|jmke ||3rd January 2004 17:27 |
AMD nForce2 memory timings explored
Piotke clears the confusion that might be there with memory timings on your nForce2 setup
A couple of days ago, I read on a Dutch forum a question concerning AMD timings. This week I saw almost the same question on our forums. It was a question about the fastest memory timings for an AMD nForce2 setup. Nobody could give a clear answer, so started to test the most common used timings... Here are the results you´ve always wanted to know!
|teki ||3rd January 2004 18:55 |
Now u got the hardware, might it be possible to test those 2-2-5 / 2-2-11 on your p4 system.
I wonder what the difference is compared to my 2.5-3-3-7 settings
this is the best my apacer pc3200 can handle.
It's almost impossible to get 50b chips in standard belgian shops.
Went to tones/forcom/digita/pcpartner and comtechnology but they only sell 50D
|The Senile Doctor ||3rd January 2004 19:03 |
piotke will be getting a p4 cpu soon, so we'll get a follow up on that soon enough...
|jmke ||3rd January 2004 19:03 |
the 2-2-11 will only perform better on a nForce2 setup
for the P4 lower timings = better
|piotke ||3rd January 2004 19:05 |
I allready have a p4 setup right now, but only one stick of BH5, so no dual channel..
I intend to buy 2 sticks of mushkin memory BH5, not sure yet :)
|jmke ||3rd January 2004 19:09 |
you can do the test in SC, but 2-2-5 will outperform 2-2-8 (max value I think, no 11 available)
|The Senile Doctor ||3rd January 2004 20:04 |
sc is not an option for comparison on a dual channel setup... you can never extrapolate the difference in single channel towards dual channel
|jmke ||3rd January 2004 20:14 |
only nForce2 shows increased performance with 2-2-11 as stated before. 2-2-8 on Intel will be slower then 2-2-5.
|teki ||3rd January 2004 20:35 |
will the difference between 2-2-2-5 and 2.5-3-3-7 on an Intel rig be comparable with the difference on and amd rig?
Because the difference between them is =~ 430 3dmarks and it seems like everyone is selling their oc memory at this moment, so for a very small price, my rig could be running a lot faster and more stable than with this crappy apacer ram.
|The Senile Doctor ||3rd January 2004 20:42 |
you could get an increase of some percents, however... in realworld applications you won't notice anything...
|quarantaine ||3rd January 2004 23:35 |
bought mushkin pc3500 lvl 2 a while ago (2*256Mb) and it never worked good. (max stable +- 180)
i have lately seen people that have also troubles getting there fsb above 200mhz.
and my geil did +- 210 stable on the same board (till it was broke)
that sucks tbh , you buy seriously expensive ram and the performance goes down.
the question for me is know, downgrade on ram or upgrade on board(that dfi infinity looks cool) . damnit, oc'ing sometimes does suck ;) (and is expensive)
|Xploited Titan ||4th January 2004 10:22 |
I got 2*512Mo DC BH-5 memory (I'm still waiting for my CPU however)...
But like the P4 setup, I shouldn't assume that the 2-2-2-11 timings are the fastest in Dual Channel, or did I understand wrong?
|cR00zIFI><3r ||4th January 2004 14:11 |
I've found a little error in the article:
The CPU was running at 2.2GHz (10 * 200 fsb), with the memory on a 1:1 divider. Meaning it was running also on 200 Mhz = synchronous.
10*200 != 2200
I have 2*256MB TwinMos pc3200 (BH-5) and an epox 8RDA+
Would 2-2-2-11 still be better than 2-2-2-5 in Dual Channel, or would it be different?
|piotke ||4th January 2004 14:57 |
Originally posted by cR00zIFI><3r
I've found a little error in the article:
10*200 != 2200
|Xploited Titan ||4th January 2004 15:16 |
Originally posted by koensa
ras to cas is most imprortant mem speed factor,then CAS.
i've run 2 2 2 5 dcddr and 2 2 2 11dcddr,and sisoft gave me a 110mb increase
11 is ONLY!faster on nforce˛ platform.
What do you mean? is 2-2-2-11 DC still faster than 2-2-2-5 DC?
|jmke ||4th January 2004 15:21 |
|jmke ||4th January 2004 15:54 |
"AMD nForce2 memory timings explored"
|teki ||4th January 2004 16:42 |
One more little question regarding a P4 setup:
I have 2*256 pc3200 ram running with a p4 2.6c
it currently runs smooth @231Fsb BUT! it is on a 5/4 divider, so the memory runs only @ 180fsb, however: when i place it higher for exampe on 240/190 something craps out immediately, is it the cheap ram? or a crappy p4c? mobo is Richbastards p4p800vm which ran over 300fsb so that shouldn't be a problem
PSU is only 365watt however...
|jmke ||4th January 2004 17:12 |
ram should be able to run @ 200Mhz with "normal" timings
|Unregistered ||4th January 2004 21:56 |
i hope that these are not silly questions but i am curious how many times these tests were run? are these scores based on one run? or multiple runs with the final results just an average score? also, is there any conclusion given as to why 2-2-2-11 would run faster? and, would it be safe to assume that 2-2-2-11 would be faster on ANY nforce2 mtb? or it only pertains specifically to just for the one that was used in these tests?
|Unregistered ||4th January 2004 22:08 |
I have a question What happens when you try and overlclock two memmory sticks with diffrent FSB like some 2100 and 2700? How about overclocking with two sticks in Dual Channel mode? How identicle do the sticks have to be? Is there anyway to overclock two sticks seperately?
|Eyemaster ||4th January 2004 23:12 |
From what I have read it seems to be any Nforce2 motherboard. 2-2-2-11 is fastest on my ASUS A7N8X Deluxe v1.04.
|Unregistered ||5th January 2004 03:11 |
2-2-2-5 is a b.s. setting according to Mushkin. Should be 2-2-2-6, wonder if that's why the performance is degraded vs. 11? You think?
"Now imagine someone closes the book you are reading from in the middle of a sentence. Right in your face! And does it over and again. This is what happens if tRAS is set too short. So here is the really simple calculation: The second burst of four has at least to be initiated and prefetched into the output buffers (like you get a glimpse at the headline in a book) before you can close the page without losing all information. That means that the minimum tRAS would be tRCD+CAS latency + 2 cycles (to output the first burst of four and make way for the second burst in the output buffers)."
|Unregistered ||5th January 2004 03:12 |
Any tRAS setting lower tRCD + CAS + 2 cycles will allow the memory controller to close the page “in your face!” over and again and that will cause a performance hit because of a truncated transfer that needs to be repeated. Along with those hassles comes the self-explanatory risk for data corruption. That one is not a real problem as long as the system is kept running but in case it is shut down and the memory content is written back to the hard disk drive, the consequences can be catastrophic. For the drive, that is.
|Unregistered ||5th January 2004 06:15 |
I wish you would have explained the meanings of the numbers in more detail, especially in layman's terms too :)
But otherwise, thanks for taking the time to experiment. I'll go try some timings myself on some hyperx's i just bought :|
|Unregistered ||5th January 2004 14:30 |
Mushkin 256MB PC3200 Level II is definitely more expensive than generic memory but not as expensive as 109,18 Euro per stick. In Europe you can buy it from e.g. www.jes-computer.de
for 74.99 Euro.
|jmke ||5th January 2004 14:33 |
due to higher taxes in Belgium the lowest price was €109 at the time of writing.
|Unregistered ||5th January 2004 14:45 |
I live in Denmark where VAT is 25%. That's why I buy from Germany. As long as you stay within EU you only pay the local taxes/VAT. :)
|Xploited Titan ||5th January 2004 16:11 |
I paid for my 1Go DC OCZ PC3500 Platinum Limited Edition (the last and best revision with BH-5 memory modules) about 314€ (transport included, I had it coming from France)...
|Unregistered ||6th January 2004 20:08 |
OK. Don't laugh. I have a i865-based Dell with a P4/2.4Ghz processor running PC3200 at 3-3-3-8 timings. The BIOS does not allow for changing the timings and obviously takes it's values from the RAM. So I am looking for PC3200 that not only will run at 2-2-2-5 but also REPORT 2-2-2-5. What can I get?
|jmke ||6th January 2004 20:32 |
Mushkin will run at CL2 2-2-5 through SPD I thought.
|Sarcastro ||8th January 2004 19:04 |
Same as kingston hyperx, the non A revision that is.
|Unregistered ||8th January 2004 23:02 |
Has anyone got a picture of the Twinmos/Twinmos 50B chip?
My supplier wants to know what to look for.
|Unregistered ||8th January 2004 23:42 |
Thanks for the fast response, but codes in that picture are (almost) unreadable.
Has anyone got a better picture?
|jmke ||8th January 2004 23:54 |
at the end of the 3rd line on the 1st chip you can make out "50B" :)
that's what your supplier needs to look for!
Powered by vBulletin® - Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO