Frequency vs latency question

[JDocs]

Hi all.

I've got a question about my ram, after using Sandra SisSoft (not the best benchmark tool I know but it does sustained memory speed vs the rated burst speed found on your memory chips spec) and I was wondering which of the following would be the best settings for my RAM.

800mhz @ CL3 : +/- 7000 @ 110% efficiency
940mhz @ CL4 : +/- 8000 @ 80% efficiency
1066mhz @ CL5 : +/- 8500 @ 60% efficiency

Does their "efficiency" rating mean anything as I never see it brought up in benchmarks and so on?

[Wolvebain]

Topic has been covered, to some extent before...

Here's just one thread I remember of it: http://www.pcbuyersguide.co.za/showthread.php?t=5873

Regarding the `efficiency` aspect... I would assume that it is the enegry / power efficiency [and not bit efficiency ] ... sorta like saying the RAM is designed for 2.0v operation... it can run as 1.8v @ 800MHz, giving it a higher efficiency % than rated... or 2.2v @ 940MHz, giving it a lower efficiency %... ?

Just an assumption here... it may be best if you provide the source of your quoted statz / spex:

800mhz @ CL3 : +/- 7000 @ 110% efficiency
940mhz @ CL4 : +/- 8000 @ 80% efficiency
1066mhz @ CL5 : +/- 8500 @ 60% efficiency

[JDocs]

The source being SisSandra on my machine at the different frequencies / latencies. Trying to figure out which one is best.

[Wolvebain]

The source being SisSandra on my machine at the different frequencies / latencies. Trying to figure out which one is best.

HmMmm...
I would then assume it is bandwidth efficiency then:

Q: Why do some systems show close to 95% bandwidth efficiency and others less than 80%?
A: While the code for all CPUs is heavily optimised, the effciency depends on chipset performance and memory settings. Only the most aggressive settings may yield > 80% efficiency, thus anything higher is a bonus.

source: SiSoftware's Q & A - Memory Benchmark page

Since you already have the software installed, you prob. would get better info / answers by searching the HELP file....

[JDocs]

I saw that but I'm not sure how much it means hench I am asking about it. I'm worried it could be something like X2900XT which does wonders in 3DMark06 but in the real world isn't as good as some of the cards it beats in 3DMark06.

Is it relevant and worth considering or just an artificial number (specially considering 110%)?

[Wolvebain]

I saw that but I'm not sure how much it means hench I am asking about it. I'm worried it could be something like X2900XT which does wonders in 3DMark06 but in the real world isn't as good as some of the cards it beats in 3DMark06.

Is it relevant and worth considering or just an artificial number (specially considering 110%)?

Okay.. firstly you need to understand what testz are being run by the benchmark tool...
-- You need to understand what objective of the test is, and how the result is calulated / gained. This will assist you in understanding what, why, and howz of the test. As I said before, check the HELP file for this info.

Then what, for you, are your real world performance dimensions...
-- There are `Real-World`performance dimensions for different applicationz, environmentz, systemz, servicez, etc... Not all gamez, like benchmarkz, are designed the same nor are they all optimized for evey platform [H/W & S/W]...

Unlike console platformz like the Sony Playstation where one cannot really `upgrade` the componentz, which forced [along with the competion factor] game developerz to uber optimize their gamez for a spesific platform --any one here played & compared GT2 and GT4 on PS2 :blink:...

So a benchmarking tool on that platform would both 100% acurate & 100% useless:
- 100% acurate: Platform stayz the same, so benchmark tool is optimized for it, as well as the fact that the benchmark will include testz for console gaming experiance, as that is what the platform is there for;
- 100% useless: if everythingz the same, then the resultz would be the same as wel [duh].

PC`z on the other hand are highly versitile... So before deciding, based on someone else`z view on whether the benchmark resultz are vaild or have any weight to you... you need to see what you use it for and teh performance dimensions for these taskz.... Alternatively, if you want to, you could use the uBer dumbed-downed Ms Windows Experiance Index that Vista providez for the average dumbed-downed end user.. and hope that application designerz and developerz also limit their work / design and developement technique based on this, as well as technology does not progress beyond this rating