800 --> 1066

R

rsgunter

[H]ard|Gawd
Joined
Feb 24, 2007
Messages
1,325
Well now that ram prices have gone down quite a bit, I'm thinking about making a move from 800 to 1066. Currently I have Ballistix 800 and at the time it was the ram I could afford.
I know very little about ram, and if you were to ask me what I didn't know about, I'd first say PSUs then Ram. :p
So anyway, if I move to Dominator 1066 ram am I going to see a performance gain? I will be overclocking. I'd still call myself a newb to understanding how ram affects the OC process.

Is it worth the upgrade or no?
 
With your current ram you can run 1:1 up to 400x9 and the ram will still be at stock speed.
You don't need any faster ram unless you think your cpu will go faster than 3600 mhz (400x9).
IMO your money is better spent elsewhere.
 
Indeed, my thoughts were leaning towards that.

I think after 3.6 I'm going to get diminishing returns anyway right?
 
For what it's worth I saw remarkable increase in mem bandwidth according to sisoft when I went from 800 to 1066, but my personal benchmarking, at the time with Oblivion, using the same run with FRAPS was fairly variable with no consistant gains...maybe minimal but not noticeable to my eye.


Everything I have read tells me this is a worthwhile but last step in your overclocking as you see the least gains here...I would max the cpu then fsb and only then creep up on mem...but I guess that is not the question...as far as running 1:1 etc...at least with the mobo I have I see no difference at any standard or custom ratio...but this may be mobo idiosyncratic as I believe the 680i is an odd bird when it comes to how it handles this..

Now the real question is...does craptastic mem at 1066 beat the new Kingston monster 800 DDR2 at 3-3-3 timing???

http://www.kingston.com/press/2007/memory/05a.asp
 
Wow, this is some monster stuff. :) Tempting but I'd rather get a 8800GTS at this point lol.
 
Lou Cypher said:
With your current ram you can run 1:1 up to 400x9 and the ram will still be at stock speed.
You don't need any faster ram unless you think your cpu will go faster than 3600 mhz (400x9).
IMO your money is better spent elsewhere.
Click to expand...

Every part of this post ^ is fact.
 
whats better 3-3-3-8 RAM running at 800MHz, or 5-5-5-15 RAM running at 1066?
 
Is that a rhetorical question? If not, the answer would then depend on whether it was still at 1:1. Either way, I'd say 3-3-3-8 would definitely beat out the slackened timings at 1066.
 
tdiddy said:
Is that a rhetorical question? If not, the answer would then depend on whether it was still at 1:1. Either way, I'd say 3-3-3-8 would definitely beat out the slackened timings at 1066.
Click to expand...

hmm interesting...i always thought speed was more important for gaming
 
I posted most of this in a thread I started about the results I got, but those here might be interested in the results I left off that thread.

I have Micron D9 chip based PC2-8500 from OCZ and Kingston that I am using in my machine. I was able to get Sandra Memory tests on:
1:1 settings, 733Mhz (w/ my 1466.7Mhz FSB) @ 4-4-4-12-1T & 16TRC (would boot at 3-3-3-9-1T, but not run Windows for >60sec w/o crashing until back at 4-4-4-12)
916.7Mhz @ 4-4-4-12-1T & 17TRC
1066.7Mhz @ 5-5-5-15-2T & 22TRC

The 916.7Mhz/1T and 1066.7/2T got me essentially identical SSE bandwidth for Int and Float, respectively. The 733Mhz runs were closer than I would have expected, but not good enough to convince me to use it. The voltage difference between stable at 916.7 vs 1066.7 left me running with the higher bus speed/higher latencies. Here is the whole enchilada.

I am happy with the DDR2-1066 I am running, but would have spent less on decent DDR2-800 if I had been paying my own (cash) money for my parts at the time I built this system. As it was, however... I am happy I got what I got. :D
 
tdiddy said:
Is that a rhetorical question? If not, the answer would then depend on whether it was still at 1:1. Either way, I'd say 3-3-3-8 would definitely beat out the slackened timings at 1066.
Click to expand...

So the answer doesn't depend on whether it was still at 1:1 then, huh? Certainly the frequency decrease in the former would negate any performance increase obtained from the lower timings, wouldn't it?
 
You must log in or register to reply here.
Back
Top