Timings vs Actual Speed

Wiggins · 10 Jul 2007 at 11:34

HI

I am now satisfied that I can quite happily overclock my processor and have enough knowledge to do so safely.

The next thing I need to get my n00b head around is memory.

I have a Dual Core 930 @ 3.6Ghz - FSB 960 on a P5N-E Motherboard and 4x512mb sticks of Samsung DDRII 533Mhz Memory in dual channel.

Overclocking the CPU to 3.6 gave me a speed of 640Mhz on the Ram with 4-4-4-10 timings.

I unlinked the FSB to Memory and dropped the memory back to 533Mhz with 3-3-3-8 timings.

Which one would realisticly be the fastest? I cannot increase the memory speed above 533Mhz with 3-3-3-8 timings.

Your help would be appreciated

Cob · 10 Jul 2007 at 11:37

640mhz 4-4-4-10 should be quicker.

But benchmark both speeds to see for yourself.

Wiggins · 10 Jul 2007 at 11:39

Thnx Cob

What's the best software to benchmark it? There are sooooo many out there.

Cob · 10 Jul 2007 at 12:04

Something as simple as Super-Pi will show the difference. Tho Everests or Si-Sandras memory benchmarks would give better results.

Wiggins · 10 Jul 2007 at 13:25

Well here are some of the changes I made to the memory and benchmarked using PCMark05 on the memory test.

4-3-3-8 T2 @ 533Mhz - 4392

4-3-3-10 T2 @ 640Mhz - 4572

4-4-4-12 T1 @ 640Mhz - 4585

4-4-4-10 T1 @ 660Mhz - 4667

I couldnt tighten the timings anymore @ 660Mhz as it wouldnt boot.

I suppose thats not bad for a set of Samsung Modules although I am looking at maybe getting the GeIL 2GB PC6400C4 800MHz Ultra Low Latency kit.

Does anyone have a set of these?

Wiggins · 10 Jul 2007 at 13:29

I used SIS Sandra and tested the memory bandwidth and got this. My score is the red. Is that good????

Wiggins · 10 Jul 2007 at 13:36

And I got this on the Latency benchmark on Sandra.

split · 10 Jul 2007 at 13:37

It's good compared to say this

Wiggins · 10 Jul 2007 at 13:44

split said:
It's good compared to say this

Do USB peripherals really make a difference plugged in when benchmarking?

split · 10 Jul 2007 at 14:18

Wiggins said:
Do USB peripherals really make a difference plugged in when benchmarking?

Me and my mate both had early DDR boards (ASUS A7M266) and no matter what I did my benchmarks where always slower than his.
I was on ADSL with a USB modem (early ADSL days), he had dial up. (PCI modem)
I found that out by trial and error. It was a bug within the USB controller and that board. Had many arguments with Alcatel over that, but they admitted it in the end.

It was fixable with a PCI to USB card (NEC chipset) but it just shows how some benchmarks do highlight problems. I wasn't the only one to have same problems...see here. wow this is old stuff.
I think it's all got better since then though. I assume it has, but hey.