>>> Corsair XMS 7 stick shootout! <<<

[naddicott]

Originally posted by: THUGSROOK

i just want to clarify that all testing i did in this thread was using 2.6v and 3:4 for the "avg joe overclocker".

i realize that recently things have changed for us, but i didnt want to invalidate my previous results in any way.

All those tests in the first post are a great reference. Enforcing 2.6v on the 2 stick tests may be limiting them more than what 'joe overclocker' could safely accomplish, however.

At 2.7v I was able to get 400DDR (2-3-3-6) with 2x512 XMS3200C2 memtest86/prime95/3dmark2k1 stable on a P4PE board (150fsb, 3:4 ratio). Corsair covers anything up to 2.8v, so I'm not terribly worried about running at 2.7v. For closer comparison to this thread's tests, someone over at asusboards using a P4B533-E runs 400DDR (2.5-3-3-7) with 2 sticks using 2.7v - more FSB or tighter timings gives them memtest errors.

2x512 Corsair XMS thread at Asusboards

 

[Mikki]

I retested mine also...3:4 2.6v ddr434 was the highest I could get, same as the last time I tested. Seems slightly odd that on my current oc, I can only get ddr432, and that's with 2.7v...

 

[ErikaeanLogic]

THUGS - you were right about the XMS3500. My guess is that they are binning the best XMS stuff for their new XMS3200LL sets:disgust:.

 

[THUGSROOK]

ErikaeanLogic ~ yep holographic decals

tstrike ~ cas222 is king

naddicott ~ i figured 2.6v vdimm was good voltage for the "mild" overclocker. the hardcore guys could scale higher voltages from there.

 

[STiSev]

Pentium 4 2.53Ghz@3Ghz with Arctic Silver III, Alpha 8492 and Panaflo M1A
Abit IT7 Max2 ver2 w/ Onboard Realtek LAN
Corsair XMS3500 DDR RAM
ATi Retail Radeon 9700 @325/317
Antec True430
(Black) Chieftec case w/ four Panaflo L1A (one L1A cools the HDs)
Custom built (black) 4x 3-piece switch (12v-off-7v) for fan speed control.
2xWestern Digital 80GB SE (8MB) with (2x) Serial-ATA adapters [One main drive & one emergency drive)
Lite-On 52x24x52
Creative 52x
Internal ZIP 100 Drive
Pioneer 16x DVD
Turtle Beach Santa Cruz
AverMedia TV Tuner
Windows XP Professional
All rounded cables to provide for better cooling.
Logitech Z-560s
Toshiba PCX-2500 Cable Modem w/ 3000Mbps/800Mbps network connection.
Sharper Image Ionic filter working 24/7 (to prevent computer dust buildup)
HP 950c Printer.
IBM KB-9910 Keyboard.
Microsoft IntelliMouse Optical v3 w/ Everglide Giganta mouse pad.
21? Sony CPD-G500.
Soon: Seagate Cheetah 15K.3 15K RPM SCSI HD w/ LSI Logic U160, Ultra160 LVD SCSI-3

Overclock details

P4 2.53@2.98 Ghz

Internal Clock: 2973.40 Mhz.
System Clock: 156.49 Mhz
System Bus : 625.98
Multiplier: 19
Cas Timings: 2-5-3-3
RAM Voltage: 2.5
RAM Speed: 417.3 Mhz
Divider: 3:4
CPU core voltage: 1.675v
Idle Temps: Case: 24* CPU: 39-42*
Torture testing: Case: 26* CPU: 52*

Status: Passes 26 hours of Prime95
Passes 16 hours of Sandra Burn
Passes 18 hours of Toast.exe




It wont go any higher.
What do you guys recommend?

 

[Mangler]

sweet....new thread and more corsair testing! Go THUGS

 

[THUGSROOK]

 

[THUGSROOK]

since there is still a HUGE demand for this thread (2.6v testing) i will continue to update it as long as i still have this motherboard.

ive got a couple more sticks of XMS3000 (512mb) coming for a few "customers".
hopefully ill find the time to toss them in my system and add their results to both threads.

i know you guys wanna see more results from the XMS3000 sticks....

 

[LastRide]

Be nice to see if these newer sticks of XMS 3000 are the same as the one's you have or if Corsair changed them in any way.What I mean by changed is still getting those great overclocks from them.

 

[THUGSROOK]

i personally think its the 6ns chips that makes the xms3000 better.

my xms3500 needed 2.7v to run stock specs.
the xms3000 was able to run xms3500 specs on 2.6v.

or this could all just be the preference of my motherboard.



we'll see

 

[Soulkeeper]

I'll be picking up two sticks of 512mb xms soon
i was thinking about getting their new c2 fast timings stuff

dang this stuff is expensive heh
400 bucks for memory outta hurt for some time

 

[Mikki]

It'd be great if, between all of us, we could test all the different memory out there...

 

[Soulkeeper]

hahahaha yeah
anand might even link us in the news section on the website if we could make graphs and stuff
but i'm lazy and stuff.....

 

[Mikki]

hehe...I'll make the graphs if you provide the memory...

 

[Mikki]

bump for Soulkeeper

 

[Soulkeeper]

heheh thanks

 

[Soulkeeper]

Ok i just read all the stuff in this thread

i'm a low latency kinda guy

but I do want the mem to clock well
I guess my ideal memory would be one that could do: 420 @ 2-2-2-7 1t @2.6
but that's prob dreaming a tad bit

I also don't like any of the athlon chipsets enough to buy one (unless I get a refurbished kt400 board for cheap)
guess I should sell my athlon 2400+

I wouldn't want to try differen't mem and run the risk of having to return them
I am already having to return a GF4 cause it didn't work

 

[THUGSROOK]

...its a 7 stick shootout now

 

[RaynorWolfcastle]

Originally posted by: THUGSROOK
...its a 7 stick shootout now

Hmmm... I don't know if you just got some duds or if I struck gold, but the stick of PC3200C2 that I got before Christmas goes up to 400 MHz, default voltage (2.5V, 2.56V actual) @ 2-2-2-5 on my epox 8k3a (160 w/5:4 ratio = 200 MHz).

That's pretty much as far as far as it'll go with those timings though; bumping up to 203 MHz gives me 5-6 memory errors per pass errors in (memtest86) test 6 w/2.6V (2.66V actual).

I'll see what I get by dropping the timings to 2-2-2-7

 

[THUGSROOK]

different boards will give different results.

these results are only directly comparable to Intel i845D and E chipset motherboards.

 

[RaynorWolfcastle]

Originally posted by: THUGSROOK
different boards will give different results.

these results are only directly comparable to Intel i845D and E chipset motherboards.

I agree, especially that the SiSoft scores for KT333 chipsets are significantly lower than the ones for Intel chipsets.
Also, I dropped the timings to 2.5-3-3-6 (most conservative settings on this board) and it didn't complete a memtest86 loop @ 418 MHz 2.6V, seems a bit low IMHO but then again I usually sync my FSB and memory at 182 MHz so this is kind of a moot issue until I move to Prescott + Springdale.

 

[THUGSROOK]

good point~

these tests were run in 3:4 mode. (since thats what most P4 users run)
follow up testing using my p4m cpu in 1:1 ratio shows that memory can run higher when synced.
the stick of XMS3000 that can do 445ddr in 3:4 mode, can run 450ddr in 1:1 mode.

follow up tests

 

[ArmchairAthlete]

I have two sticks of Corsair XMX pc3200.

What are good timings?

I'm doing 5-2-2-2 right now. Is that too much? Could it be causing instability?

 

[THUGSROOK]

using an AMD?
id have to let an AMD guy answer that

you should be using memtest86 to verify your system is stable tho....

 

[ArmchairAthlete]

Ok let me be more detailed if that'll help:

Powercolor Radeon 9700 Pro
AMD Athlon 1700+ (OCed to 2200+ or greater, 166Mhz x11)
Asus A7N8X Deluxe Motherboard
512MB Corsair XMS DDR400/pc3200 (2x 256mb sticks)
Western Digital WD800BBRTL 7200RPM 80GB Hard Drive
Mitsumi CR-4802TE 12x read, 4x write CD-RW
Lite-On LTD-166S16X 16x DVD read, 48x CD read
Windows 2000 Professional

Current timings "5-2-2-2T". I forget what each one is, but thats the order they appear in the BIOS for my board (Phoenix Award BIOS, A7N8X Deluxe mobo).

I'll try that memtest86. I've been using Prime95, not sure if that can serve as a memory tester too.