Sandra MAX3

[Prof.Wizard]

Has anyone tried the newest version?

 

[Prof.Wizard]

Few changes but still the best all-around benchmarking tool.

 

[honold]

*cough*

for all-around benchmarking i think several purpose-built, good tools is appropriate. when i benchmark my systems i use 3dmark2001 se, quake 3 1.32 demo four, winrar compression of a 200mb file, superpi 4m (calculates 4 million digits of pi), aida32, atto, and winbench 99

if i had etestinglabs' content creation and business use benchmarks i'd probably use those as well. if i wanted some kind of all-arounder, i'd go for pcmark.

aida32 is great software for system info, you guys should check it (and cpu-z) out.

 

[Prof.Wizard]

...for all-around benchmarking i think several purpose-built...

I would call that oxymoron.

 

[honold]

i would call that missing the point

the goal is to get a valid all-encompassing metric, and the means is to use multiple, specific tools to come up with an average (as opposed to a singular averaging benchmark that doesn't measure any individual thing particularly well)

 

[Prof.Wizard]

i would call that missing the point

So you were prolly talking to yourself...

the goal is to get a valid all-encompassing metric, and the means is to use multiple, specific tools to come up with an average (as opposed to a singular averaging benchmark that doesn't measure any individual thing particularly well)

If I need to measure with objectivity a certain component of my system I will download the best tool, be sure about that. But to measure all things with one app, I'll go Sandra, even if I won't be able to bet my life on the 100% accuracy of the results.

Moreover, Sandra is also great for the sysinfo that it gives. Like aida32.

 

[honold]

have you used pcmark?

 

[blakerwry]

I'm wondering what would be faster for the SS51G

Currently using a Celeron Northwood 2.0GHz CPU and Kingston value RAM PC2700...

should I

a) run at 100MHz (sync) with ultra fast timings
b) run at 133MHz (pc2100) with fast timings
c) run at 166MHz (PC2700) with normal timings


I currently have it running at the same setting I ran the RAM at on the 845GE chipset (266MHz, fast timings) But this was because the 845 is limited to a maximum RAM frequency of 133 with a 100MHz FSB. But my instincts tell me that running with normal timings at 166MHz should be the fastest.

How can I test this accurately, I know Sandra has a multimedia bench and a Memory bench that both help to get an idea of raw RAM speed, but is there anything more sophisticated that will mimic real life performance or are the Sandra tests reflective of real life performance?

 

[honold]

funny you should mention...

http://storageforum.net/forum/viewtopic.php?t=2400

 

[honold]

are the chips on your value ram winbond bh-6? if so, with enough volts, you could easily run pc2700 at tight timings

 

[Prof.Wizard]

have you used pcmark?

No never. But I've heard of that.
Is it supposed to be better?

Browsing on Futuremark's site it doesn't seem to have as many features as Sandra MAX3 Professional.

 

[blakerwry]

well I'm glad it's an intel chipset...

Nforce seems to run best in sync than out of sync... but the FSB of the northwood is QDR and thus the closest thing to sync might be DDR400 (assuming the proper buffers)...which i don't think is supported.

I tend to think that the 845 would perform similar to 875, but not sure about the SiS... I think I'll try a benchmark comparison like yours and get my own results.

.. I am, however, expecting that my results will agree w/ yours despite the different setup.

 

[honold]

if i were you my aim would be to get the highest ddr clock possible, and then using that ddr clock get the tightest timings possible

 

[blakerwry]

if i were you my aim would be to get the highest ddr clock possible, and then using that ddr clock get the tightest timings possible

I noticed your tests show CAS 2 as the fastest, but my 845 shuttle suports 1.5 CAS... does the 875 not support CAS 1.5?

 

[honold]

nope