• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Redline, way to go

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
for instance, im positive my ram doesnt like my cpu/mobo combo at all.it required a bios that was released just a month back just to get it to run properly at stock.since i dont want to switch out mobos i can either downgrade my cpu(to an older core revision thats not as fussy about what ram i use, like winchester for instance)or change out my ram(to something that isnt tccd or hynix based.) leaving only bh5, utt or micron based ram as upgrade options.
 
Last edited:
largon said:
1. CAS1.5 = CAS2. A64 doesn't support CAS1.5.
2. If the ram can do xxxMHz @ 2T it can do the same @ 1T. Memcontroller is the only thing that matters when command rate is 1T. It's simply not an ram issue.

First of A64 does support 1.5CAS and it's a mather of weather your mobo can support it.



RAM can do higher freq at 2T then at 1T and there is no way on Earth that you can do as high as freq with 1T as you would with 2T.

Memory controller plays a big part when it comes to OC'ing and when command rate is defaulted at 2T it's because memory controller is under higher load. That's why when using more then 2 sticks at a time you can't do 1T, since there is more power draw.

So what I'm trying to explain to you is that when it comes to running under 1T RAM does matter as much as memory controller. Certain chips simply can't do high freq without using 2T and prime example are Micron chips.
 
Last edited:
RAMMAN said:
what if a memory controller doesnt like a particular type of ram? or if the 2 sticks are not in the right slots?

Yes Venice procs when they were released didn't play nice with TCCD's and depending on which mobo you are running on BIOS update could fix the problem.

Also try playing with the timings and slightly higher Vdimm, that could help.
 
largon said:
Zebbo,
Eikö tRCD tosiaan mene kahteen? Entäs tWTR ja tRTW, meneekö 1?

Ei se mene, tullut yritettyä vähän kaikkien latenssien kanssa leikittyä mutta kahteen se ei mene. tWTR on varmaankin mahdollista saada 1 :)

and back to English :)
 
Zebbo,
sorry for this massive offtopic.
RedDragonXXX said:
First of A64 does support 1.5CAS --
AMD doesn't think so.

Differences between 1.5 and 2 were measured by "EMC2" @ Xtremesystems:
This pic shows that CAS1.5 on AMD A64 = CAS2.
http://img211.imageshack.us/my.php?image=cas15set100mhz0qt.png

The whole thread is worth reading.
RAM can do higher freq at 2T then at 1T and there is no way on Earth that you can do as high as freq with 1T as you would with 2T. --
I must disagree. :)

Command Rate is a memory controller latency:
"DRAM Command Rate is synonymous for address and command decode latency of the chipset. Briefly, behind this parameter is the selection of the proper physical bank within the entire memory space in any given system."
http://www.lostcircuits.com/advice/bios2/7b.shtml

Summary:
Dropping Commad Rate from 2T to 1T halves the amount of cycles during memory controller is supposed to decode address' for a bit that needs to be read. It doesn't affect the functions of the module itself. -> If the MC is *good* enough, it can run any ram at 1T.

1T simply isn't something that RAM is able to do. The memory chip itself doesn't "know" or care wheather 1T or 2T is used.
 
Last edited:
^^^^Read this!

Also if you think that RAM can do as high freq at 1T as it would on 2T, then you should try out Balistix. You are right about command rate no question about that (read my last post) but you need to understand that not all chips respond the same. That was my whole point.

I'm telling you this from my personal experience cause I been there done that. Same thing with dividers.
 
Last edited:
Yes?
I'm aware of the performance effects of Command Rate.

btw:
I’m not sure as of yet if ECC/registered RAM can do 1T with more then 2 sticks
Multiple registered DDR-SDRAM dimms can easily be driven @ 1T. That's because memory controller has to address only the registry chip on the DIMM. That chip does the rest.

ECC on the other hand is just as lethal to performance as 2T:
The error checking takes 1 extra cycle. There's nothing that prevents using 1T though.
 
But how many people do you see using ECC RAM, what's the point since most of todays boards don't support more then 4GB's :p

However I have yet to try ECC/registered sticks, but it would be nice to see how they perform.

And yea Zebbo, sorry for taking your thread off topic :D
 
Zebbo,
have you tried higher than 3.6 volts?

RedDragonXXX said:
But how many people do you see using ECC RAM, what's the point since most of todays boards don't support more then 4GB's
I thought wanted to know how reg./ecc dimms behave with command rates? That's what you wondered in the thread you linked.

Boards without memory controllers don't actually care how much ram is installed. Once again it's a proc's mc limitation and mobo mfgs. can't promise more than AMD (plus 1024MBit DRAM is, let's say, rare).
 
Last edited:
largon said:
I thought wanted to know how reg./ecc dimms behave with command rates? That's what you wondered in the thread you linked.

I know, but that's a different thread.

largon said:
Boards without memory controllers don't actually care how much ram is installed. Once again it's a proc's mc limitation and mobo mfgs. can't promise more than AMD (plus 1024MBit DRAM is, let's say, rare).

AMD's have built in MC, that's why their boards don't have a northbridge.

Anyways, we are getting far off topic on this thread with every post. If you want to discuss this some more then AIM me (under my sig) :thup:
 
That's ok, don't mind about as long as you're talking about memory :)

I only tried 3,6V + Enable 0,03V as for max. I know there is no sense to go with higher voltage with ambient temp I had. Higher volts simply would not give any sort of advantage.
 
Zebbo said:
I only tried 3,6V + Enable 0,03V as for max. I know there is no sense to go with higher voltage with ambient temp I had. Higher volts simply would not give any sort of advantage.

Just for kicks, have you tried 2-4-2-5?

I know that's way off for UTT chips but it would be interesting to see how it responds, if you even make it to post screen.
 
I did some 3DMark03 and SuperPi8M benching

Trcd 4
SuperPi 8M
3DMark03

Trcd 3
SuperPi 8M
3DMark03

So, as you can see, Trcd 3 is ~2 seconds faster in SuperPi 8M and 3DMark03 gave approx. 40 points more compared to the Trcd 4. This was just benching between the differences of Trcd 4 and Trcd 3.

But you can see here also that ~100MHz more overclock for CPU is much more effective than running memory with 1:1 divider and with lower CPU frequency.

I will check the difference between overclockability later :)
 
Last night I tried the advantage of the Trcd 4, but it really end up doing like Trcd 3 so I didn't see any need of posting results for it, I'm sorry RedDragonXXX :)

By the way, the first little overclock I did gave a nice "burn-in" for my modules, before I could run these modules at 270MHz max. and single MHz more gave me error right away. Now, they are all fine and fully stable at 274MHz with 3,5V :)
 
Going for the high end Redline memory next? That is a great OC you have there. I was able to hit 280 on 1.5V core and 3.2V VDIMM. I am using some old (6 months) Mushkin Lvl 2 ram I bought. I migrated it from my DFI NF3 board to my new NF4 Ultra-D using the same CPU as you. I also have the latest beta bios and the CPu stepping is a week 0524. The only hitch is I have to run a divider and it takes a long time to load Windows.. almost 30 secs. No problems within Windows though and I played a game online to see if it choked but no such thing happened. I am not a proponent of dividers as every time I have used one, I am notice the performance hit. Then again, I am probably not adjusting something else down the line to make up for it? I am too new to OC'ing ram to even begin to know that.. :)
 
Back