Good luck :toast:Quote:
Originally Posted by derektm
Printable View
Good luck :toast:Quote:
Originally Posted by derektm
actually it is, taken from the stress prime site,Quote:
Originally Posted by Rufus7
"SP2004 uses a redesigned user interface that calls Version 23 Prime95's torture test code. This should provide the same results as Prime95, just displayed in a different manner."
might not be the latest version of prime but it works just as well.
Thanks for backing me up :toast:Quote:
Originally Posted by Delirious
derektm, what water setup you using? ill have a dc opty soon, and im curious to see how mine'll do. Ive got a swifty kit in the mail, and an x2 3800 now.
Im using 1/2" tubing with these parts: Dangerden D5 (aka Laing D5) pump, Swiftech Storm block, and a dual heatercore with 2 120mm Sanyo Denkis fans @ 7volts. Handles heat really well. Good luck with your setup! :toast:Quote:
Originally Posted by Sorrow13
Well.. maybe he knows something..
I have no doubt on derktm's results, but do have a bit of reservation on SP2004. Say, if I were to buy a CPU based on someone's OC results, I'd rather like to see the original Prime95 results. There *is* a way to abuse SP2004. ;) (I'm not telling you, so don't ask. :P )
Like I said, I totally trust derek and envy his awesome CPU. (Will buy if he wants to sell!) I'm just talking about SP2004.
Heres for you Rufus...Quote:
Originally Posted by Rufus7
http://tinypic.com/jzwbbp.jpg
damn nice cpu there
I wish I could luck into something that nice
No. It is what Prime95 recommends.Quote:
Originally Posted by Delirious
I personally had a lot of overclocks failing in the window between 24 and 36 hours.
derektm you better be running 2.9 Ghz RIGHT NOW :). I wanna see how many volts it takes. 1.475?
Hey all,
Any news on the CCB1E chips? (Other than the ocforums link). I saw a couple of people were getting them, any news so far?
My other option would be to get a CCBWE 0546SPMW chip. I'm kind of inclined to go with the CCB1E as it doesn't have retail packaging and i need to ship it to the UK, but I'd like to see more results from the CCB1E, or even the 0546SPMW chips first.
Any suggestions?
well i finally got my 170 0530 tpmw up and running. was able to get 273x10 at 1.5v, and 275x10 at 1.536v. I'm currently working on burning the cpu in, but I was not very shocked at how good my chip clocked, especially when compared to derektm's chip, which does 280x10 at less voltage
any suggestions? or is that all i can get out of my chip (when i get home from work and get windows reinstalled i'll post pics)
Quote:
Originally Posted by Martin Cracauer
yah i know i was jokeing.
I understand the debat that everyone has over how long to run it.
Personally i run about 8 minimum hours mainly cause, i wont be putting 100% load on my machine for that period of time.
when i get a chance im goin to run mine for 24hours, just need to use the comp, so dont really have a chance.
Try lower volts? Try 1.4 * 104% in bios. Play around with this because different settings give you a more stable vcore. I find 1.4 * 104% to be perfect for 2.8ghz.Quote:
Originally Posted by terrordemon
Quote:
Originally Posted by Delirious
When I ran it for 48 hours, just as an example, I still played Counter Strike Source just to see if I could. No problems with lag or anything so you should be able to run it while your doing what you normally would on your computer.
derektm: so youre saying that if i try lower volts i may be able to get higher clock speeds? interesting, i'll give it a shot
Yep. Worked for me. Good luck!Quote:
Originally Posted by terrordemon
did you see vcore flucuate in bios readings or winxp readings....Quote:
Originally Posted by derektm
what vcore settings seem less reliable in your experience? I see you like 1.4 *104%= 1.456...
For me it doesn't equal that amount. I have checked with a DMM. I didn't see much vcore flucuations no. 1.4 * 104% is the best so far. I tried a setting that gave me 1.5 volts but it must not of been stable because my chip wasn't. I'm using a newer revision of my Ultra-D now though. 1.4 * 104% is all I have tried on this particular board.Quote:
Originally Posted by nealh
so what does the DMM report for 1.4 * 104%....the 1.456 is just 1.40 x 1.04=1.456...Quote:
Originally Posted by derektm
for example:
for me if I set in the bios 1.35 *110%=1.485..bios reads 1.46, winxp reports 1.44
I guess I should get a voltmeter..just not sure what to buy and I am afraid to fry something in the testing...does your mobo overvolt, undervolt or deliver what the bios sets
what u set in the bios is what u get, and i would trust that over any reading (bios, windows, cpuz, mbm5 etc..)other than a digital multimeter.Quote:
Originally Posted by nealh
The rule of thumb is:
On the popular DFI NF4 Ultra-D/SLI-DR you get exactly what you set in the BIOS in idle. Under full overclocked (dual-)prime95 these boards slightly raise the voltage, by about 0.025 V. The Voltage sensors report about 0.07 less than what you set in the BIOS and get in idle. The Voltage sensors do not show the increase that you get between busy and load. The Voltage sensors show fluctuations that do not exist in reality during load.
On the Infinity boards the voltage is constant over busy/load and idle, always about 0.0125 higher that you set. Very constant and stable overall. The Voltage sensors are completely BIOS-hosed on the boards, for all OSes I tried.
On my A8N-SLI premium I always get a solid 0.05 V over what I set in the BIOS, stable for busy and idle. The sensors are very accurate, they underreport a little, but they are closer to the truth than to what you set in the BIOS.
0.05V?? I guess you mean 0.005V. :)Quote:
Originally Posted by uOpt
I just got a 165 CCBWE 0550 UPMW.
The problem I am facing is that I can't get memory prime stable,
I have got 2* 512 TCCD's, they were prime stable at 310MHz
on my 146 CABYE 0540 FPBW, at the same timings. About 10-20h minutes
of dual prime renders me an error, I get this if I am running from 250 to 300 MHz on the memory the clock speed of the CPU is then 250*9 to 300*9.
It seems to me that it isn't the CPU maxing out but most likely some
memory timing. Has anyone else had this problem or have any timing suggestion?
The timings are 2.5-4-3-7 and I have tested to change it to 2.5-4-4-8 without improvement. I increased the preamble from 5mS to 6.5 and it seems as if it improved the prime stability a little bit I now get the error after 40 minutes.
Any ideas would be greatly appreciated
No :)Quote:
Originally Posted by bobov
The Asus overshoots quite a bit.
Delirious and UOpty..thank you very much for make it very clear and concise
excellent info
I will not bother with a DMM..still not sure what to buy anyway..and I am sure I would fry something with my luck
derektm: just to let you know i tried 1.4 * 1.04% and i couldn't pass sp2004 for more than 2 minutes at 2750 :( I guess that around 2700 is gonna be my 24/7 speed, now just gotta find the lowest voltage itll work at. Kinda dissapointed with this 0530 tpmw :(
One more thing..I have found interesting
I had some issues getting my new opt 170 0546xpmw to overclock..I popped it and and tried right for 2.7GHZ and did horrible
I had to initial set bios to 1.55v to get stable..ran dual prime for 10 hrs no issues....so I then waited a day and reset bios(also tried 704-2bta with no help reflashed back to 623-2) to 1.51v and ran 8hrs @2700...
last night dropped to 1.5v and I went 13hrs without a hitch when I stopped it
Also found the cpu temps dropped but my PWM which usually never hit 60C even with 1.55v on my opt 165 did hit 60C and stayed at 62C
Seems my 170@2700 creates a higher PWM temp than my opty 165
cpu temps at 1.5v are 48/50C underload
I think 1.5v is probably going to be lowest vcore to get 2700 as winxp seems to go from settings screen/logon screen to winxp and AV loaded slightly slower than 1.51v(but temps are 2 C better at 1.5v)
In my very limited experience this usually means another vcore drop will not be stable...Also do not think I will get much more of an OC...not a great chip...its ok
on a side note about lowering voltages:
is there any way to easily change the voltages in windows? i know clockgen, but for some reason when i load up clockgen the voltages tab is greyed out. is that because my CPU VID is on 1.5v and not on auto?
I use RMClock. Set up a couple of p-states and change the VID on either one as you need. Takes immediate effect.Quote:
Originally Posted by terrordemon
thanks for the tip, i'll check it out when i get home from work
use crystalcpu , I liked it better.Quote:
Originally Posted by terrordemon
erm... link? i cant find it in english that i can understand
http://crystalmark.info/download/
click on the image
after instalation , run it and hit CTRL-K to change multi and vcore
My opty is in the mail, will post with updates. Retail from Zipzoomfly.
Was gonna go tankguys but they raised the price and went oos.
Here is my fresh 165. Popped in this morning and booted @9x300 right away. (stock VCore 1.312V) It wasn't however stable so I upped the volt a bit and started initial burning. It's amazing. :)
<a href="http://imageshack.us"><img src="http://img71.imageshack.us/img71/7302/2700smallfft14hrsa3cw.jpg" border="0" width="1200" alt="Image Hosted by ImageShack.us" /></a>
<a href="http://imageshack.us"><img src="http://img71.imageshack.us/img71/1223/2700smallfft14hrsb3mg.jpg" border="0" width="1200" alt="Image Hosted by ImageShack.us" /></a>
Currently working on 2799MHz(9x311)@1.4V. If I'm lucky (Well, I already am, I guess) I might get 2900MHz out of it.
sorrow 13 what steppin is that 165 and what motherboard? :toast:
The 0546SPMW is a good stepping. I have my running 10x260 no problems, and I know I am cooling limited now. I swapped out my panaflow for a tornado and was able to take it up to 2760 stable at 1.5. I am positive that with water I can hit around 2850 ish with this chip.Quote:
Originally Posted by unclean
ill update my sig when i get info. Im in the market for an ultra-d, but for now ill have to use the inifinity. Depending on how the sale of my current hardware goes (x2 and mobo) ill upgrade ram (2x1g gskill) and power supply (antec tpII 550).
Was supposed to get the cash yesterday but you know how that goes :(
I just bought a Opty 165 CCB1E 0550 VPMW. Dunno how these clock, but there's only one way to find out no? :p:
Should be here by Monday ;)
question: if i run sp2004 on small ftt, is there ANY way that the memory could be making me fail? or does it JUST test the cpu
My CCBWE arrived.
http://www.m0px.net/A100/new_card/IMG_0058.JPG
then get ocing! what are you doing wasting time posting pictues! :p
Cause I don't have an S939 mobo yet. Duh.Quote:
Originally Posted by terrordemon
<--- owned
ROFL :DQuote:
Originally Posted by terrordemon
Has anyone gotten there CCB1E and done some overclocking? I'm getting one soon so you guys could help a brotha out I'd really appreciate it.:D
0530 TPMW
stock voltage
big typhoon
A8N SLI Premium
http://www.supload.com/thumbs/default/43 hours.PNG
way too hot, but I wonder how accurate these temps are
see these threads for more info
http://www.xtremesystems.org/forums/...ad.php?t=86045
http://www.xtremesystems.org/forums/...ad.php?t=83894
Just got my 165 in from tank guys. ccbwe 0547 tpmw. Right now running dual sp2004. So far looks good, slowly upping with clockgen every few mins. At 1.45v its currently running 2766 and i later up volts and clocks to see what its got in it.
Looks like on this board (9nda3+) its gonna top out at around 2800mhz with 1.5-1.55v. Note this is only preliminary prime stable (45mins), and to be honest i dont really care about it being stable in prime as long as its stable in games. Hopefully a new motherboard and de-ihsed will bring me up some. Cooling is water with storm block.
Quote:
Originally Posted by terrordemon
Haha :lol:
Mine's at 1035 mzh overclock x 2 cores = 2070 mzh above stock
What's the best overclocking chip ever? I think the 165 is a contender.
:fact:If that is 100% torture stable then you have an awesome chip there :clap: :toast:Quote:
Originally Posted by NeezyDeezy
It seems that the 165 overclocks just as well as the 170 ~ 175 but with just the lower multi so certainly the best bang for buck on the planet ATM
What's your definition of stable? :)
Quote:
Originally Posted by NeezyDeezy
Dual Prime95 ver 24.14 large FFT for at least 8 hours :)Quote:
Originally Posted by rick_fx
its the 3000 venice of the DC world.
Actually, 2870 mzh was windows stable but the system would reboot under load...
To date I've never gotten an error in superpi or prime with this chip, it is truly a diamond among gold.
I'm running at a conservative level because I need my computer to never crash so it's at 2835 now. It has passed 24 hour dual prime at this setting.
Reason I don't go higher with volts is because temps go nuts. Right now load reaches 56C if it's hot inside. I may buy OCZ's new phase change cooling when it comes out if I have a better paying job by then. In the meantime, I have no GOOD reason to change a single thing.
What does this mean? :) You didn't happen to do a screen shot of that 24hr prime stable? ;)Quote:
Originally Posted by NeezyDeezy
It means he can boot to windows suvvessfully but attempting to run demanding apps or benching will cause errors and reboot.
the rebooting might be memory related. You might try relaxing your ram timings.
Then it's NOT stable.......... Follow situman's lead and good luck.... :)Quote:
Originally Posted by FZ1
well, lets just say i shed blood for my computer :)
http://thecowsaysmoo.org/budwise/pic...al/Naked-2.jpg
http://thecowsaysmoo.org/budwise/pic...al/Naked-1.jpg
Ready for the Storm!
http://thecowsaysmoo.org/budwise/pic...al/Naked-3.jpg
Results:
Before
2700 1.49vcore loaded at 38C (max 24/7 OC)
It just wouldnt be happy at 2800 no matter what i tried
After:
2800 1.53vcore loaded at 36C (prime testing and running great)
I would say although i scared myself doing it, it was definately worth it...
:D
Dual instances of StressCPU at High Priority for 24 hours :)Quote:
Originally Posted by rick_fx
Please stop quoting me.:)Quote:
Originally Posted by Super Nade
I was only asking NeedyDeezy (since I was quoting him)
I have decided im going to de-IHS and also lap the die like mad mikee does. Just ordered 3micron and 0.05 micron powder to do so, which should get here next week some time. Wish me luck.
Does anyone see any value in removing the IHS and refitting it using AS5 ? If yes, how might you re-fix the IHS ? I really don't like the thought of having the core exposed, at least not without a shim.
Is the core as robust as AMD chips that don't has an IHS ?
BTW HERE is a guy who killed his Opty while removing the IHS. Look and learn!
Quote:
Originally Posted by Zim
There have been tons of people who have killed thier processors removing the IHS.
Its just a matter of you being willing to take a risk, voiding your warranty(if you have one) in the process and possibly killing.
For some its worth it for others its not, then there are those who have money to blow and dont care if they kill it cause they'll just get another.
Removing the IHS bought me another 150mhz and a 10c drop in temps. Of course with anything its a risk, just gotta be calm and take ur time.
Anyone got a CCB1E 0550 VPMW yet :stick:
Mine's coming on monday/tuesday, but the suspense is killing me :stick:
Theres a few things about removing the IHS. First, if you have a TT Big typhoon or something huge like that youre taking a much bigger risk IMO than if you have water cooling, because you have a whole lot more weight on top of the proc. Second, dont use a utility or an exacto, use the razor blade from a safety razor like a $1 two blade bic. Just break the razor head and take one of the blades and be ultra careful. Those razors are paper thin and ultra sharp and cut through the silicone holding the IHS on like butter. Also those razors are flexible so theres less chance of taking a capacitor off of the proc.
Uh, yeah...that was me :doh:Quote:
Originally Posted by Zim
It's a risky business. But leaving it on wouldn't be very extreme now would it? :fact:
Ok so let me get this straight, according to DFI-Street the process of figuring out your max cpu clock is:
1. Lower LDT Multi, Divider, Set CPU Multi, Etc.
2. Run SuperPI 32M in 10-15 Mhz incriments add voltage if neccessary until you find the highest point of stability.
3. Run OCCT to check for stability and decrease by a few MHZ if not until you find the right Mhz.
4. Run prime for 8-24 hours as a final test.
I know I need to run two instances of prime, what about OCCT and SuperPi?
OK, I successfully removed my IHS and found that it is just a piece of 1/16" heavy plate steel. As expected, the core was too low to use with a standard HSF, so I replaced the IHS using AS5 and a dab of super glue at each corner. My temps have not improved, and in fact are worse. However, as the AS5 cures, the situation should improve. That may take a week or two though.
@Zim: Just run her naked bro... You've taken the step to void your warranty already so might as well run her w/o the IHS. I bought myself 275mhz after popping off my IHS and dropped my load temps about 12*c. I saw HUGE improvements.
It was definately worth it in my case :)
@foodfightr: You will need to definately run your two instances of prime. OCCT doesn't allow for more than one instance at a time so I don't think that can be done yet. SuperPi also needs to be run in two instances w/ affinity's set through the task manager.
It's not steel, I can promise you that. I believe it's nickel coated copper. I made a keychain out of my dead Opteron 146 CABNE 0530 IHS and it is almost entirely copper.Quote:
Originally Posted by Zim
OK, that's good!Quote:
Originally Posted by wickedld9
I refitted the IHS a second time and got much better results. I made sure to remove all of the silicone sealant and ended up getting the IHS closer to the core. Now I'm basically back to where I started as far as temps go, maybe a little lower at full load. When the AS5 cures I might get a little more benefit. Overall, I don't recommend that anyone do this. The original AMD thermal compound seems pretty good.
BTW, can anyone suggest a silicone sealant similar to the original AMD one? Without it, the IHS is likely to come off when i next remove the HSF.
im not sure why you went to all that trouble just to put it back on...
I had read from several sources that AMD often made a poor job of bonding the core to the IHS. I wanted to re-do it myself to see if I could improve my mediocre temps. It was interesting to do and I didn't do any harm. Now if only I could find some of that silcone sealant I would be 100% happy.Quote:
Originally Posted by Budwise
Just a note of warning for anyone with air cooling who may want to remove the IHS - don't do it. Removing the IHS leaves the surface of the core below several parts of the socket, and a large HSF place won't be able to make contact with the core.
I got my Opty 165 CCB1E 0550 VPMW today.
Booted at 2.7 GHz with stock vcore, but failed prime in 15 minutes. :(
Howeve, i think i know why. I'm off to do some tweaking :D I have very high hopes for this chip :cool:
Looking forward to your results, maybe in the market for one of these. Hope tht stepping turns out to be a good oneQuote:
Originally Posted by K.I.T.T.
:)
You on air cooling?
Thanks mate :)Quote:
Originally Posted by mongoled
Yes, I am on air cooling, using an Arctic Cooling Freezer 64 Pro. It has been priming at 2.65 GHz at 1.31v (had to raise it slightly to give a higher voltage to my memory).
The reason for prime failing was due to the PWMs heating i think, but i found a ghetto way of cooling them :D
http://img.photobucket.com/albums/v6...9/P1040120.jpg
I'll post some prime screeies later :)
You can build a non-ghetto RAM / MOSFET cooling system pretty easily with the layout of the DFI boards. The Ultra-D and SLI-D(R) layout is a bit easier than with the Expert but it takes about 3 hours of spare time, a clear sheet of acrylic, some 2" standoffs, a drill, and a dremel tool or scroll saw.
This is what I have so far on mine... Unfortunately I've accidentally cracked this one so I'm going to whip another one on monday when my new t-square and 1.5" hole saw comes in :)
http://www.fragtek.com/pix/ram-cooler-1.jpg
http://www.fragtek.com/pix/ram-cooler-2.jpg
Taa!
Hehe. Fragtek, that just looks awesome. Nice one. :clap:
:toast: nice Frag
I've got all of the measurements here logged. I'm thinking about talking to a company that could possibly mass produce a product like this for a very decent price. It doesn're require many cuts and could be done on a CNC machine in about 3 minutes time. We'll see how the final product turns out :)
good luck!! :toast:
Update:
http://www.freespaces.com/kittrulez/...yprime1651.jpg
:toast:
Runs a bit on the hot side though :(
Hey Kitt,
how much ram u running with, is it wht u got in your sig? Nice results so far, could u try priming with large FFt's as this will stress the mem controller on the CPU
Yeah, its a gig of GeIL BH-5.
Unfortunately core 0 stopped after 50 minutes :(
It needs burning in i think. Its only 4 hours old :p:
Am running large FFTs now at the same speeds/volts :)
You should test with Custom and FFTs in place changing every 15 Minutes!
thats the ultimate Test for the CPU!
heres some benchs of my opty
http://www.xtremesystems.org/forums/...4&d=1137255401
You can get a similar product here and it's cheap! :) http://www.koolance.com/shop/product...0ad3cdcbf8e167Quote:
Originally Posted by FragTek
So far so good :)
http://www.freespaces.com/kittrulez/...yprime1652.jpg
I know its not the best, but I'm happy :fact:
What is PWM IC?
How hot is too hot for it?
How do I cool it?
Kitt, your PWM IC is burning hot :eek:, u should put a fan over tht, im sure you will get more stabilty by doing this, good results by the way. From wht ive seen 2.7Ghz with the volts u putting through it is very good. Does seem to be running hot, i know if tht were my chip the IHS would be off by now, LOL.
Whats the best stepping for a 170/175?
0530 TPMW :)Quote:
Originally Posted by Slaytanic
I think you would be better of running on the 9:th multipier,Quote:
Originally Posted by Rayman
I am getting 8000+ at 2700 MHz with my TCCD at 9*300
2,5-4-3-7.
I am now at 314*9 giving mem scores in 8300 range.
Yeah that's not a bad product but poorly designed due to the fact it's made to fit multiple boards. I'd be looking at making two models, one of the Ultra-D / SLI-DR and one for the Expert with perfect fan positioning and clearance to allow for all waterblocks and phase change systems.Quote:
Originally Posted by pcdoc1
The koolance one isn't bad if you have a setup that will allow it to fit.
I'm certain your design is superior as it's specific to a particular board. I just posted it because a lot of peeps won't have the tools, patience, or the ability to make it themselves. :)Quote:
Originally Posted by FragTek
Yup, im not trying to shoot it down. Just pointing out that it's not feasible for a lot of different scenarios that us hardcore overclockers seem to get ourselves in to :)
Plus the koolance looks like it would make it difficult to water cool with .
I really think temps are holding me back with this proc. With dual prime it gets up into the 50's with the IHS off and a storm on it. I think lapping the die to get the core flatter will help drop my temps and hopefully let me clock it up. Again if anyone has seen the thread where mad mikee lapped his, his load temps are in the high 30s. Anyone interested here is a site that has lapping powders similar to what mad mikee used on his and where i ordered my abrasives:
https://www.us-products.com/sitehtml...s/compslur.php
The sample jars are $15. Will update after lapping.