The Affect of Memory Settings to Performance

[VorLonUK]

I came across this article over at Tomshardware. It is over 2 years old now but focuses on DDR2 memory and uses Everest as a benchmarking tool. (Which seems to be popular on this forum).

I've linked to page One of the article as there is a wealth of information on every page throughout it. Further into the article there is more specific timings vs volltage vs performance, which is something that is frequently discussed on here.

I'm also still testing my system at different settings (mainly memory wise) and building a spread sheet on performance vs stability.

The article is here...
In Search of True DDR2 Bleeding Edge Memory : Optimal DRAM For Overclocking - Tom's Hardware

[BongWizard]

Re: The Affect of Memory Settings to Performance

That's a pretty good article Vorlon... only problem is its a little bit dated, but still has some useful information for those looking to get the most out of their system... nice find!

[bilbat]

Re: The Affect of Memory Settings to Performance

Excellent info! Will be looking for that speadsheet...

This info is making me lazy - think I can pretty much just 'dial-in' my DQ6 when I finally buy it!

[dylan0451]

Re: The Affect of Memory Settings to Performance

looking at the attached fan/header setup, i didn't realise you could use the power supply fan header as a normal header - i thought it was only there to monitor the psu fan speed?!

[Lsdmeasap]

Re: The Affect of Memory Settings to Performance

Depends on the Power supply really. I have not seen one for a long time with a wire that monitored speeds alone. Nowadays if your PSU had a RPM output the data is transfered thru the 24Pin cable

[VorLonUK]

Re: The Affect of Memory Settings to Performance

Depends on the Power supply really. I have not seen one for a long time with a wire that monitored speeds alone. Nowadays if your PSU had a RPM output the data is transfered thru the 24Pin cable

All the Enermax Psu's I've had (and I've had several) have had a 3 pin "Tacho" out cable for speed monitoring.

Even a more recent ELT500awt (Enermax Liberty) has the connector too (3 pin)

[Lsdmeasap]

Re: The Affect of Memory Settings to Performance

I have a few Ultra's with 3pin out, but for powering fans as there is 3-4 plugs on them

I guess I should check, maybe they also output RPM's. But I do know I can see them without it as well

[bilbat]

Re: The Affect of Memory Settings to Performance

"looking at the attached fan/header setup, i didn't realise you could use the power supply fan header as a normal header - i thought it was only there to monitor the psu fan speed?!"

If you'll look again, there's only two wires going to that header - ground and speed sense (yellow) - the CPU_FAN header handles the control for both...

[dylan0451]

Re: The Affect of Memory Settings to Performance

ahh, yes i see now... thats a good idea

whats the max current load you can put on one of these headers?

[bilbat]

Re: The Affect of Memory Settings to Performance

I recently posted this over in cooling:

Artic Cooling's multi-fan FAQ says:

"In most cases, we recommend users to connect maximum 5 PWM fans. With 5 PWM fans running at full speed, the current consumption will still remain within the motherboard specification. Recent mother boards (i.e. P965) supply up to 3.0Amps current from the CPU fan socket. In this case, one Arctic Fan 12 PWM fan consumes 0.13Amp during maximum speed operation. When up to 5 fans are connected: 0.13Amp x 5 = 0.65Amp. It is still far below the specification. Some older boards (i.e. i865) supply 1.8Amp current from the CPU fan socket, still sufficient enough to handle. For more information please consult your motherboard manual to obtain the maximum current your motherboard can supply to the CPU fan."

I am assuming that, as a manufacturer, Artic has access to MOBO data that I don't. I haven't done a really thorough search for specific board info, and am generalizing from the nine or ten MOBO manuals I DL'd and read in the process of eliminating the finalists for my next development system (which will be a Gb GA-X48-DQ6). I certainly could have missed something, as it's a total of more than a thousand pages of PDF's, but it seems to me that actual technical data (unless it pertains to OC'ing) is scarcer than hen's teeth. That said, though, if you think about it, the PWM spec's 1.5A seems pretty likely to be accomodated: my little 'handy cheat sheet' says I can put 1.4A through a 28Ga wire, and that's less than thirteen thousandths in diameter, or roughly two or three times the diameter of a black human hair! I don't know offhand what the current thickness of board traces is, but I would guess they have at least that cross-sectional area, and therefore at least that current capacity...

If anyone locates these spec's for THEIR MOBO, please post 'em here so we can make some better informed guesses!

BTW: the PWM 'spec' is 1.5A continuous; 2.2A start-up surge...

[VorLonUK]

Re: The Affect of Memory Settings to Performance

I can put 1.4A through a 28Ga wire, and that's less than thirteen thousandths in diameter, or roughly two or three times the diameter of a black human hair! I don't know offhand what the current thickness of board traces is, but I would guess they have at least that cross-sectional area, and therefore at least that current capacity...

Not sure if I've read your wording right there Bilbat..

I've used 0.22mm(squared) 7/0.22 which is much thicker than a human hair and is rated at 1.4amp for internal equipment "hookup".

Also Motherboard PCB's use a multilayer approach, so you might not be able to see the "Print" (copper traces) that are being utilised anyway.

The Print "sandwich", is often formed like:-

Signal
Supply
Signal
Signal
Ground
Signal