Opteron 144 CABNE 0544 EPMW

[thesurge]

Nice OC. What RAM?

 

[rahzel]

these CABNE 0544's seem to be hit and miss. i need 1.58v to run stable at 2673mhz, and i see others at 2.7 with just over 1.4v. i dunno how much you guys have worked in your processors, but mines a week old.

also, my DFI Ultra-D NF4 seems to be limiting me past 297 HTT which is sad for a DFI board.

 

[Triggerhappy007]

Originally posted by: rahzel
these CABNE 0544's seem to be hit and miss. i need 1.58v to run stable at 2673mhz, and i see others at 2.7 with just over 1.4v. i dunno how much you guys have worked in your processors, but mines a week old.

also, my DFI Ultra-D NF4 seems to be limiting me past 297 HTT which is sad for a DFI board.

Is your 0544 EPMW or BPMW? I think the B's don't OC as well. Maybe your PSU is not strong enough.

thesurge, I'm using this RAM. Got it for $93AR. Max it will go is 290MHz on Auto (3-4-3-7-1T), not bad for $93.

 

[rahzel]

Originally posted by: Triggerhappy007

Originally posted by: rahzel
these CABNE 0544's seem to be hit and miss. i need 1.58v to run stable at 2673mhz, and i see others at 2.7 with just over 1.4v. i dunno how much you guys have worked in your processors, but mines a week old.

also, my DFI Ultra-D NF4 seems to be limiting me past 297 HTT which is sad for a DFI board.

Is your 0544 EPMW or BPMW? I think the B's don't OC as well. Maybe your PSU is not strong enough.

thesurge, I'm using this RAM. Got it for $93AR. Max it will go is 290MHz on Auto (3-4-3-7-1T), not bad for $93.

its an EPMW too, and i have a 450w OCZ Modstream which should be strong enough.

 

[TrueWisdom]

rahzel--I'm in your boat. I can't get stable at 2700MHz even all the way up to 1.57v. I fail prime in less than five minutes. I run at 2400 with only 1.45v, but considering how other people are doing with this chip, I'm really disappointed. My Venice 3000+ got the same OC on higher voltage. I just don't have good luck, I suppose.

 

[TrueWisdom]

Actually, rahzel, check out this thread: http://www.dfi-street.com/forum/showthread.php?t=26642

Apparently large FFTs have gone from failure to success after lowering the RAM timings... since I already have RAM that is very picky with my board (OCZ PC3700 Rev. 2) I may work on tweaking those timings a bit more. I am a bit skeptical, as increasing my CPU volts seemed to make a difference in the time it took for my system to crash, but I gotta keep hope alive.

 

[thesurge]

i dont really know that much about oclocking yet but why dont you just run a divider. AMDs arent that dependent on fast ram.

 

[TrueWisdom]

The divider doesn't always fix the problem--since the memory controller of the A64 is on-chip, sometimes at high frequencies the timings need to be loosened regardless of what the speed of the RAM is. (At least that has been my experience thus far.)

 

[chexi]

I'm going to play more with RAM settings to see what the true cpu max, but I'm happy running at 2.4 - 2.5 with cpu temps under no load lower than my ambient case temps! At first I thought it was a misreport, but both the bios and Asus progam confirm that I'm usually 2 degrees cooler on the CPU than on the mobo. Under heavy load I barely crack 40c, and that means I can start finding ways to make this machine quieter, something more important to me than top speed considering it could do everything I needed it to do, including games, at 1.8ghz.

 

[rahzel]

Originally posted by: TrueWisdom
Actually, rahzel, check out this thread: http://www.dfi-street.com/forum/showthread.php?t=26642

Apparently large FFTs have gone from failure to success after lowering the RAM timings... since I already have RAM that is very picky with my board (OCZ PC3700 Rev. 2) I may work on tweaking those timings a bit more. I am a bit skeptical, as increasing my CPU volts seemed to make a difference in the time it took for my system to crash, but I gotta keep hope alive.

ill give it a try. i was thinking it was either the memory/memory controller, or i maxed out my HTT. I was leaning towards my memory, because once i set my OCZ Plat memory to DDR400 using a divider, i was able to raise my HTT by 6-7 stable. Before i could only run 291*9 stable, but now with a 2/3 divider, i can run 297 stable.

im running 4 sticks of memory, so i have to use 2T anyway, but i set my read preamble/max asynchronous latency to 9/7.5 (thats what your link says opterons are most sensative to). so far so good, i can atleast BOOT into 300x9. before i would get BSOD as soon as my AV guard loaded.

thanks for the link.

 

[TrueWisdom]

No problem--it made a slight difference for my CPU as well, but ultimately I think I got a dud. I can't get prime stable at 300x9 even at 1.58v... I may give 1.6v a shot but I don't think I'm comfortable going that high. My 275x9 is rock-solid at 1.47v, (maybe even lower, I'll check later on) but I don't think I'm going to get to 2.7GHz on this chip, which is really upsetting.

 

[chexi]

Well, I was able to get 2.7 ghz (301 x 9) at 1.55 v stable (at least for the hour I ran prime this morning). The key for me was either switching my RAM timing to 2T or putting the divider down to 50% (200 on the Asus a8n-SLI Deluxe), I did both, so I haven't isolated which one yet. I'm still dumbfounded as to why the RAM with no divider running at 200/400 can run 1T with no problem, but if I up the CPU and use the 1/3 divider that results in virtually the same (slightly lower 198/396) RAM speed, I am not stable or won't boot. Nevertheless, at this point it appears to be either a RAM issue or a motherboard RAM divider issue. The cpu seems to be able to handle at least 2.7 ghz. I will do further testing tonight and report back.

Oh, Temp at max load is 49 C. Also, I put in 1.55 v out of the box, so lower voltage may very well work. I'll find that out tonight too.

 

[rahzel]

same here, it helped a little, but its still not stable at 300x9, even with 1.6v, i can only run SP2004 for about an hour. ive also palyed with several different memory timings and using different dividers, but still not stable at 300x9, but atleast i dont get BSOS's anymore. hopefully it just needs to be worked in a little more, as its only 1.5 weeks old still. I guess we got duds, because ive heard a few peopel with the same steppings able to reach 300x9 out of the box, with lower voltage =\. im ALWAYS unlucky when i buy computer hardware, i always get poor overclocking hardware =[.

i heard that opterons have BETTER memory controllers, but it seems like they have WORSE memory controllers (compared to A64's).

 

[thesurge]

so i guess its hit or miss.

 

[chexi]

I tested it a bit a lunch today and believe that I isolated the problem to the divider. When I upped it to a .66 divider (for 198/396), I crashed out of prime in about 2 minutes. Going from 1T to 2T did not crash prime during the test (which was only 40 minutes). That said, the only things left to do are to go back to older revisions of the bios or to change the CPU speed and multiplier and try to get the best combination of settings. I can't remember if I have half stepping multipliers in the bios, so I might be trying 7 x 400 at HT x 2 or 8 x 350 at HT x3 and see if my HT can handle the extra 50, which I highly doubt. It's times like this I really wish I had the 146 and the 10x multiplier! :|

 

[benbaked]

i have this stepping, out of the box i ran it at 1.8g and default vcore for 18 hours folding...then jumped the vcore to 1.5 and it runs at 2.7ghz fine. prime smallfft and largefft and f@h stable. it will run at 311x9 with 1.5v but is not prime stable and bsod's after some decent usage.

not the best stepping of course, but not bad, even after the online retailers jacked up the prices

 

[thesurge]

My goal is to run FSB at 270 stock vcore.

 

[rahzel]

my goal is to run at 2.7Ghz (atleast). when i bought it, i was hoping for 2.8 =\. the only reason i bought an opteron is because i could get better performance for paying little to none compared to the venice 3200 i had, which maxed out at 2.56ghz. so now i only gained 113mhz and 1m cache vs 512k.

 

[TrueWisdom]

Well, I've been tweaking it some more, and I'm running at 2.475GHz now at 1.4v. (Have to get home to see if it passed prime95.) So, that's actually a big improvement for me, as 2.4GHz at 1.4v is a blessing. I'm going to push up a bit more slowly after this (maybe try 2.5GHz and 2.6GHz) and see what my max overclock is while staying under 1.5v. My Venice hit 2.4GHz but I had to throw too much voltage at it; maybe this chip won't run at 2.7 GHz :sobs: but if it runs at 2.6 at 1.4v or so, I'd be a happy camper.

And rahzel, I totally sympathize with you. I have the same terrible luck. We should form a club.

 

[TrueWisdom]

Actually, rahzel, I just had a thought--try taking two sticks of your RAM out, and populating only the 2nd and 4th slots. See what that does for your OC?

 

[rahzel]

well i tried before, but i didnt tweak my memory timings at all. I think i was stable at 300*9 with only 2 sticks of memory, i dont remember if i primed it. I'm trying to sell my 4x512mb OCZ PLatinum rev.2 memory, so i can get a 2x1GB kit.

 

[Maluno]

OK, I've been neglecting this thread for a while... I was able to hit 2610 Prime stable, as I said before, but more volts did not really help at all... The thing is, I was priming it at 2610 when my ram was running the lowest divider possible, so it was running at about 100 something MHz. When I try to go 290 x 9 and set the ram on a 133 divider, my board goes into safe mode, and resets the HTT to stock 200. I'm using the Epox 9NPA+ Ultra, for reference, and I have 2 Adata Vitesta 1Gb sticks I got for $155 on sale.

I tested the ram and it is good up until about 210 at stock timings of 3-4-4-8 1T, but with the divider with a higher HTT, I have no luck. So, I'm still messing around, I'm gonna try 2T like some of you said helped, (though I really can't relax the timings any more because they are already really loose; this WAS value ram...).

Will report back later!

 

[TrueWisdom]

rahzel--

Do that and tweak the timings. I know for a fact 4xanything with the A64 memory controller causes problems. I'm not saying you have to run your machine like that, but if you do it and it works then you'll know that the RAM and not the CPU is holding you back.

 

[Maluno]

I can't seem to get my ram to cooperate, if I try anything above the DDR200 divider, which makes it 145MHz while CPU is 9 x 290, my motherboard goes into safemode and resets the HTT to 200.

I was able to tighten the timings to 3-3-3-8 1T at 145.. I have the Dimm at 2.9v. I am confused what to try next... ideas?

I see that we are all having the same divider problem. Maybe the memory controllers are faulty on these new 144s. I hope we resolve this issue, though!

 

[TrueWisdom]

Have you tried fiddling with the Max Async/Read preamble settings? The Opterons are VERY picky with those two settings--check my link a while earlier for some tips. 1T is also a possible source of errors for you, so drop it to 2T and see where you end up. Also relax to 3-4-4-8--really, say to hell with the RAM for now, because once you figure out where your CPU is stable, you can slowly bump up RAM timings until memtest or prime fails.

Also, an update on my chip--I can run 2.4GHz at 1.39v stable, so that makes me much happier. Getting to 2.565GHz (285x9) took me up to around 1.47v, and right now I'm seeing if that will hold me into 290x9. So, 2700 may be out of reach, but if all else fails, running 2.4GHz at essentially stock voltage makes this chip much more appealing to me.