So after having the new Opteron 165 and my new ram (also listed in sig) for just about two weeks now, and having run them at stock (with a few quick checks at 250 HTT and 275 HTT just for kicks) I figured it was time to start throwing curve-balls at the hardware.
The CPU and ram were both bought from NCIX.com, and apparently I've gotten a lucky break as both seem pretty damned flexible. This week/stepping of the 165 has been good to me, and feedback from a handful of others confirms the same.
I had originally decided I was going to wait for my Antec P180 case to show up - which is replacing a Lian Li PC-61 that's just a wee bit on the noisy side. The case is now two weeks in transit, and I'm bored - so let's just get to it.
As mentioned I had already tested the machine at 250 HTT with the ram running 1:1. I did this only to ensure I had a base-point for stable, and sure enough it was. Stock on everything it ran 1:1 like a champ. A few things (PWM IC and Chipset to be precise) got a little warmer than I'd like, but I can console myself with the fact that I'll have a new case and better airflow for the components very shortly.
So this morning I set my alarm a little early (6:30am) and made myself a pot of coffee. Restarted my machine and made changes to four - and only four things. HTT pushed up to 300, RAM multiplier set to 9:10 instead of 1:1 (therefore 270mhz on the ram), and bumped both the dram and vcore voltages up a single notch (CPU to 1.375, DRAM to 2.7) just to give them a bit of overhead.
Now it's pertinent to mention here that I totally forgot to set my HTT multiplier down - and it's currently set to auto. I'm not sure if my board will use the 'auto' setting to keep it within spec, or if it's currently running at x5 (which would have been stock if I was still at 200 HTT). If the former, this may be a sign that HTT multi is not quite as important as people make it out to be - if the latter, good to know, makes my job easier.
So just about a half hour later, two instances of Prime95 running stable (one on each core). I noticed CPU1 chews through the prime work much quicker than CPU0 - and I've noticed this each and every time I've ran this test, but I can't explain it.
PWM IC's a little toasty at 66C - but I've had a couple of people reassure me that the PWM's are capable of much more than that, so not to worry.
Chipset's back within reasonable levels (46C) now that I started up SmartGuardian and it ramped up the fan-speed.
CPU's at 47-48C load - which is REALLY impressing me.
Here's a screenshot.
I'm going to leave this benchmarking for a while, but may stop it when I go to work (so as not to crash the machine and prevent myself from VNC'ing in today). Once I get the new case I'll try pushing it even further - and between now and then go for a 12hr prime stable proof.
Here's a screenshot - click for larger.
The CPU and ram were both bought from NCIX.com, and apparently I've gotten a lucky break as both seem pretty damned flexible. This week/stepping of the 165 has been good to me, and feedback from a handful of others confirms the same.
I had originally decided I was going to wait for my Antec P180 case to show up - which is replacing a Lian Li PC-61 that's just a wee bit on the noisy side. The case is now two weeks in transit, and I'm bored - so let's just get to it.
As mentioned I had already tested the machine at 250 HTT with the ram running 1:1. I did this only to ensure I had a base-point for stable, and sure enough it was. Stock on everything it ran 1:1 like a champ. A few things (PWM IC and Chipset to be precise) got a little warmer than I'd like, but I can console myself with the fact that I'll have a new case and better airflow for the components very shortly.
So this morning I set my alarm a little early (6:30am) and made myself a pot of coffee. Restarted my machine and made changes to four - and only four things. HTT pushed up to 300, RAM multiplier set to 9:10 instead of 1:1 (therefore 270mhz on the ram), and bumped both the dram and vcore voltages up a single notch (CPU to 1.375, DRAM to 2.7) just to give them a bit of overhead.
Now it's pertinent to mention here that I totally forgot to set my HTT multiplier down - and it's currently set to auto. I'm not sure if my board will use the 'auto' setting to keep it within spec, or if it's currently running at x5 (which would have been stock if I was still at 200 HTT). If the former, this may be a sign that HTT multi is not quite as important as people make it out to be - if the latter, good to know, makes my job easier.
So just about a half hour later, two instances of Prime95 running stable (one on each core). I noticed CPU1 chews through the prime work much quicker than CPU0 - and I've noticed this each and every time I've ran this test, but I can't explain it.
PWM IC's a little toasty at 66C - but I've had a couple of people reassure me that the PWM's are capable of much more than that, so not to worry.
Chipset's back within reasonable levels (46C) now that I started up SmartGuardian and it ramped up the fan-speed.
CPU's at 47-48C load - which is REALLY impressing me.
Here's a screenshot.
I'm going to leave this benchmarking for a while, but may stop it when I go to work (so as not to crash the machine and prevent myself from VNC'ing in today). Once I get the new case I'll try pushing it even further - and between now and then go for a 12hr prime stable proof.
Here's a screenshot - click for larger.