gmac is slower than hmac?

[viperscape]

Using ghash to create a gmac I noticed is slower than an hmac using sha256; I was under the impression gmac would be faster when c is left blank (no cypher). Am I using this incorrectly, or is this expected? Thanks

test msg::tests::ghash  ... bench:      18,495 ns/iter (+/- 1,067)
test msg::tests::sha256 ... bench:       8,883 ns/iter (+/- 930)

    #[bench]
    fn ghash(b:&mut Bencher) {
        let key = [random::<u8>();16];
        let d = [random::<u8>();1400];

        b.iter(||{
            let mut ha = Ghash::new(&key);
            ha.input(&d[..]);
            let mut gmac = ha.result();
        });
    }

    #[bench]
    fn sha256(b:&mut Bencher) {
        let key = [random::<u8>();32];
        let d = [random::<u8>();1400];

        b.iter(||{
            let mut sha = Sha256::new();
            sha.input(&d[..]);
            let mut hmac = Hmac::new(sha,&key[..]).result();
        });
    }