View Full Version : A RAM speed vs timings problem
i an not a nerd
05-12-2005, 08:34 PM
Well, seems that I don't know as much about RAM that i thought that i did...
A friend of mine is upgrading ram, and asked me for some advice. The two picks that i've come up with have the followings specs:
Latency 2-2-2
Frequency 400MHz
vs.
Latency 2.5-4-4-8
Frequency 550MHz
I unfortuantly don't know enough about timings to know which will perform better...
Any help in determining which stick will perform better would be appreciated.
Thanks,
pop pop
05-12-2005, 10:50 PM
oh boy.... Saph has a couple of books written here on this stuff. The math escapes me now. I'll hazard a non technical-no math guess and see what he says later. I think in general, the faster clock is better. I don't think that latency difference there will make up for the difference in the speed.
i an not a nerd
05-12-2005, 11:06 PM
O yes, i have seen some of his posts... can't remeber where, but i have seen them.
Anyways, thanks, I wasn't sure whether the higher speed would counterbalence the higher latencies.
I have 2 sticks of 512mb RAM that requires 3.3v, runs at 500 mhz and has 2-2-2! Very good performance, but it requires active cooling!!! (on stock speeds) For OCing i had to build a custom waterblock. http://www.mushkin.com/doc/products/memory_detail.asp?ID=196
heres the site for the sticks :http://www.mushkin.com/doc/products/memory_list.asp?ID=3
saphalline
05-12-2005, 11:33 PM
Is the RAM in question DDR or DDR2? It doesn't really matter, I'm just curious. ;)
For pure performance, the faster RAM will be better. However, this assumes that the faster RAM can be run at that speed. And I'm not just talking about running it at 550MHz per the BIOS, I'm also talking about the CPU in question. It's all fine and dandy to pair up a P4's 800 FSB with dual-channel 550 RAM, but the RAM won't see its full potential if the FSB stays at 800. Fast speed or not, it just won't be busy. If the CPU for this machine can keep up with the RAM, it's worth the price. If not, save the money and get the slower RAM with better latencies.
EDIT - Looking at that Mushkin link, I would actually suggest the "Redline" stuff. It hits 2-2-2 at 500MHz. What's to say it couldn't at least hit 2.5-4-4-8 at 550MHz? Or heck, at that point, I'd try it at 2-3-3 at 533MHz! :D
But that's just what I'd do... ;)
i an not a nerd
05-13-2005, 02:24 PM
Ah yes, but look closely at the redline page, it requires active cooling. I don't mind spending time, effort and $$$ on cooling, but my friend is a "IT MUST BE QUIET AND NOT COMPLICATED!!!!" kind of guy. So the redline would prolbly not work for him.
The RAM is DDR, not DDR2 and its running on an Intel machine with 800mhz FSB. Thanks, i will advise the cheaper ram.
Ya, my redline ram is running at 500 2-2-2 (stock speeds) I would OC, but my custom waterblock needs some redesigning. I'm trying to get it to stay secure in one spot, and without holes in the mobo near the ram slots, this is turning out to be a challenge! :) I personally find this fun. I'm thinking of securing the block with the case, or on the non-used RAM slots. I'm trying to design something that can snap into the non-used RAM slots as if it were RAM, but there would be no metal, so then the block could be secure in the right spot.
saphalline
05-13-2005, 04:02 PM
I'm trying to design something that can snap into the non-used RAM slots as if it were RAM, but there would be no metal, so then the block could be secure in the right spot.Ooh! That would be so cool! Don't forget that RAM modules themselves are pretty tough, so even though you can't use metal don't use something flimsy. Some type of plastic would probably work the best, but if you want to be exotic...
Find yourself a really really old mobo (I know a place where I could get my hands on several 386/486 mobo's) and cut out RAM-shaped sections from it! As long as there's no exposed traces on the surface (where the RAM slot contacts the RAM) there won't be any shorts! :D
i an not a nerd
05-13-2005, 04:54 PM
Find yourself a really really old mobo (I know a place where I could get my hands on several 386/486 mobo's) and cut out RAM-shaped sections from it! As long as there's no exposed traces on the surface (where the RAM slot contacts the RAM) there won't be any shorts!
Wow, i didn't think of that... i was thinking along the plastics line. The mobo idea would work better for heating issues as well, thanks for the idea! if i ever patent the cooler, i'll cut you in on profits lol.
vBulletin v3.6.1, Copyright ©2000-2012, Jelsoft Enterprises Ltd.