Xmicro 9800PRO Xtreme ram @ 398mhz!

I've always comtemplated putting ramsinks on my video cards but never got around to it. But today while benching my Gigacube 9800PRO Xtreme (default 398/351x2) on my A64 3000+/AN50R setup i decided why not.

I was going to use Vantec IceberQ Pro 4s copper ramsinks on my 9800PRO Xtreme but thought to myself what about my 128MB Xmicro GF4 Ti4200 VIVO's ramsinks that i removed 7+ months ago at http://www.fileshosts.com/vantec/ice...cro4200Vivo/1/ ?

You can see the ramsinks at

http://www.fileshosts.com/vantec/ice...rgPro4_01.html

http://www.fileshosts.com/vantec/ice...rgPro4_03.html

http://www.fileshosts.com/vantec/ice...rgPro4_05.html

http://www.fileshosts.com/vantec/ice...rgPro4_07.html

Well see the black sticky tape at http://www.fileshosts.com/vantec/ice...rgPro4_18.html on one of the ramsinks well all of them are still very sticky even after 7 months being off

The mod
So what i did was just stick 3 of the 4 xmicro ramsinks onto my 9800Pro's ram chips so that each ramsink covered 2 ram chips each using the same black sticky tape already on the ramsinks that Xmicro implemented. I was blown away by the results!

Before stock 9800PRO Xtreme without ramsinks
stock ram = 351mhz x 2 = 702
oc'd ram = 380mhz x 2 = 760 repeatable and benchable i could sometimes do 382mhz x 2 but with artifacts and with Cat 4.3 the VPU recover kicks in.

After 9800PRO Xtreme with Xmicro ramsinks - the surprise
stock ram = 351mhz x 2 = 702
oc'd ram = 393mhz x 2 = 786 and can barely just bench at 398mhz x 2 = 796 with artifacts and VPU recover is happy and doesn't show it's ugly head! :eek: :thumbup: :D

Using infra-read thermometer i measured the ramsinks under 3dmark load to over be between 33-36C :)

so with just stickying on old Xmicro TI4200 Vivo's ramsinks on 6/8 9800pro ram chips, my ram max oc went from 380x2 mhz to 393+ x 2 mhz! no volt mods, no extra thermal tape need at all!

Whoa Eva looking good, gonna haev to show me it in person nah jkz.

sometimes memory just gets better over time

before my dog :banana: :banana: :banana: :banana: radeon mem wouldnt even do 360, now it does 380 perfectly fine with no artifacts

and the core is gooood :) 530 on straight water

hehe ... well just finished some benching (same windows session) and

before ramsinks:
max oc'd = 483/760

after ramsinks:
max oc'd = 486/792 :cool:

gonna reboot and do a fresh benchie or two :)

Update: just did a same sessions windows bench for 3dmark2003 340 build @ 483/786 = 7,300
http://service.futuremark.com/compare?2k3=2340206

broke my previous high score by nearly 200 points :)

Update 2: rebootd pc and ran @ 483/792 = 7,306
http://service.futuremark.com/compare?2k3=2340335

actually I dont think 400mhz is quite spectacular on newer revisions of 9800pro board.. mine does 406 without heatsinks and 418 with a tad of artifacting (no heatsinks and mods)

this card is well over 6 months old hehe and room temp here is 33C :)

wonder if i can get mine up to 400mhz without volt mods hehe

latest run @ 483/792 = 23,142
http://service.futuremark.com/compare?2k1=7730882

this is with DX9.0b and Cat 4.3 :)

suffering low car scores sigh

Quote:

---

Originally posted by eva2000
this card is well over 6 months old hehe and room temp here is 33C :)

---

ahh sounds pretty nice then!

lol... something funny has happened, i decided to put the 4th remaining xmicro ramsinks on the front side lower ram chips on 9800pro xtreme

Put them on and slotted the card into AGP slot only to find out the ramsink aligned to snuggle itself right on top of the AGP retention lever in the lock position ...

now i'm having a hard time getting the card out of the AGP slot as the ramsink prevents the lever from going back up/out LOL... guess have to remove the ramsink :o

edit: decided to leave the 4th ramsink on and leave the card in the AGP slot for some benchies. The 4th ramsink installed as allowed my ram to benchmark @ 405mhz x 2 now :eek:

well 4th ramsink really did the trick

3dmark2003 @ 486/805 = 7,353
http://service.futuremark.com/compare?2k3=2340848 :D

ranked #8 for single score per user results on 3dmark2003 ORB for A64 + 9800PROs with futuremark approved drivers (Cat 4.3 + DX9.0b) :cool:

Last run before off to work - 4 ramsinks made a hella of a difference!

488/805 = 23,215
http://service.futuremark.com/compare?2k1=7731277

http://fileshosts.com/AMD/boards/Shu...tr_488-402.jpg

i've also noticed an increase in performance with the ram sinks ...increase of ~ 4 mhz ! hehe.

but when i put my MCW50-T on the core ...the ram also benched at 405 mhz! ...i think i can squeeze a few more out (after i've finished upgrading my WC setup to 1/2" tubing) :D

happy camper.. at 23,304 right now and raising http://service.futuremark.com/compare?2k1=7732917 :)

Anyone tried watercooling the memory?

Im thinking about it but im having troubles designing a waterblock that will fit with the fusion hl on the core....

Quote:

---

Originally posted by kommando
Anyone tried watercooling the memory?

Im thinking about it but im having troubles designing a waterblock that will fit with the fusion hl on the core....

---

http://www.wassergekuehlt.de/Shop/de/dept_8.html

Hmmm.... Thanks, but do i want to lose this fusion hl for a new block?

Ill make a new thread in liquid cooling section.

water cooled mem sounds fun but i love these xmicro ramsinks new high score = 23,426 http://service.futuremark.com/compare?2k1=7732999 and still on CAT 4.3/DX9.0b! :)

When are you going to do the Vmod eva?.... you know you want too...

I'm going to be doing mine soon!

i'm stoked with the card as it is ... not gonna risk it right now with volt mods hehe - hey winter coming so maybe a few more mhz lol

Well be as brave as General_Rommel and something interesting might happen in terms of overclocking :D

Nice pickup EVA2000

Did you try and put another fan blowing across the card too? Might be able to squeeze more Mhz out.

un-voltmodded right?

thanks FUGGER yup no volt mods at all!

haven't tried additional fans as room temp is 31-33C just be blowing hot air around LOL

funny think i got a bugged run and topped the 3dmark2003 ORB with a 'mother' of a mother nature score LOL

3dmark2003 = 12,962 3dmark2003s :eek:

with only AMD64 3000+ @ 227HTT hehe

http://service.futuremark.com/compare?2k3=2344651

http://fileshosts.com/AMD/boards/Shu...-402-cg226.jpg

rank #1 for a few moments lol
http://fileshosts.com/AMD/boards/Shu...2962_rank1.jpg

vs #2 macci score hehe
http://fileshosts.com/AMD/boards/Shu...k_vs_macci.jpg

http://fileshosts.com/AMD/boards/Shu..._vs_macci2.jpg

LMAO EVA, dude is that hot at your house. Its quiet chiller here :)

yeah very hot and stuffy in my house.. probably when ya have all pcs on 24/7 folding lol

Wow, that's reall cool :)

What IS the black sticky tape made out of, do you know? I wouldn't think you would get as good as results as you would with thermal paste/adhesive. That's interesting.

Anyway, nice scores :)

not sure what it's made off, it's like a paper thin layer of less sticky seal string which even after 7+ months in the open stuck the ramsinks to the ram chips fine!

maybe thermal transfer not so good ? as the very first oc'd benchie i always do it starts with slight artifacts then subsequent runs there are no artifacts ??