Is any other PII and PIII had ECC cache? Information out there is not clear.

Cheers,

Some server boards require ECC, desktop boards not so much.

I don't have direct experience with these exact CPUs, but some boards have a BIOS option to toggle cache ECC on/off. I'm not sure if the "EC" suffix here means that cache ECC is forced on all the time, or if it just means support exists and can be enabled with BIOS support.

> Will ECC version work in all slot 1 motherboards?

It should, since it's a function internal to the CPU and I don't see how the motherboard would reject the CPU based on this feature.

If you're up for testing and the CPUs are otherwise identical, running some cache tests could be useful for making a decision. Under DOS there is CACHECHK and SPEEDSYS which could reveal slight differences in cache access time. Under Windows 9x I like AIDA64 which has a memory benchmark option that includes all cache speeds and latency. On top of this, running some standard 3D benchmarks like 3DMark99 and 2000 could be useful.

In my retro builds I sometimes opt for ECC main memory just for that little bit of extra theoretical stability, even if it costs a clock cycle. I plan to run ECC + registered memory in my 440BX box to make it more workstation-y. I rarely turn on cache ECC but I could see it being valuable if you're seeking that little bit of extra stability. If the cache is working fine it probably doesn't help much, but it might one day save you from a random cosmic ray crash during your retro gaming session. :)

I got some snippet rattling round from back in the day in my head about "If use ECC CPU, gotta use ECC memory" but not sure what that was enforced on. I know "just don't use it if not supported" used to be an option but then Intel went and built iron curtains between market segments and didn't want people to be able to use consumer stuff with biz stuff, for no technical reasons really, so some stuff has a spite block of "You're not using that here".

Cosmic wrote on 2023-10-05, 18:31:

I don't have direct experience with these exact CPUs, but some boards have a BIOS option to toggle cache ECC on/off.

I have one of such motherboards, Gigabyte GA-6BXS. Seems like workstation oriented motherboard, because of integrated SCSI controller, and SCSI connectors onboard.
I really like to know, if Pentium II ECC is slower from standard Pentium II, if ECC L2 Cache is off by motherboard. Or , if it is slower ONLY if ECC l2 Cache is on in BIOS.
If the latter is the case, then Pentium II EC seems better option, as it is same fast in case ECC L2 Cache off, but have extra option that may come handy.

Wasn't it that the non-ECC L2s can only cache 512MB of main memory rather than 4GB? - not sure on performance differences though

These are handy quick references for L2 type...

PII - https://web.archive.org/web/20010211135408/ht … /pentiumII/qit/

PIII - https://web.archive.org/web/20010124040600/ht … pentiumIII/qit/

I do have a PII-400 CPU with ECC support for the L2 cache, the CPU name during post (at least in Award BIOS) is slightly different when enabled. In practice, performance is slightly slower than when disabled.

PC Hoarder Patrol wrote on 2023-10-07, 00:25:

Wasn't it that the non-ECC L2s can only cache 512MB of main memory rather than 4GB? - not sure on performance differences though

Yes and no. The ECC version always carry the TP6-e TAG chip for caching, thus being able to cache 4GiB. The non-ECC versions can be with T6 TAG, caching only 512MiB, but it's made worth TP6-e as well ...

This processor was manufactured during week 24 1998, so it's underclocked Pentium II 300, i.e. faster L2 cache. It should be better than genuine Pentium II 233.

The Serpent Rider wrote on 2023-10-07, 12:47:

This processor was manufactured during week 24 1998, so it's underclocked Pentium II 300, i.e. faster L2 cache. It should be better than genuine Pentium II 233.

As far as I know, when you underclock Pentium II 300 to 233mhz, you are slower, than actual Pentium II 233 mhz. Because, timings on cache on higher models were worse, to reach that increased speed
https://www.tomshardware.com/reviews/intel,65.html

Anyway, I didn't know, they were underclocking Pentium II 300 in late 1998. Interesting information. But if they used L2 cache from Pentium II 300 model, set to be able to reach 150mhz, than these Pentium II 233's were actually slower, than original.

W.x. wrote on 2023-10-07, 13:00:

As far as I know, when you underclock Pentium II 300 to 233mhz, you are slower, than actual Pentium II 233 mhz. Because, timings […]
Show full quote

The Serpent Rider wrote on 2023-10-07, 12:47:

This processor was manufactured during week 24 1998, so it's underclocked Pentium II 300, i.e. faster L2 cache. It should be better than genuine Pentium II 233.

As far as I know, when you underclock Pentium II 300 to 233mhz, you are slower, than actual Pentium II 233 mhz. Because, timings on cache on higher models were worse, to reach that increased speed
https://www.tomshardware.com/reviews/intel,65.html

Anyway, I didn't know, they were underclocking Pentium II 300 in late 1998. Interesting information. But if they used L2 cache from Pentium II 300 model, set to be able to reach 150mhz, than these Pentium II 233's were actually slower, than original.

I would guess that if sold as 233, Intel had configured the cache for 233 timings, thus being equally fast as any other 233 ...

The Serpent Rider wrote on 2023-10-07, 12:47:

This processor was manufactured during week 24 1998, so it's underclocked Pentium II 300, i.e. faster L2 cache. It should be better than genuine Pentium II 233.

From what date Intel started rebranding Pentium II 300's to 233. Do you know to provide any more information about this?

Somewhere around 1998, maybe very late 1997. That's not rebranding per say, just unified manufacturing for financial reasons, so all late Klamath steppings should be identical inside. Genuine PII-300 can have cache chips marked as 5ns, which are probably cherry picked factory overclocked 133 Mhz memory. I'll make a photo later.