TRCD and TRP are far more important than 1T/2T.Quote:
Originally Posted by texuspete00
You guys may be interested in this thread: http://www.xtremesystems.org/forums/...ad.php?t=65804
Printable View
TRCD and TRP are far more important than 1T/2T.Quote:
Originally Posted by texuspete00
You guys may be interested in this thread: http://www.xtremesystems.org/forums/...ad.php?t=65804
i think thats wrong for gaming tho i have seen benchmarks that with gaming 2-2-2-5 2t is worse at the same speed then 3-3-3-8 1t for gaming!!Quote:
Originally Posted by STEvil
________
Marriage counseling
I just decided to put all 4 of my OCZ pc3200 rev2 TCC5's in my ultra-D tonight and they seem to be doing pretty good with 2T.. 1T is a real no go..
http://img158.echo.cx/img158/8760/test9fg.jpg
I'll run some 3dmarks too.. ;)
Edit..
Here is some more with the 2gig's.. Maybe not totally xtreme but I think the tcc5 is doing decent..
http://img168.echo.cx/img168/5886/untitled1uq.jpg
Another thing important to note is that bandwidth efficiency goes UP with a higher cpu multi. You may want to take that into consideration also. It may not be the highest fsb clock that yields u the best performance, but the tighest timings+ a high multi + a moderate HTT, ie 250-260.
I don't think it is the multi that causes bandwidth to go up, it is just raw CPU speed (which just so happens to increase when the multi goes up :p: )Quote:
Originally Posted by fareastgq
Hmm Nice, I wanted to see something like that. What volts are you using?Quote:
Originally Posted by THunDA
What can these tcc5's of yours do at 512mb x2 (are both kits equal in their speed if used by themselves, and how much is their speed difference... I.E. can you get 300mhz at 2.5-3-3-X with just two sticks?)
no, I tested it. bandwidth effiency is constant for each multi. it does not go up. think of it as car lanes and cars, u have 10 lanes, but only 5 cars, not effiecient use of your lanes, but if u change the # of cars to 10, then the efficiency goes up. it goes up about 4-5 % per 1 whole # multi jump. ( this is according to sandra)Quote:
Originally Posted by J-Mag
Ok I see what you are saying. I was thinking in total bandwidth terms and not efficiency. However, I think Sandra is probably not the best thing to rely upon. At the same time effeciency is only just a peice of the total bandwidth and is easily outshined by HIGH CPU or MEM clocks.Quote:
Originally Posted by fareastgq
Anyway there is no way to test this for real world results because it is impossible to have the same CPU, Mem, and HTT speeds with different Multi's. Either way the multi has VERY LITTLE impact upon a real world performance. It is the CPU speed which is most important BY FAR over every other tweaking factor (I am excluding GPUs from this).
That's true, sandra is not the most reliable, but it does make sense, hopefully this big assumption off of a bench prog is somewhat correct, heh. Almost everything we do we are assuming the programs are reading correctly, we kind of have to or we can't really have a basis for anything.Quote:
Originally Posted by J-Mag
THunDA,
Can you post your entire screen shot of A64 tweaker. I would really appreciate it!
I'm a AMD noob duno how to tweak the mem settings yet. I got 4 stcks of TCC5 and get them to post above 200fsb :(
eyepiz - run 2T.
This white paper from Corsair
(PDF format) http://www.corsairmemory.com/corsair...erformance.pdf
measures the the relative effect of increasing the various timings by one clock cycle. Yes, the command rate is the most important, but not by a large margin. 2-2-2-5-2t easily beats 3-3-3-8-1T.
Edit: I originally posted the wrong link. Correct link above, incorrect but also good link below.
http://forums.anandtech.com/messagev...&enterthread=y
good comparison, but in general everyone alrdy knows that low latency (2 2 2) + 1t + high HTT is what the goal is. still, good info to see. Not to be one sided or anything, but that linked post seems a bit crybabyish. Yes u can get value ram, but value isn't mushkin redline, nor will it perform the same, Also, u CAN always get the best vid card AND the best ram for the best system. Budget? what budget? :p: Most ppl here want the best, period. thanks for the info though. :)Quote:
Originally Posted by Otter
I received my 2GB of ballistix this morning, will see how it does when my X2 arrives. Also waiting on 2gb of G.Skill LE.
Sorry about that, i accidentally posted the wrong URL. I meant to linkQuote:
Originally Posted by fareastgq
http://www.corsairmemory.com/corsair...erformance.pdf
which actually does relate to my post. :rolleyes:
On the other link, I think you're missing the point of Zebo's thread. He acknowledges that better RAM does result in slightly better performance, and his tests certainly show that. The point is that high-end RAM doesn't give you that much extra bang for your buck. I'm new to this forum and maybe y'all are filthy rich and don't care about such things, but for those of us who do, it's good to know that we don't need killer RAM. For me, it made the difference between a San Diego and a Toledo CPU for my next system. :)
well, I got the filthy part of that covered anyways..
Just got the full game of Battlefield 2 ans it's hungry for RAM. So I just ordered 2GB of Crucial Ballistix PC4000, hope it does the job :)