From: Andrey G. G. <A.G...@in...> - 2011-01-17 16:39:43
|
Hello *, Running times of the maxima testsuite by various lisps on my pentium E5300 box are sbcl-1.0.45 165 cmucl-20b 172 gcl-2.6.8_pre 172 ccl-1.6 322 ecl-11.1.1 344 clisp-2.49 552 For the first time, sbcl became fastest in this benchmark (in recent times, cmucl was faster). But... Now there is a new testsuite failure with sbcl. It's rtest16, problem 385. It's about floating-point calculation of zeta(%i+3) (%i is the imaginary unit). sbcl produced a resulr with error of order 3*10^(-9) instead of expected 10^(-15). Earlier versions of sbcl did not have this failure. Looks like a regression. All the other lisps listed above also don't have this particular failure (ecl has 2 failures in rtest8, this is another story). Looks like something is calculated in single precision when it should be done in double. It would be interesting to trace which change in sbcl has caused this regression. Many thanks for this excellent lisp system (I use it for maxima and fricas), Andrey |
From: Andrey G. G. <A.G...@in...> - 2011-01-17 16:35:33
|
P.S. ecl-10.4.1 had the same failure (plus in the previous problem 384). But in 11.1.1 they are fixed. Maybe, ecl changelog is of some help. Andrey |
From: Adam M. <ada...@o2...> - 2011-01-30 16:53:40
|
Dnia Mon, 17 Jan 2011 22:35:21 +0600, Andrey G. Grozin napisał(a): Hi, Thx for informations. Could you describe your memory settings ? ( How big memory do you use ?) Adam Do > P.S. > > ecl-10.4.1 had the same failure (plus in the previous problem 384). But > in 11.1.1 they are fixed. Maybe, ecl changelog is of some help. > > Andrey > > ------------------------------------------------------------------------------ > Protect Your Site and Customers from Malware Attacks Learn about various > malware tactics and how to avoid them. Understand malware threats, the > impact they can have on your business, and how you can protect your > company and customers by using code signing. > http://p.sf.net/sfu/oracle-sfdevnl |
From: Nikodemus S. <nik...@ra...> - 2011-03-23 18:42:48
|
On 17 January 2011 18:15, Andrey G. Grozin <A.G...@in...> wrote: > Now there is a new testsuite failure with sbcl. It's rtest16, problem 385. > It's about floating-point calculation of zeta(%i+3) (%i is the imaginary > unit). sbcl produced a resulr with error of order 3*10^(-9) instead of > expected 10^(-15). Earlier versions of sbcl did not have this failure. > Looks like a regression. All the other lisps listed above also don't have > this particular failure (ecl has 2 failures in rtest8, this is another > story). Looks like something is calculated in single precision when it > should be done in double. Do you recall what's the latest version in which that test passed? (It fails with 1.0.44 as well.) Alternatively, if a Maxima wizard can reduce a test-case that doesn't involve Maxima -- just shows an SBCL operation that returns the wrong value -- what would help too. Cheers, -- Nikodemus |