Another Opteron 165 Thread 0547 tpmw

[mrkun]

Originally posted by: Lark888
One other question! Why do some individuals select a VID*overvoltage factor versus the same voltage available a direct VID selection?

Some people have found it to be more stable than just setting the desired voltage directly.

 

[lopri]

Originally posted by: Chumster
Ambient temperature sits right around 20C, though my wife has been known to bump the temp from time to time. I may consider going higher, but my next step is coaxing my RAM to run at 9/10 (~270Mhz). BTW, lopri, you'd apprecaite my current setup.

What a classy setup you've got there. A level higher than mine!

Does that 3-blade fan blow ANY air? It's amazing that you can do away with that fan with this burning dual-core CPU.

 

[lopri]

Oh and the *magic* worked.
http://img29.imageshack.us/my.php?image=lopri2779smallfft18hrs1xf.jpg

 

[OatMan]

I have this same stepping and week sitting sealed in a retail box. It keeps calling to me, but I should be happy with my current 165 that runs 2.6 at 1.50v stable. As in fascist DFI-Street stable.

BTW memtest errors can often indicate incompatibility issues between perfectly good components, not just simply "Bad Ram". I've had enough evidence swapping hardware among multiple systems to convince myself of this.

2.7 at 1.4v is insane. I'm doing air cooling, high end air should be safe up to 1.55v.

I just hope I don't screw up my OC by swapping my beloved BH5 for Ballistx...

anyhoo nice OC you have sounds like you have a golden chip there.

 

[lopri]

Originally posted by: OatMan
As in fascist DFI-Street stable.

How do you mean?

 

[mrkun]

Originally posted by: lopri

Originally posted by: OatMan
As in fascist DFI-Street stable.

How do you mean?

Read the requirements.

 

[Lark888]

Funny chip. Runs 36 hrs Prime95 dual cores at 302*9 (2720 Mhz) and 1.45v. Bump it up to 304 in bios at 1.45v and it won't boot to windows. If I change the HTT using ClockGen, it will run at 311 Mhz at this voltage but error out on Core0 after 20 minutes or so.

I am impressed with the Opteron as much as my old Celeron 300a chips. A 50% overclock is beyond my overclocking expectations on Air cooling for most chips these days.

The smoothness of Dual core cpu at this speed is delightful. Even running dual prime95s in the background, web surfing is glichless. That includes having all the extra stuff running at the same time (Anti-virus, firewall, spysweeper, MBM5...etc.)

I'm settled in on 2720 Mhz and will probably stop pushing cpus any harder for a while. Thanks to the OP and other posters.

 

[lopri]

Congratulations on your great OC, Lark. Have you tried one of those newer HSFs? (Zalman 9500, TT Big Typhoon, Scythe Ninja, etc.) With lower temps, you'll probably run the same speed with lower vcore, or run a bit faster with same vcore. These chips seem to be very sensitive to temperatures.

 

[nealh]

Jeez..I am having such strange problems using Dual Prime 95 and my opterons...I had a similar issue with both
165@2610. ..ran dual prime 95 for 12hrs then used for a few days and reran and it failed..changed a memory timing to slighlty looser(rolled back to backup image) and small ffts passed 15hrs twice prior to stopping

so I have an opteron 170@2700 1.5v..ran 15hrs of small fft no issues at 1.5v...large fft failed at 12.5 hrs..so I loosened timings abit and reran nextday 15hrs just fine

anyway the next day I thought I had made a bios change to 2750 but in reality I did not save it..so I was rerunning dual large fft Prime 95 at 2700/same settings that the day before worked for 15hr..crashed after 4hrs on core 1..which is usually a memory issue from my experience, since cpu 0 general fails in small fft when I push the CPU speed

so WTF..why were settings stable for 15hrs failing the next day...I recheckd bios and everything seems ok...so I retried last night now I am failing at 1hr

Mind you..I am getting no BSOD, no errors or program crashes etc...

I have no idea what the issues is now...I had placed Opty 170 in the database because I felt 15hrs small fft/large fft was good to go

so I am rolling back to "originial image" that seemed allow things to be ok after issues on my 165...

Just not sure whether so driver is getting corrupted after repeated Prime 95 use and screwing around...Not sure if it is video driver issue(as I have felt 5.13 cat with ccc caused some issues so I used 5.13 with old CP)

Very odd situation..I fell confident the cpu which has passed several small fft runs at 2700 1.5v is ok..but seems the ram-cpu is an issue..although I have run large fft twice for over 12hrs...

I am so sick of this testing inconsistencies....

anyone else experience crap like this......

 

[mrkun]

Originally posted by: nealh
Jeez..I am having such strange problems using Dual Prime 95 and my opterons...I had a similar issue with both
165@2610. ..ran dual prime 95 for 12hrs then used for a few days and reran and it failed..changed a memory timing to slighlty looser(rolled back to backup image) and small ffts passed 15hrs twice prior to stopping

so I have an opteron 170@2700 1.5v..ran 15hrs of small fft no issues at 1.5v...large fft failed at 12.5 hrs..so I loosened timings abit and reran nextday 15hrs just fine

anyway the next day I thought I had made a bios change to 2750 but in reality I did not save it..so I was rerunning dual large fft Prime 95 at 2700/same settings that the day before worked for 15hr..crashed after 4hrs on core 1..which is usually a memory issue from my experience, since cpu 0 general fails in small fft when I push the CPU speed

so WTF..why were settings stable for 15hrs failing the next day...I recheckd bios and everything seems ok...so I retried last night now I am failing at 1hr

Mind you..I am getting no BSOD, no errors or program crashes etc...

I have no idea what the issues is now...I had placed Opty 170 in the database because I felt 15hrs small fft/large fft was good to go

so I am rolling back to "originial image" that seemed allow things to be ok after issues on my 165...

Just not sure whether so driver is getting corrupted after repeated Prime 95 use and screwing around...Not sure if it is video driver issue(as I have felt 5.13 cat with ccc caused some issues so I used 5.13 with old CP)

Very odd situation..I fell confident the cpu which has passed several small fft runs at 2700 1.5v is ok..but seems the ram-cpu is an issue..although I have run large fft twice for over 12hrs...

I am so sick of this testing inconsistencies....

anyone else experience crap like this......

I experienced similar randomness with Prime when I was running my 3000+ Venice. In my case (no pun intended), it was also memory related. I finally got my RAM stable at the current settings in my signature (been too lazy to try OCing the RAM anymore than what it was on my previous CPU), but before that I could run Prime three different times with the same settings and it would fail within anywhere from one minute to four hours. The RAM would always pass Memtest though, so it seems like this was a memory controller issue.

 

[nealh]

experienced similar randomness with Prime when I was running my 3000+ Venice. In my case (no pun intended), it was also memory related. I finally got my RAM stable at the current settings in my signature (been too lazy to try OCing the RAM anymore than what it was on my previous CPU), but before that I could run Prime three different times with the same settings and it would fail within anywhere from one minute to four hours. The RAM would always pass Memtest though, so it seems like this was a memory controller issue.

the odd thing is this memory will do 250mhz at 3-3-3-7-1t with my venice at 2750, but my opty both needed loosened timings

It seems the memory controller I am guessing is really get stressed when oced so the ram while stable at high speeds on single core really needs some loosening to work with the DC...also I have noticed as I tweak the ram and it works better with the cpu the system seems just as crisp and fast(if not even more so) not that things are "optimized"..I clearly see no difference with looser timings

But the odd thing just not sure why it would work one day and fail the next....mind it passed 15hrs(not 1hrs or even 4hr but 15hr and I stopped it it was not a crash on Prime 95)

 

[lopri]

nealh, I had so many occasions like yours, it's not even funny. (Although my issue was SuperPI)
One day it passes whatever I throw at it (1M, 2M, 8M, 32M, anything), and the next day fails on everything. Now I gave up on extracting most out of my memory. I just run it @relaxed timing. It's gotta be memory controller, or motherboard, or memory itself. Very annoying indeed.

 

[nealh]

Originally posted by: lopri
nealh, I had so many occasions like yours, it's not even funny. (Although my issue was SuperPI)
One day it passes whatever I throw at it (1M, 2M, 8M, 32M, anything), and the next day fails on everything. Now I gave up on extracting most out of my memory. I just run it @relaxed timing. It's gotta be memory controller, or motherboard, or memory itself. Very annoying indeed.

I would be happy to just find the timing now that works with DC opty...I thought I had it but again even after rolling back it died 9hrs into large fft last night


so I am going to try another setting...very frustrating indeed

all I want now is just a stable setup...

Wondering if I need to go back and rerun small fft again too

My ram is crucial valuram(not binned ballistix, missed that ) ran 250mhz with my venice at 3-3-3-7-1t with auto everything, level 8, level 2, 8x

now I have passed 225mhz 3-3-3-8-16-1t auto everything except level 6, level1, 8x. this was able to run 12 and 15 hrs of large fft...but then failed
so changed to 3-3-3-8-16-1t, auto everything except level 6, level1,8x, and tried 16 and 256 on idle cycle

so now I will set to 3-3-3-10-16-1t..auto everything except level 6,level 1, 8x,16 idle cycle timer

I would except it was the memory but it worked with same except setup except I had a venice before only difference is swapped cpu....

so I believe it is the opty.. when cpu speed is maxed memory controller is a bit "weak" and needs some help but this is a pain to figure out