• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

*** Official Opteron 170 Ocing Thread ***

w3bbo said:
Are you testing CPU or memory? If you are testing how far you can stretch your CPU you really ought to take memory out of the equation as its much simpler that way. Once you have found your max CPU(FSBxMulti) then start on the memory. Doing both is problematic when trying to find out whats causing instability.
Click to expand...

What would you say is the best way to test JUST the cpu?
 
Put your memory on the lowest divider possible (100) with slack timings, say 3-4-4-8 at 2T. Put your HT to x3 (600). I would also just use 1 stick of memory rather than 2 (dual). That more or less takes memory out of the equation although it will obviously still be used, just it wont be strained. There is no way you can test CPU alone without memory but there are ways you can test CPU with as little memory use as possible. I use Prime to test.

When using P95 or SP2004, select small fft's as this stresses the CPU with very little memory usage in comparison to blend or large fft's, you also should consider setting the priority level to high (8 or 9) to ensure that no other programs you have running in the background interfere with the test.

Assuming you have already tested to find the max FSB of your board (low multi+high FSB) then start from stock volts and slowly increase when you find prime throws out an error - retest again. Keep doing this until you feel temps are becomming a problem. You have now found your max CPU speed.

There are obviously other factors to take into consideration such as vcore fluctuation, motherboard limits etc but these can only be solved with hardware upgrades. Dunno if this is the best method but its the one I use.
 
Agent WD40 said:
W3bbo - I now know about the .5 multis. My memory has been running about 10Mhz less than it should be on the dividers so atleast I know why now... I'll up the FSB again and go down to a 9x multi.
As for temps. I'm at about 44 - 46C load at 2.9ghz with 1.44v. It tends to fail prime close to 50C.

Agent.
Click to expand...

Well whattaya know.... :rolleyes:
...I rest my case.

"Just sick and tired beeing sick and tired"
 
Easy clint / Easyrider I don't want you two kicking off again!! ;)

I want this thread to be just about clocking opty 170s. And you are both right. Some fail closer to 60C others fail cloder to 50C. Its really down to the chips.

But hay, if you wanna have a bu* fight about temps. Start a different thread... :p

Agent. :D
 
Agent WD40 said:
Hi Teggs, thanks for the swift reply.

I'm running the latest bios (29/3 I think) and the RAM is on a 166 divider, but for time time being its on a 9/10 'cos i'm testing my ram... :D

Apart from upping the Vcore, any other voltages I should think about bumping?

Agent :cool:
Click to expand...

Try using a lower multi. With my DFI and various chips (3 x Venice 3000+, Opty 144, Opty 146) I've only used the 133, 140 and 150 dividers. Any higher and I've had to sacrifice core clock which is NOT a good exchange for memory bandwidth.
 
My new PC is sitting behind me (-graphics card due to Nvidia NEVER having stock). and the 170 has this code on it.

OSA170DAA6CD
CCBWE 0609FPAW
1429464C60345


With my Zalman what do you reckon I will get overclock wise, A Google for CCBWE 0609FPAW shows some good and bad results, as long as I get stable FX60 speeds I will be happy, can't complain about getting a £700 chip. :D
 
undervolt to overclock :confused:

hi guy's its been a while! I've been having another go at getting my 170 higher than 2.8ghz, my progress is good but it doesn’t seem to make much sense to me. I started by upping my clock to see if i could get it prime stable at 2.85ghz but it wouldn’t have any of it! would only prime for an hour and a bit. I thought maybe if I upped the vcore to 1.475v that would help matters, well pretty much instantly it failed prime, dropped it to 2.8ghz again and it wouldnt even run stable, just like 2.85ghz it failed instantaneously. temps seemed fine at around 50 load. I dropped it back to 1.4v and it ran stable at 2.8ghz again like b4, I though why not have a stab at an even lower vcore. I tried 1.36 and to my amazement it ran stable at 2.85 and then 2.9ghz. but would fail 2.95mhz after about 15mins. again i tried a lower vcore of just 1.34 volts and I managed to get it to x2prime at 3ghz until it stated to get dark outside :) heres where im at...



anyone had similar experience with there dual core?? this goes against what I believe about overclocking cpu's. at one point i was stuck at 2.7ghz with this 170, nothin seems straight forward with this chip!!
 
Wow! I think thats the highest clock ive seen on such low volts either here or at xs forums. Gratz!

I think it must be down to heat, they say it really helps to keep your temps under 45 deg and yours are only 38 deg now instead of 50deg. Maybe using they chip over the last few months had burned it it somewhat though ive never seen 200Mhz extra heh.
 
Last edited:
Hesky82 said:
hi guy's its been a while! I've been having another go at getting my 170 higher than 2.8ghz, my progress is good but it doesn’t seem to make much sense to me. I started by upping my clock to see if i could get it prime stable at 2.85ghz but it wouldn’t have any of it! would only prime for an hour and a bit. I thought maybe if I upped the vcore to 1.475v that would help matters, well pretty much instantly it failed prime, dropped it to 2.8ghz again and it wouldnt even run stable, just like 2.85ghz it failed instantaneously. temps seemed fine at around 50 load. I dropped it back to 1.4v and it ran stable at 2.8ghz again like b4, I though why not have a stab at an even lower vcore. I tried 1.36 and to my amazement it ran stable at 2.85 and then 2.9ghz. but would fail 2.95mhz after about 15mins. again i tried a lower vcore of just 1.34 volts and I managed to get it to x2prime at 3ghz until it stated to get dark outside :) heres where im at...



anyone had similar experience with there dual core?? this goes against what I believe about overclocking cpu's. at one point i was stuck at 2.7ghz with this 170, nothin seems straight forward with this chip!!
Click to expand...

Wow, nice clock man. I have the same stepping as you. I am fitting a new 120.3 rad in my rig so hopefully that will help me get even lower temps at load. Gonna try the lower vcore as well. Its funny your situation sounds just like mine. . .
 
Hesky82 said:
hi guy's its been a while! .. heres where im at...



...nothin seems straight forward with this chip!!
Click to expand...

Hesky, what cooling are you using mate? And have you taken the IHS off?

I'm blatently going to try dropping the volts later, possibly down to 1.32v.

I'd laugh so hard if it works! ;)

With this logic, maybe it will do 4 ghz with .7v? :confused:

Agent.
 
Gonna try this when I've finished with this 146. Currently got it to 3gig (0602 GPAW) so my attention is here atm but this PC is soon to be sold to a mate so then I'll get back on my own PC and give Hesky's theory a try on my 170. V.Nice work m8.

Mind posting up your settings for the DFI D with the HZ sticks, just bought some and it would save me a lot of time tweaking them.

Cheers
 
Agent WD40 said:
Hesky, what cooling are you using mate? And have you taken the IHS off?

I'm blatently going to try dropping the volts later, possibly down to 1.32v.

I'd laugh so hard if it works! ;)

With this logic, maybe it will do 4 ghz with .7v? :confused:

Agent.
Click to expand...

Well the thing is I started at a relatively high vcore of about 1.41 volts which my 146 ran best at, in turn I think my dual core simply ran a tad to "warm", not hot. and this capped my overclock from the outset. From what I've found it simply hit a wall in prime if I kept the temps around 50c, not hurting the cpu but hindering its max clock, which in my case was around 2.8ghz. I never thought about lowering the vcore because I thought 50c was an ok temp for any cpu.
At the moment I've been trying to get it around 8-10c+ lower than it was running before at 2.8ghz to sort of simulate the results you get from taking the ihs off. I did this by adding 2 huge 120mm fans running flat out surrounding imo a rather rubbish zalman 9500 cpu cooler, the 120mm fans and lower vcore help it run as much as 15c lower than before!! Atm my cpu still has its hat on.
Basically I'm trying to justify removing the ihs and going wc or better air, and I think I have.
If I can keep the cpu lower than 40c on load it will prime at a much higher clock, Ive found 1.34v is the max vcore I can run until anything higher pushes it over the 40c mark. Also its worth noting core 0 for me is effected most by temp, core 1 simply goes on and on, I thought it was the weaker core but I think it is simply running at a slightly higher temp than core1.
 
w3bbo said:
Gonna try this when I've finished with this 146. Currently got it to 3gig (0602 GPAW) so my attention is here atm but this PC is soon to be sold to a mate so then I'll get back on my own PC and give Hesky's theory a try on my 170. V.Nice work m8.

Mind posting up your settings for the DFI D with the HZ sticks, just bought some and it would save me a lot of time tweaking them.

Cheers
Click to expand...

I used a trick Clint taught me to get a fast stable dram setting quicky.
what he told me....
first use "Set Optimised Default" in BIOS, THEN turn up the HTT to the rated speed of your ram with the cpu on a lower multi if need be, do this before booting into windows and then in windows write those settings from A64 Tweaker down and set them in bios manually next boot...otherwise you will never get a good start.

that did the trick for me and I'd have to say these are the key settings that worked for me once I was up and running.

Max Async Latency - 08.0 Nano Seconds (some people can run 9ns with there uccc's but my hz just wont have it)

DRAM Drive Strength - 5 (weak 3)

And try these 3 relaxed tref settings out and see which works best for you, 1168 work best for me...

0908= 100mhz(7.8us)
1032= 133mhz(7.8us)
1168= 166mhz(7.8us)

I bet Clint has found some better dram setting now though :D

I know you can go tighter with the tref setting but this allowed me to run my 146 200mhz higher and I'm pretty sure its done the same for my 170.
tbh non of my dram settings are that tight, I still havent tweaked them yet.
 
thats really interesting to know, trying out a few things with my fx60 now, on a fzr64 so still gets very hot

trying 1.328v at the moment and slowly increasing fsb, though its 46c dual prime at the moment though LOL
 
turbotoaster said:
thats really interesting to know, trying out a few things with my fx60 now, on a fzr64 so still gets very hot

trying 1.328v at the moment and slowly increasing fsb, though its 46c dual prime at the moment though LOL
Click to expand...

Take the side of your case off and get some fans blowing all around the CPU cooler. That should lower your temps if nothing else...

Agent.

P.S. Turbo, whats the stepping on your FX 60? I've got one at home still in its box and I don't know if I should open it or send it back...
 
You must log in or register to reply here.
Back
Top Bottom