• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Slot Processors

Gamingphreek

Gamingphreek

Lifer
I think this is an appropriate place for this, either that or CPU but i want a more specific response.

My question is straight forward, why did we go to Slot processors for a while. Was there interference with traces on the M/B that didn't allow a socket or does a slot provide more power?

Pros, COns.

-Kevin
 
To go into a little more detail, we needed a lot of cache (I think before the P2 / Athlon, caches were <32kb), and on-die caches reduce yields (die area goes up a little -> cost goes up a lot). By using a slot package (which adds its own costs), you could manufacture the cache separately (and buy from the lowest bidder if you don't fab it yourself) and get better yields. Eventually, the relative pros/cons of die area vs a cartridge changed, and you could do better with on-die cache.
 
Well, the "slot" processor (L2) cache was faster (speed-wise) than leaving it on the motherboard, because we couldn't put it on-die yet.
Take the Katmai PIII (I owned one) as an example. 500 MHz on a 100 MHz FSB, with 512 K L2 cache.
Would you rather your 512 K of cache run as a add-on card to the motherboard at 100 MHz? Or would you rather your 512 K of cache run on its own PCB with the CPU at 250 MHz?
Obviously, faster cache is better.

The PII and the early PII Celeron were interesting however - The PII had more cache [512K to 128 K] but the PII's cache ran slower than the Celeron's on-die cache. [150 MHz compared to 300 MHz] and the Celeron ran faster than the PII did.
 
Yeah, and just point it out again, The P1 had no L2 cache onboard, even though you might think it did. It was in a slot like RAM on the motherboard.
 
You all seem to have forgotten Intel's marketing ploy.

The P2 slot was developed to deliberately be incompatible with Pentium motherboards. Then Intel got a patent on it so no other processors could use it.

This was to make life harder for other CPU manufacturers (AMD etc).

The other CPU manufacturors continued with the Pentium socket (I think it was called socket 7) and extended it (eg faster FSB, AGP support etc). An AMD K6 500 motherboard would support a Pentium 166.

The debacle of the Celeron 300A (as mentioned in prev post) proves this. The Celeron was actually faster than the P2 because it had an on-die cache, plus it had better overclocking potential.

Then Intel eventually reverted back to a socket when the gains of locking out competitors was reduced to almost nothing, and the extra expense of the slot interface was not worthwhile.

If my memory is wrong then please state your view!
 
Originally posted by: tom1000000
You all seem to have forgotten Intel's marketing ploy.

The P2 slot was developed to deliberately be incompatible with Pentium motherboards. Then Intel got a patent on it so no other processors could use it.

This was to make life harder for other CPU manufacturers (AMD etc).

The other CPU manufacturors continued with the Pentium socket (I think it was called socket 7) and extended it (eg faster FSB, AGP support etc). An AMD K6 500 motherboard would support a Pentium 166.

The debacle of the Celeron 300A (as mentioned in prev post) proves this. The Celeron was actually faster than the P2 because it had an on-die cache, plus it had better overclocking potential.

Then Intel eventually reverted back to a socket when the gains of locking out competitors was reduced to almost nothing, and the extra expense of the slot interface was not worthwhile.

If my memory is wrong then please state your view!
Click to expand...

CTho9305 has the right explanation. At the time, cache was extremely expensive to manufacture on the same die as the CPU because of lower yields and overall costs. Intel did not just switch through 3 interfaces to "make life harder for other CPU manufacturers". Slot 1 CPU cartidges were more expensive to produce than socket based CPUs because of the PCB, packaging materials, the physical interconnect system on both the CPU and mainboard, etc but was offset by cheapening the actual CPU + cache cost. Later, when it became more economically feasible to produce the cache on-die, Intel (and AMD additionally) reverted back to the socket design.
 
Originally posted by: Fox3a
Yeah, and just point it out again, The P1 had no L2 cache onboard, even though you might think it did. It was in a slot like RAM on the motherboard.
Click to expand...

yeah, it was called a COAST socket.
 
intel wanted to merge the ppro and pmmx, which is what the p2 is. they had problems integrating the mess of cache as in yields were too low.

starting with the celeron 300a, they finally got yields to a point as to reintroduce socket form.
 
Originally posted by: Mday
intel wanted to merge the ppro and pmmx, which is what the p2 is. they had problems integrating the mess of cache as in yields were too low.

starting with the celeron 300a, they finally got yields to a point as to reintroduce socket form.
Click to expand...

P-pro was a bit different. It had 3 integer pipes wheras a pentium-II only had 2. The P-mmx was pathetic because it would take too many clock cycles to switch back from mmx mode to normal and vice-versa(somebody told me 50-60 clock cycles), and they rectified this problem in the p-2 (2-3 clock cycles).
The P-pro was wayy faster than the p-II at the same clock speed.....yeah and it was difficult in those days to have an on-die cache.
 
Technically it was like it's been said - moving the caches onto their own "backside bus" on the CPU allowed two things: Much faster L2 cache, and also faster frontside bus (because the L2 cache chips were off it).

Early attempts at putting the L2 cache onto the CPU were clock limited - the Pentium Pro very much so, the K6-III a little less but still considerably worse than the L2-less K6-2.

With chip manufacturing improving enough to put lots of L2 cache directly onto the CPU die, the step back from the expensive slot solution to socket CPUs was only logical.

So there's your reasons why Intel took the low end back to sockets first - because this is where cost matters and clock speed doesn't.
 
Short-term, we're looking at sockets. There are lots of R&amp;D dollars going into packaging, so it's pretty hard to say what the future holds. I have heard of some promising new interposer technologies, but those are still two generations out, and they are still a socket-type interface.

If Intel has their way, the next change is going to be the BTX form factor. I heard a lot about it earlier this year, but not too much lately. I wonder when that is going to hit the market...
 
Originally posted by: tom1000000
You all seem to have forgotten Intel's marketing ploy.

The P2 slot was developed to deliberately be incompatible with Pentium motherboards. Then Intel got a patent on it so no other processors could use it.

This was to make life harder for other CPU manufacturers (AMD etc).

The other CPU manufacturors continued with the Pentium socket (I think it was called socket 7) and extended it (eg faster FSB, AGP support etc). An AMD K6 500 motherboard would support a Pentium 166.

The debacle of the Celeron 300A (as mentioned in prev post) proves this. The Celeron was actually faster than the P2 because it had an on-die cache, plus it had better overclocking potential.

Then Intel eventually reverted back to a socket when the gains of locking out competitors was reduced to almost nothing, and the extra expense of the slot interface was not worthwhile.

If my memory is wrong then please state your view!
Click to expand...

Ummm, the original athlon was SlotA. 😕
I guess I see where you were going with the whole Socket3->Socket7 thing but Socket7 also added on-die cache. The entire reason was because of the cache, as stated above.

Anyways I have many examples of processor packages hagning on my wall. If I had a decent digi I would take some pictures. Got an AMD 486DX2-80 with DIP packaged on motherboard cache, a pentium 75 with COAST.
 
I had just opened up my ancient pentium-II machine and the cache memory is manufactured by NEC japan.

I also have a celeron-266 slot-1 lying around, and I am planning to solder some cache from a burnt off pentium-II 300 onto the celeron module. Wish me best of luck guys, I hope I don't burn off the cache mem while soldering.....
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top