• Please review our updated Terms and Rules here

Socket 5/7 problem

L

Llamarama

Member
Joined
Jan 2, 2013
Messages
24
Location
Tyneside
Hi Everyone,

Recently picked up a very nice Elite Group SI54P socket 5 Baby AT motherboard, it's very nice :) However I am having some problems finding a processor for it. The usual searches online show up lots of processors marked at 5/7, and I know you can run a socket 5 processor in a socket 7 board.

Is there any easy way to tell socket 5 from 7 at a glance (pin count, layout, batch codes, copyright dates, etc?) or in failing that, Can a socket 7 processor run in a socket 5 board with extra cooling since it'll be running at 3.3v? One other thought would be to try and find a processor that was only released as socket 5 (I think early WinChips were socket 5 only, but cannot find anything to back this up other than an incomplete list on wikipedia.)

Thanks, Mike :)
 
The major difference from Intel is that Socket 7 chips tend to include MMX while the non-MMX versions are likely to be Socket 5. Pentium 90 or 100 should be Socket 5.

A socket 7 chip can not run in Socket 5. The one extra pin should prevent it from being inserted but Socket 5 does not support the lower voltage needed.
 
Differences in sockets

Note that Socket 7 looks like Socket 5, but for an extra pin on the right side of the 5th row from the bottom. I've seen board labeled as Socket 5 that had the extra pin hole, so go figure.

The problem is that the Socket 5 board won't accept much beyond a 133MHz Socket 7 CPU. I've heard of people successfully running S7 chips in S5 boards, so YMMV.
 
Last edited:
Hi everyone, thanks for the replies! :) I followed a few links and got a list of possibly compatible processors and found a few old ones at work that were destined for the bin but may have a second life yet! :D
 
Chuck(G) said:
The problem is that the Socket 5 board won't accept much beyond a 133MHz Socket 7 CPU. I've heard of people successfully running S7 chips in S5 boards, so YMMV.
Click to expand...

There were interposer upgrades that could get Socket 5 up to 200 MHz. They're pretty rare today though and probably fetch scalper Ebay pricing.
 
I wonder if they're worth it. I've got PowerLeap Slot-1 to Socket 370 adapter that allows for use of a 1.4GHz Tuallie in a P2-equipped board. It's faster, but not much faster than a 500MHz slot 1 CPU. Oddly, it's identified by the BIOS as a PPro.

Sometimes it's a matter of lipstick on a pig.
 
GiGaBiTe said:
There were interposer upgrades that could get Socket 5 up to 200 MHz. They're pretty rare today though and probably fetch scalper Ebay pricing.
Click to expand...

I bought a Powerleap PL-ProMMX when they were still new around 2000 to upgrade a Packard Bell system. They would get you up to about 333mhz but I found that by then even with 64mb ram and windows 98SE the board itself was really holding you back at those speeds. It also required a MrBios upgrade.
 
Chuck(G) said:
I wonder if they're worth it. I've got PowerLeap Slot-1 to Socket 370 adapter that allows for use of a 1.4GHz Tuallie in a P2-equipped board. It's faster, but not much faster than a 500MHz slot 1 CPU. Oddly, it's identified by the BIOS as a PPro.

Sometimes it's a matter of lipstick on a pig.
Click to expand...

It was well wroth it back in the day if your board had a bios update to support tualatin's. I had a Abit BE6-ii. I still miss that system.
 
Chuck(G) said:
I wonder if they're worth it. I've got PowerLeap Slot-1 to Socket 370 adapter that allows for use of a 1.4GHz Tuallie in a P2-equipped board. It's faster, but not much faster than a 500MHz slot 1 CPU. Oddly, it's identified by the BIOS as a PPro.

Sometimes it's a matter of lipstick on a pig.
Click to expand...

I have two Slot 1 to PGA370 adapters, but they don't work in any of the three Slot 1 motherboards I have with any PGA370 CPU I have. If your adapter is that performance crippled, it must be running on a 66 MHz FSB or something. I've seen some benchmarks of modded 440BX boards with a 133 MHz FSB running 1 GHz+ PIIIs and getting good results. I don't know how long those systems lasted because the 133 MHz FSB was pushing that 440BX chipset pretty hard, and IIRC it also had to be volt modded for stability.

Another issue could be that the BIOS doesn't have the correct microcode for the Pentium 3, that will really cripple performance since it doesn't know about SSE and other improvements.

NeXT said:
I bought a Powerleap PL-ProMMX when they were still new around 2000 to upgrade a Packard Bell system. They would get you up to about 333mhz but I found that by then even with 64mb ram and windows 98SE the board itself was really holding you back at those speeds. It also required a MrBios upgrade.
Click to expand...

I had a Gateway 2000 system which originally came with a Pentium MMX 200, and later upgraded it with an Evergreen Spectra AMD K6/2 400. It had the same problem of the performance difference not being that noticeable because it was hobbled on a 66 MHz FSB with PC-66 SD-RAM. It's probably why Evergreen technologies didn't add multipliers to support faster chips, the bus speed crippled performance gains past that speed.

Many years later, I got a real Super 7 system and the same AMD K6/2 400 and the performance difference was night and day.
 
GiGaBiTe said:
I have two Slot 1 to PGA370 adapters, but they don't work in any of the three Slot 1 motherboards I have with any PGA370 CPU I have. If your adapter is that performance crippled, it must be running on a 66 MHz FSB or something. I've seen some benchmarks of modded 440BX boards with a 133 MHz FSB running 1 GHz+ PIIIs and getting good results. I don't know how long those systems lasted because the 133 MHz FSB was pushing that 440BX chipset pretty hard, and IIRC it also had to be volt modded for stability.

Another issue could be that the BIOS doesn't have the correct microcode for the Pentium 3, that will really cripple performance since it doesn't know about SSE and other improvements.
Click to expand...
I did not have to up the voltage on my BE6-ii to use 133mhz FSB. But I could see some needing to. The big thing was running the AGP bus out of speck. If I recall at 133mzh FSB the AGP port runs at 82mhz and not 66mhz. That could cause all manner of stability problems with some AGP cards.
And yes bios support and 133mhz ram is a big thing. if your running at 66mhz bus its not going to preform well. And a lack of bios support can make a system as slow as a 286.
I put a 1.4ghz Piii-s in a gigabyte board once and without a bios update it was horrid.
 
Many boards at the time cascaded the FSB clock to get the AGP clock with a 1:1 or 2:3 divider, so yeah AGP would run alarmingly out of spec at ~87 MHz.

There were a few weird motherboards out there though, like the FIC PA-2013 which had a fixed AGP clock and cascaded PCI and memory clocks off of it, leaving the FSB to be independently clocked. ISA was further cascaded off PCI with a 1:3 or 1:4 divisor.
 
GiGaBiTe said:
I have two Slot 1 to PGA370 adapters, but they don't work in any of the three Slot 1 motherboards I have with any PGA370 CPU I have. If your adapter is that performance crippled, it must be running on a 66 MHz FSB or something. I've seen some benchmarks of modded 440BX boards with a 133 MHz FSB running 1 GHz+ PIIIs and getting good results. I don't know how long those systems lasted because the 133 MHz FSB was pushing that 440BX chipset pretty hard, and IIRC it also had to be volt modded for stability.
Click to expand...

Nope, just saying that memory bandwidth and the basic FSB speed are really the limiting factors. Yes, it's a Compaq 440BX board with a 100MHz FSB. You'd think that going from a 600MHz Slot 1 P3 to a much faster socket 370 chip would result in a dramatic improvement in performance, but it really doesn't. You're saddled with what's on the board, aside from the CPU. The Powerleap slockets were the best of the bunch, IMOHO--even today, they don't last long on that auction site when they come up.

I never found that a 440BX could be relied upon for 133MHz performance. 120MHz, maybe, but I've found that 133 was just too unstable, even with added heat sinks.

I think that P4s got a similar bad rap--it wasn't the CPU so much as the supporting logic and memory that crippled them.
 
Chuck(G) said:
I never found that a 440BX could be relied upon for 133MHz performance. 120MHz, maybe, but I've found that 133 was just too unstable, even with added heat sinks.
Click to expand...

I had a lot of 440bx boards over the years and have found that most of the good ones can do 133mhz FSB. But many do have trouble over 100mhz. You need a AGP card that can hale the higher buss speed and 133mhz ram as well.

It all really comes down to the mix of parts and the board. Heat really is not the problem.
With a good board you can pull off something like a 170mhz fsb.
I was able to push it more but then the PCI bus ran way to fast and I had to use a isa video card and run the system off a floppy disk.

And it was stable at 158mhz FSB.
image_id_1648327.bmp

That is just under 1hr and a half of full load stress., it took about 2 and a half hours to complete whit the CPU at 750mhz.
 
Last edited:
My experience is otherwise. For example, I've got a Supermicro 440GX board with two 1GHz/133 Slot 1 matched CPUs. Memory is registered server RAM, rated at 133MHz. Video is a high end 8x AGP card. Will run for a time, but eventually, crashes, given a good workload. Run at 120MHz FSB, pretty stable; at 100MHz, rock-solid.

SM, on the website of the time, stated that "Customers have reported being able to run at 133MHz, but we do not warranty our products for that speed".

The improvement in speed is not worth the risk of a crashing system, is the bottom line for me.
 
Chuck(G) said:
I think that P4s got a similar bad rap--it wasn't the CPU so much as the supporting logic and memory that crippled them.
Click to expand...

No.. the Netburst arch was just a crap design. You could have the same everything and compare a 1.4 GHz Tualatin with a 1.4 GHz Willamette and the Tualatin would absolutely crush it. My dual 1.4 GHz Tualatin system can hold ground with a 2.8 GHz P4 in everything but memory bandwidth.

Intel made the mistake of continuing the MHz race started back in the PIII days with AMD. They made the conscious decision of going with a weaker and less efficient architecture that could clock higher, and believed that more MHz would offset the inefficient architecture. I still remember the Intel press releases in the early 2000s where they said the Pentium 4 would eventually reach 10 GHz, before having to backpedal a couple of years later when Prescott and Gallatin cores were hitting clock walls, thermal limits and drawing ridiculous amounts of power when overclocked. I can't find it now, but there was an article by Toms hardware I think of overclocking a Pentium D and the CPU started pulling something like 300W by itself.

AMD made the right choice with the Duron/Athlon/Athlon XP of making a more efficient core over continuing the MHz race with Intel. An Athlon XP 3200+ was 1 GHz slower than the Pentium 4 3.2, but in benchmarks was only about 15-20% slower.
 
I'm not saying "the same everything"; that'd be silly--AFAIK, I've never seen a 440BX P4 design, if that was even possible. What I am saying is that the chipsets (e.g. 845) were really miserable--and the RDRAM-based 850 was just ridiculous, given the system cost.
 
Mr. Horse said:
Agreed, what I posted was a bench marking setup. But I do know of a handful of 440BX board listed as sporting 133mhz FSB
Click to expand...

Well, that's exceeding Intel's spec, so take that for what it's worth. I do have a couple of P3 non-440 around here that run 133 FSB, but they're rated to do so. One ridiculous one is an 820 board, complete with slot 1 CPU and 1 GB RDRAM. I keep it around because it's got a really good FDC onboard. Has a couple of ISA slots, too, but ISA DMA doesn't work right. The other, I think is a VIA chipset with Socket 370 CPU and not particularly noteworthy.

Right after the P4 era, I converted to AMD and haven't looked back since.
 
You must log in or register to reply here.
Back
Top