386SX wrote on 2022-01-23, 13:05:

I understand the cache point but I wonder if like in the past would have been that difficult to design an external fast L2 cache on sockets like the early Pentium that would have solved the L2 problem and still living on old style Socket...

The PowerPC 750 (aka G3) and very early PPC 7400 (basically the Katmai of the G4 line 😜) had half-speed L2 chips mounted on a mini PCB that plugged into the CPU ZIF socket. Definitely an interesting looking setup, with the uncooled L2 chips peeking out at you from under the heatsink.

386SX wrote on 2022-01-23, 13:05:

Tetrium wrote on 2022-01-23, 11:21:

My first PC was a Slot 1 one. Back then the Slot 1 concept was sold as the way forward, as progression. Because this way it was […]
Show full quote

386SX wrote on 2022-01-21, 17:15:

Thanks, I'll see if I can find anything wrong, I didn't check the back side of the mainboard being installed right now but I suppose soon I'll put the 430VX I had back in this case. The slot idea for the CPUs never felt IMHO like the best one seen in consumer electronics considering how heavy and powerful CPUs became and how much power they needed... even back in those times I didn't even have a Slot based mainboard. Jumped directly from the K6-2 to the later Duron 750 socket version.

My first PC was a Slot 1 one. Back then the Slot 1 concept was sold as the way forward, as progression. Because this way it was way easier to scale up CPU frequencies because the cache didn't need to be on-die anymore.
CPU power dissipation had increased somewhat, but at that time noone really saw the huuuge leap in CPU power consumption coming, so this was not something that was even a consideration when the CPU slot was designed. That and for a newer CPU it was quite typical a new cooling solution would be designed anyway, it just didn't matter at the time Slot 1 was released.
And tbf, Slot 1 was actually capable of mounting substantially beefier coolers compared to Socket 7. Some Slot 1 CPU HSFs were basically quite literally just 2 Socket 7 heatsinks with 2 5cm fans mounted on top, made suitable for mounting on a Slot 1 CPU instead of on a Pentium 1 ZIF socket.
Also swapping out CPUs is usually much easier compared to Socket 7 where the HSF will need to be unmounted and remounted every time a CPU using a corretly installed CPU HSF is swapped. So there is definitely something to say for Slot 1 being the superior solution compared to Socket 7, which is a comparison which was made somewhat commonly back in the day. Of course from hindsight things may look very different.

CPU slot design was more expensive, so as soon as the L2 cache could be put on the CPU die again with good enough yields (which happened when Intel made Mendocino), Intel made Socket 370 and soon enough the desktop Pentium 3 CPU went over to s370 entirely.

I understand the cache point but I wonder if like in the past would have been that difficult to design an external fast L2 cache on sockets like the early Pentium that would have solved the L2 problem and still living on old style Socket when until the cpu beside the 1Ghz models with those huge heatsink, they didn't seems impossible to use some heavier heatsink on the socket. Let's see what they did on the Socket 462 and the power demanding various Athlon (1400.. XP 3200 etc..) and heatsink really heavy and large. I understand that the early Pentium II seems to have quite increased the power demand with the early core and that probably felt like a problem compared to the past cpu solution. But once new cores (P2-300/350? I don't remember) were released the temperature problem came back into a acceptable range imho.

What you're saying here is totally true. Later CPU sockets were able to run much heavier heatsinks (especially handy for Socket A since these heatsinks make it relatively easy to find a beefy CPU cooler for the other sockets of the same size (which are s370, s7, s5 and s3)). But Socket 7 for instance was not made with heavy duty heatsinks in mind. Often you'll find there's motherboard components preventing you from actually mounting one of the much larger heatsinks.
Also with the heavy heatsinks I'd say it's advisable to make use of all 3 tabs and Socket 7 often omitted the 3rd CPU tab.
In some cases Socket 3 (which has the same physical size as Socket 7, funfact 😜 ) doesn't have any of those socket tabs at all.

Pentium 2 was often sold with integrated CPU cooling.
Yes, with Deschutes the power dissipation was lowered significantly compared to Klamath.

I don't know maybe I never really like those as a concept, while I understand the theorical positive side of having a cpu that couldn't have broken pins or similar but at least the Slot plastic bus should have been attached to the mainboard in so many points that should have been impossible to make any force on that. Also a different plastic guides logic migh have helped into removing them (like a Socket with a mechanism to help the user). For example similar thing happened over the years with the video card bus, now in metal, double slot etc.. But the video card logic make the bus less problematic to install or uninstall, those cpu had to stay there in a vertical position and possibly suffering also case vibrations or casual hit. Lately talking to a PC store (one of those few old stores remained having worked into this market since the 80's) manager, talking about the Slot cpus immediately said about the "problems with contacts" I didn't even know about but now I'm discovering cause not having Slot board in those times. To say that if even a common local repair store remembers that 20 years later maybe was a discussion subject for those who had them. 😉

Regarding the Slot 1 mounting mechanism, I'm pretty sure that Intel did provide detailed installation guidelines and tbf installing a Pentium2 cartridge is much more straight forward than installing a Pentium 1 as no heatsink needs to be installed and no pins can be damaged like on a ZIF socketed CPU.
To be fair, back when Slot 1 was new there was virtually no mention of any problems with the Slot 1 mechanism being easy to damage and having build dozens of Slot 1 systems myself (both privately as well as for others) I've never come across this problem...which of course doesn't exclude the possibility that I did in fact meet this problem and did not recognize it properly. But the same I can say about socketed CPUs, except we did in fact break stuff more easily. Mostly bent pins, broken Pentium 4 plastic brackets which in fact were quite fragile (but still only if mistreated or being careless) and the pushpins which could be broken which caused throttling issues until we figured out the issue (actually has this P4 which ended up having 1 broken pushthrough pin which caused the heatsink to not be mounted properly, causing the CPU to throttle all the time).

Also, usually the Slot 1 cartridge is not very heavy and after insertion actually seemed to make the PCB more sturdy where with a ZIF socket the board would tend to flex instead (especially and mostly with Socket A HSFs though, never noticed such a thing with Socket 7 stuff).

And the Slot 1 is essentially a lengthened AGP slot flipped onto its side and AGP slots don't break this easily either (even though AGP slots can also break, just like Slot 1 slots can still break).

Regarding the case vibration, the original mounting mechanism of Pentium 2 would lock the CPU into place and I doubt this mechanism could break with just vibrations.
With sudden shock force (like dropping a case), I consider this as mistreatment and usually PC hardware is not designed with sudden shocks in mind.

And regarding the repairman, this seems unlikely to be the case imo. If a Slot 1 board was defective, it's up to the motherboard manufacturer or the system builder to enable the warranty procedure. If the boards were out of warranty, then these must have been older boards. And when properly installed, they don't 'just break' if the insides are left alone.
It seems just so implausible to me.
I mean, if Slot 1 is supposedly very fragile then essentially AGP should be as well, and afaik this is simply not the case.
So I wouldn't worry about this too much.

Of course if mothervoards are thrown around and pulled out of piles of other boards, this is also mistreatment and has essentially nothing to do with the way Slot 1 was produced, nor could it be considered a flaw of the Slot 1 design.
With mistreatment virtually anything can be broken.

Standard Def Steve wrote on 2022-01-23, 18:18:

386SX wrote on 2022-01-23, 13:05:

I understand the cache point but I wonder if like in the past would have been that difficult to design an external fast L2 cache on sockets like the early Pentium that would have solved the L2 problem and still living on old style Socket...

The PowerPC 750 (aka G3) and very early PPC 7400 (basically the Katmai of the G4 line 😜) had half-speed L2 chips mounted on a mini PCB that plugged into the CPU ZIF socket. Definitely an interesting looking setup, with the uncooled L2 chips peeking out at you from under the heatsink.

I had no idea about this, it sounds kinda funny 😜

TrashPanda wrote on 2022-01-23, 12:25:

I'm getting it from all the slot based boards I had to try and repair and usually replaced, Intel boards had good reliability of the solder joints cheaper boards did not, the whole slot design was stupid from the start. It was never needed and initially was just a gimmick by Intel to stop AMD from using its sockets, in the end it worked and ever since AMD and Intel have gone their own way separating the market. (Which is what Intel wanted)

I dont think anyone would disagree with the reliability of slot based designs as they have aged, mostly the boards you find now are the ones that had more robust solder used the cheaper boards are all e-waste by now either from Capacitor Plague or the slot died from shit solder joints. (Solder from that era in general was terrible, once it dries out its only a matter of time and force before it fails)

This sounds like a solder problem though and not a Slot 1 problem.
If this is the case, other slots like AGP should also be affected.

Sorry, but this doesn't seem very plausible to me. And besides the bad solder point which I am not aware of, this has nothing to do with Slot 1 anyway.

About it not being needed, it was either this or continue making Pentium Pro CPUs which would end up being much more expensive. Intel tried this and they had huge yield issues as the L2 cache and CPU could not be tested separately.
Some info on this can be found in this thread:
https://forums.anandtech.com/threads/why-did- … -socket.431403/

What Noriaki wrote is correct. With Modus's explanation I respectfully disagree - and not just because I work for Intel, but also because what he wrote doesn't really make sense.

The Pentium Pro used Socket 8 which uses GTL+ signalling and is not backwards compatible with socket 7. Switching to a slot didn't changing the signalling method or protocol, and had a negligible impact on pin level timing. There's no reason why, if a competitor could create a socket 8 part, they couldn't switch to a slot instead. It's just a form factor change, not a signalling change.

The reason why the Pentium Pro caching method wasn't on the Pentium II was cost. The Pentium Pro was targetted at the server/workstation market and was (and still is) quite expensive. Part of the reason for this cost was that the MCM (multi-chip module) used on the Pentium Pro was more expensive to produce and introduced yield issues compared to conventional single-chip packages. There's a reason why no one else has ever produced a MCM part for CPU's, memory, chipsets or video cards which is targetted at the home market and this is cost. A cost-effective method was needed that put the cache very close to the CPU for high-bandwidth, low latency operation and this lead to slot 1.

A cost effective method was needed, so yes, Slot 1 was needed 😜

TrashPanda wrote on 2022-01-23, 13:23:

Look at socket 8, it had the first Pentium 2 on it as a Pentium II overdrive, it worked fine but Intel being Intel saw a way to remove AMD and others from its platform, the slot1 based design which thanks to the chipset was locked to Intel CPUs and Intel refused to share it with AMD, AMD went and did their own slot format in slot A but from that point on the two never again used the same socket or platform, I cant say for sure if that was a good or bad thing but it certainly hasn't done either of them any harm.

I own a Pentium II overdrive, its not huge or incapable of being cooled and aside from the socket 8 platform only ever seeing major use in server boards it works perfectly fine.

The one thing I would have loved to see on the Slot connector was through board retention with a backplate, a way to take stress off the connector itself and apply it to the supporting structure which would have been stronger and not connected to the board by weak solder joints. As for damaging slots .. seen that way to many times with AGP and PCIe GPUs when people use a little to much lateral force and snap the slot right off the board or use to much force when removing it because they failed to unlock it which ends up tearing the slot off the board.

There was actually a through-board mounting mechanism for Slot 1, though it was used to mount the 2 plastic mounting clips that the CPU cartridge would use to click into, immobilizing the cartridge while inserted into the Slot 1 slot.

Slots by their very nature are more fragile than sockets.

No, they're not. And there's just no evidence of this 😀
And the one who states something will need to provide the evidence of this.

PEBCAK exists and should be respected.

I'm just gonna insert a bunch of semi-random smileys here 🐤 🌵 🌮 🧊
And as I mentioned before, if slots get damaged by abuse, it's not the slot's fault. That's called blaming the victim 😜

A bit of interesting reading can be done here btw: Why Intel dropped the Slot processor form?

Update (maybe).. it looks sort of strange that with the Pentium III Katmai the L2 cache reading problem never happened so I tried downgrading the bios of the mainboard and it seems like with the previous versions v3.3 v3.3e and v3.4 the L2 cache missing after some time don't happen at least until I'm writing. The latest version v5.6 was a beta one with 40GB disk support. I've also seen that the v3.4 have problems with reboot hanging at the bios logo/before che cpu reading every time I rebooted, while the v3.3 version has problems with the IDE second channels using two drives disable one in the o.s. I found that the v3.3e version seems the most stable for now.
I'll update later if the problem still appear. For testing I'm using a Celeron 333 (Mendocino), ATi Rage Pro Turbo, 128MC PC100, 4,3GB IDE disk and Win 98.

Update.. NO.. it still happen but I found that even without moving the case or the cpu, changing the cpu to a 400Mhz Pentium II it happened again, but after a shutdown and starting it without touching the case or the cpu, the L2 cache appears again.

I try other cpu again but I'm sure this problem also happened with those in the past. Strangely the Pentium III Katmai never showed this problem not one even one time.

I am seeing a very similar problem in my retro PC. The motherboard is a FIC VL-601 (with the 440LX chipset), and the CPU is a Pentium II 333 SL2S5. 4 out of 5 times when the PC is powered on the cache is there, but sometimes it disappears, and then the CPU presents itself as a Celeron 333.

It does not suffice to reset the PC. I have to power it down, and power up again, and the cache comes back. No playing with the socket, shaking the PC etc. required. The CPU sits firmly in the socket, and the motherboard is (at least visually) in a great condition, and has been very stable otherwise.

I first noticed this issue after installing two Voodoo 2 cards in the PC, albeit this may be coincidental. The PSU is a Fortron with 20A on the 5V rail, which should be enough for a Pentium II system... but know knows.

I am planning to do some tests with a different PSU, and with a different 440LX board. Unfortunately, I don't have another Pentium II at hand.