From: Mark Booth Newsgroups: comp.sys.transputer Subject: Re: ST450 Date: Thu, 18 Feb 1999 18:57:15 +0000 Organization: OxIM Ltd. Message-Id: References: <36c03424.0@glitch.nildram.co.uk> Reply-To: Mark Booth Mime-Version: 1.0 Xref: ukc comp.sys.transputer:9053 In article , Wayne Fisher writes >In article <36c03424.0@glitch.nildram.co.uk>, GP wrote: >>Does anyone know how the ChanOutTimeFail function works? Source code?! >If you're trying to get this function working on a ST20450, you may be >out of luck. We've noticed that the LSC version of ChanInTimeFail does >not work correctly on the ST20450. Using it would occassionally - >about 1% of the calls to it - result in corrupted data. Interesting. We were getting corruption on channel communications when using normal Virtual channels. I wonder whether this is related. >It looks like either enbt or dist >is the culprit. I still find it difficult to believe that SGS could screw up the Transputer so badly. It would have been so much nicer if the 20450 had been a 40MHz T805 or even T425. >We gave up at that point and re-thought what we were trying to do. Yup, we have just thrown out all of out 20450 based boards, and thrown away all of the development that went with them. All in all it has been a complete disaster. If SGS had told us 'this processor doesn't work, and we will never make it work' two years ago we could have implemented it using T425s and had the who system working a year ago. >I thought I'd >give you a warning. I would warn anyone away from using 20450s. Just junk the boards, it will save you time and expense even in the short run. Take care, Mark......... (Bitter and twisted, moi? *8') -- Mark Booth - My ReplyTo: address is valid, other addresses are for tracking UCE -------------------------- markbSig@oxim.demon.co.uk -------------------------- OxIM Ltd. Registered Office 12 Kings Meadow, Ferry Hinksey Road, Oxford OX2 0DP Voice +44(0)1865 204881, Fax +44(0)1865 204882 Registered in England 3619439