E5200 overclocking - P5k/EPU + Core-Contact Freezer
siruspernot
TN
Like I said in muddocktors thread, when I got my CCF I was going to attempt to explore the e5200. Below is what I've been able to come up with so far.
Equipment
Antec 900 Case (120mm fans on medium - 200mm fan on high)
ASUS P5K/EPU P35
Intel E5200 Wolfdale 45nm
Kingston ValueRAM 2x2GB DDR2-800 @ 667Mhz
Sunbeamtech Core-Contact Freezer
Tuniq TX-2 thermal grease
Testing Software
Orthos (Small FFT torture test)
Real Temp 2.70 (Tj max set at 100)
CPU-Z
On this setup I did note that Vdrop under load was on average .0250 for all settings. The voltages I list here are the voltages after Vdrop and not what I selected in the motherboard. Just add .0250 to the voltage to get the motherboard BIOS setting.
A memory clock of 667MHz is used for each setting to maintain the magic 1:1 ratio. Your system will perform at its absolute best if you stick to this formula. Since the FSB is running at 1333MHz the memory will run at 1:1 at 667MHz. In this case underclocking is better.
Note: All listed settings and voltages were tested as stable by going for over an hour each using Small FFTs in Orthos, these are not simply bootable settings but tested stable. Your system may not respond to the same settings even using identical hardware as each setup will differ. Use these settings as guidelines to start from and what ballpark you should expect to be in.
All settings are set to auto in the BIOS for Jumperless Configuration except for:
CPU multiplier
FSB frequency
PCI-E frequency
Memory Frequency
Load Line Calibration Enabled
CPU Spread Spectrum Disabled
PCIE Spread Spectrum Disabled
Stock Intel Heatsink performance:
200x12.5 (2.5GHz) @ 1.040V: Idle = 38-36c Load 55-53c
333x9 (3GHz) @ 1.1760V: Idle = 39-37c Load 56-55c
333x10 (3.3GHz) @ 1.2V: Idle = 39-37c Load 66-66c ( stopped before it got hotter )
Summary
As you can see the stock cooler holds up pretty well with a 500MHz overclock. The was unheard of years ago. I didn't let the stress test run for very long at 333x10 because at about 2 minutes the temps were hitting 66c and this is too hot for me. 3GHz is the safe limit for every day needs on the stock cooler.
Sunbeamtech Core-Contact Freezer performace
Important Notes: First off, there isn't much information on the net about whether or not this cooler will work on the P5K/EPU motherboard. It does work but the clearances for the mounting ring are unbelievably small and the EPS ATX plug sits under the fan after it's mounted. To unplug the plug you have to remove the fan and possibly the cooler. Mine is oriented to point straight up through the 200mm top fan.
200x12.5 (2.5GHz) @ 1.040V: Idle = 38-36c Load = 47-46c
333x9 (3GHz) @ 1.1760V: Idle = 38-37c Load = 51-50c
333x10 (3.3GHz) @ 1.2V: Idle = 38-37c Load = 53-53c
333x11 (3.67GHz) @ 1.272V: Idle = 38-37c Load = 58-58c
I tried 333x12 (4Ghz) @ 1.328V but temps shot up past 65c. Anything lower than this voltage would fail orthos instantly. I'm sure with a more exreme cooling setup someone could run 4GHz as an everyday clock. I also tried 333x11.5 (3.8Ghz) but didn't like my temps being in the low 60s under load. This may be acceptable to some users.
Summary
The CCF surprisingly did little for Idle temps. I imagine this has a lot to do with my CPU IHS needing to be lapped. It is also stated that idle temps cannot be accurately judged due to the core temperature sensors being designed for detecting thermal limits and not producing accurate temp results. The Load results are great though. While I was previously unable to attempt 333x10 due to temps I was able to go beyond that with a very managable 58-58c core reading with 333x11. To apply the thermal grease I laid 3 thin lines of grease along the aluminum gaps in between the 4 heat pipes. When I tested this on my benchtop and lifted the processor, the grease had spread very evenly directly around the area of the CPU IHS. I would say the Tuniq TX-2 is a very nice product. I then cleaned them both off with alcohol and re-applied the same way in the system.
Pictures
My stable everyday overclock
This one is just for fun, it wasn't stable but I just wanted to boot at 4GHz
Thank you and have fun.
Equipment
Antec 900 Case (120mm fans on medium - 200mm fan on high)
ASUS P5K/EPU P35
Intel E5200 Wolfdale 45nm
Kingston ValueRAM 2x2GB DDR2-800 @ 667Mhz
Sunbeamtech Core-Contact Freezer
Tuniq TX-2 thermal grease
Testing Software
Orthos (Small FFT torture test)
Real Temp 2.70 (Tj max set at 100)
CPU-Z
On this setup I did note that Vdrop under load was on average .0250 for all settings. The voltages I list here are the voltages after Vdrop and not what I selected in the motherboard. Just add .0250 to the voltage to get the motherboard BIOS setting.
A memory clock of 667MHz is used for each setting to maintain the magic 1:1 ratio. Your system will perform at its absolute best if you stick to this formula. Since the FSB is running at 1333MHz the memory will run at 1:1 at 667MHz. In this case underclocking is better.
Note: All listed settings and voltages were tested as stable by going for over an hour each using Small FFTs in Orthos, these are not simply bootable settings but tested stable. Your system may not respond to the same settings even using identical hardware as each setup will differ. Use these settings as guidelines to start from and what ballpark you should expect to be in.
All settings are set to auto in the BIOS for Jumperless Configuration except for:
CPU multiplier
FSB frequency
PCI-E frequency
Memory Frequency
Load Line Calibration Enabled
CPU Spread Spectrum Disabled
PCIE Spread Spectrum Disabled
Stock Intel Heatsink performance:
200x12.5 (2.5GHz) @ 1.040V: Idle = 38-36c Load 55-53c
333x9 (3GHz) @ 1.1760V: Idle = 39-37c Load 56-55c
333x10 (3.3GHz) @ 1.2V: Idle = 39-37c Load 66-66c ( stopped before it got hotter )
Summary
As you can see the stock cooler holds up pretty well with a 500MHz overclock. The was unheard of years ago. I didn't let the stress test run for very long at 333x10 because at about 2 minutes the temps were hitting 66c and this is too hot for me. 3GHz is the safe limit for every day needs on the stock cooler.
Sunbeamtech Core-Contact Freezer performace
Important Notes: First off, there isn't much information on the net about whether or not this cooler will work on the P5K/EPU motherboard. It does work but the clearances for the mounting ring are unbelievably small and the EPS ATX plug sits under the fan after it's mounted. To unplug the plug you have to remove the fan and possibly the cooler. Mine is oriented to point straight up through the 200mm top fan.
200x12.5 (2.5GHz) @ 1.040V: Idle = 38-36c Load = 47-46c
333x9 (3GHz) @ 1.1760V: Idle = 38-37c Load = 51-50c
333x10 (3.3GHz) @ 1.2V: Idle = 38-37c Load = 53-53c
333x11 (3.67GHz) @ 1.272V: Idle = 38-37c Load = 58-58c
I tried 333x12 (4Ghz) @ 1.328V but temps shot up past 65c. Anything lower than this voltage would fail orthos instantly. I'm sure with a more exreme cooling setup someone could run 4GHz as an everyday clock. I also tried 333x11.5 (3.8Ghz) but didn't like my temps being in the low 60s under load. This may be acceptable to some users.
Summary
The CCF surprisingly did little for Idle temps. I imagine this has a lot to do with my CPU IHS needing to be lapped. It is also stated that idle temps cannot be accurately judged due to the core temperature sensors being designed for detecting thermal limits and not producing accurate temp results. The Load results are great though. While I was previously unable to attempt 333x10 due to temps I was able to go beyond that with a very managable 58-58c core reading with 333x11. To apply the thermal grease I laid 3 thin lines of grease along the aluminum gaps in between the 4 heat pipes. When I tested this on my benchtop and lifted the processor, the grease had spread very evenly directly around the area of the CPU IHS. I would say the Tuniq TX-2 is a very nice product. I then cleaned them both off with alcohol and re-applied the same way in the system.
Pictures
My stable everyday overclock
This one is just for fun, it wasn't stable but I just wanted to boot at 4GHz
Thank you and have fun.
0
Comments
I tried setting the NB strap to 400 when I tried this. Since my board is rated to do 1600FSB easy I'm left wondering where the hold up is. The only thing left could be the chip but using the Vcore setting from my 333x11 profile (1.29V) I figured 6x400 should have been an easy target. What areas should I look at for improvement?
EDIT: I've been googling this for a little while and the general consensus is that we're lucky to get 333MHz as some people can't get past 330. Others with cheaper boards get even less. Seems my only option for more performance is increasing the multi, voltage, and cooling ability. I read that the IHS on these chips are glued on and cheap Zalman TIM is used between the core and the IHS. I tried to use a disposable razor blade to separate the two but was unable. My attempt was to be replacing the TIM with TX-2 or AS5 and check for a difference. I will be lapping the IHS however as soon as I can source some proper wet sand paper. I'll post updates here
Thanks to the mods for letting me keep this journal here. Feedback is welcome.
Are you saying that intel has locked the FSB to 500 or in his case 330 to340? Or is it just a case where the performance can't go any higher.
Thanks for posting your results and being so detailed with them.
Gotchya... hrmm.. I wonder what the deal is. I mean isn't this basically the same chip as a core duo, just stripped of some it's cache and optimizations?