[Chinasaur]

I do ~60 nodes per second on my Dual 1.8.

Can any of you smart people estimate nodes per second with a Quad 2.5?

TIA.

[Lateralus]

4x 2.5GHz / 2x 1.8GHz = 2.77~
~60 x 2.77~ = 162~
And say minus 10% for latency = 145.8

Yayz, I win!!!11

[OneMacGuy]

Originally Posted by Lateralus

4x 2.5GHz / 2x 1.8GHz = 2.77~
~60 x 2.77~ = 162~
And say minus 10% for latency = 145.8

Yayz, I win!!!11

Given the speed quotes I have seen on some news sites, specifically that a Dual Core 2.3 was faster than an existing dual processor 2.5, I think it "might" be around 10-20% higher. Don't forget that the new machines also have significantly faster memory I/O. They are now using DDR2 memory.

[Lateralus]

The only benchmarks I have seen done on the new machines were Xbench. Xbench hardly ever translates to real world performance.

[Karl Schimanek]

Originally Posted by Lateralus

And say minus 10% for latency = 145.8

Yayz, I win!!!11

I don't think so.

First:

System controller
This fast application-specific integrated circuit (ASIC) uses a lower-latency memory subsystem with support for ultrafast DDR2 SDRAM.

Source: Apple

Second: 1MB Level 2 cache

Thrid: Quad Benchmark *drool*
dual 2.8 Xeon: 49 seconds
dual 2.5 G5: 38 seconds
quad 2.5 G5: 17 seconds

Karl

[Lateralus]

We'll see.

[Karl Schimanek]

Yes we will
I second my statement: Faster Chip-to-Chip communication
And DDR2, see XBench (XBench isn't that bad )

Karl

[Lateralus]

The new G5's may have a faster memory controller, but DDR2 533 isn't any faster than DDR 400 because of the higher latency. To acheive any real performance gain with DDR2 over DDR 400, you need to go with DDR2 667.

Which Apple didn't do, for whatever reason...

[Karl Schimanek]

I know, but there is a memory available with very good latency.
http://www.ocztechnology.com/product...d_dual_channel

And the XBench results:
2.3 Dualcore: 121,99
2.0 Dualcore: 110,32
2.5 Dual: 103,73