DFI NFII Ultra: Mean Green Dream Machine
by Wesley Fink on July 30, 2003 6:25 PM EST- Posted in
- Motherboards
DFI NFII Ultra: BIOS and Overclocking
DFI uses the familiar Award BIOS on the NFII Ultra. Our test board was equipped originally with the 6/05/2003 Performance BIOS. While we were impressed with the incredible array of options available in this BIOS, we had some problems with the stability of the BIOS in some extreme overclocking situations. The later 6/25/2003 BIOS proved to be fast and very stable, and thus, benchmarks were run using this BIOS.DFI later supplied an updated 7/10/2003 Performance BIOS, which proved to combine the excellent stability of the 6/25/2003 BIOS with the incredible selection of tweaking options available in the 6/05/2003 BIOS.
The Performance BIOS has the most incredible array of overclocking options that we have ever seen on an nForce2 board.
FSB - 100-300 in 1MHz intervals
AGP - Auto or FIXED at 1Mhz intervals from 50-100
CPU Ratio - 5X-22X
vCore - Auto, 1.10-2.05 in 0.025v increments
vAGP - Auto, 1.5-1.8 in 0.1v increments
Chipset Voltage - Auto, 1.6-1.9 in 0.1v increments
vDIMM - Auto, 2.5-2.8 in 0.1v increments
RAM Ratios - Auto, 2:1. 5:3, 3:2, 4:3, 5:4, 6:5, 1:1, 5:6, 4:5, 3:4, 2:3, 3:5, 1:2
T(RAS) - 1-15
T(RCD-Read) - 1-7
T(RCD-Write) - 1-7
T(RP) - 1-7
T(RC) - 9-31
T(RFC) - 9-31
Super Bypass - Enable, Disable
Sync Mode Memory Bypass - Auto, Enable, Disable
Data Scavaged Rate - Fast, Normal
DIMM 1 Driving Strength - Auto, 1-15
DIMM 1 Skew Rate - Auto, 1-15
DIMM 2 Driving Strength - Auto, 1-15
DIMM 2 Skew Rate - Auto, 1-15
DIMM 3 Driving Strength - Auto, 1-15
DIMM 3 Skew Rate - Auto, 1-15
CAS Latency - 2.0, 2.5, 3.0
The options available in the Performance BIOS are certainly a demonstration that DFI does know what is involve in producing a board squarely aimed at the computer enthusiast. How can you not be impressed with an enthusiast's dream collection of tweaking options in the Performance BIOS available for this board?
The voltage adjustments, FSB adjustments, and ratio controls are in a menu item DFI calls “Genie”.
We can tell you the added options do make a difference in squeezing every bit of performance from the NFII Ultra, but we have only scratched the surface in mastering the available tweaks. The speed and overclocking abilities with this BIOS are impressive. A word of caution here also: you can get yourself into trouble with the available options if you're not careful. If you don't understand these options, choose AUTO; otherwise, you may be reformatting your hard drive. The 6/25/2003 release version of the BIOS includes a full array of options, but does not include all these options; rather settling for optimum values for most situations. But AMD tweakers will love this Performance BIOS.
The NFII Ultra PC Health section includes System and CPU temperatures, CPU and Chassis fan speeds, Vcore, and Power Supply readings for all rails. There is also an adjustable setting for Maximum CPU temperature that will automatically shut down the system, if the preset temperature is exceeded. DFI includes the familiar Winbond Hardware Monitor for keeping tabs of voltages, temperatures and fan speeds in Windows.
Voltage adjustment options for the DFI NFII Ultra are very well-balanced. We were pleasantly surprised to see vCore adjustments to 2.05V – much higher than offered by most other nForce boards, except those from Epox. The vDIMM adjustment range, while much better than the DFI 975PRO LanParty, tops out at 2.8V. Although this should be adequate for most users, we are now seeing new high-speed memory from OCZ and GEIL that offers warranty support for use at voltages up to about 3.0V. We would have preferred to see the vDIMM reach to the 3.0V to 3.2V range, but in reality, we will not be running nForceII boards at DDR500 as we might on an Intel 875P board. Considering the practical overclocking limits of the nForce 2 Ultra 400, we would conclude that the 2.8V maximum voltage for memory is probably a good choice. Besides, we have chipset voltage adjustments on the NFII Ultra – all the way to 1.9V. In testing overclocks on the NFII Ultra, we found that we often attained better results getting beyond walls by increasing the chipset voltage a bit, rather than increasing vDIMM. Certainly, the wide chipset adjustment range coupled with the decent vDIMM adjustment range gives the overclocker many very effective options.
FSB Overclocking Results
For FSB overclocking, the following setup was used on the NFII Ultra:| Front Side Bus Overclocking Testbed | ||
| Default Voltage | Maximum Overclock | |
| Processor: | Athlon XP 2500+ (Barton Core) | Athlon XP 2500+ (Barton Core) |
| CPU Vcore: | 1.65V (default) | 1.775V (default) |
| Cooling: | Thermalright SK-7 with 80mm Thermaltake Fan | Thermalright SK-7 with 80mm Thermaltake Fan |
| Power Supply: | Powmax 350W | Powmax 350W |
The above overclocking setup at default voltage allowed us to reach a stable FSB of 223MHz. This makes the DFI NFII Ultra the best overclocking nForce2 Ultra 400 board that we have tested at AnandTech. With this kind of performance at default voltage, we were curious just how far we could push the Barton 2500+ with a bit more voltage. We were able to achieve complete stability at 227MHz using 1.775V. This represents a true CPU speed of 2.502GHz from a chip designed to run at 1.83GHz – a 37% overclock. Since the 3200+ runs at 2.2GHz, this would represent a performance rating in the range of 3600+, if such a rating existed.
The DFI NFII Ultra has a very wide range of multiplier adjustments, available from 5.5X all the way up to 22X. The whole range of multipliers appeared to work well on our Barton 2500+. We did not see any missing ranges in the multiplier settings, as have been reported on other nForce2 boards.
Facebook
Twitter
RSS
46 Comments
View All Comments
Anonymous User - Friday, August 1, 2003 - link
This article is useless because it doesn't help anyone that is looking for a high-end board.
1.Readers want to see how it compares in games! Even though it is stated that different video cards are used, these numbers do not help anyone.
2.Readers want to see how it compares against other top of the line boards! (why not compare it against a P3, it would be just as usefull!)
3."Performance tests for the DFI NFII Ultra LanParty were run with the ATI 9800 PRO 128MB video card with AGP Aperture set to 32MB"
AGP Aperture set to 32MB??? Most people would set this to 128MB! Is there a compatibility problem that should of been stated?
This article should either be fixed or removed from AnandTech's website as it is damaging to their reputation.
If nothing is done about this article then it shows how much AnandTech listens to it's readers.
-no insult intended towards anyone-
Anonymous User - Friday, August 1, 2003 - link
This article is a bit low on quality. Visiting the forums I know there are tons of qualified guys/gals that would love, including myself, to write/work at anand. I can't believe that this new guy was the best thing they found. I agree with other comments posted. I also hate with a passion the new benchmark result that are used in some of the article, some were flash... I restrict what runs on my browser because I hate to see a woman f****** a cow when I browse. What happened to the plain jpeg/gif of the past? Would much prefer that since otherwise no point of reading the articleAnonymous User - Friday, August 1, 2003 - link
just wanted to respond to number 18, who quoted one of my comments.Actually man, I have been a NVIDIA fan and own several NVIDIA products. However as with most of us gamers you have to go where the speed, performance and quality is. I do own a 9700 pro and currently use it for my main gaming, but then again why not? does NVIDIA produce anything that compares?
NVIDIA has had their last 2 product lines fail, along with there cheats and shortcuts to produce good numbers. I seem to remember another company that did those things, can we say 3DFX? Who baught them???? NVIDIA.
my point was not that i am biased, but that ATI is currently the top of the line for speed, quality and performance, besides the fact it supports the new directx9.
which is better to do tests on after all, outdated and slow technology, or up to date top of the line technology?
Anonymous User - Thursday, July 31, 2003 - link
Jeez, why would anyone even bother writing articles for such ungrateful SOBsAnonymous User - Thursday, July 31, 2003 - link
I may have missed it, but how much is this thing gonna cost?justly - Thursday, July 31, 2003 - link
There is an explanation about the scores, at the bottom of the preformance test configuration page you can read this"Many benchmarks show widely different results with different video hardware, so we have indicated benchmarks run with the ATI Radeon 9800 PRO with an asterisk. Benchmarks without an asterisk were run with the nVidia Ti4600."
Iam glad to see the move to the ATi 9800 Pro, this eliminates any video bottleneck and allowes for 8X AGP compatibility testing.
Anonymous User - Thursday, July 31, 2003 - link
This article really needs fixin!First major complaint is the use of old graphics cards. Compare apples to apples. Not apples to oranges to peaches.
Second, the Asus A78NX was not tested either. This IS the gold standard with AMD enthusiasts. How can we make a good comparison?
Third, where are the game tests and 3D Mark scores?
Anonymous User - Thursday, July 31, 2003 - link
In response to #6(Wesley):I'm not saying your numbers are wrong, I'm saying your description seems wrong. As #21 points out, that description makes absolutely no sense, and as far as I've read, it's wrong. RAID 1.5 is a RAID 1 mirror with "optimized" reads; nothing more. Check Tom's Hardware, I believe they have a good article on this.
Anonymous User - Thursday, July 31, 2003 - link
Umm....Striping and mirroring with 2 drives only?OK...Let's look at this the way he explained it:
Take 2, 80 GB drives...
Half of each contains is striped, the other half used to mirror the stripe.
In what way is this useful? If 1 drive fails, you've lost the stripe AND the useless mirror...
Please explain to me why this is a good thing, Anandtech....
-Phil Green
LM Information Systems
Anonymous User - Thursday, July 31, 2003 - link
ok i wont comment the different video cards used, im sure you realise this isn't right... especially for the games benches.hopefully you'll make up, testing the other boards with this raddy too... also all tests that were done on just this board shall be done to the rest of the boards in time... when you fix all these things i hope you'll put up some notice on the main page.
i got a major complaint however... the idea to use flash for displaying the graphs isn't good at all. the newest flash plugins for mozilla are incredibly slow, and almost make my pc freeze as i open multiple tabs with your articles(with at least 2 flash adds on each page) so i prefer to disable my flash plugin. i know i dont represent the majority of your readers here, so i'm not important... but yet i think you should consider simple gifs for your graphs.
now another thing... why does the forum open in such a weird window? i mean, no addressbar and toolbar, etc. thats kinda annoying.
now a question about the test results... i find some specview results quite weird. in a couple if tests the dfi scores quite less than the rest, and then there's that test where dfi scores 5 times more than the rest.... i'd like to read your comment on these tests, hopefully you have some explanation.
bye folks, and forgive my bitching :) inspite it, i do like your site and thank you for the articles. you're doing a great job