I don't believe anyone is complaining about 24-bit addressing mode lacking the overclock. People who haven't tried it yet are just curious. And that's why I strongly encourage everyone to give it a try to see how great it is. And while Mustermann did kick off the project, you brought it to full completion, @phipli. Without your incredible work on the software, most people wouldn't be able to enjoy it at all.

So when people ask, I answer in a matter-of-fact way. It's a fact the overclock doesn't work in 24-bit mode. The reason is nice to know, I guess, but it doesn't really matter. For as you just said, the only thing that really NEEDS 24-bit mode is the IIe Card, and most people don't have one of those. System 6 also needs it, but I can't run System 6 on my LC575 board. Not sure about the 475, but even if you could run System 6, it's written in assembly language and flies even on processors with low clock speeds. Besides, a lot of the games you'd want to run with a very high clock speed only work on System 7 anyway, and they need RAM too, so you'd want 32-bit addressing ON for them.

But in terms of compensation, I do apologize for my oversight...

Folks, you can Donate to @phipli via his website here:

I donated just now. And I don't say that to seek praise. I'm happy to support your work and hope my donation will inspire others to respond in kind.

>> JDW said:

I don't believe anyone is complaining about 24-bit addressing mode lacking the overclock. Click to expand...

>> JDW said:

I donated just now. And I don't say that to seek praise. I'm happy to support your work and hope my donation will inspire others to respond in kind. Click to expand...

Yes, sorry, I was just trying to explain that there are diminishing returns on the investment of time, I wasn't begrudging it, but there gets a point where the benefits are so small that it isn't worth the time it takes (which would likely be much less time for someone else, but this sort of thing isn't my bread and butter). Thank you for your generosity, I'm not looking for payment, nobody should feel obliged, there is a significant reward in seeing people getting use out of it as is.

Generally I recommend using the Control Strip module, it gives instant speed changes and is more readily on hand. It makes changing speed a bit more like changing monitor bit depth.

The reason the spicy o'clock benchmarks faster in RAM, ROM and VRAM (which shows up especially in the video benchmarks) is that it keeps the default timings your computer is set to (for most of us, that means 33MHz timings). This means you're pushing stock hardware harder. The software overclock has been designed to give more of an official, released machine, style experience - it aims to adjust timings to keep them in specification for the LC 475 hardware timings as much as possible - i.e. for RAM ROM and VRAM. This should mean it is compatible with more machines. With this in mind, the LC 475 set to 40MHz scores almost identically to the Quadra 840av in CPU performance - the only official 40MHz Quadra Apple released.

With regards to using the Advanced settings, I'd say "your mileage may vary". By changing those settings, most users would be pushing their hardware outside of the specification. Good if you've replaced the soldered RAM and fitted fast VRAM and ROM, but likely to cause instability for those of us that have soldered 80ns RAM and less than stellar VRAM.

The default settings do unfortunately target the LC 475 as I don't have a 575 and actually didn't originally realise they were 33MHz or that they had 70ns soldered RAM - I'm not overly familiar with the AIO scene - my SE/30 is my newest non-iMac AIO.

>> phipli said:

I don't have a 575 and actually didn't originally realise they were 33MHz or that they had 70ns soldered RAM Click to expand...

I have the opportunity to purchase some OKI brand 60ns rated MSM514400D-60SJ RAM chips that would allow me to desolder my 70ns KM44C1000BJ-7 chips and do a swap. Not being sure if a mere 10ns of faster speed would really matter, I sent Mustermann a PM to gather his thoughts. If he feels it's probably not worth it, I won't bother. There's always a risk of damaging something when you desoldering that many chips using hot air, but I suppose there is something to be said about taking the risk "for science." Goodness knows I did that when I desoldered an EEPROM on my rare 10MB MMI drive in my HyperDrive Mac in order to help out a fellow community member repair his non-functioning drive. Sometimes you just have to boldly go where no man has gone before! [raised-hand-with-par] And that's precisely what you did with your work on Soft 475 Overclock, so thanks again!

>> JDW said:

I have the opportunity to purchase some OKI brand 60ns rated MSM514400D-60SJ RAM chips that would allow me to desolder my 70ns KM44C1000BJ-7 chips and do a swap. Not being sure if a mere 10ns of faster speed would really matter, I sent Mustermann a PM to gather his thoughts. If he feels it's probably not worth it, I won't bother. There's always a risk of damaging something when you desoldering that many chips using hot air, but I suppose there is something to be said about taking the risk "for science." Goodness knows I did that when I desoldered an EEPROM on my rare 10MB MMI drive in my HyperDrive Mac in order to help out a fellow community member repair his non-functioning drive. Sometimes you just have to boldly go where no man has gone before! And that's precisely what you did with your work on Soft 475 Overclock, so thanks again! Click to expand...

I'd recommend just desoldering what is there rather than putting faster RAM back. But you still have the risk of melting some of the plastic in the surrounding area.

I desoldered the onboard on my Centris / Quadra 610 and on one of my 650s. I probably will on my 475 at some point. I'd like to get to the point that I have a fast ROM, fast VRAM and fast RAM just to see where the limits are.

Do you have a fast ROM in yours?

>> phipli said:

Do you have a fast ROM in yours? Click to expand...

The stock 575 boards don't have a ROM socket, but the good news is you can solder one in because the pads are there! So I did. And I did that because I received a compatible ROM SIMM designed by @Jockelill , which is rated for 55ns. It has some quirks in that I can't seem to boot from it, but I did confirm it kills the RAM checking at cold boot. Even so, the RAM checks normally finish before the CRT lights up anyway, so that doesn't matter so much. It's mainly to get a faster access time than the soldered ROM.

>> JDW said:

Even so, the RAM checks normally finish before the CRT lights up anyway, so that doesn't matter so much. It's mainly to get a faster access time than the soldered ROM. Click to expand...

That is actually something to keep in mind actually - I've found that my Centris 650 wont run at 40MHz if I use my 60ns 128MB SIMMs at all. With other (smaller) SIMMs, I can get 50MHz.

It is always worth trying a few different SIMM, although the limit you (JDW) specifically are hitting doesn't sound like what I've seen as RAM limits. When my RAM isn't fast enough I tend to have sad mac (way too slow), freeze just before the first extension loads, or random freezes (including ones where the mouse keeps going but nothing responds). Also worth noting that some of my 70ns SIMMs are happy to run faster than my 60ns SIMMs, and my single 50ns 72pin SIMM actually isn't any faster than the best of my 60ns SIMMs. I've heard that timings are tighter on EDO SIMMs, so while macs don't need EDO (and some aren't compatible with them (although the 475 and 575 are)), it might be a small advantage.

But more generally, what I mean is, if someone is seeing those sort of symptoms, try a different RAM SIMM.

Updated version of the Control Strip Module available here :


Most changes (other than the typo I'd made in the menu) are in the background, but this new version seems to be more stable at higher speeds due to refined timings.

I've added a couple more speeds to the menu for people that want to live on the edge.

I've made it use different timings for computers that shipped fitted with 70ns RAM.

One impact of the changes is that if you run it on an LC 575 at 40MHz, the CPU score now exceeds the Quadra 840av by a small amount.

Sad news.

Based on the fact that the Performa 630 is a derivative of the LC 475, and its memory controller, the F108 is apparently based on the MEMCjr... plus it uses resistors to set the clock speed as on the LC 475 using a clock generator of the same brand... sounds hopeful...

I made a special test version of the software overclocking software. Because I knew that the video subsystem was very different, I made multiple options that variously ONLY changed the clock speed by overriding the strap resistors, others that didn't change video settings, and lastly left the normal options in the list as well. I then made sure the software functioned as expected on my LC 475 (it did)... and moved it onto my fairly stock Performa 630...

Sadly running it just triggers a Type 1 error and the speed doesn't change. Looks like Apple removed the ability to program the clock from the memory controller, or possibly moved the registers around. If it is the latter, we're stuck unless someone is able to get a copy of the F108 ERS document, or has other document describing the F108 registers.




A real shame, because this would have opened up software overclocks for the 630, 580, 6200, 5200, 6300, 6300 and related machines.

I also checked out the ERS for the memory controller on the LC III, and the register that sets the speed is listed as Read Only, which doesn't sound promising. This means software overclocks aren't happening wrt the LC III, IIvi, IIvx or P600. The chip only seems to support 16, 25 and 33MHz anyway, so the LC III+ mod is still the best solution for this. (Mods to the IIvx family machines are way more complex as you have to add in the whole clock doubler circuit to convert a IIvi to a P600, and the cache circuit to convert to a IIvx).

Off topic - the IIvx isn't as bad as people remember - yes it was too expensive and replaced by the cheaper and way, way better Centris 650 a few months later... but it actually benchmarks slightly ahead of a IIci stock, and the QuickDraw performance is better than a IIci, and it even supports Thousands of colours with a VRAM upgrade. As a retro computer, they're a good alternative to a IIci. The slower bus means that if you upgrade them with processor accelerator card you wont quite get the same performance as with the same upgrade in a IIci, that is the only real disadvantage (because lets be honest, you don't need 128MB of RAM).

If you have a IIvx/IIvi/P600 schematic, I'd be really interested btw. I'd love to see if it is possible to bump the bus to 25MHz or 33MHz (I suspect it wouldn't run).

The video is now out! Let the good times roll.


Humble thanks to @phipli , Mustermann and the original contributors, for the incredible work that brought 475 Overclock to the masses!

>> phipli said:

If it is the latter, we're stuck unless someone is able to get a copy of the F108 ERS document, or has other document describing the F108 registers. Click to expand...

A lot more work, but might it be possible to work backwards through the firmware?

Find the references to the known register in the LC475 firmware. Identify the surrounding code, used to set clock speed.

Then search for similar code in the Q/LC/P 630 firmware and then look at which register it uses?

>> JDW said:

The video is now out! Let the good times roll. Humble thanks to @phipli , Mustermann and the original contributors , for the incredible work that brought 475 Overclock to the masses! Click to expand...

Thank you for this. I had totally missed this development and only found out about it from your video. I was able to clock my Mystic CC to 45Mhz without any stability issues.

>> trag said:

A lot more work, but might it be possible to work backwards through the firmware? Find the references to the known register in the LC475 firmware. Identify the surrounding code, used to set clock speed. Then search for similar code in the Q/LC/P 630 firmware and then look at which register it uses? Click to expand...

I'm afraid that the 475 ROM code doesn't call it to my knowledge - it was a silent, unused feature. The clock chip has baked in default speeds set by the voltages it sees on its pins, the serial functionality is perhaps left over from when they were doing development before they got the custom clocks with the custom speeds baked in.

As shipped the speed is set on power up by the resistors directly, no serial comms needed.

I made the video announcement on FaceBook as well, and one gentleman posted his video in reply:


He posted that to tell me that even though we both downloaded Doom II from Macintosh Garden, he gets FPS info in the upper left corner of the game. Also, while it's VERY hard to see due to his blown out menus in the video, it seems he has different options than I do too. He said he doesn't get any option to change the Graphics Detail either, like I do.

The differerence? He's using an LC475. I'm using an LC575.

I'm posting about that in this thread so that when people test 475 Overclock on their 475 and 575 machines and then run Doom II (downloaded from Macintosh Garden), perhaps we can finally figure out why some people get FPS and Graphics Detail options while others do not.

>> JDW said:

I made the video announcement on FaceBook as well, and one gentleman posted his video in reply: He posted that to tell me that even though we both downloaded Doom II from Macintosh Garden, he gets FPS info in the upper left corner of the game. Also, while it's VERY hard to see due to his blown out menus in the video, it seems he has different options than I do too. He said he doesn't get any option to change the Graphics Detail either, like I do. The differerence? He's using an LC475. I'm using an LC575. I'm posting about that in this thread so that when people test 475 Overclock on their 475 and 575 machines and then run Doom II (downloaded from Macintosh Garden), perhaps we can finally figure out why some people get FPS and Graphics Detail options while others do not. Click to expand...

Hum, my understanding was that Doom and Doom II showed FPS on mac, but you had to reduce the size from full screen.

>> phipli said:

Hum, my understanding was that Doom and Doom II showed FPS on mac, but you had to reduce the size from full screen. Click to expand...

Believe me, I tried every size the game offers, down the to teensiest tiniest size you can hardly even see. FPS is a no-go on my machine. I am desperate to know why.

Has anyone tested current draw and/or voltages on the 5V line at different frequencies? I thought @JDW measured voltage at various frequencies (maybe for Spicy O Clock), but I can't find a video with that information.

>> David Cook said:

I thought @JDW measured voltage at various frequencies (maybe for Spicy O Clock), but I can't find a video with that information. Click to expand...

Not at different frequencies, no. I just confirmed at my primary Spicy clock speed that all was well with voltages.

But if any voltage measurements are done by anyone, it's important to use a scope (not a handheld DMM meter) and test while idle and while the CPU is being loaded by Norton System Info or maybe the Daystar PowerDemo. That's because the higher you overclock, the more voltages to the CPU tend to drop. And by drop I mean very short, almost instantaneous drops in voltage that are very spiky in nature (transient drops). This is why low ESR OS CON caps played a role in allowing myself and Kay Koba to overclock to 50MHz. Prior to using OS CON (roughly 20mΩ ESR), Kay had solid Tantalum capacitors (roughly 700mΩ ESR, mind you) on his LC575 board and he told me he couldn't overclock much beyond 45MHz or so.

>> JDW said:

Not at different frequencies, no. I just confirmed at my primary Spicy clock speed that all was well with voltages. But if any voltage measurements are done by anyone, it's important to use a scope (not a handheld DMM meter) and test while idle and while the CPU is being loaded by Norton System Info or maybe the Daystar PowerDemo. That's because the higher you overclock, the more voltages to the CPU tend to drop. And by drop I mean very short, almost instantaneous drops in voltage that are very spiky in nature (transient drops). This is why low ESR OS CON caps played a role in allowing myself and Kay Koba to overclock to 50MHz. Prior to using OS CON (roughly 20mΩ ESR), Kay had solid Tantalum capacitors (roughly 700mΩ ESR, mind you) on his LC575 board and he told me he couldn't overclock much beyond 45MHz or so. Click to expand...

If I remember, you shouldn't expect big differences between when the CPU is busy and when it isn't with this era of chips, although I'd be interested to know for certain.

My PSU isn't recapped, so my machine isn't a good example.

>> phipli said:

My PSU isn't recapped Click to expand...

[face-with-finger-cov] Shhh!

He didn't mean that folks. Just a late April fools joke.

>> JDW said:

And by drop I mean very short, almost instantaneous drops in voltage that are very spiky in nature Click to expand...

That's very interesting. Originally, my thoughts were just as to whether or not the power supply was becoming strained, and how much extra heat was being dissipated by the total board. But, your comment makes me wonder whether I need to learn more about measuring power at across frequencies. I suppose I can still measure the AC input regardless, but I had hoped to measure the 5V rail directly.