Potential speedup using decimal floating-point hardware | IEEE Conference Publication | IEEE Xplore