Conroe Buying Guide: Feeding the Monster
by Gary Key & Wesley Fink on July 19, 2006 6:20 AM EST- Posted in
- Motherboards
ASUS P5N32-SLI SE
Basic Features
We reviewed the ASUS P5N32-SLI back in October 2005 in its initial release to the market. At that time the board revision and BIOS we tested with offered excellent performance, stability, and compatibility across the board. It was also the first Intel board to offer dual X16 support for NVIDIA SLI and with the right Pentium 4 or Pentium D processor the P5N32-SLI offered very good gaming performance.
Since that time the original board has either been loved or hated by its owners. Over the course of time the board was revised with changes to the resistors to improve VRM performance along with numerous BIOS revisions that just never lived up to the results of our pre-release 0047. Issues ranged from poorer than normally accepted FSB overclocking results with premium components, memory incompatibilities with 4GB installations, RAID issues with various BIOS releases, and other items that generally tarnished the reputation of the board.
ASUS did listen to the user complaints and additional component changes were made on the board to improve stability - but the most significant changes occurred in the BIOS. While the options generally remain the same, the overall performance of the board (along with stability) has improved over the previous 1.02 version of the board with the 0310 BIOS.
We are still testing the board with a wide array of components but we are glad - actually relieved - to report that the majority of user issues reported to us have been solved. The differences between the two boards are of course compatibility for the Core 2 Duo series, a couple of minor layout changes with the 4-pin 12V Molex connector being removed, and additional capacitors being added to the board for improved stability. The SATA ports are still in the same location which means SLI users will want to plug in their drives before installing the second video card. The basic features and layout of the ASUS P5N32-SLI SE remain nearly the same as the original board.
Basic Performance
Overall, the performance of the P5N32-SLI was excellent. It is also the only board in our Buyers Guide to officially support NVIDIA SLI. It not only supports SLI but offers full dual X16 capability to each PCIe graphics slot. The board was very stable with our X6800, E6700, and E6600 Core 2 Duo processors while providing asynchronous operation between the front side bus and memory controllers. This ability allowed us to dial in extremely low memory latencies at specific speeds in order to maximize the bandwidth of our Corsair DDR2 modules. Like the 975X boards, the nForce4 Intel Edition chipset allows the user to adjust the X6800 CPU ratios up or down.
This ASUS board had a tendency to get very warm during testing and requires a case with good air circulation to operate at normally accepted temperatures. Although the board never failed during overclocking or extended testing, our fingers on more than one occasion wanted to be iced down after touching the passive heatsinks on either the MCP or SPP.
We anxiously await the production release of the nForce 500 Intel Edition chipsets in a few weeks that will mainly bring benefits to the MCP such as additional SATA ports, improved networking features, HD Audio, and some general refinement. Users need to realize that the nForce 590 SLI Intel Edition will still use the C19A SPP will be used that is on this board. Although the chipset is now at a revision C1 and has undergone several months of fine tuning, do not expect the performance or overclocking results to be improved greatly with the nForce 590 SLI boards. In our internal testing we have noticed some minor but measurable improvements, but nothing revolutionary.
Overclocking
This is not the board to own if you expect or require high FSB overclocks with fixed multiplier Intel CPUs. While the FSB results are in alignment with other NVIDIA chipset boards they do not match overclocking results with Intel chipset boards at this time. We expect to see this change when NVIDIA releases their revised chipsets in several months. However, the board did match the overclock ranges of the Intel 975X board when increasing the multiplier on the X6800, and that allowed far greater control/tuning of the memory speeds due to the asynchronous operation of the controllers.
Although we wish the board provided additional memory voltage settings, we were able to run our memory timings at higher speeds with lower latencies than on the Intel chipset boards. That resulted in improved performance of the board when overclocked. Our only issue with the memory controller at this time is that the BIOS does not fully support memory timings over DDR2-1000 - although this should change before release. We will provide overclocked performance results in an upcoming article along with a comparison to the AM2 SLI boards. This board should sell in the $190 range, significantly lower than the first available Intel 975X boards with stock performance that matches or exceeds those boards. If dual X16 SLI is important to you and high FSB overclocks are not, then this is currently the board to have.
Basic Features
ASUS P5N32-SLI SE | |
Market Segment: | High-End/Enthusiast |
CPU Interface: | Socket T (Socket 775) |
CPU Support: | LGA775-based Pentium 4, Pentium XE, Celeron D, Pentium D, Core 2 Duo |
Chipset: | nForce4 Intel SLI Intel Edition (C19 revC1) nForce4 Intel SLI Edition (CK-804) |
Thermal Design: | 8-phase power Fan-less Heatpipe Cooling ASUS Stack Cool 2 for OC |
Bus Speed Support: | 1066/800/533MHz |
Bus Speeds: | 533 to 1600 in 1MHz Increments |
Memory Speeds: | 400 to 1600 in 1MHz Increments |
PCIe Speeds: | 100 to 148.4375 in 1.5625MHz Increments |
PCI: | Fixed at 33 |
AI Tuning: | Manual, Auto, Overclock Profile, AI N.O.S. |
Core Voltage: | Auto, 1.2250V to 1.7000V in 0.0625V increments |
PEG Link Mode: | Auto, Disabled, Normal, Fast, Faster |
CPU Clock Multiplier: | Auto, 6x-20x in 1X increments if CPU is unlocked |
DRAM Voltage: | 1.8V to 2.4V in .05V or .10V increments |
1T/2T Memory: | Auto, 1T, 2T |
DRAM Timing Control: | Auto, 6 Options |
NB HT: | Auto, 200MHz, 400MHz, 600MHz, 800MHz, 1000MHZ |
SB HT: | Auto, 200MHz, 400MHz, 600MHz, 800MHz, 1000MHZ |
NB Voltage: | Auto,1.4V, 1.5V, 1.6V |
SB Voltage: | Auto,1.5V, 1.6V |
Memory Slots: | Four 240-pin DDR2 DIMM Slots Dual-Channel Configuration Regular Unbuffered Memory to 16GB Total |
Expansion Slots: | 2 - PCIe X16 1 - PCIe X4 2 - PCIe X1 2 - PCI Slots 2.2 |
Onboard SATA/RAID: | 4 SATA 3Gbps Ports - NVIDIA 2 SATA 3Gbps Ports - SI3132 (RAID 0,1,0+1,JBOD) -NVIDIA (RAID 0,1) - Silicon Image 3132 |
Onboard IDE: | 2 Standard ATA133/100/66/33 Ports (4 drives) |
Onboard USB 2.0/IEEE-1394: | 10 USB 2.0 Ports - 4 I/O Panel, Six via Headers 2 Firewire 400 Ports by TI TSB43AB22A |
Onboard LAN: | Dual Gigabit Ethernet Marvell 88E1115 PHY, Marvell 88E8053 |
Onboard Audio: | Realtek ALC850 8-channel Codec |
Power Connectors: | ATX 24-pin, 8-pin EATX 12V |
I/O Panel: | 1 x Parallel 1 x S/PDIF Out (Coaxial + Optical) 1 x PS/2 Keyboard 1 x PS/2 Mouse 2 x RJ45 4 x USB 2.0/1.1 1 x External SATA 8-Channel Audio I/O |
BIOS Revision: | AMI 0121 |
We reviewed the ASUS P5N32-SLI back in October 2005 in its initial release to the market. At that time the board revision and BIOS we tested with offered excellent performance, stability, and compatibility across the board. It was also the first Intel board to offer dual X16 support for NVIDIA SLI and with the right Pentium 4 or Pentium D processor the P5N32-SLI offered very good gaming performance.
Since that time the original board has either been loved or hated by its owners. Over the course of time the board was revised with changes to the resistors to improve VRM performance along with numerous BIOS revisions that just never lived up to the results of our pre-release 0047. Issues ranged from poorer than normally accepted FSB overclocking results with premium components, memory incompatibilities with 4GB installations, RAID issues with various BIOS releases, and other items that generally tarnished the reputation of the board.
ASUS did listen to the user complaints and additional component changes were made on the board to improve stability - but the most significant changes occurred in the BIOS. While the options generally remain the same, the overall performance of the board (along with stability) has improved over the previous 1.02 version of the board with the 0310 BIOS.
Click to enlarge |
We are still testing the board with a wide array of components but we are glad - actually relieved - to report that the majority of user issues reported to us have been solved. The differences between the two boards are of course compatibility for the Core 2 Duo series, a couple of minor layout changes with the 4-pin 12V Molex connector being removed, and additional capacitors being added to the board for improved stability. The SATA ports are still in the same location which means SLI users will want to plug in their drives before installing the second video card. The basic features and layout of the ASUS P5N32-SLI SE remain nearly the same as the original board.
Basic Performance
Overall, the performance of the P5N32-SLI was excellent. It is also the only board in our Buyers Guide to officially support NVIDIA SLI. It not only supports SLI but offers full dual X16 capability to each PCIe graphics slot. The board was very stable with our X6800, E6700, and E6600 Core 2 Duo processors while providing asynchronous operation between the front side bus and memory controllers. This ability allowed us to dial in extremely low memory latencies at specific speeds in order to maximize the bandwidth of our Corsair DDR2 modules. Like the 975X boards, the nForce4 Intel Edition chipset allows the user to adjust the X6800 CPU ratios up or down.
This ASUS board had a tendency to get very warm during testing and requires a case with good air circulation to operate at normally accepted temperatures. Although the board never failed during overclocking or extended testing, our fingers on more than one occasion wanted to be iced down after touching the passive heatsinks on either the MCP or SPP.
We anxiously await the production release of the nForce 500 Intel Edition chipsets in a few weeks that will mainly bring benefits to the MCP such as additional SATA ports, improved networking features, HD Audio, and some general refinement. Users need to realize that the nForce 590 SLI Intel Edition will still use the C19A SPP will be used that is on this board. Although the chipset is now at a revision C1 and has undergone several months of fine tuning, do not expect the performance or overclocking results to be improved greatly with the nForce 590 SLI boards. In our internal testing we have noticed some minor but measurable improvements, but nothing revolutionary.
Overclocking
ASUS P5N32-SLI SE Overclocking Testbed |
|
Processor: | Intel Core 2 Duo E6700 Dual Core, 2.67GHz, 4MB Unified Cache 1066FSB, 10x Multiplier |
CPU Voltage: | 1.525V (default 1.2V) |
Cooling: | Tuniq Tower 120 Air Cooling |
Power Supply: | OCZ GameXStream 700W |
Memory: | Corsair Twin2X2048-PC2-8500C5 (2x1GB) (Micron Memory Chips) |
Hard Drive | Hitachi 250GB 7200RPM SATA2 16MB Cache |
Maximum OC: (Standard Ratio) |
318x10 (4x HT, 3-3-3-9 1T) 3180MHz (+19%) |
This is not the board to own if you expect or require high FSB overclocks with fixed multiplier Intel CPUs. While the FSB results are in alignment with other NVIDIA chipset boards they do not match overclocking results with Intel chipset boards at this time. We expect to see this change when NVIDIA releases their revised chipsets in several months. However, the board did match the overclock ranges of the Intel 975X board when increasing the multiplier on the X6800, and that allowed far greater control/tuning of the memory speeds due to the asynchronous operation of the controllers.
Although we wish the board provided additional memory voltage settings, we were able to run our memory timings at higher speeds with lower latencies than on the Intel chipset boards. That resulted in improved performance of the board when overclocked. Our only issue with the memory controller at this time is that the BIOS does not fully support memory timings over DDR2-1000 - although this should change before release. We will provide overclocked performance results in an upcoming article along with a comparison to the AM2 SLI boards. This board should sell in the $190 range, significantly lower than the first available Intel 975X boards with stock performance that matches or exceeds those boards. If dual X16 SLI is important to you and high FSB overclocks are not, then this is currently the board to have.
123 Comments
View All Comments
Gary Key - Wednesday, July 19, 2006 - link
We are looking forward to the value SLI boards in early August. Prices will range on average from $95 to $120. As soon as we can post a review up on these boards, it will be done. :)EODetroit - Wednesday, July 19, 2006 - link
Can Anandtech max out the memory and make sure these systems are stable in 64 bit Windows OS'? I'd really like to make sure that there's no time bombs if I buy 8GB of ram that will force me to RMA a lot of stuff. If you don't have 2GB sticks, at least test with 4x1GB. The motherboards advertise that they support 8GB, but no one ever seems to check them on it. If Anandtech could do that, it would be a great help.Thanks!
Genx87 - Wednesday, July 19, 2006 - link
Not bad for an older chipset to win the SLI benchmarks.I am curious what the new chipset can do!
supremelaw - Wednesday, July 19, 2006 - link
Dear Gary and Wesley,I'm thinking back to about 8 months ago,
when we first started assembling our
ASUS P5WD2 Premium motherboard
(which has recently become our primary
production machine).
We offered our assistance at the ASUS
User Forum, because a LOT of users
were stumbling over the IT8211F IDE
controller, which requires a device driver
to be extracted from the Support CD.
I fear that P965 motherboards are headed
for the same serious problems, particularly
if motherboards add an on-board IDE
controller that is NOT "native" e.g. JMicron.
Here's the scenario: a less-than-expert
user sees a PATA IDE port, and thinks
he can use (or recycle) a PATA optical
drive to run Windows Setup. And, he's
probably read (or heard) the stories about
SATA optical drives that just don't work
with Windows Setup.
Is this user headed for major problems?
I think so.
And here's why ...
If the BIOS has not been modified to
support native PATA / IDE optical devices,
a Catch-22 results: you need the device
driver from the Support CD, but you can't
read the Support CD without the device
driver -- not if the optical device is wired
to that on-board IDE controller.
If you want confirmation of this problem,
check out the ASUS User Forum for the
P5WD2 Premium, particularly the numerous
complaints Users were posting about the
ITE IT8211F on-board IDE controller.
To make this problem even more exasperating,
the User Manual failed to mention that the
F6 sequence will load the ITE driver during
Windows Setup, BUT one can STILL not
run Windows Setup from an optical drive
wired to that ITE controller. The device driver
can be added AFTER Windows XP is
successfully installed.
Fortunately, the P5WD2 Premium has a
BLUE native IDE port as well, and we
avoided all of these problems by running
Windows Setup from a PATA optical drive
wired to that BLUE native IDE port.
Thanks for all the great reviews!
Sincerely yours,
/s/ Paul Andrew Mitchell
Webmaster, Supreme Law Library
http://www.supremelaw.org/">http://www.supremelaw.org/
Gary Key - Wednesday, July 19, 2006 - link
Hi Paul,The scenario you listed is a very real possibility and as you have noted has existed in past boards. In fact, I was on the phone with Wes when our first P965 was fired up and the Optical Drive was not recognized. Of course, I had a few choice words to say about the situation. The only way to load the new OS image and Driver CD was through the Optical drive and that was not going to happen in this case. I ended up loading a new image on a drive in another system, installing the inf and network drivers in a folder, and then moving this drive to the new machine. The issue was a very early bios that did not support the hooks from the external IDE chipset into the ICH8. We received an updated bios a few days later and all was well from that point forward.
Since Intel has basically left a "lane" open in the ICH8 to support IDE (much in the same way as the LAN controller logic, being real simple here to keep it short), then the only issue is to ensure the bios has support for the IDE link. We have not seen this issue at all in the latest boards that we have received and have been told it will not occur in shipping boards. While most suppliers are going with the JMicron solution, Biostar included the VIA VT6410 that turned out to offer excellent performance in our upcoming storage tests. I hope this helps and thank you for your comments today.
:)
Andy4504 - Wednesday, July 19, 2006 - link
I was most surprized by the poor BadAxe (X975XBX) Overclocking. Because the memory controller isn't directly tied to the FSB speeds, the fact that you cannot incrase the memory voltage without hardware modification should make little / no difference in CPU overclocking.I personally own a X975XBX with an 805D. I've found that the best overclocking isn't done by selecting +30% OR + any percent for that matter, but rather choosing the higher bus speed, then selecting an underclock from that higher speed.
With full access to the memory multiplier range, most any ratio could be set.
Wesley Fink - Wednesday, July 19, 2006 - link
Conroe runs at 1066 FSB speed. 1333 support has been in and out of the different BIOS revisions. So with Conroe you can select no higher bus speed at worst, or a modest 1333 at best. 805D runs at 533 (166 quad) so you have differnt options. It really isn't possible to select higher bus speeds and clock down with Conroe on the BadAxe.Paladin165 - Wednesday, July 19, 2006 - link
In the review you mention that the 7600GT would work with the cheap ASrock board, I was thinking about going with this setup (if another ultra-cheap board doesn't come out soon). I was wondering, how much impact would the 4x PCIex speed have on the 7600GT? Are there any situations where it would choke off performance? Does it provide enough power?This cheap board seems like a good buy because new motherboards are going to be coming out so rapidly over the next six months it doesn't make sense to drop $250 on a bleeding-edge board.
Gary Key - Thursday, July 20, 2006 - link
The 7600GT works fine. I am trying to procure a 7600GS PCIe and AGP cards to directly compare the video performance on the board. Hopefully, I will have both cards before the full review goes up. We also have two other ASRock boards that are under $75 arriving shortly. I think the performance with the 7600GT will be fine unless you like to play Oblivion and even with the PCIe x4 interface you will not notice a real difference with this card.Paladin165 - Thursday, July 20, 2006 - link
"I think the performance with the 7600GT will be fine unless you like to play Oblivion"!!!
Oblivion is exactly what I want to play! What is it about this setup that hurts Oblivion performance?