Error analysis and precision estimation for floating-point dot-products using affine arithmetic | IEEE Conference Publication | IEEE Xplore