[geeks] Interesting read

[Peter L. Wargo]

On Friday, February 22, 2002, at 09:01 PM, Joshua D Boyd wrote:

> From what I hear it was more caused by the SRAM cache problems than 
> other bugs,
> but still.  The SRAM cache problem in E10Ks was pretty bad.

Actually, is was more or less like any other problem.  I can comment on 
this, as I was a customer when it broke, and we had not had E$ problems 
at that time.  Other problems, yes. but not E$.

The actual number was not huge, but the problem seemed to "cluster" at 
certain sites, where it never hit at others.

IMO, Sun didn't plan well enough for large cache sizes, as the smaller 
cache CPU's were bulletproof.  I had an E4000 with 250MHz/1M modules 
that was rock-solid.  Year+ uptimes.

BTW, the same modules were used in the E3000-E10000.

-Pete