Institut ³Jozef Stefan², Ljubljana, Slovenia 1001 Ljubljana, p.p. 3000 / Jamova 39 / Tel.: n.c. 1773900, Fax: 1235 400 Vas znak: nas znak: datum: February 15,1996 Dr. Jason. S. Aaron, Director, Genscan U.S. P.O. Box 28754 Santa Fe, NM 87592-8724 Dear Dr. Aaron, I understand your concern with deadlines for the CEPH linkage calculations, however, we are unable to run the Parallelised Fastlink package yet. We are in constant contact with both your linkage group and Convex SPP team, but have not been able to pinpoint the source of stack overflows that crash our test runs. At the moment, we have been unable to secure definite assurances from either Convex or your group as to when we can expect stable operations. I must therefore propose that you reconsider the schedules for complete linkage computation of all current families and subsequent statistical analysis of psychological data. My personal belief is that we must wait for the Mach V3.4 release for improved cache consistency (which may be the prime cause for stack overflows) which is scheduled for end April; we have made arrangements to test run the beta version (end March). Providing we can achieve a stable environment for Fastlink by the end of April, the delay in these calculations should not exceed two months. Best regards, Vladimir Alkalaj Head of National Supercomputing Centre