- Oct 13, 1999
- 22,377
- 2
- 81
ECS NFORCE4-A939 MINI REVIEW
CLIFFNOTES VERSION:
Very inexpensive as a Fry's bundle and works exactly as it should. This board is a reasonable value for the price. For being "free" with CPU makes it a very worthwhile board. Not as attractive if purchasing on its own with better boards in similar price bracket. Decent overclocking. No Smart Fan.
OVERVIEW:
The point of this "mini review" is to give people an idea of what they can expect when purchasing one of these since most reviewers tend to ignore "bottom feeder" motherboards such as this one.
I went to Fry's Electronics on 4/25 and picked up this board along with an OEM Paris core Sempron 3100+ for $79.99 plus tax. I don't know if they just didn't have many of these boards or if they're selling quick because it was just in the ad from the day before and I had to wait about 10 minutes while two guys searched high and low for the board, finally finding one hidden behind some other boards. I'm accustomed to Fry's having stacks of the "combo sale" motherboards and indeed the other boards that are typically on sale in the combos were stacked all around. I also picked up a PNY 512MB PC3200 module with it for $39.99 - $20MIR. The board detected the PNY memory at DDR400 CAS 2.5.
MOTHERBOARD FEATURES:
Nforce4-4X chipset
1X PCIe 16X
2X PCIe 1X
3X PCI
3X DDR
4X SATA
5.1 Audio
10/100/1000 NIC
3 sets of USB headers plus 4 ports in rear, total 10 ports.
Phoenix/Award BIOS v1.0b (shipped version)
The board is a bit above basic. No firewire, no 7.1 audio, but most other things people need/want.
FIRST IMPRESSIONS
This board reminds me of the NFORCE4-A939 and indeed there is much resemblence between the two with similar chipset, similar model designation, similar layout and similar BIOS (including limitations, more later).
This board takes the nice layout of the socket 939 version and makes it even better. Similarities: The location of the 4 pin ATX +12v plug is right behind the PS/2 ports right at the edge of the PCB so that the wires do not have to cross over the CPU area. The 24 pin ATX plug is along the edge of the PCB by the RAM and just below it, also along the edge are the IDE ports. The three sets of USB headers are right along the bottom edge just past the front panel headers.
Differences: The Nforce4-4X chipset may be just a hair farther away from the PCI-E 16x slot which is nice because on the socket 939 board the chipset HSF is at an angle and one corner of it almost hits the video card. This board puts the chipset HSF a bit farther away. Also, the SATA ports are between the chipset and the edge of the PCB as with the socket 939 board, but since this board is narrower by a hair the SATA headers are even closer to the edge. Last, for fans of floppy drives, the FDD port is along the edge of the PCB between the SATA ports and the bottom corner. Not the ideal place, but much better than the location of the FDD port on the socket 939 board (under the last PCI slot).
The chipset fan is pretty noisy and spins at 5800RPM. The good news is that it easily clears the PCIe 16x slot so if your video card doesn't have anything sticking out the back of it, you can replace the HSF with a large passive heatsink. Note that this heatsink is held on by wire hooks like those found on some Pentium 4 boards that look like the end of a paper clip sticking through the board. I want to mention this because it seems as if most of the chipset heatsinks are held on by push pins with springs and this one is different.
The CPU socket is far enough away from the top edge of the board and away from the chipset HSF (with PCIe 16x slot below the chipset HSF). The closest tall object would be the first RAM stick, which is about 1" away from a retail box HSF (from Sempron 2600+).
The board can be used with a 20 pin ATX power supply. Indeed there's a sticker over the extra 4 pins that has to be removed to use a "proper" 24 pin power supply.
BIOS:
The board comes with BIOS version 1.0b with a limit of 250MHz HTT/FSB. Latest on the ECS web site is 1.1a so I flashed to latest version. Latest version raises the FSB/HTT limit to 400MHz, a higher number than any stock board can attain. I can't see any other major changes in CMOS setup.
The BIOS is the typical Phoenix/Award type that many boards use. It has all the "typical" settings and doesn't explicitly try to hide settings like some boards I've seen. I thought the previous ECS board I used, the NFORCE3-A had a lot of "enthusiast" options for an ECS but this one has even better voltage settings.
Advanced Chipset Features:
- clock speed from 200-400MHz in 1MHz increments
+++ (0.5MHz increments between 201-210MHz, 2MHz increments between 230-248MHz and perhaps other oddities)
- HT Frequency from 1-4x (default 4X)
- Memclock settings of 100/133/166/200MHz
- 1T/2T timing setting
- CAS latency 2/2.5/3
- CPU voltage from default to +375mV in 25mV increments
- DDR voltage defaults to 2.63v and goes from 2.55v to 3.11v in 0.08v increments
Power Management Setup:
-AMD K8 Cool'n'Quiet contro Auto/Disable [SIC]
-Hammer Fid control StartUp, 4-9 in ½ multiplier increments (9x is default on my CPU)
Yes you read this right. The Vcore will go to 0.375v above default in small increments, meaning a 1.40v CPU can be pushed to 1.775v!!! Not something you'd want to do on an everyday basis. Somewhat more useful is memory voltage to 3.11v, good for running RAM at full 1:1 speeds and low latencies on an overclock.
OVERCLOCKING RESULTS:
My goal was to see first how high HTT the board can go, and second how high MHz my processor can go.
First, high HTT. To achieve this, I minimized all other aspects by lowering memory to 133MHz setting, lowering HT Frequency to 3X and lowering Hammer Fid to 6X. I verified that system booted at a 1200MHz underclock with memory running DDR266. Knowing that furballi had a 289MHz HTT POST limit in his BIOS, I had a goal. Testing at 260MHz and successfully POSTed. Bumped to 280MHz and no POST. Okay. Reset BIOS and bump to 270MHz and no POST. Huh? Back down to 260MHz and bumping it up 5MHz at a time and it surpasses 270MHz. Seems like it doesn't like to have a high HTT slapped on it. Gotta sweet talk it a bit before it'll give me the goods. 275MHz worked. 280MHz, no POST. Strange, even underclocked I was unable to reach over 275MHz HTT POST (in 5MHz increments). Fiddled with various settings, no joy... until I raised CPU vcore.
I remembered that the PC Health Status reported CPU voltage as being a bit under spec, so I bumped it an arbitrary 100mV so the CPU gets 1.50v, or a reported 1.45-1.47v. Now I was easily able to get higher HTT. My only explanation for this is that the CPU needed voltage for the overclocked interface regardless of core clock. So now I worked my way up to 285MHz, then 289MHz HTT. At 6X Hammer Fid the CPU is still underclocked a hair at 1734MHz. Time to try 290MHz.
Nope. Just as furballi experienced, no POST at 290MHz HTT and above. Since we find this behavior on more than one board and it is at such an exact setting, the logical explanation would be a BIOS limitation.
The next step is to find maximum clock on the CPU that came with the board (though this review is on the board). Leaving settings of HT at 3X and memory at 133MHz, letting Hammer Fid go full blast at 9X and CPU overvolted, the first step was an easy 240MHz HTT. I did find another oddity in the BIOS. +125mV gave me the same voltage as +100mV in PC Health Status, but +150mV gave me 0.05v over +100/125mV. So, I left it on +150mV which detects at 1.50-1.52v.
My plan is to find highest POST speed, and then start testing for stability. 260MHz HTT, 2340MHz core. 270MHz HTT for 2430MHz and no POST. Time to get medieval on the voltage! Bump up to +175mV and again, doesn't detect at any more voltage than at +150mV, plus saving on exit and system powered off. Okay, NEED MORE VOLTAGE!!! +200mV gives me a detected 1.55-1.57v. 265MHz HTT for 2385, no sweat. Go to 270MHz, save and exit and board powered off again. Powering on, no POST. Looks like this Paris core CPU doesn't like to go over 2.4GHz, just like the one before it (came with NFORCE3-A bundle, tested on excellent Biostar Tforce6100 board).
Final settings:
HTT 260MHz
Core 2340MHz (detected 2347MHz)
+150mV vcore
Memory 166MHz (detected DDR426, CAS 3)
2.79v vDIMM
Tests stable at these settings. Note that I had occasional strange occurances of the board shutting itself off while exiting BIOS or hitting the reset switch. This only happened when highly overclocked even if the overclock seems stable (running various software torture tests).
POSSIBLE FUTURE TESTING
Might be stable at lower voltages or lower latencies.
Haven't tested for possible USB 2.0 lockups/BSOD under Windows as experienced on the socket 939 version.
CONCLUSION:
Very inexpensive as a Fry's bundle and works exactly as it should. This board is a reasonable value for the price. For being "free" with CPU makes it a very worthwhile board.
Unless you are an extreme overclocker or have a CPU with a low multiplier, this board is a reasonable choice.
For quiet freaks the placement of the Nforce4 chipset allows for easy replacement using a large passive heatsink though the hoop mouning instead of pin mounting makes it less universal. Also, lack of Smart Fan is an undesireable oversight.
If you are wanting a budget Nforce4 motherboard and do not need frills, this is a great choice. People wanting a budget build and getting this with a CPU as a cheap Fry's bundle need not shop for another motherboard.
If you were out shopping for the board by itself, makes for a tougher sell because you can get the decent Biostar Geforce6100-M7 board for around the same price which will give near the same overclock plus have decent integrated graphics, or for around $10 more you can get the most excellent Biostar Tforce6100 board which will absolutely overclock better (as a board, CPU dependent) and has features like user programmable Smart Fan. Also, the Biostar boards are passively cooled.
Note that clockgen will get around the BIOS HTT/FSB limitations.
CLIFFNOTES VERSION:
Very inexpensive as a Fry's bundle and works exactly as it should. This board is a reasonable value for the price. For being "free" with CPU makes it a very worthwhile board. Not as attractive if purchasing on its own with better boards in similar price bracket. Decent overclocking. No Smart Fan.
OVERVIEW:
The point of this "mini review" is to give people an idea of what they can expect when purchasing one of these since most reviewers tend to ignore "bottom feeder" motherboards such as this one.
I went to Fry's Electronics on 4/25 and picked up this board along with an OEM Paris core Sempron 3100+ for $79.99 plus tax. I don't know if they just didn't have many of these boards or if they're selling quick because it was just in the ad from the day before and I had to wait about 10 minutes while two guys searched high and low for the board, finally finding one hidden behind some other boards. I'm accustomed to Fry's having stacks of the "combo sale" motherboards and indeed the other boards that are typically on sale in the combos were stacked all around. I also picked up a PNY 512MB PC3200 module with it for $39.99 - $20MIR. The board detected the PNY memory at DDR400 CAS 2.5.
MOTHERBOARD FEATURES:
Nforce4-4X chipset
1X PCIe 16X
2X PCIe 1X
3X PCI
3X DDR
4X SATA
5.1 Audio
10/100/1000 NIC
3 sets of USB headers plus 4 ports in rear, total 10 ports.
Phoenix/Award BIOS v1.0b (shipped version)
The board is a bit above basic. No firewire, no 7.1 audio, but most other things people need/want.
FIRST IMPRESSIONS
This board reminds me of the NFORCE4-A939 and indeed there is much resemblence between the two with similar chipset, similar model designation, similar layout and similar BIOS (including limitations, more later).
This board takes the nice layout of the socket 939 version and makes it even better. Similarities: The location of the 4 pin ATX +12v plug is right behind the PS/2 ports right at the edge of the PCB so that the wires do not have to cross over the CPU area. The 24 pin ATX plug is along the edge of the PCB by the RAM and just below it, also along the edge are the IDE ports. The three sets of USB headers are right along the bottom edge just past the front panel headers.
Differences: The Nforce4-4X chipset may be just a hair farther away from the PCI-E 16x slot which is nice because on the socket 939 board the chipset HSF is at an angle and one corner of it almost hits the video card. This board puts the chipset HSF a bit farther away. Also, the SATA ports are between the chipset and the edge of the PCB as with the socket 939 board, but since this board is narrower by a hair the SATA headers are even closer to the edge. Last, for fans of floppy drives, the FDD port is along the edge of the PCB between the SATA ports and the bottom corner. Not the ideal place, but much better than the location of the FDD port on the socket 939 board (under the last PCI slot).
The chipset fan is pretty noisy and spins at 5800RPM. The good news is that it easily clears the PCIe 16x slot so if your video card doesn't have anything sticking out the back of it, you can replace the HSF with a large passive heatsink. Note that this heatsink is held on by wire hooks like those found on some Pentium 4 boards that look like the end of a paper clip sticking through the board. I want to mention this because it seems as if most of the chipset heatsinks are held on by push pins with springs and this one is different.
The CPU socket is far enough away from the top edge of the board and away from the chipset HSF (with PCIe 16x slot below the chipset HSF). The closest tall object would be the first RAM stick, which is about 1" away from a retail box HSF (from Sempron 2600+).
The board can be used with a 20 pin ATX power supply. Indeed there's a sticker over the extra 4 pins that has to be removed to use a "proper" 24 pin power supply.
BIOS:
The board comes with BIOS version 1.0b with a limit of 250MHz HTT/FSB. Latest on the ECS web site is 1.1a so I flashed to latest version. Latest version raises the FSB/HTT limit to 400MHz, a higher number than any stock board can attain. I can't see any other major changes in CMOS setup.
The BIOS is the typical Phoenix/Award type that many boards use. It has all the "typical" settings and doesn't explicitly try to hide settings like some boards I've seen. I thought the previous ECS board I used, the NFORCE3-A had a lot of "enthusiast" options for an ECS but this one has even better voltage settings.
Advanced Chipset Features:
- clock speed from 200-400MHz in 1MHz increments
+++ (0.5MHz increments between 201-210MHz, 2MHz increments between 230-248MHz and perhaps other oddities)
- HT Frequency from 1-4x (default 4X)
- Memclock settings of 100/133/166/200MHz
- 1T/2T timing setting
- CAS latency 2/2.5/3
- CPU voltage from default to +375mV in 25mV increments
- DDR voltage defaults to 2.63v and goes from 2.55v to 3.11v in 0.08v increments
Power Management Setup:
-AMD K8 Cool'n'Quiet contro Auto/Disable [SIC]
-Hammer Fid control StartUp, 4-9 in ½ multiplier increments (9x is default on my CPU)
Yes you read this right. The Vcore will go to 0.375v above default in small increments, meaning a 1.40v CPU can be pushed to 1.775v!!! Not something you'd want to do on an everyday basis. Somewhat more useful is memory voltage to 3.11v, good for running RAM at full 1:1 speeds and low latencies on an overclock.
OVERCLOCKING RESULTS:
My goal was to see first how high HTT the board can go, and second how high MHz my processor can go.
First, high HTT. To achieve this, I minimized all other aspects by lowering memory to 133MHz setting, lowering HT Frequency to 3X and lowering Hammer Fid to 6X. I verified that system booted at a 1200MHz underclock with memory running DDR266. Knowing that furballi had a 289MHz HTT POST limit in his BIOS, I had a goal. Testing at 260MHz and successfully POSTed. Bumped to 280MHz and no POST. Okay. Reset BIOS and bump to 270MHz and no POST. Huh? Back down to 260MHz and bumping it up 5MHz at a time and it surpasses 270MHz. Seems like it doesn't like to have a high HTT slapped on it. Gotta sweet talk it a bit before it'll give me the goods. 275MHz worked. 280MHz, no POST. Strange, even underclocked I was unable to reach over 275MHz HTT POST (in 5MHz increments). Fiddled with various settings, no joy... until I raised CPU vcore.
I remembered that the PC Health Status reported CPU voltage as being a bit under spec, so I bumped it an arbitrary 100mV so the CPU gets 1.50v, or a reported 1.45-1.47v. Now I was easily able to get higher HTT. My only explanation for this is that the CPU needed voltage for the overclocked interface regardless of core clock. So now I worked my way up to 285MHz, then 289MHz HTT. At 6X Hammer Fid the CPU is still underclocked a hair at 1734MHz. Time to try 290MHz.
Nope. Just as furballi experienced, no POST at 290MHz HTT and above. Since we find this behavior on more than one board and it is at such an exact setting, the logical explanation would be a BIOS limitation.
The next step is to find maximum clock on the CPU that came with the board (though this review is on the board). Leaving settings of HT at 3X and memory at 133MHz, letting Hammer Fid go full blast at 9X and CPU overvolted, the first step was an easy 240MHz HTT. I did find another oddity in the BIOS. +125mV gave me the same voltage as +100mV in PC Health Status, but +150mV gave me 0.05v over +100/125mV. So, I left it on +150mV which detects at 1.50-1.52v.
My plan is to find highest POST speed, and then start testing for stability. 260MHz HTT, 2340MHz core. 270MHz HTT for 2430MHz and no POST. Time to get medieval on the voltage! Bump up to +175mV and again, doesn't detect at any more voltage than at +150mV, plus saving on exit and system powered off. Okay, NEED MORE VOLTAGE!!! +200mV gives me a detected 1.55-1.57v. 265MHz HTT for 2385, no sweat. Go to 270MHz, save and exit and board powered off again. Powering on, no POST. Looks like this Paris core CPU doesn't like to go over 2.4GHz, just like the one before it (came with NFORCE3-A bundle, tested on excellent Biostar Tforce6100 board).
Final settings:
HTT 260MHz
Core 2340MHz (detected 2347MHz)
+150mV vcore
Memory 166MHz (detected DDR426, CAS 3)
2.79v vDIMM
Tests stable at these settings. Note that I had occasional strange occurances of the board shutting itself off while exiting BIOS or hitting the reset switch. This only happened when highly overclocked even if the overclock seems stable (running various software torture tests).
POSSIBLE FUTURE TESTING
Might be stable at lower voltages or lower latencies.
Haven't tested for possible USB 2.0 lockups/BSOD under Windows as experienced on the socket 939 version.
CONCLUSION:
Very inexpensive as a Fry's bundle and works exactly as it should. This board is a reasonable value for the price. For being "free" with CPU makes it a very worthwhile board.
Unless you are an extreme overclocker or have a CPU with a low multiplier, this board is a reasonable choice.
For quiet freaks the placement of the Nforce4 chipset allows for easy replacement using a large passive heatsink though the hoop mouning instead of pin mounting makes it less universal. Also, lack of Smart Fan is an undesireable oversight.
If you are wanting a budget Nforce4 motherboard and do not need frills, this is a great choice. People wanting a budget build and getting this with a CPU as a cheap Fry's bundle need not shop for another motherboard.
If you were out shopping for the board by itself, makes for a tougher sell because you can get the decent Biostar Geforce6100-M7 board for around the same price which will give near the same overclock plus have decent integrated graphics, or for around $10 more you can get the most excellent Biostar Tforce6100 board which will absolutely overclock better (as a board, CPU dependent) and has features like user programmable Smart Fan. Also, the Biostar boards are passively cooled.
Note that clockgen will get around the BIOS HTT/FSB limitations.
.
Facebook
Twitter