SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Rambus (RMBS) - Eagle or Penguin -- Ignore unavailable to you. Want to Upgrade?


To: grok who wrote (30707)9/26/1999 7:15:00 PM
From: Bilow  Respond to of 93625
 
Hi KZNerd; Re "Heisenbugs", reminds me of the problem they used to have with super computers built out of the uncompensated ECL families, i.e. 10K versus 10KH.

The logic levels were very dependent on temperature (as well as power supply). So when you removed the skins, the temperature environment changed, and some of the problems went away...

In fact, the Rambus channel architecture reminds me of a lot of techniques from super computers. At Scientific Computer Systems (which made a Cray clone), we had technicians tweak the clock lines in order to minimize skew (and thereby allow higher speed clocking). They used different lengths of coax to speed or slow the clocks on the back plane. This is a more traditional technique for supercomputer construction than people realize. It is one of the reasons that supercomputers tended to be so expensive... (Where are those companies now?)

I am sure that you could get Rambus Channels into production and running pretty well if you had a technician check out each box with the best scope money could buy. If the data eyes weren't right on, he would tweak a passive somewhere to correct for it.

The problem is that this sort of manufacturing technique is unsuitable for mass production of (relatively) low cost '86 clones. I am really looking forward to reading the reports that drift out of these companies (i.e. DELL, INTC, CPQ, &c.) with regard to this fiasco. It will make highly entertaining reading.

Funny thing. Zeev's Micron "scenario" is looking more and more reasonable, what with their RDRAM abandonment announcement. Makes them look very smart.

P.S. I like the term "Heisenbugs", it is now officially part of my vernacular. We should have a term for another type of bug, the one that never repeats, because it is so rare...

-- Carl



To: grok who wrote (30707)9/27/1999 4:37:00 AM
From: John Walliker  Respond to of 93625
 
KZNerd,

Carl, I can't remember who but someone named these bugs "Heisenbugs" after Heisenburg's Uncertain Principle which says you can't observe something without changing it. Just putting a scope probe on these lines will change operation so you can't troubleshoot it.


Very true. Much lateral thinking is needed to devise tests for problems like this.

John