Intel P965: Mid-Range Performance Sector Roundup
by Gary Key on October 20, 2006 9:00 PM EST- Posted in
- Motherboards
Firewire and USB Performance
After looking at many options for Firewire and USB testing, we finally determined that an external USB 2.0, Firewire 400, and Firewire 800 hard disk would be a sensible way to look at USB and Firewire throughput. We utilize a RAM disk as our "server", since memory removes almost all overhead from the serving end. We turn off disk caching on the USB and Firewire side by setting up the drives for "quick disconnect" so our results are consistent.
We use 2GB of system memory with timings of 3-3-3-9 and set our RAM disk to 450MB with system memory at 1550MB. Our standard file is the SPECviewPerf install file, which measures 432,533,504 bytes (412.4961MB). After copying this file to our RAM disk, we measured the time for writing from the RAM disk to our external USB 2.0, Firewire 400, or Firewire 800 drive utilizing our internal Windows based timing program. The copy times in seconds were then converted into Megabits per second (Mb) to provide a convenient means of comparing throughput. Higher Rates therefore mean better performance in this particular test.
Possibly the most interesting finding in our Firewire and USB throughput tests is the outstanding performance of an external hard drive connected to Firewire 800. Our benchmarks show Firewire 800 is up to 42% faster than a drive connected to the more common Firewire 400, and about 11% faster than the fastest USB 2.0 solution.
We see our Intel ICH8 chipset finally overtake perennial champion NVIDIA in USB 2.0 performance. We see the USB performance on the ATI SB600 still trailing the other solutions but it has vastly improved compared to the SB450. The Firewire soltuion from VIA is still slightly faster than the TI solutions normally used.
Ethernet Performance
The current motherboard test suite includes LAN performance measurements. All of these boards utilize PCI or PCI Express based controllers with the only difference being the supplier of the core logic.
The Windows 2000 Driver Development Kit (DDK) includes a useful LAN testing utility called NTttcp. We used the NTttcp tool to test Ethernet throughput and the CPU utilization of the various Ethernet Controllers used on the Intel motherboards.
We set up one machine as the server; in this test, an Intel system with an Intel CSA Gigabit LAN connection. Intel CSA has a reputation for providing fast throughput and is a logical choice for our Gigabit LAN server.
On the server side, we used the following Command Line as suggested by the VIA whitepaper on LAN testing:
The CPU utilization performance favors the Marvell Gigabit controllers with the Realtek solutions having the highest utilization numbers. The throughput numbers also favor the Marvell Gigabit controllers with the Realtek options close behind. ASUS recently switched to the Attansic L1 PCI Express based controller chip with it posting decent results in each test. We do not understand ASUS's decision to utilize a PCI based Gigabit controller on the 570SLI when the NVIDIA chipset has native support for Gigabit operations. This decision results in the worse throughput and average CPU utilization numbers. However, even with throughput at 646Mb/s it still exceeds what most home networks are capable of and certainly any DSL or Cable based Internet connection.
After looking at many options for Firewire and USB testing, we finally determined that an external USB 2.0, Firewire 400, and Firewire 800 hard disk would be a sensible way to look at USB and Firewire throughput. We utilize a RAM disk as our "server", since memory removes almost all overhead from the serving end. We turn off disk caching on the USB and Firewire side by setting up the drives for "quick disconnect" so our results are consistent.
We use 2GB of system memory with timings of 3-3-3-9 and set our RAM disk to 450MB with system memory at 1550MB. Our standard file is the SPECviewPerf install file, which measures 432,533,504 bytes (412.4961MB). After copying this file to our RAM disk, we measured the time for writing from the RAM disk to our external USB 2.0, Firewire 400, or Firewire 800 drive utilizing our internal Windows based timing program. The copy times in seconds were then converted into Megabits per second (Mb) to provide a convenient means of comparing throughput. Higher Rates therefore mean better performance in this particular test.
Possibly the most interesting finding in our Firewire and USB throughput tests is the outstanding performance of an external hard drive connected to Firewire 800. Our benchmarks show Firewire 800 is up to 42% faster than a drive connected to the more common Firewire 400, and about 11% faster than the fastest USB 2.0 solution.
We see our Intel ICH8 chipset finally overtake perennial champion NVIDIA in USB 2.0 performance. We see the USB performance on the ATI SB600 still trailing the other solutions but it has vastly improved compared to the SB450. The Firewire soltuion from VIA is still slightly faster than the TI solutions normally used.
Ethernet Performance
The current motherboard test suite includes LAN performance measurements. All of these boards utilize PCI or PCI Express based controllers with the only difference being the supplier of the core logic.
The Windows 2000 Driver Development Kit (DDK) includes a useful LAN testing utility called NTttcp. We used the NTttcp tool to test Ethernet throughput and the CPU utilization of the various Ethernet Controllers used on the Intel motherboards.
We set up one machine as the server; in this test, an Intel system with an Intel CSA Gigabit LAN connection. Intel CSA has a reputation for providing fast throughput and is a logical choice for our Gigabit LAN server.
On the server side, we used the following Command Line as suggested by the VIA whitepaper on LAN testing:
Ntttcpr -m 4,0,‹server IP› -a 4 -l 256000 -n 30000
On the client side (the motherboard under test), we used the following Command Line:Ntttcps -m 4,0,‹client IP› -a 4 -l 256000 -n 30000
At the conclusion of the test, we captured the throughput and CPU utilization figures from the client screen.The CPU utilization performance favors the Marvell Gigabit controllers with the Realtek solutions having the highest utilization numbers. The throughput numbers also favor the Marvell Gigabit controllers with the Realtek options close behind. ASUS recently switched to the Attansic L1 PCI Express based controller chip with it posting decent results in each test. We do not understand ASUS's decision to utilize a PCI based Gigabit controller on the 570SLI when the NVIDIA chipset has native support for Gigabit operations. This decision results in the worse throughput and average CPU utilization numbers. However, even with throughput at 646Mb/s it still exceeds what most home networks are capable of and certainly any DSL or Cable based Internet connection.
Facebook
Twitter
RSS
62 Comments
View All Comments
Gary Key - Saturday, October 21, 2006 - link
The DS4 will be reviewed. Gigabyte has stated they will not bring it into the US but we are still trying to convince them (really more like begging and calling our marketing rep at home on the weekends to beg some more) to release it in the States. The copper backplate can be removed off the DQ6 and I really did not see any benefit with it on in testing. It makes for a good rebate with the price of copper today. ;-)The AHCI issues stay the way they are at this time. It is frustrating to say the least. I was being a bit sarcastic in my statement but it is a little harder than it should be to enable AHCI on the ICH8R.
Ryan Norton - Sunday, October 22, 2006 - link
I'm in Taiwan, so the DS4 is all over the place :)Capt Caveman - Saturday, October 21, 2006 - link
Not sure if you live in the US or not but Gigabyte is not bringing/selling the DS4 to the US.Also, the copper backplate for the DQ-6, can be removed with a tworx(sp?) screwdriver. At XS, many just went to Home Depot and got longer screws for their HSF.
lopri - Saturday, October 21, 2006 - link
Excellent review that'll help potential buyers enormously. A couple things:1. Error in the chart (page 14): There is a discrepency between the chart (3-4-3) and the commentary (3-4-4) :)
2. In memory review Wesley always put tRP ahead of tRCD, while Gary does the opposite. It'd be nice to have a consistency for less experienced users!
3. Gary, did you test the P5W-DH with wirless module installed or without? I recently found out the wirless module could skew CPU/memory-sensitive benchmarks on this board big time. I'm not sure if my finding is true in general, but if it is, then the comparison between a board with such feature and a board without it can be unfair.
4. Can Sandra Unbuffered be really an indication of general performace? @400FSB, setting memory ratio 4:5 (DDR2-1000/4-4-3) boosted the score by a whooping 400~500 MB/s from the ratio 1:1 (DDR2-800/3-3-3), which never realized for other tests in a meaningful way. Oh this is about my own testing. :D
Thanks for the great review.
Gary Key - Saturday, October 21, 2006 - link
Thanks for the comments.1. The timing error is corrected. Jarred and I were editing at the same and we found out after the article went live that our saves to the final copy would overwrite each other. Bad timing for several other mistakes that have been cleared up now.
2. Wes is wrong. Just kidding, we will get on the same page. :)
3. I turned off the WiFi on the PSW-DH. The scores were even worse with it on. Not that they are bad but the board runs a little looser timings in order to overclock at the high end. DFI also does this with their boards targeted for the overclocking market.
4. Sandra Unbuffered can be an indication of performance in apps that are memory sensitive. This is not always the case but it is one of the better yardsticks available at this time.
Lothar - Saturday, October 21, 2006 - link
Do you plan on comparing the Gigabyte DS3 vs the S3 version for us to know if there are performance/overclocking issues with the S3?The only difference so far between the DS3 and S3 is "All Solid Capacitors".
Are there any other difference I'm missing?
The S3 is $110, and the DS3 is $150.
I have a hard time to justify paying an extra $40 for only "all solid capacitors"
The term is nothing but marketing to me so far.
I haven't seen any proof of a performance/overclocking issue between the two boards.
If you or anyone else plan on testing the differences or can provide something(Ex: any review link) stating otherwise, that would be great.
Nakazato - Monday, October 23, 2006 - link
In theory, cleaner power.... but aside from the theory, the onboard sound does start flaking out the higher you go. This has been true on 2/2 boards I've tried it on. So an add-in card is needed for the higher overclocks... 460+ish.goinginstyle - Monday, October 23, 2006 - link
No issue here with the Biostar board at 500FSB and the Realtek ALC-883. It sounds fine but a X-FI is still the way to go for gaming.Gary Key - Saturday, October 21, 2006 - link
Yes, the S3 will be in part two. :)
Lothar - Saturday, October 21, 2006 - link
The only other differences I found were RAID support and 2 extra USB ports...It's not worth the $40 price difference if performance and overclocking results are the same IMO.