• 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 ***

Does your board have a 9:10 ratio for the memory, like the DFi's ??
This would bring it to 250 odd memory speed
Click to expand...
Ah, yeah thanks, actually there is. It's called a 183/200 divider, which is 0.915, which would bring the memory up to 247 or so. After posting the above I realised there is another approach I could try: Use an HTT of around 310, the 9x CPU multiplier (=2799MHz), the same 166/200 (0.83) memory divider, rely on the fact the memory controller sometimes picks a different memory divider than perhaps you expect, pump 2.8v through the memory, and loosen the CAS ever so slightly to 2.5. RAM is running 254 at 2.5-3-2-6-1T. All this has been through yet is ~50 minutes of blend Prime (memtest86 running right now) on one core and CPU Burn-in on the other, but that's a decent start as normally if the memory is at fault Prime fails within 30 minutes.

So it looks like I might be able to take advantage of the desired CPU speed (2.8) as well as a fast and tightly timed memory. Sweet.
 
Last edited:
Some regular visitors will probably already be aware of this but there is a program over at XS forums >>>>LINKY

This programme will tell you the maximum temperature that your CPU will run at without getting errors. i.e. my opteron 170 comes out with a value of 49ºc, this means if i start pumping the volts through my cpu and the temp goes above 49ºc then i will start to get errors, BSOD etc.

Just use it as a guide when overclocking with voltages as these dual core opterons run hot.
 
Boah said:
Ah, yeah thanks, actually there is. It's called a 183/200 divider, which is 0.915, which would bring the memory up to 247 or so. After posting the above I realised there is another approach I could try: Use an HTT of around 310, the 9x CPU multiplier (=2799MHz), the same 166/200 (0.83) memory divider, rely on the fact the memory controller sometimes picks a different memory divider than perhaps you expect, pump 2.8v through the memory, and loosen the CAS ever so slightly to 2.5. RAM is running 254 at 2.5-3-2-6-1T. All this has been through yet is ~50 minutes of blend Prime (memtest86 running right now) on one core and CPU Burn-in on the other, but that's a decent start as normally if the memory is at fault Prime fails within 30 minutes.

So it looks like I might be able to take advantage of the desired CPU speed (2.8) as well as a fast and tightly timed memory. Sweet.
Click to expand...

I have the exact same CPU (CCBWE 0550 408l), memory and motherboard...will be keeping a very close eye on your progress m8y ;)
 
are all you running 2.7ghz and above got yout HT set to x3?

I think I have corrupted windows :eek:

In device manager all my mobo devices have alert signs by them somehow the NF4 drivers have uninstalled themselves.

I can't work it out I played half life for 2 hrs last night without a hitch :confused:

Turned it on this morning and this happened!

I did make an image of my windows drive using v2i and then build a raid 0 array before with 2 raptors and slapped this image back on the array so maybe that has something to do with it.

I'm gonna do a clean install I think just to make sure.
 
Last edited:
Yeah my Ht is set to 3x. Have you tried installing the raid drivers again?

Not had a problem with windows corruption at all with the Nvidia raid0 setup I got.
 
Call me a noob! can I save my save games?

So i can go back to where I was after insalling windows and games again.

I'm at blacklist 5 in need for speed and would like to carry on at this point after reinstalling. :eek: :p
 
You lost me m8. I take it you can't get windows to boot up anymore? The only way I know of if windows wont boot up is to install a fresh copy of windows on a separate partition/HD then go into your program files of the corrupt version and hunt down the games 'saved games' folder. Copy it over to the new copy of windows and then install the game and copy the 'saved game' into the correct instalation path.

Oh and its a great idea to install windows etc at 'stock' settings imo and make a disk image BEFORE you start to OC.

/me shuts the door after the horse has bolted
 
Cool, I just sold my Opteron 146 CABNE 0545 BPMW (oem) on auction for £185 :cool:

So tonight I have to wap that out of my system and get it wrapped up for shipping and at long last I can get the mofo Opteron 170 0550 UPMW installed!

Gunning for 2.6GHz *Bomb-Proof* stable, anything higher is a bonus!!

large-explode.jpg

"A shot of Big.Wayne HQ during intense Opteron 170 stability testing!"
 
2.6ghz will be a breeze with that stepping.

100 % stable.

2.6 ghz won't be even pushing that chip in my experience.But you could get a dud and have a weaker core.And then you mght have to push up the volts to strngthen the weaker core.

good luck with it:)
 
Clairvoyant said:
Well Redoing the artic silver 5 on my Opty, has just lowered the temp from 55 to 54. I have the ThermalRight SI-120 & Akasa AK-183-L2B Ultra Quiet 120mm Fan.
Click to expand...
That's similar to me (52-53C if it's hot in the room in a TT Big Typhoon), after re-doing the AS5 twice, so either I'm doing something consistently wrong, or it's just the CPU. I do think the fan perhaps got just a bit too low RPM to help the temps, and in hindsight I should probably have left the Big Typhoon stock fan on.

w3bbo said:
I have the exact same CPU (CCBWE 0550 408l), memory and motherboard...will be keeping a very close eye on your progress m8y
Click to expand...
I actually created a small and simple table with all the settings and resulting speed as a reference and as a reminder of which settings to double-check when I change something ;)

I ran one pass of memtest86 before going to work this morning, with no errors, will do more testing tonight when I get home.
 
Last edited:
Hi Guys,

Whats the best stepping to get atm then - or is it still a bit hit and miss and more to do with your overall setup and ** ability to tweak? Any SPAW-like steppings?

ps3ud0 :cool:
 
Funnily enough, I don't have an Opteron but I do have a good SPAW stepping. It's an 0546SPAW doing 2.6hz on 1.35v. Anyway, sorry - you want Opti's

0550UPMW and 0550VPMW's are the good 'uns. In fact, there is a thread at Xtremesystems with the best steppings, if I find it again, I'll pos up :).

Another one which springs to mind is the 0546XPMW.
 
Hmm. I've been "struggling" to get the memory running as fast as possible and be Prime stable at the same time. 310HTT seemed to run with with Prime95 Blend on one core and CPU Burn-in on the other, however it failed when running Prime95 Blend one one and Prime95 Small FFT on the other - so that combination obviously stresses the system more. If anyone is interested, so far the table I've got down is this:

Code: 
#|HTc|HTm|HTs |CPUm|CPUv|CPUs|RAMm (div)|RAMd|RAMs|RAMv|RAMt        |
-|---|---|----|----|----|----|----------|----|----|----|------------|-
1|310| 3x| 930|  9x|1.39|2790|0.83 (166)| 11 | 253| 2.8|2.5-3-2-6-1T|Fail
2|305| 3x| 915|  9x|1.39|2745|0.83 (166)| 11 | 249| 2.8|2.5-3-2-6-1T|Fail
3|300| 3x| 900|  9x|1.39|2700|0.83 (166)| 11 | 245| 2.8|2.5-3-2-6-1T|Test
4|280| 3x| 840| 10x|1.39|2800|0.83 (166)| 12 | 233| 2.7|2.5-3-2-6-1T|Work
5|275| 3x| 825| 10x|1.39|2750|0.83 (166)| 12 | 229| 2.7|2.0-3-2-6-1T|Work
6|270| 3x| 810| 10x|1.39|2700|0.83 (166)| 12 | 225| 2.7|2.0-3-2-6-1T|Work

HTc = HyperTransport clock (CPU FSB Frequency)
HTm = HyperTransport link multiplier (K8 to NB Frequency)
HTs = HyperTransport speed (HTc*HTm).
CPUm = CPU Multiplier
CPUv = CPU Voltage (vcore/VID)
CPUs = CPU Speed (HTc * CPUm)
RAMm = RAM Multiplier (BIOS xxx/200 divider)
RAMd = Actual RAM CPU Divider chosen by the memory controller.
RAMs = Actual RAM Speed (CPUs/RAMd).
RAMv = RAM Voltage.
RAMt = RAM Timings (CAS-TRCD-TRP-TRAS-CMD).

"Work" = working as opposed to "succeeded", as none the setups really have completed as long tests as some people here prefer them to.

At the moment #3 is in testing and has run nearly 5 hours with Prime95 Blend one one core and Prime95 Small FFT on the other. No Memtest86 test completed, but this was succeeding at higher speeds, so I think it's the combination of the two tests that cause a problem (unless there's actually a conflict between running the two on the same system, which isn't impossible).

The question is of course if it's worth loosening the timings further (perhaps to CAS 3, not sure if I'd have to loosen TRP as well) to gain the extra 100MHz CPU speed and 9MHz memory speed? Or if it's worth taking another hit on the memory speed (down to 233) and run the CPU 100Mhz faster at 2.8GHz?

It's also possible to bring up the argument what is "stable" and what is "prime95 stable"? If it looks stable with Prime95Blend+CPUBurnIn and Memtest, is that stable enough?
 
Last edited:
Boah said:
It's also possible to bring up the argument what is "stable" and what is "prime95 stable"? If it looks stable with Prime95Blend+CPUBurnIn and Memtest, is that stable enough?
Click to expand...

I go on if the system is stable on every test, dual instances of prime then thats a stable system.

If I get errors its not stable IMO.

When testing I can fail prime at certain settings and yet game for hours at these settings.

The fact that my pc is not calculating 100% is enough for me to not to call it stable.

Error free prime and tie this in with error free memtest and how much more stabilty could you want?

Example I could game for hours with my ram at 270 mhz no problems what so ever.But In memtest I would get errors in test 5.

So I now run my ram on a divider at 260mhz (error free in memtest) and my cpu at 2.8ghz.

At least I know all data is getting processed correctly.

Gaming doesn't even stress the system as much as running two instances of prime.

Neither does encoding or anything else for that matter.

If you cpu can number crunch and heat up to the max and the temps are fine then thats stable.

People also go on a system being gaming stable.And doing a few 32 mb super PI's.

IMO this is not "stable"

Unless I'm error free for 12 hrs at least , thats when am I happy.

IMO.
 
Last edited:
IMO this is not "stable"
Click to expand...
I do agree, I was just very annoyed that the tests done with CPU Burn-in counted for "nothing" :)

Unless I'm error free for 12 hrs at least , thats when am I happy.
Click to expand...
But what tests do you run to be satisfied? A combination (ie 12 hour runs of different tests)?

Just curious that's all.

When I get home from work tonight, assuming the test succeeded, I might try to 280HTT/2800CPU with a 183/200 memory divider (254MHz) and try to find a timing that works.
 
Boah said:
I do agree, I was just very annoyed that the tests done with CPU Burn-in counted for "nothing" :)


But what tests do you run to be satisfied? A combination (ie 12 hour runs of different tests)?

Just curious that's all.

When I get home from work tonight, assuming the test succeeded, I might try to 280HTT/2800CPU with a 183/200 memory divider (254MHz) and try to find a timing that works.
Click to expand...

To start with ,two instances of small FFT and run memtest.
Once this is stable I run two instances of all tests combined.
 
Digital Punk said:
Just got hold of my new opty 170. :)

CCBWE 0551 UPMW

Was supposed to be OEM, but I got sent a Retail package for OEM price. :D
Sweet.

Can't wait to get clocking it.
Click to expand...

i also have that stepping, any news on its capabitys before i open it :D
 
You must log in or register to reply here.
Back
Top Bottom