|
|
|
Thread Tools | Display Modes |
#11
|
|||
|
|||
old processor -- 8085
"g. beat " @spam protected wrote in message
. .. With the availability of the Intel Pentium processor (space rated) -- I am curious what space application would warrant such a "new processor" -- used for a sub-system? I don't know the new chip, but I would guess that one advantage might be lower power consumption than later chips that use way more circuitry than required by many applications. Jim McCauley |
#12
|
|||
|
|||
old processor -- 8085
|
#13
|
|||
|
|||
old processor -- 8085
In article ,
frédéric haessig wrote: : duplicate the CPU pipeline multiple times and have a voting system between pipeline stages... That's one solution used to reduce probability of upsets ( not only for radiation, BTW )..... provided your voting system itself is not susceptible to radiation-induced failure ( which is a design mistake I saw more than once )... Another "provided" is "provided you have some way to test the replicated pipelines independently, and without the test circuitry adding radiation vulnerabilities too". (I understand that was the really tricky part of Actel's rad-hard FPGAs, which have voted triple redundancy in every cell and hence give transparent voting-out of single-bit upsets.) -- spsystems.net is temporarily off the air; | Henry Spencer mail to henry at zoo.utoronto.ca instead. | |
#14
|
|||
|
|||
old processor -- 8085
rk wrote: On Mon, 24 Jul 2006 23:50:23 GMT, (Henry Spencer) wrote: In article , frédéric haessig wrote: : duplicate the CPU pipeline multiple times and have a voting system between pipeline stages... That's one solution used to reduce probability of upsets ( not only for radiation, BTW )..... provided your voting system itself is not susceptible to radiation-induced failure ( which is a design mistake I saw more than once )... I was talking to a guy from Intel last week, and he said that as the feature size on processors falls below 45nm the soft error rate is becoming significant across the system. "We're not going to need rad-hard latch registers yet, but..." I thought he was joking, but he wasn't. Standard commercial processor designs are going to need a significant (as in 10 percent) amount of real estate given over to detecting and correcting errors, and probably quite soon. rupert -- rk, Just an OldEngineer "The number of people having any connection with the project must be restricted in an almost vicious manner. Use a small number of good people." -- Kelly Johnson in Skunk Works |
#15
|
|||
|
|||
old processor -- 8085
|
|
Thread Tools | |
Display Modes | |
|
|