I'd like some feedback on my temps/voltages on this OC.
I'm in the process of tuning/testing/benching the setup listed in my sig, with the core components of an E6420, P5K Deluxe Wifi, and Ballistix PC2-8500 Ram (4x1GB) in an Antec 900.
Cooling is Ultra 120 Extreme with 2xScythe F in push/pull config plus other various fans in the case (SilenX 90 CFM blowing out on back slot). For testing/benching purposes, had all fans set to max speeds. Bios version: 0501 with some commonly tweaked items enabled/disabled to help with stability (Spread Spectrum, etc.) but non-Vcore/DRAM voltages left on Auto.
Before I got the U120X, stock speed (2.13) and stock HSF idle temps were about 33 c and 44 under load.
I started out a little disorganized, but jumped right in and after a little random tweaking hit a tentative target of 3.6 @ 8x450 with 5:6 (PC1080). It was stable for at least 30 minutes in Orthos and ran 3DMark06 fine (I know that doesn't mean it's really stable, but it's a start). Temps were over 60 c though and it took 1.55 volts Vcore to get there. I'm not sure I want to run 24/7 that hot, even if it ends up running Orthos/Memtest stable for a few days, and want to find a good heat/performance balance.
Before bed, I found what seemed like a good balance at 3.52 (8x440) at 1.45 vcore and running the RAM at 4:5 for 1100 mzh. Orthos ran stable overnight and Memtest86+ ran 3 passes with no errors, but temps topped out at 60c under load.
So I went back and started a more systematic approach, removing the DRAM as a limiting variable by going 1:1 and setting the DRAM voltage to 2.2 so I knew that wouldn't be a likely limiter for a long time up the FSB testing chain.
I started at 375 FSB and started working my way up. I'm planning to get a stable clock first and then adjust RAM timings later then compare OC'ing RAM vs. 1:1 with tighter timings. Following are some results I've noted so far and I'm curious if these temps at these higher voltages look right or if I may need to look into changing some other settings for a solid base clock at a reasonable temp. Unless noted, these are the minimum settings it took to load Windows and run a quick Orthos test for 5 minutes:
Speed - FSB --- Vcore/DRAM - Idle - Load
2.13 -- 8x266 - Auto/Auto --- 27 - 36
3.00 -- 8x375 - 1.30/2.20 --- 30 - 41
3.20 -- 8x400 - 1.30/2.20 --- 31 - 43
3.40 -- 8x425 - 1.35/2.20 --- 33 - 48
3.52 -- 8x440 - 1.45/2.20 --- 34 - 60 *(4:5 - 1100)
3.60 -- 8x450 - 1.55/2.20 --- 36 - 64 ** (5:6 - 1080)
* Orthos stable 8 hours and Memtest86+ stable for 4 hours (manually stopped both - no failures).
** Only 30 minute run of Orthos - fully stable for that time (manually stopped).
I'm concerned about the FSB increases above 425 resulting in such high voltage requirements and the higher temps.
Are these within expected ranges, or should I look at the seating of the HS? I used about a full rice grain worth of AS5 centered on the IHS and let the HS spread it on contact.
I'll be "OK" with running @3.4 for everyday use, but if I can push it a bit more up to 3.6 with better heat results, I'd be happier. I'm fine with changing multis, but would the higher FSB actually require more vcore and result in a higher NBridge core/possible instability? I can post some pics of my BIOS settings later at home, too.
Is there something I may be missing?
I'm in the process of tuning/testing/benching the setup listed in my sig, with the core components of an E6420, P5K Deluxe Wifi, and Ballistix PC2-8500 Ram (4x1GB) in an Antec 900.
Cooling is Ultra 120 Extreme with 2xScythe F in push/pull config plus other various fans in the case (SilenX 90 CFM blowing out on back slot). For testing/benching purposes, had all fans set to max speeds. Bios version: 0501 with some commonly tweaked items enabled/disabled to help with stability (Spread Spectrum, etc.) but non-Vcore/DRAM voltages left on Auto.
Before I got the U120X, stock speed (2.13) and stock HSF idle temps were about 33 c and 44 under load.
I started out a little disorganized, but jumped right in and after a little random tweaking hit a tentative target of 3.6 @ 8x450 with 5:6 (PC1080). It was stable for at least 30 minutes in Orthos and ran 3DMark06 fine (I know that doesn't mean it's really stable, but it's a start). Temps were over 60 c though and it took 1.55 volts Vcore to get there. I'm not sure I want to run 24/7 that hot, even if it ends up running Orthos/Memtest stable for a few days, and want to find a good heat/performance balance.
Before bed, I found what seemed like a good balance at 3.52 (8x440) at 1.45 vcore and running the RAM at 4:5 for 1100 mzh. Orthos ran stable overnight and Memtest86+ ran 3 passes with no errors, but temps topped out at 60c under load.
So I went back and started a more systematic approach, removing the DRAM as a limiting variable by going 1:1 and setting the DRAM voltage to 2.2 so I knew that wouldn't be a likely limiter for a long time up the FSB testing chain.
I started at 375 FSB and started working my way up. I'm planning to get a stable clock first and then adjust RAM timings later then compare OC'ing RAM vs. 1:1 with tighter timings. Following are some results I've noted so far and I'm curious if these temps at these higher voltages look right or if I may need to look into changing some other settings for a solid base clock at a reasonable temp. Unless noted, these are the minimum settings it took to load Windows and run a quick Orthos test for 5 minutes:
Speed - FSB --- Vcore/DRAM - Idle - Load
2.13 -- 8x266 - Auto/Auto --- 27 - 36
3.00 -- 8x375 - 1.30/2.20 --- 30 - 41
3.20 -- 8x400 - 1.30/2.20 --- 31 - 43
3.40 -- 8x425 - 1.35/2.20 --- 33 - 48
3.52 -- 8x440 - 1.45/2.20 --- 34 - 60 *(4:5 - 1100)
3.60 -- 8x450 - 1.55/2.20 --- 36 - 64 ** (5:6 - 1080)
* Orthos stable 8 hours and Memtest86+ stable for 4 hours (manually stopped both - no failures).
** Only 30 minute run of Orthos - fully stable for that time (manually stopped).
I'm concerned about the FSB increases above 425 resulting in such high voltage requirements and the higher temps.
Are these within expected ranges, or should I look at the seating of the HS? I used about a full rice grain worth of AS5 centered on the IHS and let the HS spread it on contact.
I'll be "OK" with running @3.4 for everyday use, but if I can push it a bit more up to 3.6 with better heat results, I'd be happier. I'm fine with changing multis, but would the higher FSB actually require more vcore and result in a higher NBridge core/possible instability? I can post some pics of my BIOS settings later at home, too.
Is there something I may be missing?