The System
Case: Antec Sonata II
Power: Antec 450W SmartPower
Mainboard: Epox Ultra 9NPA+
CPU: Athlon64 Socket989 3200+ (2.0Ghz, 10x multiplier)
CPU HSF: Thermalright SI-120, 120mm Nexus fan, AS5
GPU: eVGA Geforce 7800GT
GPU HSF: Zalman ZF700-AlCu, AS-Ceramique
Memory: 2x1GB OCZ DDR PC-3200 Platinum
Hard Drive: WD 320.0 GB SATA
Not Relevant: Audigy 2 ZS, Dell 2405FPW, NEC 3540A, Logitech X-530
The Overclock
I was able to get my CPU / System to overclock about average: 2.5 Ghz (from 2.0Ghz). Temperatures are well within acceptable limits at full loads on Prime95 with no errors (voltage is auto-set at 1.60 V). I'm a little dissapointed that it didn't go much higher, but it's not a bad overclock.
I have pretty good memory (2-3-2-5), but I haven't really tweaked it much. I've played around with it some, but there wasn't an obvious performance improvement and there were too many variables. It's currently running at 250 Mhz with 2T/3-4-6-3 and an auto-set voltage of 2.76 V. Do you think that further tweaking could result in a >5% performance gain (in gaming, not synthetic memory benchmarks)?
My GPU performed rather exceptionally. The core easily reaches 509mhz (from 400mhz) and the memory reaches 1.20Ghz (from 1.0Ghz) on stock voltage. Temperatures are also well within acceptable ranges at full load with rthdribl.
Some Comments
I started by going through the "isolation" process described in the Quick & Dirty A64 Overclocking guide, but didn't follow it rigorously. Is the chipset ever the limiting factor? It was useful to remove the memory from the process -- I forced it to a 1:2 multipler and tested the CPU with Prime95 at increasing clockspeeds until errors occured.
The GPU is a bit more unique. The "auto-detect" function actually places the clock speed higher than what I decided was stable (it puts it in the 520mhz range!) And, interestingly enough, I can run rthdribl without problems for hours on end at that range. However, even though rthdribl brings the GPU to it's max temperature, 3dMark05 will freeze after several loops even when rthdribl doesn't show any problems. Any idea why?
The Benchmarks
These are all rounded because there's too much variation to give it down to the specific digit. Format is: Overclocked (vs. Stock System).
3dMark05: ~8,200 (vs. 6,700)
3dMark03: ~17,000
3dMark01: ~26,300
Aquamark: ~82,000 (vs. 69,400)
How do these compare to your scores if you have a similar setup? Any suggestions or thoughts?
Case: Antec Sonata II
Power: Antec 450W SmartPower
Mainboard: Epox Ultra 9NPA+
CPU: Athlon64 Socket989 3200+ (2.0Ghz, 10x multiplier)
CPU HSF: Thermalright SI-120, 120mm Nexus fan, AS5
GPU: eVGA Geforce 7800GT
GPU HSF: Zalman ZF700-AlCu, AS-Ceramique
Memory: 2x1GB OCZ DDR PC-3200 Platinum
Hard Drive: WD 320.0 GB SATA
Not Relevant: Audigy 2 ZS, Dell 2405FPW, NEC 3540A, Logitech X-530
The Overclock
I was able to get my CPU / System to overclock about average: 2.5 Ghz (from 2.0Ghz). Temperatures are well within acceptable limits at full loads on Prime95 with no errors (voltage is auto-set at 1.60 V). I'm a little dissapointed that it didn't go much higher, but it's not a bad overclock.
I have pretty good memory (2-3-2-5), but I haven't really tweaked it much. I've played around with it some, but there wasn't an obvious performance improvement and there were too many variables. It's currently running at 250 Mhz with 2T/3-4-6-3 and an auto-set voltage of 2.76 V. Do you think that further tweaking could result in a >5% performance gain (in gaming, not synthetic memory benchmarks)?
My GPU performed rather exceptionally. The core easily reaches 509mhz (from 400mhz) and the memory reaches 1.20Ghz (from 1.0Ghz) on stock voltage. Temperatures are also well within acceptable ranges at full load with rthdribl.
Some Comments
I started by going through the "isolation" process described in the Quick & Dirty A64 Overclocking guide, but didn't follow it rigorously. Is the chipset ever the limiting factor? It was useful to remove the memory from the process -- I forced it to a 1:2 multipler and tested the CPU with Prime95 at increasing clockspeeds until errors occured.
The GPU is a bit more unique. The "auto-detect" function actually places the clock speed higher than what I decided was stable (it puts it in the 520mhz range!) And, interestingly enough, I can run rthdribl without problems for hours on end at that range. However, even though rthdribl brings the GPU to it's max temperature, 3dMark05 will freeze after several loops even when rthdribl doesn't show any problems. Any idea why?
The Benchmarks
These are all rounded because there's too much variation to give it down to the specific digit. Format is: Overclocked (vs. Stock System).
3dMark05: ~8,200 (vs. 6,700)
3dMark03: ~17,000
3dMark01: ~26,300
Aquamark: ~82,000 (vs. 69,400)
How do these compare to your scores if you have a similar setup? Any suggestions or thoughts?
Facebook
Twitter