Synchr030/S 256MB SDRAM Accelerator for the SE/30

[Huxley]

This looks incredibly cool, and like a fantastic companion to the Booster I already purchased from you! I do have a couple beginner-level questions:

• Physically speaking, is it possible to install this Synchr030 + the Booster + a MacCon Ethernet card into an SE/30, or will I be running out of room inside?
• My Booster is version 1.1, but I see some references online to a version 2.0. Is there any compatibility issue or other difference between 1.1 and 2.0 that I should be aware of?

 

[zigzagjoe]

This looks incredibly cool, and like a fantastic companion to the Booster I already purchased from you! I do have a couple beginner-level questions:

• Physically speaking, is it possible to install this Synchr030 + the Booster + a MacCon Ethernet card into an SE/30, or will I be running out of room inside?
• My Booster is version 1.1, but I see some references online to a version 2.0. Is there any compatibility issue or other difference between 1.1 and 2.0 that I should be aware of?

I make the booster accelerators. A friend lent me one of the synchro cards for testing. It's a neat card!

However, the synchro is only compatible with PDS versions of the booster, and only when the booster is the first card in the pds slot, with the synchro on top. Socketed boosters will not work at all, nor will PDS versions plugged into the top slot of the synchro (without modifications). I haven't released the v2 boosters yet, but the compatibility with regards to synchro is unchanged.

Booster pds + macron + synchro is going to be really tight mechanically. It may work but it will be close if it does. I have found the original MacCons have some strange effects on the bus electrically and 3 card combos involving MacCons are often iffy. I wouldn't recommend it unless @ymk has tested that scenario.

 

[ymk]

However, the synchro is only compatible with PDS versions of the booster

What happens with the socketed version?

Physically speaking, is it possible to install this Synchr030 + the Booster + a MacCon Ethernet card into an SE/30, or will I be running out of room inside?

I haven't tested this configuration. I have the v1.2 Booster and can confirm that (unmodified), it must be on the bottom. If you're willing to cut a clock trace on the Booster and insert an RC filter, it will work in either position.

Note that if you place Synchr030 on top, it should be shielded from the flyback wire or the card could be damaged.

 

[zigzagjoe]

What happens with the socketed version?



I haven't tested this configuration. I have the v1.2 Booster and can confirm that (unmodified), it must be on the bottom. If you're willing to cut a clock trace on the Booster and insert an RC filter, it will work in either position.

Note that if you place Synchr030 on top, it should be shielded from the flyback wire or the card could be damaged.

It's the same behavior as if a pds booster was plugged into the top of the synchro (which would track as that's essentially the same relationship). Pretty sure the synchro needs the little bit of delay the buffered clock on the booster introduces, and the RC filter ends up adding a similar delay. I'm not set to easily introduce a delay to the cpu clock inside the CPLD, so I can't immediately test this. The newly implemented /STERM logic also doesn't agree with the synchro, though it works with micron xceed and iisi, but that may be my fault.

I personally make metal support brackets to both support the card stack and also shield it from the anode connection.

 

[ymk]

Pretty sure the synchro needs the little bit of delay the buffered clock on the booster introduces, and the RC filter ends up adding a similar delay.

If that were the case, Synchr030 wouldn't work by itself. I think the longer (unbuffered) clock path through Synchr030 breaks the Booster. The RC filter doesn't delay the clock to the Booster, it advances it, counteracting the longer signal path.

It looks like your metal shield should work as long as it doesn't contact the card.

 

[zigzagjoe]

If that were the case, Synchr030 wouldn't work by itself. I think the longer (unbuffered) clock path through Synchr030 breaks the Booster. The RC filter doesn't delay the clock to the Booster, it advances it, counteracting the longer signal path.

It looks like your metal shield should work as long as it doesn't contact the card.

The part that gives me pause is that the socket version is taking a direct unmodified clock off the LB. And same symptom. Could be a marginal timing on control signals also, where the delayed clock to the synchro or advanced clock to the booster is causing them to rise earlier. I have to send the card to its owner so I don't anticipate getting to the bottom of this.

For my use cases I put an offset bit with a threaded hole to pick up the hole on the card. So that's the only part that makes contact. But these brackets have to be made to suit each particular combo of cards in a stack, this is just an example.

 

[ymk]

The part that gives me pause is that the socket version is taking a direct unmodified clock off the LB. And same symptom. Could be a marginal timing on control signals also, where the delayed clock to the synchro or advanced clock to the booster is causing them to rise earlier.

That's a good point. The P33 has the opposite issue; Synchr030 works only on the bottom and not downstream of the riser's clock buffer. It could be that Synchr030's signals are borderline early for the Booster and late for the P33. Getting both Booster configurations to work without mods might break the P33.

Did you test the socketed Booster with /DSACK or /STERM?

I may add clock tuning options to the next rev.

 

[zigzagjoe]

That's a good point. The P33 has the opposite issue; Synchr030 works only on the bottom and not downstream of the riser's clock buffer. It could be that Synchr030's signals are borderline early for the Booster and late for the P33. Getting both Booster configurations to work without mods might break the P33.

Did you test the socketed Booster with /DSACK or /STERM?

I may add clock tuning options to the next rev.

Same with either /DSACK or /STERM. Seems like /STERM isn't actually used until the extension loads. On one of the new cards that supports /STERM, it locks when the extension loads. But I need to verify that I've got timing right as I only have the IIsi and Micron Xceed as /STERM test cases.

It could also be that Synchro requires strobes to be perfectly inline with standard 68030 timing (ie. AS transitioning halfway thru a clock cycle, etc). I know booster is a bit loose on those, it also will issue /AS again quicker than standard 68030 timing, and it also does not at all delay /DS by a cycle on writes.

 

[ymk]

Seems like /STERM isn't actually used until the extension loads.

Right. That's when 2-cycle access begins.

I know booster is a bit loose on those, it also will issue /AS again quicker than standard 68030 timing, and it also does not at all delay /DS by a cycle on writes.

With accelerators, only /AS timing matters. /DS is ignored.

 

[Joopmac]

I cant get the combination of carrera040+adapter yet to work, both ways…
The n-shaped way is not possible with Original PSU, with a picopsu fits

 

[croissantking]

View attachment 81970I cant get the combination of carrera040+adapter yet to work, both ways…
The n-shaped way is not possible with Original PSU, with a picopsu fits

The Synchr030 hasn’t been tested with an 040 upgrade yet, I don’t think.

It most likely would, as long as the extension loads. In 24-bit mode, the extension operates on the 030 MMU directly, which probably won't work on an 040. The accelerator would also need to work with the BMOW ROM.

Mechanically, it fits, though it places the card very close to the flyback wire, so shielding would be needed. However, starting up with this configuration results in death chimes and simasimac. I don't know if there's a riser setting that would help.

@ymk tried the card with Bolle’s riser and it death chimed, so it may not be compatible.

 

[ymk]

I cant get the combination of carrera040+adapter yet to work, both ways…
The n-shaped way is not possible with Original PSU, with a picopsu fits

Thanks for trying it out.

I'm working on a PDS to cache adapter for Bolle's PowerCache P33 that enables both fast SDRAM and L2 cache. Here are some early Speedometer 4.02 benchmarks:

 

[Bolle]

Looks like it only boosts FPU scores a bit.
Have you made sure the cache actually was active? Does toggling the cache on and off in the control panel change the results?

 

[ymk]

Looks like it only boosts FPU scores a bit.
Have you made sure the cache actually was active? Does toggling the cache on and off in the control panel change the results?

No FPU is installed in the P33.

The control panel does make a difference. I use the MacBench 3.0 CPU benchmark as a quick test:

Memory	L2 Cache	Score
DRAM	Disabled	1.13
SDRAM	Disabled	1.32
DRAM	Enabled	2.10
SDRAM	Enabled	2.33

The P33 doesn't cache slot-C space, so the adapter translates addresses.

 

[ymk]

MacBench 3.0 results for the P33. Disk and graphics results are expanded. The 100% system is a P33 with cache enabled and DRAM.