Decent RAM OC: 166 to 220

VIAN

Diamond Member
Aug 22, 2003
6,575
1
0
I've always heard that RAMsinks never do anything. That they are just there for extra safety. Well, if that is true that is total horse dung.

I got some nice RAMsinks from COMPUSA. They are copper, chrome plated. Put them on my 2x 512MB of PC2700 Corsair Value Select. I was able to go from 190MHz, 2-2-3-3 to 220MHz, 2.5-3-4-4. My Memory bandwidth, according to Sisandra used to perform like the kt400 option, but now surpasses the best non-overclocked Athlon XP machine. I am amazed because I didn't expect these kind of results.
 

Soulkeeper

Diamond Member
Nov 23, 2001
6,732
155
106
yeah the only time they don't do anything is if you get cheesy aluminum onces with a bad finish and paste them on with a quarter inch of cheap epoxy or thermal tape

i also had good success with copper rammsinks (i prefer the onces that aren't plated with anything)
i used artic silver epoxy and seemed to get atleast 20-40mhz more out of my GF4 memory
 

EarthwormJim

Diamond Member
Oct 15, 2003
3,239
0
76
Which one's did you buy? I might think about doing this since my Samsung memory seems to top out at 190MHz.
 

VIAN

Diamond Member
Aug 22, 2003
6,575
1
0
Just go to COMPUSA and get the chrome plated, copper core RAMsinks. The Aluminum ones are blue so they will be easy to distinguish.
 

kylebisme

Diamond Member
Mar 25, 2000
9,396
0
0
Originally posted by: VIAN
I was able to go from 190MHz, 2-2-3-3 to 220MHz, 2.5-3-4-4.

i'm guessing the looser timeings got you more mhz than the sinks.
 

Peter D

Diamond Member
Oct 28, 2002
3,603
0
0
I think when people say RAMsinks suck, they mean for Video Card RAM. System memory will get stressed more and hotter in some cases so of course some heatsinks will help. Congrats nonetheless :)
 

jagec

Lifer
Apr 30, 2004
24,442
6
81
RAMsinks are nice for video card RAM too (especially those toasty 9800pros...)
Before RAMsinks, I could only get up to 360MHz RAM on my 9800pro. After putting on some RAMsinks, I got up to 375. It would even work at 380, but start artifacting after about 10 minutes at 100% load. (Using ATITool)
 

Soulkeeper

Diamond Member
Nov 23, 2001
6,732
155
106
yeah artifacts sucks, be carefull i fryed a GF2 gts from not bumping the mem down an extra few mhz to stop the accasional artifacts

most ddr bga memory seems to get atleast 10-20mhz from adding good heatsinks
system memory is worth adding good heatsinks if you can pull it off
heck i would subzero cool my memory if i could
imagine what it'll be like when we get 1ghz ddr system memory :)
 

Regs

Lifer
Aug 9, 2002
16,666
21
81
Impressive! Good buy. That's a success story important to note for the future.
 

ponyo

Lifer
Feb 14, 2002
19,688
2,811
126
Ram sinks for DDR system memory is a waste. It is however useful for Rambus memory.
 

VIAN

Diamond Member
Aug 22, 2003
6,575
1
0
quote:
Originally posted by: TheSnowman

quote:
Originally posted by: VIAN
I was able to go from 190MHz, 2-2-3-3 to 220MHz, 2.5-3-4-4.



i'm guessing the looser timeings got you more mhz than the sinks.



Pretty safe bet...
Unfortunately, no, because, I was able to bump up the clock speed up more. I only had to change the timings when I got to about 210 or 215. I might be able to squeeze 5MHz more out of this because I think I made a wrong assumption while I was searching for stability.

Soon, I'm gonna get a new heatsink for my 2700+ only stable at 2.2MHz. I can't go passed that because it gets too hot and the X-Dream can't handle it. I am able to get up to 2.3MHz, but I need to go up to 1.8v and the heatsink can't handle that. It failed at 2.4 and refused to run because it was running at 74C. Heat is the enemy.

I also need to get some VGA silencer + some copper RAMsinks for my 9700 Pro and try to get it up, because I cannot go any higher than 360/333 and I know it's possible to go higher. Plus the VGA Silencer will help keep the case cool and possibly make room for better overclocking results with other components.
 

JBT

Lifer
Nov 28, 2001
12,094
1
81
uhh why do you run the TRAS so LOW??? 2-2-3-3 and 2.5-3-4-4 is EXTREMELY LOW for that!!! Do u suffer from data corruption at all? Why don't us use like 7-11??? which has proven to be faster on NForce2 boards and less stressful on your ram.
 

VIAN

Diamond Member
Aug 22, 2003
6,575
1
0
I was able to get it up to 225MHz @ 2.7V. I know I can't go over that.

uhh why do you run the TRAS so LOW??? 2-2-3-3 and 2.5-3-4-4 is EXTREMELY LOW for that!!! Do u suffer from data corruption at all? Why don't us use like 7-11??? which has proven to be faster on NForce2 boards and less stressful on your ram.
My RAM is really weird, it doesn't like to do 3cas or 5 for the second digit or above 7 for the last digit. If I do that, it will not boot. All I did was use the lowest timings at stock speed and then up the clockspeed by 5MHz each time and uped the timings when necessary. at 230MHz a voltage increase or timings increase will not help its stability and at 235 it cannot go into windows and it needs high timings and gives screen corruption and freezes.

How would I tell if I suffered from data corruption? Most I do is play video games. I run a couple of benchmarks. 3dmark2001, sisandra. 3dmark2001 happens to stress the graphics card and the cpu-RAM more than 3dmark2003 or sisandra respectively. I played COD for an hour or two fine. That may be the most I'll play in one sitting, other than that, I leave it on all day - if it can do that, it is stable for me.

Sisandra benched.

Int 3250MB/s
Float 2950MB/s

this is with 225x9.5. CPU gets too hot at multi 10.
 

ponyo

Lifer
Feb 14, 2002
19,688
2,811
126
Vian, Download and run Prime95. Tell us the results. I've a feeling your machine will fail the test in less than 5 minutes.
 

BigJelly

Golden Member
Mar 7, 2002
1,717
0
0
Originally posted by: i82lazyboy
Originally posted by: TheSnowman
Originally posted by: VIAN
I was able to go from 190MHz, 2-2-3-3 to 220MHz, 2.5-3-4-4.

i'm guessing the looser timeings got you more mhz than the sinks.

thats what i thought right away. For my p4, video editing with 2.5,3,3,6 was 94.6 fps and with 2,3,3,6 i get 124.6 fps with the same processor speed/FSB.
 

bjc112

Lifer
Dec 23, 2000
11,460
0
76
Originally posted by: TheSnowman
Originally posted by: VIAN
I was able to go from 190MHz, 2-2-3-3 to 220MHz, 2.5-3-4-4.

i'm guessing the looser timeings got you more mhz than the sinks.

Exactly...

Try and runn exact same timings and then see the increase... I'm sure you had some gains, but 30mhz is dramatic.
 

VIAN

Diamond Member
Aug 22, 2003
6,575
1
0
Well, yeah the timings helped, but the heatsink did at least get me 10-15 more MHz.
 

VIAN

Diamond Member
Aug 22, 2003
6,575
1
0
It's weird. I run Prime95 on torture test.

Sometimes I would get errors around the 11th time and sometimes it would just keep going. When I tried it the first time, it almost finished. But then something happend, where my bus speed was changed to 200 and the program completely shut down. Weird. Any ideas.

It can't be heat because when I increased the voltage the errors stopped. Then when I decreased the voltage back to where it was, there were no errors.

Could it be the northbridge, it's not too hot to touch. Could it be the PS? CPU is working at decent speed, underclocked by 20MHz.

Then I found out I was running an old Prime95 and downloaded the newest version.
 

THUGSROOK

Elite Member
Feb 3, 2001
11,847
0
0
yea, ummm....
ramsinks dont do chit.

oh, and please improve your testing proceedures before posting stuff like this. TY
 

Jeff7181

Lifer
Aug 21, 2002
18,368
11
81
Originally posted by: THUGSROOK
yea, ummm....
ramsinks dont do chit.

oh, and please improve your testing proceedures before posting stuff like this. TY

Way to be blunt =)
 

pspada

Platinum Member
Dec 23, 2002
2,503
0
0
It's true, the only way to verify that your Ramstinks are increasing fuctionality is to also try WITH the looser times but WITHOUT the ramsinks. Then once you've established that baseline, you can add the ramsinks to the mix and get a result that has more to back it up than, er, well, "Wow, RAMsinks Rule!"
 

VIAN

Diamond Member
Aug 22, 2003
6,575
1
0
Yeah yeah, will do next time.

Finally was able to do 220@2.5-4-4-4 which was stable for 12 hours. Long enough for me.