Warp Engine 3040 (for A3000) RAM configuration

[RDP]

I have a 3040 in my A3000. I have pretty well identical 32meg SIMMS in the two available slots, however (and unfortunately for MAC emu.....) I cannot get the ram to configure as it says it should according to the manual, which is as a single contiguous block, I can only ever achieve two 32meg allocations.

I'm wondering if anyone can confirm if due to the 3040 having two less slots than the 4040, and that they are labeled SIMM 1 & SIMM 3 (instead of 1 & 2, or 3 & 4) that it is actually impossible to achieve a single contiguous block of ram because of that? If anyone can prove that its possible based on their own config, I would be grateful to hear about it.

thanks,
RDP

[Hedeon]

Weird. I would expect an allocation at 0x8000000-0x8400000 or something like that. What are the allocations and do they have the same memory attributes?

[RDP]

Thanks for your reply. According to SYSINFO:

1st:

Start $0A000000
End $0BFFFFFF
Size 32.0MB
Lower bound $0A000020
Upper bound $0C000000
First address $0A520530
Largest block 33554400
Chunks 59


2nd:

Start $0E000000
End $0FFFFFFF
Size 32.0MB
Lower bound $0E000020
Upper bound $10000000
First address $0E000020
Largest block $33554400
Chunks 1


In case of input error I attach screenshots as well.


thanks,
RDP

[RDP]

Just a friendly bump.

RDP

[grelbfarlk]

Have you tried another set of 32MB SIMMs?
The Warpengine in my A4000 has the memory at 0x08000000-0x0FFFFFFF as Hedeon thought.
Are jumpers B,C off?
The manual states:
3.6 Contigious Memory
Any time you install the same type and size of SIMMs, you will get contigious memory. For example, if you install two 32MB SIMMs, you will get 64MB of contigious memory. Now, if you install a 16MB in the third socket (SIMM2), this SIMM will also be contigious, giving you 80MB of contigious memory. If you were to add a fourth SIMM module of 16MB or less, it will result in a seperate memory block that is not contiguous with the 80MB block.

The manual was for the A4000 version the A3000 addendum doesn't mention anything about this.
Looking at Chucky's guide, there is a "5th" SIMM slot on the board for the A3040 version between slots 3 and 4? Was your board originally an A4000 version that someone downgraded?
https://wordpress.hertell.nu/?p=380

[RDP]

Thanks for your reply.

I have not tried another set of SIMMS as I don't have any, I would need to purchase (should not be too hard to do that) if it comes to it.

To triple verify the jumpers I would need to open the 3000D, and my fingers are still cut up from the session I had with it earlier this week, so I'll need to defer that for awhile.... in the meantime I strongly believe my setup is:

B: OFF
C: OFF

I believe that to be correct to the manual as:

B OFF = Double sided (which would be period correct for a 32meg SIMM).
C OFF = 16/32M (and I have 32 meg SIMMS, recognized as two 32meg SIMMS).

Since I'm successful in having two 32meg SIMMS recognized I believe the jumpers to be correct.

What I would be very very very interested in is if someone with a 4040 WE with 4 SIMM slots would try their memory in only slots 1 and 3 (same as my 3040 WE) and see if it remains contiguous. if it does, then I should try other SIMMS in mine. If it becomes split it means that a 3040 can never group the memory due to SIMM 1 and 3, which should be (SIMM 1 and 2, or 3, and 4), which it never will be.

My 3040 was bought as a 3040, and came with the manual addendum.

thanks!
RDP

[trixster]

I have a Rev3 4040 converted to 3040 in my A3000, so simms 2 and 4 removed, and with two identical 32mb simms I only ever get two 32mb blocks

I’ll see about removing two 32mb simms from the Rev1 4040 in my A4000 and get back to you

[grelbfarlk]

Trixster do you have SIMM5 populated with a socket? I'm wondering if there aren't changes to the ROM or CPLD for the real A3000 version.

[trixster]

Good question, I’ll have to look at my boards to see if either have that simm5 space

[trixster]

Ok, here are my two boards (photos from a few months ago)

4040 converted to 4030 - It seems to me there is a simm5 placement there as you can see the 3 close-together-holes under the lower memory module


4040, no 5th simm slot

[RDP]

Wow, really appreciate all the feedback! Can't wait to see if a slot 1 & 3 SIMM combo on the 4040 comes back as two chunks!

Here is my setup attached: The only differences I can see in the SIMMS are some GOLD vs. SILVER in the middle, one connection made on the right SIMM and not on the left SIMM, and the ID on the memory chips themselves is C vs. B. (of the larger sequence).

Bonus Q: Does anyone know what the L jumper is for?

Trixster: Given your experience which 040 to 060 adapters would you recommend? I'm considering one myself.


thanks,
RDP

[trixster]

The 040-060 adapters have worked well for me, I have 3 of them.

In order to use one in a 3000 it needs to be very low profile so it will fit under the disk drive shelf, and there is no room for a heatsink or fan unless a hole/slot has been cut into the metalwork - this is the case with my aa3000+ (by the case’s previous owner) so I can have a fan. An 060 @50mhz is generally fine without cooling though.



With cooling an 060 on a WarEngine is great - the one in my A4000 is running at 96mhz. The WarpEngine has a fast memory controller too compared to many other accelerators. Oh, and the scsi still works because the board/cpu clock is separate from the scsi clock

[RDP]

My 3000 case already has a hole cut for the 040 fan setup, so I'm good to go on height restrictions. I take it all three work equally well for you in practice?

thanks,
RDP

[trixster]

there is zero difference between the three adapters i have other than their construction, they simply act as an interface with vrm

[RDP]

Trixster, when you do find the time I'm very interested to hear the results of you only using two SIMMS in your A4000, in SIMM 1 & 3 slots.

thanks!
RPD

[trixster]

hi, sorry for the delay, it is difficult to find time to dismantle the machine

[RDP]

No need to be sorry, I was just wanting to keep this topic on the table, I'm very pleased to be finally able to test this theory. I can be patient, I've been waiting for a chance for easily over a decade!


RDP

[trixster]

@RDP

I was thinking about this this morning and I'm not certain im going to be able to get the info you're after.

At the moment my 4040 in my Mediator A4000 produces one contiguous 128mb block with it's 4 simms installed. However, due to a conflict between the Mediator and the WE's rom, I have to use the board with the L jumper removed (see post #23 onwards here http://eab.abime.net/showthread.php?...ght=warpengine). I believe this is because the Warp Engine SCSI is a separate virtual Z3 autoconfig board and this clashes somehow with the Mediator at boot. Even removing the WE's ROM does not fix this, only removal of the jumper seems to completely disable what the WE is trying to do at boot.

The further affect of this is that if i remove a simm, any simms after that are not auto-configured by the machine at boot, and any ram thereafter has to be maunally added to the system with the addmem command. As such, if I were to remove say simms 2 and 4, simm 1 would be added as a 32mb block, and simm 3 would be ignored by the system until i manually add it as a 2nd 32mb block.

I think you need someone with a normal busboard and WE 4040 to do this test for you.

[trixster]

Further to that, I do not think putting this 4040 into my AA3000+ will help either. The AA3000+ does boot "ok" with the L jumper enabled but the WarpEngine's virtual Z3 autoconfig board causes havoc with the machine's Firebird Prometheus busboard and the ZZ9000 and i'm not sure i can get to a command prompt to run showconfig... as such I run this machine with the L jumper removed on it's WE 3040, thus ram in this machine is configured with the addmem command during the workbench boot sequence. Not ideal for helping with your request, sadly.

[trixster]

Perhaps drop a line to EAB member A4000Bear? He had at least two WarpEngines and multiple A4000s, and I think wrote this article: http://members.iinet.net.au/~davem2/overclock/warp.html