关于６.５与６.６的求解器的问题，有同时有过的进来下

xingfeng45

我前段时间用６.５建立模型分析的，６.６虽然也装了但是没有分析过同样的问题，最近重新装电脑，只装了６.６的，因为要继续进行就想看下两者结果的差别，结果我６.５中的模型在６.６中计算原来cohesive单元已经完全破坏，但是现在分析的时候竟然没有一点破坏的迹象，我原来以为是模型问题，重新建立，可是还是老样子，这样的问题大家遇到过么，怎么会事，两个我用的都是forture６.６装的

aresaran

6.6的cohesive有很大改进.

kingy

下边是和一个abaqus技术支持的通信，或许对大家有点提示。
XXXX,
take a look at the following data which is from a benchmarking I did just now. It is a fairly small problem (required memory smaller than 256M). Also I think it is a problem that accesses the disk very intensively. For 6.54, it took 652 sec. I am not surprised with this number cause I used 6.51 before. However, the 6.63 is much much faster. The I/O time is so small, I am even thinking that this new version putting small scratch file in memory. Do you have any idea?

xxxxx

      　　　　　　　 6.5-4(1)    6.6-3(1)
USER TIME (SEC)         122.2   112.6
SYSTEM TIME (SEC)       10.8    13.3
TOTAL CPU TIME (SEC)    133 　125.9
WALLCLOCK TIME (SEC)    652　134


Reply

Yes xxxx. The solver has been optimized better in 6.6. This is particularly true if there is contact in the model. ABAQUS HQ has done similar benchmarkings on a variety of problem and shown that 6.6 is better performer.

Thanks for the note.

Cheers,
xxxxxx

dava

我现在同时有６．５和６．６，xingfeng45的问题蛮有意思的，是个不错的经验，发给我看看行么？最近正好有空算东西，看来高版本的abaqus求解器有一些明显改动

xingfeng45

这两天学校检察，qq没有装，
不过模型也前曾经发给过你，不知道你有没有了
你自己也可以建立个cohesive模型，看结果一样不？是用power　low的

dava

好的，我看看。估计这个问题是因为高版本的求解器和低版本相比有了很大改进