dnetc benchmarks

[minator]

I posted a link to some benchmarks in this thread in post 228.

According to them the PA6T not only beats a G5 at a higher clock speed but it completely destroys the G4.

Looks like it is misconfigured to me.  Is the L2 switched on?  What power mode is it in?  Are both memory controllers on? etc...


BTW This is a remarkably common problem. Even the big boys do it.

[minator]

Lets put it this way: what Hyperion is going to do when PPC is finished?

Good question, and given all the PPC suppliers are switching to ARM, one they are going to have to answer fairly soon.

[minator]

Interesting opinion, but false.

Really?

Freescale - shipped low end ARM for years but have been recently been phasing out other architectures.  Yes, they are developing a high end PPC for networking but they have also licensed Cortex-A15 which will be similar performance wise.  Given the massive cost of developing CPU cores I can't see them doing anything other than die shrinks after this.

Applied Micro (AMCC) - They don't design PPC cores, they mainly license them from IBM.
They have announced a 64bit ARM for servers that is more aggressive and clocked higher than anything they have have PPC wise.

LSI - Another company doing ARM at the low end.  They had started doing PPC but they also just licensed the A15.

IBM - Go find me a PowerPC that you can actually buy as a chip that isn't a G3.
IBM don't make ARM chips but they are on Linaro's board - Linaro is a company that does ARM Linux.

No current ARM processor outperforms even the dated G5.

A high end A15 should at least equal a high end G5.

ARM is an in-order processor design

Wrong. Both the A9 and A15 are out of order.

[minator]

So I don't think it's up to misconfiguration, but rather CPU bound vs memory bound tasks. In Addition OS4 does not support the 2nd core of PA6T.

But...

Many of the benchmarks shown so far are hitting memory but none of them show any advantage over the G4.

OTOH some are AltiVec and maybe it's really fussy running that?

The memory and especially L2 benchmarks look lower than I'd expect.  My guess is the memory might be running slow and the L2 is running very slow.

Linux benchmarks should confirm it.