My new system

[learners permit]

The analogy I use is Mhz=Horsepower. Bandwidth=Torque. Hehe of course I'm an auto mechanic and this gives me an idea of their relationship.

EDIT: Hmmm. Latency= throttle response?

[Thraxz]

Actually, bandwidth is more like horsepower, mhz is RPM and that latency is the torque inverted.

 

It's really a crappy analogy.

[learners permit]

I see your point as both those measures use time as a factor.

[Thraxz]

(rpm x torque)/5250 = hp

 

Unsimplified:

 

(2pi x rpm x torque)/(550 x 60) = hp

 

The point is... it does even come close to fitting that.

[Angry_Games]

Actually, bandwidth is more like horsepower, mhz is RPM and that latency is the torque inverted.

 

It's really a crappy analogy.

actually its more like Learner's Permit said with A64's

 

Mhz is horsepower...pure and simple

 

bandwidth...is like...useless lol

 

latency...is like useless also

[Angry_Games]

Correct me if I am wrong GM, with the socket A platform as well as others... do you get higher overclocks with only 1 RAM module or not?

 

I remember something like this, never tried but knew someone... oh, I remember now, he was using two sticks of 256mb each, and in theory you can overclock more than with two sticks of 512mb. Is this true as well???

 

I wonder what the socket 939 has to say about this, since maybe loosing real dual channel will hurt a lot... but the memory controller should work well with just one stick.

 

I feel I should have read all the other messages... lol, well sorry... me no english

any platform, regardless of cpu, overclocks better with only a single DIMM, and a single-sided 256MB DIMM will overclock better than a 512MB DIMM, which will overclock better than a 1024MB DIMM (assuming all DIMMs carry the same IC, which they don't, but you get my meaning)

 

 

the less load on a memory controller (whether a separate hub like AthlonXP or Intel, or integrated like A64), the better overclock can be achieved

[Angry_Games]

ps: always keep in mind that NEO is running the old NF2 system with a separate memory controller hub...and is NOT running the A64 or Sempron cpu with the integrated memory controller (I've tried to advise him many times about trying to compare an apple XP to an orange A64 but I suppose he still hasn't heard what I've said).

 

The hard part is that until you have owned both, and have ran them side by side with practically the same hardware other than mobo+cpu, you can't truly appreciate the differences (nor can you appreciate the superiority of the A64/Sempron and it's integrated memory controller)

[Thraxz]

It would seem to me the mhz and bandwidth are as interrelated as hp and torque. One determines the other with modifiers of course.

 

200hp at 5250 rpm means your motor has 200 ftlbs of torque while 200hp at 52500rpm means you have 20 ft lbs of torque.

 

Likewise, 200 ft lbs of torque at 525 rpm is only 20 hp and 200ft lbs at 52500 rpm means you have 2000hp.

 

The closest parallel I can come up with is that MHZ are torque, modified by latencies = bandwidth

 

and vice versa

 

bandwidth modified by an inverse of the latencies = mhz

 

The key thing is, which value most closely relates to hp, Mhz or bandwidth? I personally side with bandwidth from my own experiences.

 

EDIT; Also, if we were to view mhz and bandwidth as hp/torque, different types of applications would favor one or the other.

 

A low torque high HP Honda S2000 wouldn't do very well hauling 10k lb trailer up a hill, however, a high torque low HP Dodge Ram diesel would do it easily. It is easy to make the parallel that something like superpi would prefer bandwidth over mhz or which ever it may or may not prefer and the exact opposite apply to things like video benchmarks.

 

This is starting to make too much sense to me and I need to test it out.

[Sorrento]

any platform, regardless of cpu, overclocks better with only a single DIMM, and a single-sided 256MB DIMM will overclock better than a 512MB DIMM, which will overclock better than a 1024MB DIMM (assuming all DIMMs carry the same IC, which they don't, but you get my meaning)

 

 

the less load on a memory controller (whether a separate hub like AthlonXP or Intel, or integrated like A64), the better overclock can be achieved

Thanks AG... I think I was correct but had my doubts about it.

 

 

As for RPM, Torque, megahertz, bandwith... you are all wrong, computers and cars don't have anything in common... or do then

[TheGeekMistress]

Correct me if I am wrong GM, with the socket A platform as well as others... do you get higher overclocks with only 1 RAM module or not?

 

I remember something like this, never tried but knew someone... oh, I remember now, he was using two sticks of 256mb each, and in theory you can overclock more than with two sticks of 512mb. Is this true as well???

 

I wonder what the socket 939 has to say about this, since maybe loosing real dual channel will hurt a lot... but the memory controller should work well with just one stick.

 

I feel I should have read all the other messages... lol, well sorry... me no english

the NF2 platform is a traditional NB/SB chipset. the bus from cpu to NB is single channel (or 64bit), but in the NB it has two memory controllers (2x64bit, hence 128bit mem bus)). so the system memory runs like it would as a dual channel setup, but the XP core was never optimised for that much bandwidth so it only uses a single channel bus from the dual memory controllers. all this adds up to a tad better bandwidth and better system ram to NB latency.

 

my own experiences with my NF2 IU, is that with my BH5 2x256 i could run 255mhz 10-2-2-2-2.0-12-16-4-4-6-4-3-3-5-1T stable and taking out a stick running only one stick made zero difference. with my TCCD 2x512 i could run it up to 265mhz but that wasn't always stable, that was hitting the top end of my chipset, not topping out the ram. but with the TCCD in dual channel i could run 260mhz 8-2-2-2-2.0-9-11-1-1-3-3-2-1-3-2T stable. for me, running only one stick never made any difference in timings or overclocking.

 

with the 939 nf4 platform it's changed from the nf2 platform. the memory controller is internal to the cpu and there's no NB mostly b/c the mem controller went "on die" and other functions were combined with the traditional SB to make one a one chip chipset. the 939 core is a true 128bit interface al the way. so having two 64bit memory controllers that equal 128bit bus in the 939 cpu, taking out a stick and only running one stick of ram will make no difference as each 64bit mem controller in the internal memory controller operates only one stick each when running two stick of ram in dual channel.

 

on my NF3-250 skt754 platform, it's only a 64bit bus or single channel. so loading the one 64bit memory controller with two sticks of ram, as opposed to the 939 having one stick per mem controller, the two sticks on mine loads the single comtroller down and doesn't perform as well. if i take out one stick of ram on my NF3 chipset, then that will allow for better perfromance.

 

TGM

[TheGeekMistress]

I did, I put the memory down to 2-2-2-5 1T but I had to put it on a insanely low divider that put the memory down at 180mhz. My efficiency was 94% but I could only pull ~5400mb/s and my superpi went up to 33.0xxs.

if you were to possibly find a mid point of tight'ish timings and a high'osh bus, your scores may be better than your best. but then again, you're running a 939 dual channel platform and there just might be something special to that platfrom that makes it different than the platform i've worked with.

 

TGM

[ReelFiles]

TGM, you are just a fountain of knowledge.

I was wondering if you hold any certifications or degrees. How's it going with your WC setup?

 

Btw, I think I have hit ludacris speed, as I am dual priming, surfing the web with Firefox and working in Paint Shop Pro right now. hehe.