I am currently evaluating different types of hardware to upgrade our workstations. In the course of running some control benchmarks, I've gotten some results that I can't easily explain, and I'm questioning AMD vs. Intel when I wouldn't have ordinarily done so.

System 1 = Intel Core2 Duo 6400 @ 2.13GHz, 4GB RAM, 1066FSB, 160GB SATA, Radeon HD 3870 Softmodded to Firestream 9170, XP Pro SP2

System 2 = AMD Turion ML-30 @ 1.76GHz, 1GB RAM, 800FSB, 160GB SATA, Radeon HD 3870 Softmodded to Firestream 9170, XP Pro SP2

I expected to see a 25% to 35% performance gain with the Intel System over the older AMD.

SPECViewperf 10.0 Results:

System 1 (Intel)

3DSMax - 7.88
Catia - 11.87
Ensight - 34.79
Maya - 168.67
Pro-E - 16.23
Solidworks - 38.86
TCVis - 18.56
UGNX - 53.09


System 2 (AMD)

3DSMax - 14.68
Catia - 16.54
Ensight - 38.38
Maya - 149.70
Pro-E - 16.63
Solidworks - 39.06
TCVis - 14.19
UGNX - 42.74


In summary, the faster Intel machine with 4 times the RAM bests the AMD boat anchor in only three benchmarks, and even then by very little. The old AMD machine puts up a much better showing overall when considering what the hardware is - a $20 laptop processor on a $39 clearance motherboard.

In an interesting aside, the Intel System above showed about 50% of the Disk Index score of the old AMD machine when compared with Cadalyst 2008.

Is there some 'feature' of Intel CPUs/chipsets that is inferior to the AMD offerings? Conventional wisdom is that for the last couple years, Intel has been much better bang for the buck, but now I am not so certain.

Do I just have a hobbled Core Duo system, or is this typical of what can be expected in real-world performance? In non CPU-intensive tasks the old AMD system even feels faster than the newer Intel system.

We are evaluating an Intel Quad-Core machine at the moment as well as an ATI FireGL V5600, curious if we should be checking out a system based on an AMD CPU as well.