Printable View
Quote:
Originally Posted by Repoman
Nop :nono: more gain in removing IHS on single-core
because the cores are warming themselves throug the die. :fact:
More important to find chip that go high on low Vcore.
.................................................. .......................
ur wrong all the way around man, first dual cores automatically run hotter, more heat = less stability, right? I dropped almost 10C going topless on my X2Quote:
Originally Posted by zzurf
"more important to find a chip to go high on low vcore," this is an art thats almost impossible for unskilled shoppers, and its almost always luck of the draw..
Ive seen 0550 do 3.2 and some not even do 2.7 soo...
finally decided to pop lid on 0550UPMW due to high temps even at stock! but what I did was popped lid....the took IHS put it in a small bowl and soaked it in lighter fluid....then I took CPU and put it in a small bowl and soaked it in lighter fluid as well....I soaked them both in lighter fluid because it is not a hot enough solvent to damage anything other than the black silicone glue that they put on the chips to keep IHS on...I let them soak for about 1 hours then grabbed a credit card and scraped off black silicone glue like it was butter....came off in about 1 min for IHS and CPU....then I got 2 new bowls and poured rubbing alcohol on both...let it sit for a couple of mins then took a paper towl to wipe off excess and let both of them air dry for about 10 mins..then I put some AS5 on CPU core and glued lid back on for a healty drop in temps...was idle @ 31C now 27C and load was 53C down to 40C...here is pic of naked core with all the silicone gone:
http://www.badongo.com/pic.php?file=...re.jpg&s=black
sorry for crappy pic....
What method are you guys using to remount the IHS? Do you keep pressure on it as the glue drys? What kind of glue? I think temps are the biggest reason my CCb1E 550 blows compared to others. The temps are an avg of 10c hotter than my old 543tpmw. Im not sure what to think anymore. The CCB1E was stable at 2.85ghz with 1.55v, but then became unstable. When it became unstable, no amoutn of voltage would stabilize it. I ran cpu burn over night 2 nights in a row and am now priming at 2.85ghz with 1.525v. So the burn in did help some, whether it will be stable for at least 8 hours remains to be seen. Ill keep you guys updated if anyone cares.
Quote:
Originally Posted by Hybrid
I used regular old superglue last night....but I will use better tomorrow. And yes after I glued IHS on I put cpu on black foam and put my sugar jar on top of it for weight.
Grinch..did you actually soak the entire cpu..including the pins???
It only too k10min to dry?..I would be afraid reinstall it into the mobo for risk of a short, unless it had air dryed for hours...
your results are very impressive..did the cpu overclock any better?
I just do not have the balls to remove my IHS,,afraid I will kill the cpu, i hve read many how to guides..the risk is a huge concern
BTW, very nice work
Quote:
Originally Posted by nealh
I literally had the entire cpu submerged in lighter fluid...pins and all. You use the rubbing alcohol to dry chip chip as to it evaporates fast! Was able to get stable at 2.9 ghz...on air.
I don't care about dropping temps, that is not gain for me, gain is MHz.Quote:
Originally Posted by Philosophy
I have only seen small gain on dual-cores, more gain on single cores after removing IHS. These are my experiences.
Fine, you dropped 10C but how much did you gain?
.................................................. ..........................
My Opteron 175 doesn't seem to do anything stable over 9x278 = ~2500Mhz @1,45V on a Epox 9NPA+ SLI , no matter the voltage i give it. It's burned in plenty i've used CPU-Burnin for over 3 sessions of 10 hours and atleast 72 hours of priming sofar. Ram is running at 175mhz (133 divider) Cooled by a Big Typhoon, 45C under load.
It's a 0550MPMW , is it true all 175's suck *ss and all 170's out-clock it easily ? I bought the 175 originally because i thought (hoped) they would be better chips. (and if so, what is the deal with that??)
If so, i'm selling this piece of junk and getting a 170....
Dropped temps will increase your overclock, why do you think people use watercooling??Quote:
Originally Posted by zzurf
Alright when i had m 170 on air it topped out at 3 Ghz with 1.48v.
On water it does 3 Ghz with 1.40 volts and 3050 Mhz with 1.48v. And my water wasn't even a 10c gain more like 7-8c.
Grinch i'm very interested in removing IHS and gluing it back on, so normal super glue is enough to completely fill the gap between the green pcb and IHS?
Permatex makes something called "The Right Stuff". Looks and has the same feel as AMD's adhesive. AutoZone/Pep Boys has it. I place a piece of cardboard in the stock plastic tray overnight after I re-apply the IHS.Quote:
Originally Posted by Hybrid
You can't tell the diference from new. Just dont forget to leave an air gap on the left side (with the markings at the top).
is this the right stuff?Quote:
Originally Posted by cdelong
http://www.autobarn.net/ch25223.html
Can't really go to autoparts store when i'm stuck on campus w/out any transportation.
EDIT
Overnight burn seemed to help stability at 3050 Mhz
http://img.photobucket.com/albums/v7...l/fabdf367.jpg
I will post some pics later of lighter fluid and alcohol method...:woot:
anyways here are some pics of my steup and what I am using to cool my setup:
http://www.badongo.com/pic.php?file=...nt.jpg&s=black
http://www.badongo.com/pic.php?file=...le.jpg&s=black
http://www.badongo.com/pic.php?file=...de.jpg&s=black
WTF are you talking about, you don't gain any MHz just by having an HSF on a naked core, you gain MHz from the temp drop it gives you. So you should care about dropping temps.. :stick:Quote:
Originally Posted by zzurf
hi
i just changed my previous 160 ccbwe 0534 spmw to a Opty 170 CCBWE 0550 UPMW, and this one didn't get as high as a tought it will be i played for houars at 2.93 with 1.6v but i got a problem with the memory latency acording to everest lantecy benchmark, with the 165 i got 39.6 latency and now doen't matter what i do i can no get anything under 46, if i push to hard the pc won't boot and in consecuence my super pi time is 29 sucks.i was thinking here in a memory controler problem. but what do you think?
the system is.
dfi nf4 expert. with NF4EDC07 is the last oficial one.
silverstone zeus 650 watts
corsair 2gb 3500 ll pro infineon ****
Opty 170 CCBWE 0550 UPMW @2933
(2) xfx 7800 gt
sb auridy zs
western d. 250 gb sata 2 js series
swiftetch apex 120 kit + vga blocks same brand.
What batch is that?
Too many volts man, i'm running water as well and volts beyond 1.53 does not go down well, just gets too hot. That won't explain your latency but... if you want to clock higher try to scale it up slowly starting with low volts.Quote:
Originally Posted by mc_chofo
don't know.
but at what frecuencies are you right now?
and do you know anything about what is happening with this thing.
and by the way and i don't know if it's normal, i don't remember anything like that but, when i changed the 165 for the 170 it removed in bios the memory divider that runs at 150mhz.
i've been trying to oc with low v. but it's impossible and you have another stepping, not all of them runs at same freq. and for me it's impossible to run this cpu at lower v. at those frecuencies.
my temps are. 27 iddle and 36-38 load
dfi nf4 expert. with NF4EDC07 is the last oficial one.
silverstone zeus 650 watts
corsair 2gb 3500 ll pro infineon ****
Opty 170 CCBWE 0550 UPMW @2933
(2) xfx 7800 gt
sb auridy zs
western d. 250 gb sata 2 js series
swiftetch apex 120 kit + vga blocks same brand.
Yeah I got 2 170 CCB1E 0550VPMWs 678L, 523 & 168 batches. Both initially tested @ 2.9Ghz @ 1.45 (DMM, = 1.375 in bios) on my DFIs, whereas the UPMW needed 1.5 to get to 2.7Ghz.
One is now breaking in in rosetta @ 1.375(DMM) (1.30v x 1.04bios) / 10x280, will move that up in a few days. Other one goes in during the week.
Quote:
Originally Posted by mc_chofo
Quote:
Originally Posted by mc_chofo
38c load with 1.6 volts and a dual core at 2.9 Ghz? You are running 2 instances of Prime right, with correct affinities set? And the IHS is on...?
Also, i wouldn't use 150 divider, doesn't really work for me anyway, 180 and 166 work best for me.
@Absolute_0
absolutely ridiculous clocks with nearly stock voltage
yes it a really great cooling solution, recomended. :), and yes i'm doing everything above all test and everything.
those are really great steppings, i'm looking forward for one of those, but here in mexico is hard to find them. so
please help me with the latency problem, 'cause i think could be an RMA of the mother or the cpu. i hope non of those will.
All dividers will work fine, as long as you set your MAL/RP correctly and you have a decent MB. My Ultra-D AD0 lets me use all the dividers with a bit of tweaking but the one I had before it (RMA'd, not sure what rev) only let me use 100/120/166/180 no matter what.
Also I think burn in helped my 0550VPMW as well- my burn in settings are 312x9 with 1.400v, core 0 crapped out in prime after 5 seconds and core 1 lasted 8 minutes at first. Now they do 90 secs, and 23mins.. plus 308x9 is now prime stable while before 307x9 was the absolute max. I haven't tried 309 after the burn in yet.
This is after ~8 8hr+ sessions.. it did get much worse before it got better. At one point 307x9 (stable before burn in) crapped out after 4hrs in prime
sorry i'm lost in translation of what you said that i have to set right something about mal and another thing. can you explain please? and thanks now i'm burning it at same freq as you did but with 1.5v but i'm burning with prime. what do you suggest about the burning? and what was about the other?