ATI’s Crossfire: Best Overclocker on the Market?
by Wesley Fink on September 27, 2005 12:05 AM EST- Posted in
- CPUs
Firewire and USB Performance
It is really difficult to put together a simple, repeatable, and consistent test to measure USB and Firewire Performance. Since our goal was to make this a standard part of motherboard testing, we needed a benchmark that was reasonably simple to run and that would also provide consistent results on the same test bed. We finally determined that an external USB 2.0, Firewire 400, and Firewire 800 hard disk might be a sensible way to look at USB and Firewire throughput.
Our first efforts at testing with an IDE or SATA drive as the “server” yielded very inconsistent results, since Windows XP sets up cache schemes to improve performance. Finally, we asset up a RAM disk as our “server”, since memory removed almost all overhead from the serving end. We also managed to turn off disk caching on the USB and Firewire side by setting up the drives for “quick disconnect” and our results were then consistent over many test runs.
We used just 1GB of fast 2-2-2 system memory set up as a 450MB RAM disk and 550MB of system memory. Our stock file was the SPECviewPerf install file, which is 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 or Firewire 400 or Firewire 800 drive using a Windows timing program written for AnandTech by our own Jason Clark. 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.
Perhaps the biggest surprise in our testing was the poor performance of USB 2.0 on the ATI Crossfire AMD. ATI had promised improvement in USB performance in the SB450 compared to the earlier SB400, but we could see little improvement in our USB 2.0 tests. ATI responded that our USB 2.0 tests measured sustained throughput, which is relevant in USB operations only with applications like the USB 2.0 hard drive that we use for this test. ATI claims that burst USB 2.0 throughput has been improved in the SB450 to near the same levels of competitive USB 2.0 peak performance. We could not confirm these claims with any test procedures. Perhaps more important, ATI is aware of the issues with USB 2.0 performance and these will be fully corrected – sustained and peak performance – in the SB600 south bridge. It appears that this issue of USB performance may also be corrected by combining the ATI RD480 northbridge with the just released ULi M1575 Southbridge.
Since our ramdisk/apptimer File Copy is measuring sustained throughput, we also compared performance of ATI IDE/SATA/Sil SATA2 using this same test procedure. The performance of ATI IDE remains excellent, but SATA and SATA 2 performance are very competitive, but not standout as in the iPeak storage benchmarks of the same controllers.
It is really difficult to put together a simple, repeatable, and consistent test to measure USB and Firewire Performance. Since our goal was to make this a standard part of motherboard testing, we needed a benchmark that was reasonably simple to run and that would also provide consistent results on the same test bed. We finally determined that an external USB 2.0, Firewire 400, and Firewire 800 hard disk might be a sensible way to look at USB and Firewire throughput.
Our first efforts at testing with an IDE or SATA drive as the “server” yielded very inconsistent results, since Windows XP sets up cache schemes to improve performance. Finally, we asset up a RAM disk as our “server”, since memory removed almost all overhead from the serving end. We also managed to turn off disk caching on the USB and Firewire side by setting up the drives for “quick disconnect” and our results were then consistent over many test runs.
We used just 1GB of fast 2-2-2 system memory set up as a 450MB RAM disk and 550MB of system memory. Our stock file was the SPECviewPerf install file, which is 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 or Firewire 400 or Firewire 800 drive using a Windows timing program written for AnandTech by our own Jason Clark. 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.
ATI uses the commonly available and good performing VIA Firewire 400 chip. Firewire 800 is 40 to 55% faster than a drive connected to Firewire 400, but it is still rare to find Firewire 800 integrated into motherboards.
Perhaps the biggest surprise in our testing was the poor performance of USB 2.0 on the ATI Crossfire AMD. ATI had promised improvement in USB performance in the SB450 compared to the earlier SB400, but we could see little improvement in our USB 2.0 tests. ATI responded that our USB 2.0 tests measured sustained throughput, which is relevant in USB operations only with applications like the USB 2.0 hard drive that we use for this test. ATI claims that burst USB 2.0 throughput has been improved in the SB450 to near the same levels of competitive USB 2.0 peak performance. We could not confirm these claims with any test procedures. Perhaps more important, ATI is aware of the issues with USB 2.0 performance and these will be fully corrected – sustained and peak performance – in the SB600 south bridge. It appears that this issue of USB performance may also be corrected by combining the ATI RD480 northbridge with the just released ULi M1575 Southbridge.
Since our ramdisk/apptimer File Copy is measuring sustained throughput, we also compared performance of ATI IDE/SATA/Sil SATA2 using this same test procedure. The performance of ATI IDE remains excellent, but SATA and SATA 2 performance are very competitive, but not standout as in the iPeak storage benchmarks of the same controllers.
40 Comments
View All Comments
ShadowVlican - Tuesday, September 27, 2005 - link
well i hope i can buy the Xpress200.... ATi ain't real until i can buy their motherboards readily everywhere i go (like nVidia's NF4)Beenthere - Tuesday, September 27, 2005 - link
It's no secret ATI has the ability to produce equal or better chipsets and GPUs than Nvidia - they have done this before. What ATI needs to do is get their sh*t together on the details and CUSTOMER SERVICE - Yeah, they've heard of it but evidently they don't know the MEANING OF IT ! Nvidia ain't much better, but ATI's so called Customer Support is a bad joke. Delivering what you promise is a VERY IMPORTANT CONSIDERATION when you are charging the enthusists market segment El PREMO PRICES for your hardware and you had better DELIVER THE GOODS. ATI has failed miserably and Nvidia ain't far behind despite the fact both companies have reaped fortunes from the consumer enthusiast market. Until both companies improve their CUSTOMER SUPPORT neither are getting any of our corporate dollars.yacoub - Tuesday, September 27, 2005 - link
Okay so the first good vendor (Asus or similar) to come out with a passively-cooled northbridge, the ALC-880"D" audio chip, a Southbridge with better USB performance, and the rest (it can even be a single GPU board so it's under or around $100 in price) gets my money. =Pyacoub - Tuesday, September 27, 2005 - link
Okay that wasn't supposed to reply to your post. Interesting.Live - Tuesday, September 27, 2005 - link
Does ATI crossfire support NCQ hard drives or not? Have I understood it correctly in that it is not supported trough the 4 SATA ports from the Southbridge but you can get support from the 2 ports from the included Sil controller?I have read the http://www.anandtech.com/storage/showdoc.aspx?i=24...">What's in a name? SATA II Misconceptions
But I still fail to get the facts straight. Does a SATA 2 controller, either from a SB or separate like the Sli, support all of the capabilities in the SATA 2 specs as long as the hard drive does so?
Since I plan on going dual core next, no NCQ seems like a deal breaker to me.
Wesley Fink - Tuesday, September 27, 2005 - link
The SB450 Southbridge does not support NCQ. The ULi M1573, used on the retail Gigabyte Crossfire AMD and some upcoming retail boards, DOES support NCQ. The Silicon Image 3132 on the ATI Reference Board supports both NCQ and 3Gb SATA2 on the extra SATA ports.The just introduced ULi M1575 Southbridge, which can be used with the ATI Crossfire Northbridge (as soon as it hits the market) supports 3Gb SATA2, NCQ, PCIe Gigabit Ethernet, Azalia HD audio, and features competitive USB throughput. The ATI SB600 will also implement all these features.
etriky - Tuesday, September 27, 2005 - link
Anyone have some links to shed some light on this quote."There is a lot of discussion on the web these days claiming that you can minimize the impact of the 2T setting with certain options on Revision E AMD processors."
I've done some searches and come up with nothing.
Thanks.
bigtoe36 - Tuesday, September 27, 2005 - link
forcing a burst length of 8 and forcing burst2opt can bring back some of the lost performance going to 2T, both these features are on the crossfire reference board.Also, i mamaged to talk DFI into making a direct copy of the ATI reference that will run reference bios files, we should see this board in October. This board will be released along side the board already designed by Oskar...so you will have the choice of 2 CF boards from DFI.
Palek - Tuesday, September 27, 2005 - link
On page 1, 2nd paragraph the article says:"AMD had done a particularly excellent job targeting the enthusiast for the new chipset launch, but that realization seemed to come late in the chipset development process."
I suspect that AMD should be ATI.
On page 3, 1st paragraph:
"AMD Crossfire" first, then "Crossfire AMD" later. "ATI Crossfire" or "ATI Crossfire AMD" may be less confusing.
Also on page 3, in last paragraph "AMD Reference Board" is used twice, but the board is referred to earlier on in the article as the "ATI Crossfire AMD Reference Board". "AMD Reference Board" makes it sound like AMD made it.
Question:
How hot did the northbridge get during various phases of your testing? I think a lot of users would appreciate some info regarding operational temperatures. If the ATI chipset turns out to run a lot cooler than nForce4 chipsets, I will gladly forgive the USB speed problems and go with an easier to cool motherboard.
Wesley Fink - Tuesday, September 27, 2005 - link
Corrected.See other comments for NB heat and cooling. My subjective observation is the ATI Northbridge is cooler during heavy OC than the nF4 under the same conditions. However, many of you push boards a lot further than I do so you can take that with a grain of salt. ATI designed this board for the enthusiast and extrene overclocking and temperature under stress was a definite design consideration.