

Not sure why you think an FPU is important for compiling c code...

I am making assumptions.

My mission is to (try to) use A/UX's C compiler to run GCC. For the TBD applications I may be wandering into, I am assuming the FPU will be beneficial.

But strictly with C, I am also assuming that any calculation (and I view C as a set of calculations requested by a user) would be performed with the FPU.

Please correct me if I am mistaken; This is one reason I want experience programming on vintage hardware.

It looks like A/UX requires an FPU, which, granted, is a different issue from compiling C.

smevans526 wrote on 2020-02-28, 21:53:

But strictly with C, I am also assuming that any calculation (and I view C as a set of calculations requested by a user) would be performed with the FPU.

Err...no, that's not how CPUs work. The FPU stands for floating-point unit, which performs calculations on numbers with something after the decimal point. This is only really useful for things like science and computer games (well and lots of other things TBH). But not compiling code.

Not that I am saying you should not get a CPU with an FPU!

Hey, not to further dampen your parade, but I had a funny feeling about it and did some checking with the 68kMLA OGs and they pointed me to this chart: http://aux-penelope.com/hardware.htm

A/UX is extremely picky about what hardware/chipsets it will work with. I'm afraid you cannot force the LC/Performa 6xx chipsets to work for A/UX even with a full 040 upgrade, or any sort of theoretical software workaround.

kolderman wrote on 2020-02-29, 00:16:

smevans526 wrote on 2020-02-28, 21:53:

But strictly with C, I am also assuming that any calculation (and I view C as a set of calculations requested by a user) would be performed with the FPU.

Err...no, that's not how CPUs work. The FPU stands for floating-point unit, which performs calculations on numbers with something after the decimal point. This is only really useful for things like science and computer games (well and lots of other things TBH). But not compiling code.

Not that I am saying you should not get a CPU with an FPU!

So if I am using the float or double datatypes, then their calculations are not performed by the FPU?

smevans526 wrote on 2020-02-29, 02:22:

kolderman wrote on 2020-02-29, 00:16:

smevans526 wrote on 2020-02-28, 21:53:

But strictly with C, I am also assuming that any calculation (and I view C as a set of calculations requested by a user) would be performed with the FPU.

Err...no, that's not how CPUs work. The FPU stands for floating-point unit, which performs calculations on numbers with something after the decimal point. This is only really useful for things like science and computer games (well and lots of other things TBH). But not compiling code.

Not that I am saying you should not get a CPU with an FPU!

So if I am using the float or double datatypes, then their calculations are not performed by the FPU?

When the program runs, yes. Not when compiling though. Compilation is basically text translation.