Benchmarking question using SISoftware

[antgross@pacbell.net]

I realize there are many benchmarking methods, but is there one that is more important than the other? For example, all things being equal (except CPU) is it fair to say that if I ran this test on a AMD 3800+ and got a Dhrystone ALU of 12000, would I say it's about 40% faster than the system below?

I'm trying to get a general sense of what kind of performance increase I would get in upgrading from AMD 2700 to 3800.

thanks


SiSoftware Sandra

Benchmark Results
Dhrystone ALU : 8877MIPS
Whetstone FPU : 3370MFLOPS

Performance Test Status
Run ID : ANTHONY2 on Friday, August 13, 2004 at 8:14:06 AM
NUMA Support : No
SMP Test : No
Total Test Threads : 1
SMT Test : No
Dynamic MP/MT Load Balance : No
Processor Affinity : P0
Number of Runs : 64000 / 640

Processor
Model : AMD Athlon(tm) XP 2700+
Speed : 2.16GHz
Model Number : 2700 (estimated)
Performance Rating : PR3136 (estimated)
Type : Standard
L2 On-board Cache : 256kB ECC Synchronous Write-Back (16-way, 64 byte line size)

[doctorgonzo]

What are you using the system for? Synthetic benchmarks aren't all that helpful without some context. If all you do is e-mail and word processing, an AMD 3800+ isn't going to make you type 40% faster, so you won't see much performance benefit at all.

[antgross@pacbell.net]

I'm just trying to get a sense as to which benchmark is the most important. I use chess software and databases and want to know if ALU or FPU is the most important benchmark for this kind of thing. The databases searched are often asked to look for particular position etc and it has to search millions of positions.

[doctorgonzo]

Databases are I/O and memory dependent, not processor dependent. Having a lot of fast RAM and a fast HD are far more important for databases than processor speed.

The chess algorithms probably use floating point numbers, so the FPU benchmark would be more accurate.