Intel P35: Intel's Mainstream Chipset Grows Up
by Gary Key & Wesley Fink on May 21, 2007 3:45 PM EST- Posted in
- CPUs
Disk Controller Performance
The AnandTech iPeak test is designed to measure "pure" hard disk controller performance, and in this case we keep the hard drive as consistent as possible while varying the hard drive controller. The idea is to measure the performance of each hard drive controller with the same hard drive. We report the scores as an average number of I/O operations per second so that higher scores translate into better performance. This number is somewhat meaningless as far as hard disk performance is concerned; however, the scores are useful for comparing "pure" performance of the storage controllers in this case. Results are taken from trace files of Winstones 2004. Due to compatibility issues with Vista, our IPEAK tests are run on Windows XP SP2.
The performance patterns hold steady across both Multimedia Content I/O and Business I/O with the ICH9R outperforming the Intel ICH7R, Intel ICH8R, and NVIDIA 680i chipsets in our tests. Intel told us we should see up to a 4% difference between the ICH9 and ICH8; while our tests do not reveal that, we have noticed a difference in actual usage within Vista, especially when transferring large files. We will have Vista specific tests in the near future. Although our NVIDIA boards generally score lower in these "pure" throughput tests, we find their actual performance in disk intensive applications are generally equal to other solutions during actual usage. We did not experience any RAID 0 issues with the NV controller or drivers under Vista during normal usage but users have reported their RAID arrays degrading over time. We are still investigating these issues.
Power Consumption
As always, we measured power consumption at two states: at idle sitting at the Vista desktop and under load while running our 3DMark06 and Company of Heroes benchmarks. At both settings, EIST/C1E were disabled to show power usage with the system in a normal desktop state.
The surprise is that despite a 16W TDP, the P35 boards actually consume more power than the other chipsets at both idle and load. Even though the DDR3 board has a lower power rating for the memory, it consumes more power than the DDR2 board due to additional circuitry . We can no longer say that NVIDIA is the only high performance chipset approved by OPEC.
FSB Overclocking
Our FSB overclocking results are limited to the quad core QX6700 for this article. We will greatly expand our overclocking results for the roundup that will include five different CPUs. Our ASUS boards overclock the quad core the best at this time thanks to a fairly mature BIOS and great board design. Our Gigabyte and MSI boards have improved greatly over the last week with new BIOS designs and we expect overclocking to improve before the official launch date. In limited testing with the E6600, we were able to hit a 576 FSB on the ASUS boards, 514 on the Gigabyte board, and 508 on the MSI board with the latest BIOS releases.
Facebook
Twitter
RSS
58 Comments
View All Comments
Gary Key - Tuesday, May 22, 2007 - link
X38 is basically ready, going through some fine tuning now... I understand it will be held until after the 1333CPUs are launched and DDR3 availability is a little more widespread/cost effective. I expect late August right now, but you never know with Intel. ;-)JarredWalton - Monday, May 21, 2007 - link
Technically Q3 is any time between July 1 and September 30, but if they're saying Q3 right now it probably means some time in August at best.gigahertz20 - Monday, May 21, 2007 - link
I applaud ASUS for only including 1 legacy connection on their P5K series, and not 4 like Gigabyte has chosen to do for their P35 board. Death to legacy connections!I mean really, why even include those damn legacy ports. The enthusiast that buys one of these boards will not be using them, they are a waste of space. Instead of having them, they should replace them with more USB ports or something useful.
JarredWalton - Monday, May 21, 2007 - link
I still have a parallel based laser printer that works fine for what I need, and I'm quite happy using it until it dies. There are also people that use serial devices that cost a lot of money. I don't think every board needs legacy support, but it's good that there are still options for people that *need* certain legacy devices. I've got several KVM switches that won't be useful if PS/2 ports disappear. :(yacoub - Monday, May 21, 2007 - link
Don't they offer USB or eSATA to serial/parallel convertors for those sort of situations? :)JarredWalton - Monday, May 21, 2007 - link
Sure, but I haven't had the need to try one yet. :)Truth be told, my printer has a USB port, but it behaves very poorly using a USB connection. It's a Brother HL-1240, and if the printer isn't powered on when you boot, Windows won't see it unless you unplug it and plug it back into a different USB port. It just works better with LPT, and as I said for my needs it's sufficient. The way I figure it, having the ports there isn't hurting most people. I've never seen anything to indicate they hamper performance, and how many extra transistors are "wasted" on these ports? Maybe a few thousand? Heheh. 45M transistors on the P35 is a bit crazy....
For what it's worth, between mouse, keyboard, and my LCD (which actually has four USB ports and flash memory readers), I haven't had any need for more than four USB ports on a motherboard. But then, I've got too many PCs around anyway.
TA152H - Monday, May 21, 2007 - link
I agree with you, but for another reason.I don't like USB at all, because a few years ago I ran some tests, on motherboards ranging from MVP3 based to a KT880, and USB seems to have a negative impact on performance, particularly on memory, in many cases.
It doesn't make my keyboard work any better, or my mouse, and I'm not sure why I need it for those functions at all. PS/2 ports don't do it well enough? I'm not crazy about this one size fits all approach, especially when it comes with overhead. The current ports work fine.
USB is a crappy, bloated technology. I'm not sure the "S" should stand for "serial" at all, I think there is a better word that begins with S for it.
strikeback03 - Tuesday, May 22, 2007 - link
Wake me up when Bluetooth works over PS/2.Though one reason to still include PS/2 keyboard/mouse is that it is hard to screw up support for those in Linux kernels. Same can't be said about USB.
TA152H - Tuesday, May 22, 2007 - link
Wake me up when I need Bluetooth.You could implement Bluetooth easily if USB didn't exist, but you're missing the point anyway. When you have to use USB for stuff that is handled more efficiently by PS/2 ports, it's a bad thing. Or other ports. It adds no function for these devices, and comes with overhead. It's a bad idea, but of course Intel was in the mode of making as many things as possible use CPU power so they could keep selling their latest and greatest.
It's just a rehash of some dorky Apple stuff that most people here don't remember. The original MacIntoy didn't have any slots, and you'd attach stuff to some serial bus for expansion. Naturally, it didn't work out, and they had to add slots. At least they didn't get rid of slots for USB, they just made it bloated.
DigitalFreak - Monday, May 21, 2007 - link
Man, if the P35 boards are going to be around the $250 mark, I'm not looking forward to see the price on the X38 boards.... :-(