G.Skill Cheap but Corsair seems to perform

C

computergeek22

Limp Gawd
Joined
Dec 2, 2005
Messages
136
I know this question has been asked many times by many others including myself once or twice. I've finally come down to my buying decision. The G.Skill DDR500 are 40 bucks cheaper than the corsair TWINX2048-4000PT which was reviewed at Anandtech and was able to reach DDR 580 @ 1T @ 1:1 when overclocked which is astonishing. The G.Skill however was able to reach DDR540 at 1:1 @ 1T. The thing is that I can always use a divider but I want to go 1:1 if possible and my Opteron 170 is already ordered. Considering these processors can usually reach around 2.7-2.8 w/ decent cooling, I was wondering should I save the 40 bucks or just go with the corsair?

I plan on using an A8R32-MVP mobo.
 
they're based on the same IC.. both will clock pretty similarly on average. ;)
 
It's really hard to just go by reviews due to the fact that each set performs differently.
But I would favor the g.skills against the corsair.
 
I hear and from personal experience this RAM, g.skill, and I would assume others (since same IC etc..) do NOT like dividers :( Couldn't get the ram stable @ 260mhz with a 9/10 divider (since my FSB was @ 289) but 260 1:1 was stable.

I think its my CPU mem controller, can't handle 2gb with divider? I dunno, but regardless the g.skill is an awesome deal for the money and is good RAM.
 
If you've taken any Statistics courses, you know that the larger the sample, the better the results. Chances are that Corsair sent Anandtech a cherry-picked set that was among their top performers. Meanwhile, G.Skill has been receiving rave reviews from nearly all of its owners [myself included] for excellent results.

My advice is to not go on the basis of a single review, but to rather get a larger, broader sampling. I can tell you right now that my G.Skill sticks [same ones as the ones you're considering] have safely reached DDR560 at stock voltage and timings, and I have tested them to DDR580, and they have been completely stable.
 
Bona Fide said:
My advice is to not go on the basis of a single review, but to rather get a larger, broader sampling.
Click to expand...
which is exactly why i'm trying to take the time to make a somewhat respectable memory oc database.. :p
 
(cf)Eclipse said:
which is exactly why i'm trying to take the time to make a somewhat respectable memory oc database.. :p
Click to expand...
If there's anything I can help you with, let me know. I'm home all day, and this sounds like an undertaking :p
 
hehe, thanks.. what i really need is an efficient way of managing all the entries, and strict guidelines that people will actually follow ( yeah right :rolleyes: )

what i'm thinking of so far is something akin to this:

1. split everything up by size, then speed rating within each size group
2. id the sticks.. name and model number. IC if known too
3. require proof of stability.. probably a superPI 32m shot with ram speed and timings easily visible

so if i had only my entries, it'd look like this so far:

256mb sticks
PC3200
Kingston Hyper-X 2x256mb - Winbond BH-5 - 275mhz - 2-2-2-6 @ 1T - 3.6v (link)

512mb sticks
PC3200
Crucial Ballistix 2x512mb - Micron 5b G - 302mhz - 3-3-3-8 @ 1T - 2.9v (link)
OCZ Gold GX XTC 2x512mb - Infineon UTT-BH - 261.9mhz - 2-2-2-6 @ 1T - 3.7v (link)

PC3500
OCZ Gold GX 2x512mb - Infineon UTT-BH - 258.6mhz - 2-2-2-6 @ 1T - 3.6v (link)

PC3700
OCZ Platinum EB 2x512mb - Micron 5b C - 267.9mhz - 3-3-2-6 @ 2T - 3.2v (link)


1024mb sticks
PC3200
Crucial Value 1x1gb - Micron 5b D - 263mhz - 3-3-3-8 @ 1T - 2.9v (link)

PC4000
TeamGroup Xtreem Cronus 2x1gb - Micron 5b F - 295mhz - 3-3-3-1 @ 1T - 2.73v (link)


obviously, it still needs a bit of work and refinement, but it's a start :D
 
I see where you're going with this, and it could kick ass. But yeah, if you need ANY help at all, I'm always here (literally, I'm in two full leg casts)
 
ouch, start healing bro!

and thanks for the offer :D
if i need help, it'll probably just to get word out so that people take the time to run 32m and post :D but we'll see how that turns out ;)
 
Yeah, the biggest concern there would be getting people to take the 30-40 minutes to just let their computer run SuperPI 32M. No one wants to be bothered with that. But I see what you mean, wanting some proof of stability or even just proof of truth.
 
People run 3D mark to compare their e-penor, why not Super Pi :p

I;d do it, but I can't afford (read: Can't leave house) nice ram so I have Sone Samsung TCCC Crap, I couldn't even give you a model number :/
 
Bona Fide said:
Yeah, the biggest concern there would be getting people to take the 30-40 minutes to just let their computer run SuperPI 32M. No one wants to be bothered with that.
Click to expand...
which tells me the world is full of lazy asses.. you can run 32m while doing other things.. :rolleyes: :p :D
 
oh for the love of...PI ! :p

im sure people would not have a problem letting pi run just before they go to sleep...SOME if you guys gotta sleep at SOME point in time...so just let it run then :p
 
Guys I've gotten into another fix. It seems that I won't be able to run 1:1 because I just noticed how much more expensive the opteron 170 was from when I last checked. Stupid AMD, why'd they have to raise the prices! Anyway I guess I'm stuck with an Opteron 165 and thus I will most likely have to use a divider to reach the overclocks I'm planning. How do the G.Skills work under dividers. I'm planning 166 divider so I can get it 300fsb:250ram.
 
even when running 1:1, you're using a divider, since the memory controller runs at cpu speed, the divider will be the same as the cpu multi...

so if i was doing 300x10, 1:1, the ram is actually cpu/10, for 300mhz
if i change that to 375x8 at 5:6, the ram is still cpu/10.. and gives 300mhz ;)
 
You must log in or register to reply here.
Back
Top