LeonardoWake up and smell the glaciersEagle River, AlaskaIcrontian
edited February 2006
PART III, Opening Act
Last night I assembled the computer with a new (refurbished) P5DW2. Overall, it's a nice board layout. It was satisfying to see how thick the PCB was - definitely an enthusiast's board. First step was to pull the north and south bridges off and apply Arctic Silver to the chips. I also added a fan to north bridge. (775 boards' north bridges get very, very hot under overclocking.) The south bridge sink was not completely flat and would not leave an imprint on the sink's bottom of the Arctic Silver. The south bridge doesn't get all that hot anyway, so I just re-applied some thick viscosity silicone paste. Ought to work well enough. The board has EPS12v capability, so I had already prepared for this by modifying my PSU with an additional 4-pin plug in tandem with PSU's existing plug. No problem. The P5WD2 is stock with a decent mosfet heatsink. I removed that too in order to apply Arctic Silver. I changed my mind after seeing the horseshow array of tiny mosfets. It would have been very difficult to apply AS to all the mosfets without stray paste ending up on the exposed electircal solder-traces. Maybe I'll get to that some other time. I had to make two modifications to get the Big Tyhpoon to mount on the board. First I had to file down a protrusion on the end of the top-side heatsink retension plate, as it was abutting against the mosfet heatsink. No problem - only about ten minutes of work with a file, the plate firmly held in a bench vise. Second, there were through-the-board-solder terminals on the bottom side of the board. Some of these were about four or five millimeters in length. Simple - just snipped them off with wire cutters. OK, everything was mounted. Popped the motherboard in the case, connected the cables, and voila - nothing. Hmm. Never did quite figure out what was wrong. Perhaps I had mispositioned the front panel's power switch lead or reset lead. But, after removing all the cables and reattaching, the computer fired right up.
I only spent about 45 minutes tweaking last night, so results so far are just preliminary. I figured this board was at a minimum just as capable as my previous P5PL2 was maxed out. I was right. After testing at BIOS stock settings for voltages, FSB, and CPU, I cut to the chase. The first OC I tried was a 700MHz FSB boost. 3.5GHz was instant and flawless. Just two more bumps before I went to bed, the last being at:
3700MHz
mulitplier 14
FSB 1056 (264 quad)
memory divider 3:4 yielding 353MHz (706MHz DDR2).
It ran all night at these settings running two instances of Folding.
Early motherboard observations:
- very resilient
- very smooth
- extensive tweak settings
- very stable voltages, little fluctuation under OC + full load
- I'm getting the same OC or better than before with less vCore, or at least less indicated vCore
My goal was a stable 1000MHz overclock. It's only 100MHz away. I should have time for more tweaking and testing this evening.
Leo, I want to prep my Asus A8R-MVP before setting it up. Do you think I should apply Arctic Silver 5 to the north/southbridge sinks as well?
I can never figure out how to get those little white plastic spring clips off without destroying them (or scratching the hell out of the motherboard, which scares me terribly )
0
LeonardoWake up and smell the glaciersEagle River, AlaskaIcrontian
edited February 2006
Yes and no. Northbridge - yes, do apply premium thermal paste. I doubt the ATI northbridge will heat up as much as the Intel 945P or 955X, but it probably will still get quite hot, especially if you overclock. Speaking of which, should you choose to overclock, a fan on top of the northbridge sink would be advisable. I've long been an advocate of passive cooling, but I can't do that on these Socket 775 northbridges without a massive, passive sink. The soutbridge will probably do just fine without premium paste. The current draw through the southbridge under CPU FSB overclocking is not nearly that of the northbridge. Best advice is to leave the case side cover off when the machine is under full load or overclocked, to reach and feel the north and southbridges. If they feel hot, there may not be much thermal headroom left, leaving the computer liable to instability. My (now RMA'd) Asus P5PL2's north AND southbridge's got quite hot. I had AS on both of them, as well as a fan on both of them. But then, that's the only mobo of any brand I've used where the southbridge got hot like that.
You are right. Those springloaded plastic retention thingies are never as taught as original when you reinstall them. Often I just put them in the parts bin after removal. Plastic cable ties, one head on top of the sink's hole, with another tie cinching from the backside of the motherboard works very well. The black ones are hardly noticeable. You can also use small nuts and bolts with nylon washers. That's the best method for through-the-board cooler mounting.
Caveat emptor - I've no experience with ATI northbridges. It's my guess it won't get as hot as Socket 775 board's northbridge under the voracious electrical apetite of a D series dual Prescott.
I would think it would be a good idea Brian, if you plan to do some overclocking with it. The way I usually remove those plastic spring clips is to gently squeeze the end together on the back side of the mobo with a pair of needlenose pliers while also trying to gently push it back through the hole in the mobo.
Two instances of Folding running. Graphics screen saver running. With previous motherboard, Asus P5PL2, the max stable OC was 3646MHz, and that was with a higher vCore. This was a very easy jump to 3.8GHz. I'll probably let this set until tomorrow evening before I push more.
Well I'm gonna do both with AS5. I guess I may as well do the mosfets too. The sink on the mosfets seems pretty impressive
Thanks!
Not only did I apply AS5 to all of those ...I also lapped the hell out of them too. But this board runs cool AFAIC so in hindsight it wasn't really necessary.
Great job Leo!
For anyone interested there are cpu's being given away at anandtech thru the newegg article. Check it out here ...http://www.anandtech.com/IT/showdoc.aspx?i=2694&p=6 scroll all the way to the last paragraph or so and find the link.
Two instances of Folding running. Graphics screen saver running. With previous motherboard, Asus P5PL2, the max stable OC was 3646MHz, and that was with a higher vCore. This was a very easy jump to 3.8GHz. I'll probably let this set until tomorrow evening before I push more.
Fantastic OC Leo! That is a fine piece of silicon you have there Looks like the ES turned out to be a great deal after all. Love to see what it would do under the vapo
That's a really nice overclock out of your furnace, Leo.
Also I'm taking some of your advice and I ordered a Big Typhoon the other day for my Opti system. The SLK948-U I've been using on it is a good heatsink, but I want something quieter and cools even better.
0
LeonardoWake up and smell the glaciersEagle River, AlaskaIcrontian
edited February 2006
You'll love it! It's also a great looker if you've got a windowed case. How deep is your case. It's a tall heatsink. If the case is narrow, there will only be limited room for intake on the sink's fan.
0
LeonardoWake up and smell the glaciersEagle River, AlaskaIcrontian
edited February 2006
nice overclock out of your furnace, Leo
I'm not done yet. I'm still exploring vCore. It was stable yesterday at 3.8GHz. Before I went to bed, I inadvertantly bumped the vCore down a couple notches in the BIOS (don't overclock when sleepy!). The system froze during the night. I bumped it back up this morning and lowered the FSB - didn't want to take chances while at work away from the puter. When time permits and I find the vCore sweetspot, minimal to do the job, I'll start ratcheting up the FSB again. This BIOS has so many FSB/DRAM dividers that I can keep the memory as close to default (or OC) as I wish. Cooling is still a challenge though. Either the P5WD2 runs the CPU a couple degrees hotter than my previous motherboard, or the AS5 isn't settled in yet, or the thermistor registers hotter than the other board....or something.
I just find it hard to believe the OC claims on a couple other forums of users with D820 stating they are using the stock CPU heatsink. Do the production D8XX series run cooler than the engineering samples? I'm keeping the CPU within tolerances, but still, it would be nice to have more thermal headroom.
You'll love it! It's also a great looker if you've got a windowed case. How deep is your case. It's a tall heatsink. If the case is narrow, there will only be limited room for intake on the sink's fan.
I'm running a Coolermaster Praetorian case , which has plenty of depth to it. I might have to remove the ducting plastic off the side duct, but otherwise I shouldn't have any clearance problems. The case doesn't have a side window, but that isn't a big deal to me anyways and if I would want a window in it later I could just cut the side panel and install a plexi window in it. It's a great case BTW, but it's also the most expensive case I've ever bought too.
0
LeonardoWake up and smell the glaciersEagle River, AlaskaIcrontian
edited February 2006
3800 MHz is stable:
CPU frequency - 3.8GHz
multiplier/FSB - 14X271 (FSB quad 1084)
DDR2 - 362MHz (DDR2 724); FSB/DRAM ratio 3:4
vCore - set 1.4375, indicated 1.38 (Asus Probe), only 0.01v fluctuation
vMem - default 1.9
CPU core temperature - 55*C at full load (indicated case temp 29*C)
At these settings, the computer functions perfectly and is free of crashes or errors. There is not much thermal headroom left for the CPU. The northbridge at this OC is not hot, just warm. (I mounted a small fan the NB heatsink)
I believe there is still about 200MHz headroom left with the CPU FSB. I ran the computer for about half an hours under full load at 3912MHz. It froze. I'm pretty sure this was due CPU core heat - 58*C. The Big Typhoon is a hero, but a couple things conspire against it:
third floor room; heat from downstairs flows into my office
internal case configuration is not optimal for directing cool air to the heatsink
This weekend I will experiment with an exhaust fan for the office to pump out heated air. (It doesn't help that systems 1, 3, and 4, all overclocked are chugging away in the small office, for a total space heater effect of 14.7GHz plus two monitors! ) I will also experiment with constructing a cool air duct from the case front panel to the Typhoon's fan. Leaving the computer side cover off is not an option. It is good for a 2*C CPU temp reduction, but I don't want the sound level increased and also don't want the introduction into the case of dust and lint.
The P5WD2 motherboard BIOS has higher vMem and vCore settings that are seemingly unlimited. There are no voltage limitations for overclocking. I've just got to crack this heat nut. I am so pleased with this motherboard. I'm yet another user that will confirm it is the big daddy of Socket 775 motherboards. (but then, the 975X chipset boards are out...P5WD2-E Premium)
Keep monitoring, all you jaded overclockers. I'll update the thread as developments ensue.
Comments
Last night I assembled the computer with a new (refurbished) P5DW2. Overall, it's a nice board layout. It was satisfying to see how thick the PCB was - definitely an enthusiast's board. First step was to pull the north and south bridges off and apply Arctic Silver to the chips. I also added a fan to north bridge. (775 boards' north bridges get very, very hot under overclocking.) The south bridge sink was not completely flat and would not leave an imprint on the sink's bottom of the Arctic Silver. The south bridge doesn't get all that hot anyway, so I just re-applied some thick viscosity silicone paste. Ought to work well enough. The board has EPS12v capability, so I had already prepared for this by modifying my PSU with an additional 4-pin plug in tandem with PSU's existing plug. No problem. The P5WD2 is stock with a decent mosfet heatsink. I removed that too in order to apply Arctic Silver. I changed my mind after seeing the horseshow array of tiny mosfets. It would have been very difficult to apply AS to all the mosfets without stray paste ending up on the exposed electircal solder-traces. Maybe I'll get to that some other time. I had to make two modifications to get the Big Tyhpoon to mount on the board. First I had to file down a protrusion on the end of the top-side heatsink retension plate, as it was abutting against the mosfet heatsink. No problem - only about ten minutes of work with a file, the plate firmly held in a bench vise. Second, there were through-the-board-solder terminals on the bottom side of the board. Some of these were about four or five millimeters in length. Simple - just snipped them off with wire cutters. OK, everything was mounted. Popped the motherboard in the case, connected the cables, and voila - nothing. Hmm. Never did quite figure out what was wrong. Perhaps I had mispositioned the front panel's power switch lead or reset lead. But, after removing all the cables and reattaching, the computer fired right up.
I only spent about 45 minutes tweaking last night, so results so far are just preliminary. I figured this board was at a minimum just as capable as my previous P5PL2 was maxed out. I was right. After testing at BIOS stock settings for voltages, FSB, and CPU, I cut to the chase. The first OC I tried was a 700MHz FSB boost. 3.5GHz was instant and flawless. Just two more bumps before I went to bed, the last being at:
3700MHz
mulitplier 14
FSB 1056 (264 quad)
memory divider 3:4 yielding 353MHz (706MHz DDR2).
It ran all night at these settings running two instances of Folding.
Early motherboard observations:
- very resilient
- very smooth
- extensive tweak settings
- very stable voltages, little fluctuation under OC + full load
- I'm getting the same OC or better than before with less vCore, or at least less indicated vCore
My goal was a stable 1000MHz overclock. It's only 100MHz away. I should have time for more tweaking and testing this evening.
Leo
I can never figure out how to get those little white plastic spring clips off without destroying them (or scratching the hell out of the motherboard, which scares me terribly
You are right. Those springloaded plastic retention thingies are never as taught as original when you reinstall them. Often I just put them in the parts bin after removal. Plastic cable ties, one head on top of the sink's hole, with another tie cinching from the backside of the motherboard works very well. The black ones are hardly noticeable. You can also use small nuts and bolts with nylon washers. That's the best method for through-the-board cooler mounting.
Caveat emptor - I've no experience with ATI northbridges. It's my guess it won't get as hot as Socket 775 board's northbridge under the voracious electrical apetite of a D series dual Prescott.
Thanks!
Currently:
3800MHz
vCore - 1.4375 set/1.37 measured (Asus Probe)
vMem - 1.9 set
FSB - 271/1085 (quad)
DRAM - 362 (724 DDR2)
Two instances of Folding running. Graphics screen saver running. With previous motherboard, Asus P5PL2, the max stable OC was 3646MHz, and that was with a higher vCore. This was a very easy jump to 3.8GHz. I'll probably let this set until tomorrow evening before I push more.
Great job Leo!
For anyone interested there are cpu's being given away at anandtech thru the newegg article. Check it out here ...http://www.anandtech.com/IT/showdoc.aspx?i=2694&p=6 scroll all the way to the last paragraph or so and find the link.
Fantastic OC Leo! That is a fine piece of silicon you have there
Also I'm taking some of your advice and I ordered a Big Typhoon the other day for my Opti system. The SLK948-U I've been using on it is a good heatsink, but I want something quieter and cools even better.
I just find it hard to believe the OC claims on a couple other forums of users with D820 stating they are using the stock CPU heatsink. Do the production D8XX series run cooler than the engineering samples?
I'm running a Coolermaster Praetorian case , which has plenty of depth to it. I might have to remove the ducting plastic off the side duct, but otherwise I shouldn't have any clearance problems. The case doesn't have a side window, but that isn't a big deal to me anyways and if I would want a window in it later I could just cut the side panel and install a plexi window in it. It's a great case BTW, but it's also the most expensive case I've ever bought too.
CPU frequency - 3.8GHz
multiplier/FSB - 14X271 (FSB quad 1084)
DDR2 - 362MHz (DDR2 724); FSB/DRAM ratio 3:4
vCore - set 1.4375, indicated 1.38 (Asus Probe), only 0.01v fluctuation
vMem - default 1.9
CPU core temperature - 55*C at full load (indicated case temp 29*C)
At these settings, the computer functions perfectly and is free of crashes or errors. There is not much thermal headroom left for the CPU. The northbridge at this OC is not hot, just warm. (I mounted a small fan the NB heatsink)
I believe there is still about 200MHz headroom left with the CPU FSB. I ran the computer for about half an hours under full load at 3912MHz. It froze. I'm pretty sure this was due CPU core heat - 58*C. The Big Typhoon is a hero, but a couple things conspire against it:
third floor room; heat from downstairs flows into my office
internal case configuration is not optimal for directing cool air to the heatsink
This weekend I will experiment with an exhaust fan for the office to pump out heated air. (It doesn't help that systems 1, 3, and 4, all overclocked are chugging away in the small office, for a total space heater effect of 14.7GHz plus two monitors!
The P5WD2 motherboard BIOS has higher vMem and vCore settings that are seemingly unlimited. There are no voltage limitations for overclocking. I've just got to crack this heat nut. I am so pleased with this motherboard. I'm yet another user that will confirm it is the big daddy of Socket 775 motherboards. (but then, the 975X chipset boards are out...P5WD2-E Premium)
Keep monitoring, all you jaded overclockers. I'll update the thread as developments ensue.