WTF - PC3200 2-2-2-11 faster than PC3200 2-2-2-7??!!

j@cko

Diamond Member
Feb 13, 2000
3,814
0
0
Well, I bought a pair of OCZ PC3200 Platinum Edition LE (BH-6/ 2 x 512MB) and it was advertised as 2-2-2-7. However, I couldn't get it up to run at 2-2-2-7 whatsoever (after all the BIOS tweakings and all). I consulted a guy from OCZ and he told me to try 2-2-2-11 so I did and I was able to run it @ 2-2-2-11 stably. He later emailed me that 2-2-2-11 is actually faster than 2-2-2-7 on NF2. (which is what i have, a NF2 system). I tested the above setting with only 1 module.
Why don't the OCZ modules I have run @ advertised timing? Does this cripple this overclockability of the RAM?
I still want a RMA, cu'z it wasn't able to run at the advertised timing, what do you guys think?

Cheers,

Jack

Here's his reply,
TRAS at 11 on NF2 boards is better than TRAS at 7..i wouldn't worry about this setting as TRCD and TRP are the more important timings

mobo: Epox 8RDA+ rev 2.1
BIOS Settings:
Advanced BIOS
Advanced Chipset
 

bandana163

Diamond Member
Jul 2, 2003
4,170
0
0
It might be a MoBo problem. What kind of motherboard do you have? NF7? A7N8X?
Maybe you should try it in another rig to see if the memory works at the required latencies.
 

JustAnAverageGuy

Diamond Member
Aug 1, 2003
9,057
0
76
an 11 tRAS IS better on nForce2. Even better than 5, or 6.

I thought that was a commonly known fact. I agree with the tech.

There is a long thread about it over on the AMDMB.com forums.
 

JustAnAverageGuy

Diamond Member
Aug 1, 2003
9,057
0
76
That's almost like saying, why can't I run at 2000MHz when I can run at 2050MHz? :p

the tRas Latency and raising it is merely psychological.

As for the 2-2-2-7, did you try raising the vDimm?
 

Megatomic

Lifer
Nov 9, 2000
20,127
6
81
I have done testing on an Epox 8RDA+ v1.1 and an Abit NF7-7 R2 with the same CPU and memory (2500+ and 2x 512MB PC3500) and my testing has proven to me that tRAS of 11 is indeed faster than 5, 6, and 7. I have posted about it in several threads and I even linked to a message from a Mushkin tech regarding the matter. I can dig up that link if necessary.

May I ask why it is important to you to run with a tRAS of 7? I am not clear why you'd want to when you know it is a slower setting.
 

j@cko

Diamond Member
Feb 13, 2000
3,814
0
0
well, it just bothers me that it couldn't run at the advertised speed. It just seems like there is something wrong the modules when it couldn't run @ the advertised speed.
It's like saying, "Ok, so your newly bought car only run @ 100km/h, but cannot do 60Km/h"... Wouldn't you feel weird and something wrong with it?
 

Megatomic

Lifer
Nov 9, 2000
20,127
6
81
I certainly can't explain why it WON'T run at tRAS=11 and my opinion of OCZ is biased so I'll leave my feelings about them out of this.


Is there any way you can test some other RAM to see if you can run at tRAS=11? If your system can do it with another brand you can RMA (or return) the OCZ and get something better.
 

j@cko

Diamond Member
Feb 13, 2000
3,814
0
0
u know, i've heard some pretty nasty stories about OCZ... I thought that I would take a chance and OCZ has improved quite a bit recently. My friend told me to go with Mushkin, but I didn't since I couldn't find Mushkin BH6 here in Canada.
No, I don't have other boards that I could test this on, but I did have a Corsair RAM that I could run 2-3-2-6 @ PC3200 and this one couldn't(that was withasyncrhonized cpu/mem freq. right now i'm running with synchronized cpu/mem freq), which is actually quite disturbing. These new modules are suppose to be better, not worse...
 

InlineFive

Diamond Member
Sep 20, 2003
9,599
2
0
My OCZ memory works perfectly at 2-3-3-7 or 2-3-3-11. I run it at 2-3-3-11 since I get an extra 80MB/s of bandwidth.

-Por
 

JBT

Lifer
Nov 28, 2001
12,094
1
81
I don't see what the big deal is if it can run at 11 which is faster what is the problem? I run my memory at 2.5,2,2,11 and it is a bit faster than 2.5,2,2,6 though it does run it fine... why worry when u will be getting more performance?
 

PliotronX

Diamond Member
Oct 17, 1999
8,883
107
106
I think it varies system to system but overall it does show a boost for a large portion of the NF2 systems. The simplified version of tRAS is that it's like a set time allowed for the system to read the "pages" in the memory like a book. If there isn't enough time to allow for a full page reading, it would lessen performance (mainly bandwidth, but bandwidth doesn't always=performance) and if it was way too low it could cause corruption (Mushkin explains it as anything below CAS + tRCD + 2 is risky, i.e. below 6 for 2-2-2). Although this effect only seems to show up on NF2 systems whereas with Intel chipsets, the lower is always better (but to Mushkin's guideline).

On my NF7-S with 2 512Meg sticks of BH-5 running in dual channel, tRAS at 11 shows an odd increase in certain benchmarks, or areas of benchmarks and shows actual decreases in others. For example, the really short blocks in Prime95's benchmark slow down the higher tRAS is but the longer blocks are faster with tRAS at 11 and no higher and UT2003 shows a small boost also Sandra shows more bandwidth with buffered benchmarking whereas AIDA32's memory benchmark shows a slowdown with the higher tRAS. Overall I haven't noticed any difference between the two, so if it makes your system stable I vote to stick with tRAS 11 :beer::)
 

First

Lifer
Jun 3, 2002
10,518
271
136
It does run at 2-2-2-7, just not on your board, which isn't really OCZ's fault.
 

Pelochas

Member
Dec 17, 2001
60
0
0
i dont know if i will be any help but theres good info on all these high performance ram at Toms hardware and even they had problems. overall they find that there is not a big difference in perfomance between high performance and the regular ram (both of a respectable manufacturer), it may not be true but it helps us low budget guys a speed per dollar figure and find out that the really expensive stuff is not worth it.
 

Brian48

Diamond Member
Oct 15, 1999
3,410
0
0
Originally posted by: Megatomic
I have done testing on an Epox 8RDA+ v1.1 and an Abit NF7-7 R2 with the same CPU and memory (2500+ and 2x 512MB PC3500) and my testing has proven to me that tRAS of 11 is indeed faster than 5, 6, and 7. I have posted about it in several threads and I even linked to a message from a Mushkin tech regarding the matter. I can dig up that link if necessary.

May I ask why it is important to you to run with a tRAS of 7? I am not clear why you'd want to when you know it is a slower setting.

I have a 8rda (non-Plus) v1.1 and I've found that a tRAS setting of 11 is optimal for me as well.
 

sandorski

No Lifer
Oct 10, 1999
70,486
6,034
126
I'm in way over my head here, but I'll offer a theory as why a "slower" setting is faster in this case: It might be a timing issue with the chipset/mobo. Perhaps the chipset/mobo operates in such a way that the "faster" setting doesn't synch with the chipset/mobo properly, like in the case of running the ram Asynch at a higher speed(fsb wise) than the cpu?
 

JustAnAverageGuy

Diamond Member
Aug 1, 2003
9,057
0
76
Originally posted by: sandorski
I'm in way over my head here, but I'll offer a theory as why a "slower" setting is faster in this case: It might be a timing issue with the chipset/mobo. Perhaps the chipset/mobo operates in such a way that the "faster" setting doesn't synch with the chipset/mobo properly, like in the case of running the ram Asynch at a higher speed(fsb wise) than the cpu?

not quite
 

txxxx

Golden Member
Feb 13, 2003
1,700
0
0
I think this 11 timing issue comes from an internal timing issue within the nForce chipset itself. Clearly a 7ns TRAS has an access penality at this resolution.