This is incorrect. That's like saying first gear and second gear give the same speed at 1000 RPMs. Clock speed is just ONE aspect of a CPU's performance. Right now, there are three high-end MPCore's we're dealing with;
Cortex-A8 (iPhone 3G S, 4, iPad, phones using the Hummingbird and TI OMAP SOCs)
Scorpion - (anything using a Snapdragon)
Cortex A9 - (iPad 2, likely iPhone 5, OMAP 4, Samsung's upcoming Exynos SOC, Tegra 2)
Clock for clock, Cortex A9 is roughly 20% faster than A8. Scorpion is roughly 5% faster than A8. A9 and Scorpion have support for 128-bit SIMD instructions while A8 is limited to 64-bit (main reason Snapdragon kills OMAP/Hummingbird in Linpack). All 3 MPCores can support NEON instructions sets. Anything using Scorpion or A8 DOES support NEON. Tegra removed NEON from Tegra 2's A9 to save on manufacturing costs, but Tegra 3 will have it. This kills performance in some areas. Exynos and OMAP 4 are expected to support NEON.
Then there's the multi-core aspect. Scorpion and A9 have support for up to quad-core, while A8 is limited to single core. Don't tell me that you honestly expect a 1ghz single-core to be the same speed as a 1ghz dual-core just because they are both 1ghz.
In terms of overall speed, it's A9 > Scorpion > A8. However, due to the removal of NEON, there will be some differences within the A9 segment. It should be;
A5/OMAP4/Exynos > Tegra (all indications are that A5 = Exynos, just as A4 = Hummingbird). And no one, NO ONE, in the industry expected the new PowerVR to kill Tegra 2. They were expected to be on par. We'll have a better idea when Exynos comes out and we can compare the two GPUs on the same OS.