Corsair TWINX1024-3200C2PT vs. TWINX1024-4000PRO...a Tie?!?!?

[photogearjunkie]

Hi!

This is my first post here, I'm really hoping you guru's can help me out...

I just upgraded from Corsair TWINX1024-3200C2PT to Corsair TWINX1024-4000PRO to increase my overclocking potential as well as be able to use a 1:1 divider and enable PAT/MAM. Prior to replacing the memory modules, I ran a set of benchmarks so that I could see how much of an improvement I would get. The various benchmarks included UT2003, SiSoft Sandra, Super PI, Memtest86, etc. I ran each one 3 times (rebooting after each BM), and noted down the averages. I then installed the new modules, and re-ran the battery of benchmarks. All of the benchmarks ran flawlessly, without any errors.

Afterwards, I compared the results and was stunned to find out that both sets of memory modules performed within 1% of each other?!?! The UT2003 numbers were EXACTLY the same??? What gives- I was expecting a 5% to 10% increase in speed/bandwidth??? Can you please let me know if the new modules are bad, or if I set something up incorrectly? Or, do the slower memory timings negate the increased memory bus speed? Lastly, if you could tell me why PAT/MAM is not enabled, even with the 1:1 ratio, I would really appreciate it. My configurations are listed below:


Setup #1

P2.4C running at 3.0GHz using 250MHz FSB
Asus P4P800 Deluxe
Corsair TWINX1024-3200C2PT running at 320MHz with a 5:4 Divider
Timings set to 2-3-3-6-8
PCI:AGP locked at 33:66
"PERFORMANCE" set to "STANDARD"
"MEMORY ACCELERATION MODE" set to "OFF"


Setup #2

P2.4C running at 3.0GHz using 250MHz FSB
Asus P4P800 Deluxe
Corsair TWINX1024-4000PRO running at 500MHz with a 1:1 Divider
Timings set to 3-4-4-8-8
PCI:AGP locked at 33:66
"PERFORMANCE" set to "STANDARD"
"MEMORY ACCELERATION MODE" set to "ON" (Although it was shown as 'Disabled' in CPUID)

Thanks in advance for your help, I REALLY APPRECIATE IT!

 

[Duvie]

Well you were seriously misinformed if you thought you were going to recieve 5-10% in performance...Tha fact is there are many test that showed this already...Tha main factor??? It is the fact you are cas 3 with the 500mhz ddr...That is what made this a tighter race.

5:4 @250fsb for 400dr cas 2 vs 1:1 @50fsb for 500ddr cas 3??? Yeah seems about right......

Now the question is as you push it for say like 275 could you have still used 5:4 with the pc3200??? YOu likely will have to with the PC4000 stuff but the cpu increase will outweigh your drop in speed and then some....Push forward now...Who knows at 275fsb5:4 ratio you may be able to run that memory at cas 2.5 at least which willbe better IMO...

 

[myocardia]

With that processor, that board, and that memory, why aren't you running at 300fsb? Do some benchmarks at 300fsb, and see if you don't get quite a bit more than a 5-10% improvement!

 

[photogearjunkie]

Thanks for the replies, guys! But now I have a few more questions...

1. Do you have any ideas why the PAT/MAM isn't working when it's set to 'ENABLED' in the BIOS? I have a revision 1.1 board and thought that, as long as I used 1:1 dividers, I could enable the MAM.

2. Also, the RAM wouldn't o/c past 510MHz at 1:1, even with 2.85V. (I immediately got a bunch of errors in Memtest86.) What is the next step for trying to squeeze some extra performance out of this puppy? Does the next step involve using the dividers, as Duvie mentions?

3. What is the max safe voltage that I can pump into the CPU?

Thanks again!

 

[Shimmishim]

Originally posted by: photogearjunkie
Thanks for the replies, guys! But now I have a few more questions...

1. Do you have any ideas why the PAT/MAM isn't working when it's set to 'ENABLED' in the BIOS? I have a revision 1.1 board and thought that, as long as I used 1:1 dividers, I could enable the MAM.

2. Also, the RAM wouldn't o/c past 510MHz at 1:1, even with 2.85V. (I immediately got a bunch of errors in Memtest86.) What is the next step for trying to squeeze some extra performance out of this puppy? Does the next step involve using the dividers, as Duvie mentions?

3. What is the max safe voltage that I can pump into the CPU?

Thanks again!

1. YOu have a p4p800 as I do. You lose PAT after 200 mhz fsb. There is a hacked bios by BigToe over at ABXzone that enables you to access PAT over 200 mhz.

2. The next step is to vmem mod your board. It's very easy to do and I was able to get my hyperX to 250 mhz 1:1 @ 2-2-2-6 timings

3. I wouldn't push more than 1.7 volts. The p4p800 a really bad board known for it's vcore drop under load. In bios, if you set it to 1.65, it'll idle at like 1.75 volts but will drop to 1.62 or 1.63 on load...

In order to fix this, you need to do the vdroop mod to your board so that under load, the vcore doesn't drop. I have done this mod to my board and at 1.65, it idles around 1.75 but stays very stable at 1.68 under load.

Hope that helps.

 

[photogearjunkie]

Quote:

1. YOu have a p4p800 as I do. You lose PAT after 200 mhz fsb. There is a hacked bios by BigToe over at ABXzone that enables you to access PAT over 200 mhz.

A hacked BIOS- that sounds kinda scary???

2. The next step is to vmem mod your board. It's very easy to do and I was able to get my hyperX to 250 mhz 1:1 @ 2-2-2-6 timings

huh??? you lost me???

3. I wouldn't push more than 1.7 volts. The p4p800 a really bad board known for it's vcore drop under load. In bios, if you set it to 1.65, it'll idle at like 1.75 volts but will drop to 1.62 or 1.63 on load...

In order to fix this, you need to do the vdroop mod to your board so that under load, the vcore doesn't drop. I have done this mod to my board and at 1.65, it idles around 1.75 but stays very stable at 1.68 under load.

huh??? you lost me again??? sorry for my lack of knowledge on this.