ASUS P5B-E: P965 stepping C1 versus C2, Round One
by Gary Key on October 4, 2006 9:00 AM EST- Posted in
- CPUs
Overclocking
Our BIOS settings were fairly conservative as we left the memory settings at Auto. We tried fine tuning each of ten different memory options but in the end the board responded better over the 485FSB level with memory set at Auto timings. We changed our voltage and other settings to 2.10V RAM, 1.450V FSB Termination, 103 MHz PCI Express, PCI fixed at 33.33MHz, DRAM Frequency at 1:1, Spread Spectrum disabled, Static Read Control disabled, and all power conservation settings disabled. Dialing in the proper CPU voltage was tricky at times due to the voltage droop on this board. In our final settings we found 1.5250V ended up being around 1.4875V under load with 1.5000V dropping to 1.4625V.
We were amazed by our overclocking results on this board with either MCH stepping as we figured the lack of MCH voltage adjustability and memory voltages not exceeding 2.10V would have hampered our clocks to the mid-400 FSB level. Like other ASUS boards, we had to gradually increase the FSB speeds while adjusting CPU voltages and memory settings in order to reach this level. Our retail E6300 was able to boot into Windows XP at 7x521 but we consistently had benchmark failures until we backed off to 7x515 on the C1 stepping and 7x514 on the C2 stepping. With our component choices these standard overclock levels were solid for 24/7 operation. We are currently testing the overclocking capability of our E6600 and E6400 processors for the next article update. We were not able to exceed the 525MHz FSB level when dropping the CPU multiplier to 6. We could post at 6x533 with either stepping but once again the stable limit was at 6x525. We see this limitation as a combination of not having enough MCH voltage and the inability to increase our memory voltage past 2.10V. In the end, both MCH steppings overclocked in the same manner although we found the memory and 3DMark06 scores to slightly favor the C2 stepping even with the C1 stepping system having a slightly higher overclock, maybe our Intel engineer was correct about the minor memory tweaks.
Considering the BIOS limitations we have with the memory and MCH voltages it is impossible to tell if the C2 stepping would really overclock better on this board or not; we do know for a fact that our C1 seems to be every bit the equal to our C2 stepping. This is exactly what the motherboard suppliers have been telling us and why they will not be changing the board designations when the switchover to C2 occurs. The majority of the suppliers we contacted expect this switchover to be complete by the end of October. We expect the current overclocking capability will change when a supplier introduces a new board or BIOS design that takes additional advantage of the P965 chipset. However, we have been informed that even in this case the two steppings are almost identical in performance with the board layout, BIOS, and components being the most critical areas in improving overclocking. After testing two other C2 stepping motherboards we found a difference of 12 FSB between identical boards so the luck of the draw in getting a good MCH is just as important as getting an overclock friendly CPU.
| Asus P5B-E Overclocking Testbed |
|
| Processor: | Intel Core 2 Duo E6300 Dual Core, 1.86GHz, 2MB Unified Cache 1066FSB, 7x Multiplier |
| CPU Voltage: | 1.5250V (default 1.3250V) |
| Cooling: | Scythe Infinity Air Cooling |
| Power Supply: | OCZ GameXStream 700W |
| Memory: | Corsair Twin2X2048-PC2-6400C3 (2x1GB), 2.10V (Micron Memory Chips) |
| Video Cards: | 1 x MSI X1950XTX |
| Hard Drive: | Seagate 320GB 7200RPM SATA2 16MB Buffer |
| Case: | Cooler Master CM Stacker 830 |
| Maximum CPU OC: (Standard Ratio) |
515x7 (4-4-4-12, 1:1), CPU 1.5250V C1 Stepping - 3610MHz (+93%) |
| 514x7 (4-4-4-12, 1:1), CPU 1.5250V C2 Stepping - 3602MHz (+93%) |
|
| Maximum FSB OC: (Lowest Ratio) |
525x6 (5-4-5-12, 1:1), CPU 1.5000V C1 Stepping - 3148MHz (+97%) |
| 525x6 (5-4-5-12, 1:1), CPU 1.5000V C2 Stepping - 3148MHz (+97%) |
|
Our BIOS settings were fairly conservative as we left the memory settings at Auto. We tried fine tuning each of ten different memory options but in the end the board responded better over the 485FSB level with memory set at Auto timings. We changed our voltage and other settings to 2.10V RAM, 1.450V FSB Termination, 103 MHz PCI Express, PCI fixed at 33.33MHz, DRAM Frequency at 1:1, Spread Spectrum disabled, Static Read Control disabled, and all power conservation settings disabled. Dialing in the proper CPU voltage was tricky at times due to the voltage droop on this board. In our final settings we found 1.5250V ended up being around 1.4875V under load with 1.5000V dropping to 1.4625V.
 |
| Click to enlarge |
 |
| Click to enlarge |
We were amazed by our overclocking results on this board with either MCH stepping as we figured the lack of MCH voltage adjustability and memory voltages not exceeding 2.10V would have hampered our clocks to the mid-400 FSB level. Like other ASUS boards, we had to gradually increase the FSB speeds while adjusting CPU voltages and memory settings in order to reach this level. Our retail E6300 was able to boot into Windows XP at 7x521 but we consistently had benchmark failures until we backed off to 7x515 on the C1 stepping and 7x514 on the C2 stepping. With our component choices these standard overclock levels were solid for 24/7 operation. We are currently testing the overclocking capability of our E6600 and E6400 processors for the next article update. We were not able to exceed the 525MHz FSB level when dropping the CPU multiplier to 6. We could post at 6x533 with either stepping but once again the stable limit was at 6x525. We see this limitation as a combination of not having enough MCH voltage and the inability to increase our memory voltage past 2.10V. In the end, both MCH steppings overclocked in the same manner although we found the memory and 3DMark06 scores to slightly favor the C2 stepping even with the C1 stepping system having a slightly higher overclock, maybe our Intel engineer was correct about the minor memory tweaks.
Considering the BIOS limitations we have with the memory and MCH voltages it is impossible to tell if the C2 stepping would really overclock better on this board or not; we do know for a fact that our C1 seems to be every bit the equal to our C2 stepping. This is exactly what the motherboard suppliers have been telling us and why they will not be changing the board designations when the switchover to C2 occurs. The majority of the suppliers we contacted expect this switchover to be complete by the end of October. We expect the current overclocking capability will change when a supplier introduces a new board or BIOS design that takes additional advantage of the P965 chipset. However, we have been informed that even in this case the two steppings are almost identical in performance with the board layout, BIOS, and components being the most critical areas in improving overclocking. After testing two other C2 stepping motherboards we found a difference of 12 FSB between identical boards so the luck of the draw in getting a good MCH is just as important as getting an overclock friendly CPU.
Facebook
Twitter
RSS
27 Comments
View All Comments
crash6767 - Friday, January 26, 2007 - link
ORDER PLACED 1/23/2007 9:10:24 PM FROM NEWEGG.COM.RECEIVED 1/26/2007 1:14 PM.
MODEL NUMBER PRINTED ON CIRCUIT BOARD, ABOVE DIMM A1:
P5B-E 1.02G
YYYYYYYYYEEEEEEEEEEEEEEEEEEEESSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS!!!!!!!!!!!!!!!!!!!!!
still waiting on the power supply to arrive (purchased from another retailer) so no OC numbers yet. BUT 1.02G IS ALIVE AND KICKING!!!!#!#!@!#@!@
*crossposted everywhere*
agigolo - Wednesday, October 11, 2006 - link
Ok, so I've read this and the unfortunate part (unless I missed it and I don't think I did) but when these acronyms are used (like MCH) I wish they would be defined on the 1st useage... can someone be a good soul and explain MCH please???Gary Key - Wednesday, October 11, 2006 - link
MCH - Memory Controller Hub, aka Northbridge. :)cornfedone - Sunday, October 8, 2006 - link
We've seen time and time again in the past few years where rushed out the door mobos are over-hyped and in fact don't perform as advertised. Many don't even run industry standard memory without problems which is a disgrace. These boards are way over priced and sought by naive consumers after they read the glowing online reviews that fail to mention the many defects in these products.Once the motherboard problems start to get online exposure the mobo makers move to the next model chipset and rave how it's so much better than the previous model. Naturally the mobo companies don't fix the problems with the previous products they shipped and in many cases refuse to even acknowledge the defects that become confirmed by tens of thousands of duped customers. Instead the mobo companies whip out the next trick of the week half baked mobo and make sure that hardware review sites get "special" versions for testing so the reviews are always positive despite the production board defects that exist. Unless a reviewer is buying the retail mobo from a retail outlet, they ain't necessarily getting the same mobo as all other consumers.
You gotta wonder if the gullible fanboys will ever wake up to this scam or if they will keep paying through the nose for defective, over priced mobos. As long as sheep keep buying these dysfunctional mobos the manufacturers will keep shipping garbage. There is no incentive to deliver a properly functioning mobo if the sheep will buy half baked goods at twice the price they should sell for.
The C1 / C2 chipset deal is just another example of hype yet people will believe the C2 will provide a 20% performance increase because they are so gullible. One accurate scientific test is worth much more than a thousand online opinions. The fanboys need to buy a clue instead of pissing their money away on crappy mobos and over hyped chipsets.
Binkt - Thursday, October 5, 2006 - link
Hi, nice sleuthing so far guys, thanks.One thing that concerns me is the NB temperature. My Gigabyte 965p-DS3 has a very high operating temperature and I was wondering if you had observed a difference between the steppings in this regard.
Gary Key - Thursday, October 5, 2006 - link
Hi,There were no temperature differences between the two boards on the MCH. We highly recommend that you replace the paste on the MCH heatsink with AS5 and place a 40mm on it if you plan on overclocking 24/7. The same holds true with the DS3, on my personal system I just replaced the MCH heatsink with this one - http://www.newegg.com/Product/Product.asp?Item=N82...">SwiftTech.
jambaz - Thursday, October 5, 2006 - link
http://images.anandtech.com/reviews/motherboards/a...">synthetichttp://images.anandtech.com/reviews/motherboards/a...This picture shows a CPU speed of 2.4 Ghz when really the speeds are 1.86, 3.6 and 2.4 Ghz. The "general" picture below has the correct way of showing it imho.
By the way, it would be nice if Anandtech would try to show performance of a lowbudget e6300 paired with value ram instead of this RAM they use that cost 800$..
Like the benchmarks we see now are 1:1 @ 515 Mhz = DDR1030, but what if you did 5:4 or 4:3 so ram would be less of an expense?
Thanks for great articles!
JarredWalton - Thursday, October 5, 2006 - link
I don't believe you can go lower than 1:1 with the 975X/P965 chipsets and Core 2 Duo. That was from the days of Pentium 4/D where you could run the FSB at a higher speed than the RAM. So if you want to overclock, either you pay a boatload of money on RAM, or you don't OC as far, or you get a more expensive CPU. Not great choices unfortunately.lopri - Thursday, October 5, 2006 - link
I can do memory frequency lower than FSB on P5W-DH. Of course it is not a recommended configuration.JarredWalton - Friday, October 6, 2006 - link
Must vary by motherboard/BIOS implementation. I know I've seen several boards where 1:1 (DDR2-533) is the lowest possible selection. Or maybe there was a DDR2-400 choice I missed? Meh - can't check now, since I don't have the systems anymore.