VOGONS


First post, by Stevogamer

User metadata
Rank Newbie
Rank
Newbie

A few weeks ago I stumbled upon this Japanese article while researching about using slotkets and SMP for my other thread about the Supermicro P6DGE.

Link is in Japanese
http://www9.plala.or.jp/j-fuji/rv226/smp/geta.htm

I was particularly intrigued since the article mentioned that PL-IP3/T's could use coppermines. I thought this certainly wasn't possible.
So I tested the two that I did have. I have a revision 1.1 and a 2.0.

The 2.0 as expected didn't post with a SL5QV installed and the voltage set to 1.75V. But the 1.1 curiously did work. Interestingly the old documentation for early PL-IP3/T's mention a jumper for SMP which I assume is just because Powerleap copy and pasted the PL-IP3 manual and changed the headers to say PL-IP3/T. But considering this does anybody know if SMP worked on the 1.0 revision, since that did have a jumper in place of the blank solder pads with Pin 1 and 2 bridged on Rev. 1.1.

Below are pictures of revision 1.0 and 1.1

2023-02-18_15-56-40.png
Filename
2023-02-18_15-56-40.png
File size
682.4 KiB
Views
958 views
File license
GPL-2.0-or-later
2023-02-18_15-54-45.jpg
Filename
2023-02-18_15-54-45.jpg
File size
238.85 KiB
Views
958 views
File license
CC-BY-4.0

Edit: I also recall powerleap's website early on saying it did not support Pentium III-S's so tualatin SMP probably wouldn't be an option on these adapters. And I can confirm that my system was unstable when I did try to use a Pentium III-S 1.4GHz on the Rev 1.1. I was testing it on a BH6 with the SS bios and had the FSB lowered to 100mhz. The only slot 1 board that I have that supports 133mhz is a DFI TA-64B and that didn't like the powerleap at all the FSB was way high. So I assume the Rev 1.1 just doesn't supply the correct voltage.

My Rev 1.1 came with a 1.2GHz Tueleron for reference. My Rev 2.0 interestingly came with a QEJ0ES engineering sample for the Pentium III 1000EB although would that be a Coppermine-T since the 2.0 wouldn't take my normal coppermine? CPU-Z just says Coppermine.

i-img900x1200-1647772806huuveu406698.jpg
Filename
i-img900x1200-1647772806huuveu406698.jpg
File size
161.89 KiB
Views
949 views
File license
GPL-2.0-or-later

Edit 2: My idea was to desolder the existing bridge and adding jumper pins to have selectable modes just like described in the manual https://web.archive.org/web/20021122174912/ht … ructions%20.PDF

Edit 3: It's also interesting that in a post from 2010 by retro games 100, Powerleap adapter , they say that were able to get a SL5QV working on the 2.0 which matches the Japanese article. But my 2.0 wouldn't POST at all not even with the correct voltage setting. But it worked just fine in my ABIT Slotket !!! (I had to set the voltage, AUTO didn't work) and the 1.1 (Is auto only). I never expected the SL5QV to work on a FC-PGA2 socket anyways but strange that other people did get it working.

Also the only real difference that I can see between the 1.0 and the 1.1 is the use of Chino-Excel transistors on the 1.0 (and all the other common revisions) and Fairchild transistors on the 1.1. The Fairchild transistors seem slightly inferior in spec compared to the Chino-Excel's but not to a significant difference that would make the Pentium III-S's not work. I can't rule out that it is possible that the P3S's just don't work with my BH6. I did try recapping my BH6 which didn't help.

I don't have any other boards on hand at the moment (I have a BE6-II in storage but it needs a recap on the VRM and the DFI just doesn't work with the tualatins) and my Supermicro board is in shipping.

My guess is the SL5QV probably won't work in SMP on the 1.1 since it'll be missing the 56ohm pullup resistor required for coppermines but it will work with tualatins since the spec said it didn't need the pullup resistors. I'll be testing this in a few months once my motherboard arrives but I figured this info could be useful for others as well.
https://web.archive.org/web/20210802085908/ht … ime/slotket.htm

Last edited by Stevogamer on 2023-02-19, 18:10. Edited 6 times in total.

Reply 1 of 6, by Horun

User metadata
Rank l33t++
Rank
l33t++

Interesting. Sounds like a good idea.

Hate posting a reply and then have to edit it because it made no sense 😁 First computer was an IBM 3270 workstation with CGA monitor. Stuff: https://archive.org/details/@horun

Reply 2 of 6, by PARKE

User metadata
Rank Oldbie
Rank
Oldbie
Stevogamer wrote on 2023-02-19, 00:00:

Interestingly the old documentation for early PL-IP3/T's mention a jumper for SMP which I assume is just because Powerleap copy and pasted the PL-IP3 manual and changed the headers to say PL-IP3/T. But considering this does anybody know if SMP worked on the 1.0 revision, since that did have a jumper in place of the blank solder pads with Pin 1 and 2 bridged on Rev. 1.1.

There is a reference that may be useful at the very bottom of this page:
https://forums.overclockers.com.au/threads/sl … slotket.164450/
ooooooooooo
Anyway, as far as I know the only reason why the PL-iP3/T is not SMP capable is that they forgot to connect the corresponding pins. It can be fixed with two pieces of wire.
ooooooooooo

Reply 3 of 6, by Stevogamer

User metadata
Rank Newbie
Rank
Newbie
PARKE wrote on 2023-02-19, 12:39:
There is a reference that may be useful at the very bottom of this page: https://forums.overclockers.com.au/threads/sl … slotke […]
Show full quote
Stevogamer wrote on 2023-02-19, 00:00:

Interestingly the old documentation for early PL-IP3/T's mention a jumper for SMP which I assume is just because Powerleap copy and pasted the PL-IP3 manual and changed the headers to say PL-IP3/T. But considering this does anybody know if SMP worked on the 1.0 revision, since that did have a jumper in place of the blank solder pads with Pin 1 and 2 bridged on Rev. 1.1.

There is a reference that may be useful at the very bottom of this page:
https://forums.overclockers.com.au/threads/sl … slotket.164450/
ooooooooooo
Anyway, as far as I know the only reason why the PL-iP3/T is not SMP capable is that they forgot to connect the corresponding pins. It can be fixed with two pieces of wire.
ooooooooooo

Initially, that's what I thought but it seems like it might've just been that the originally pre-release PL-IP3/T's were just modified PL-IP3's using the tualatin socket mod.

That's my guess since the pictures used in the preliminary pages match the design of the PL-IP3 Rev 2.0 shown in the manual here https://web.archive.org/web/20030614092708/ht … 2WebManual2.pdf
and the preliminary PL-IP3/T page here https://web.archive.org/web/20011110045514/ht … oducts/iP3T.htm

It's interesting though that Powerleap probably figured it out by December 2001 since the page changes to say SMP is now supported. So I guess I don't have to change anything on my 1.1 for SMP. Although it still is weird that it doesn't work with the P3-S 1400 which is 1.45v compared to the 1.475v needed for Celerons and I'm not too keen on attempting the Japanese voltage mod at least not on something hard to replace.

Reply 4 of 6, by PARKE

User metadata
Rank Oldbie
Rank
Oldbie

Yeah, but there were two versions of the PL-IP3 and according to your research there were three revisions of the 'T' version and I find it difficult to wrap my head around what they exactly did. Fact is that they released an SMP "server" edition somewhere in 2002 which should support all "S" cpu's plus SMP. Everything before that may be luck of the draw ?

Reply 5 of 6, by Stevogamer

User metadata
Rank Newbie
Rank
Newbie
PARKE wrote on 2023-02-19, 16:41:

Yeah, but there were two versions of the PL-IP3 and according to your research there were three revisions of the 'T' version and I find it difficult to wrap my head around what they exactly did. Fact is that they released an SMP "server" edition somewhere in 2002 which should support al "S" cpu's plus SMP. Everything before that may be luck of the draw ?

I believe that was a marketing thing. Since powerleap sold the adapter both with and without CPU's. The waybackmachine archives show at least as early as April 2002 that they started selling the P3-S at 1.26ghz bundled. Between Nov 15, 2001 preliminary info to the update on Nov 16,2001 they updated the page to say SMP was supported.

My rev 1.1 is from March 2002 well past the updated page. Although it should be noted that the archives show that somewhere between April and August 2002, powerleap updated their page to show that the 1.4ghz CPUs do work. Although that might be referring to the Celeron and the non-s variant. The marketing material doesn't change until Dec 2002 to say 100% compatibility. Which leads me to believe 1.1 could only do the 1.26ghz cpus.

If somebody can verify the 1.0 worked with the 1.4ghz P3-S that would be great, since that likely means it's just the Fairchild transistors weren't up to the job on the 1.1.

My guess for the jumpers existing on the 1.0 was just a formality since the PCBs were the same as the PL-IP3 Rev. 2.0 so probably the production line was already putting the jumpers in.

In April 2002 they start to mention support for the 1.45, 1.475, and 1.5v cpus. But pictures on the page changed to the 1.1 or 2.0 revision sometime after April.

The powerleap.com/support page only goes until 2003. With the original support page for 2002 not showing pl-ip3/T documentation. The illustrated guides also never update on the product page.

Edit: The 2.0 revision is quite a different in design with a jumper to select voltage instead of an auto sense function on the 1.1 (if it isn't just hard-coded for 1.45v). My 1.1 came with a 1.475v Celeron 1200 which was what they would've sold as a bundle for the before they started also selling the P3-S bundles.

The 2.0 documentation says the jumper for 1.45-1.475v are the same. So I'd assume the 1.1 probably was always set to 1.475v. But that wouldn't make sense since my SL5QV is 1.75v (it won't boot if lower 😅)

I've attached a picture of the rev 2.0 for reference too just for completeness 😀

PXL_20230218_182126777~2.jpg
Filename
PXL_20230218_182126777~2.jpg
File size
1.09 MiB
Views
807 views
File license
CC-BY-4.0

Edit: Looking at the 1.0 image, looks like somebody had switched the jumper to the Celeron SMP pins, so I guess maybe powerleap soldered on a permanent jumper wire for 1.1 to make sure people didn't accidentally disable SMP.

Edit:

Looks like the original PL-IP3/T Rev 1.0 didn't feed the correct voltage to the Pentium III-S giving below 1.45V which would explain why my Pentium III-S 1.4 GHz was unstable on my Rev 1.1. At least according to this article that was the case.
https://ascii.jp/elem/000/000/328/328562/3/
Given this information, it would be insane for me to source a 1.0 just to test if that's true. But I can slap on a PL-370/T as a sandwich which is also a bad idea but a possibility to test the Pentium III-S 1.4 GHz.

Reply 6 of 6, by envagyok

User metadata
Rank Member
Rank
Member

I have 2pcs of powerleap shower adapters.
A friend and I would like to trade them for other rare old hardware.
What is the value?
It would be nice to know what value I should consider for the exchange.
Unfortunately I have not found any for sale or sold on ebay.