MSI K8N Neo4 nForce4 Ultra Socket 939

[kenton]

No idea what PWM stands for, Power Windows Management? Google should know but I don't care ;P

I'm sure there's a way to rewire the pins, all you'd need is a pin-poker-outer-dealie, poke out I guess the PWM wire and replace it with the RPM wire (which should be yellow, but I can't check ATM)


Good news on my temps. At 2500MHz (which Primed for 21 hours by the way, I only stopped it to install my new Antec TruePower Trio 650) my temps have fallen to not peaking 50C. Something to be noted, I did have the cables all strewn crazy with the Antec 350 because I knew I'd be replacing it ASAP. I ALWAYS neaten/ziptie my cables, my case is so clean you could almost eat off of it. I do an excellent tuck job ;P

I'm about to reboot at 2600MHz stock voltage and we'll see if she can prime for another almost 24 hours.

 

[Zap]

Pulse-width modulation

PWM can be used to reduce the total amount of power delivered to a load without losses normally incurred when a power source is limited by resistive means...
PWM is also often used to control the supply of electrical power to another device such as in speed control of electric motors, volume control of Class D audio amplifiers or brightness control of light sources and many other power electronics applications. For example, light dimmers for home use employ a specific type of PWM control...
These rather simple types of dimmers can be effectively used with inert (or relatively slow reacting) light sources such as incandescent lamps, for example, for which the additional modulation in supplied electrical energy which is caused by the dimmer causes only negligible additional fluctuations in the emitted light. Some other types of light sources such as light-emitting diodes (LEDs), however, turn on and off extremely rapidly and would perceivably flicker if supplied with low frequency drive voltages.

Basically instead of using resistance to bleed off voltage (inefficient, inaccurate and creates heat relative to PWM) PWM pulses full power, thus approximating a duty cycle. Some of the fancier fan controllers use PWM like this one that I used to have. LED fans were like strobe lights on this unit so I ended up just giving it away (the fan controller, not the fans).

 

[cardiac]

Originally posted by: Zap
Pulse-width modulation

PWM can be used to reduce the total amount of power delivered to a load without losses normally incurred when a power source is limited by resistive means...
PWM is also often used to control the supply of electrical power to another device such as in speed control of electric motors, volume control of Class D audio amplifiers or brightness control of light sources and many other power electronics applications. For example, light dimmers for home use employ a specific type of PWM control...
These rather simple types of dimmers can be effectively used with inert (or relatively slow reacting) light sources such as incandescent lamps, for example, for which the additional modulation in supplied electrical energy which is caused by the dimmer causes only negligible additional fluctuations in the emitted light. Some other types of light sources such as light-emitting diodes (LEDs), however, turn on and off extremely rapidly and would perceivably flicker if supplied with low frequency drive voltages.

Basically instead of using resistance to bleed off voltage (inefficient, inaccurate and creates heat relative to PWM) PWM pulses full power, thus approximating a duty cycle. Some of the fancier fan controllers use PWM like this one that I used to have. LED fans were like strobe lights on this unit so I ended up just giving it away (the fan controller, not the fans).

Great explanation.


Thank you,

Bob

 

[GeezerMan]

Well, you guys talked me into trying this board with a Opteron 165 myself. My 165 came in with the stepping of CCBBE 0616XPMW, so it should be a good one. It came with the smaller, 70mm fan on the heatpipe. I'm assuming I would be better of using the 80mm version of the heatpipe. My theory is the lesser heatpipe was shipped with the slower speeds, like the 1.8GHz Opteron.

 

[kenton]

2600MHz at stock voltage primed for 24 hours : http://www.kenton.org/msi2600opty.JPG

I did experience an odd shutoff after I had about 3 or 4 hours into my first prime. I was using the computer while it was under stress (Firefox, maybe Winamp) and it just shut off. Now, to be fair, this did happen with the X2 4400+ I had in here briefly a time or two and it may've happened with my Opty when I was running at stock speeds for a few days.

I have no idea why it did that, but nonetheless I've got proof that I'm stable at 2600MHz at stock voltage for 24 hours.

BTW: I got 0617, everyone else seemed to get 0616, but it's CCBBE so that's all I cared about


Geezerman, does your OEM heatsink look like: http://i20.ebayimg.com/05/i/06/57/10/08_1.JPG -- I was under the impression that was the "superior" heatsink out of AMD's OEM offerings.

 

[GeezerMan]

Originally posted by: kenton
2600MHz at stock voltage primed for 24 hours : http://www.kenton.org/msi2600opty.JPG

I did experience an odd shutoff after I had about 3 or 4 hours into my first prime. I was using the computer while it was under stress (Firefox, maybe Winamp) and it just shut off. Now, to be fair, this did happen with the X2 4400+ I had in here briefly a time or two and it may've happened with my Opty when I was running at stock speeds for a few days.

I have no idea why it did that, but nonetheless I've got proof that I'm stable at 2600MHz at stock voltage for 24 hours.

BTW: I got 0617, everyone else seemed to get 0616, but it's CCBBE so that's all I cared about


Geezerman, does your OEM heatsink look like: http://i20.ebayimg.com/05/i/06/57/10/08_1.JPG -- I was under the impression that was the "superior" heatsink out of AMD's OEM offerings.

You are correct, your pic is of the better AMD HSF. Mine that came with the Opeteron is not that one, it's this one: Link to Fan Pic


I'll probably install the better one, since I have one sitting here, or my Arctic Cooling Pro 64

 

[blckgrffn]

Either of the OEM opteron dual core HS are better than the vast majority of the ones that came with any of the single cores and even the 3800+ and 4200+ X2's. You should be set.

Or just use the AC heatsink. Personally, I like that best out of the options you have

Nat

 

[GeezerMan]

How about overclocking with the stock AMD heatpipe?

 

[kenton]

GeezerMan, strange! Mine came with the regular 4 pipe, I bought two from that seller (one cools my 2.76GHz 754 Venice) and one is cooling my dad's 3400+/stock Venice and then obviously I got one with my boxed Opty.

All I can think of is maybe they ran out of the old pipe heatsink on Week 17 and started shipping these coolers.

 

[kenton]

Sorry for double posting again. I think it's this old ass version of Norton 2003 that messes up my javascripts. Old Norton + Updates > Freeware, in my opinion. When you have 2GB of RAM you don't have to worry about the 30MB Norton eats... Plus, out of all the Norton naysayers I've *NEVER* had *ONE* problem. Not one, ever!

Anyways, overclocking using the 4 pipe heatsink is wonderful at least on single core processors. I had an Arctic Cooling Alpine64 (it's a super large Al block with a 92mm fan on top, I mean the size of the heatsink dwarfs anything OEM and it was $6 at newegg) and it didn't let me Venice go past 2.6GHz without me getting worried about heat. It was getting pretty close to 55C under load and gaming and this was at 1.58-1.6V (love those Venices!!!!)

Anyways, I slapped on the four piper and I was able to overclock another 167MHz without passing 55C. There's several Venice/754 loyals on Hardforum that say 1.6v and 55C is the limit on air so I reached it happily.





Also, I'm having a goofy problem out of this Opteron.

- Prime both cores at 2600MHz -- Stable for 24 hours
- Prime ONLY CORE0 at 2600MHz -- Stable for 5 hours
- Prime both cores again -- Stable > 5 hours (we'll see what happens, if it's another 24 hour prime)


Tempted to part out and sell this machine on eBay and sell some other hardware that's collecting dust (brand new motherboards, couple of Clawhammers, some low-megabyte DDR RAM, etc) and get a CORE 2 DUO...

 

[GeezerMan]

Yours was made in week 17, mine was made in week 16. Maybe they upgraded in week 17. I just read the review on Frosty Tech again on the 2 pipe version. It was rated number 2 ( at that time)for cooling, better than the Freezer Pro 64. The huge Ultra-120 was number 1, but it's like a skyscraper

That's very Impressive. I'll just watch the temps while overclocking. I feel deja vu, I know I have discussed this before.

Frosty Link

 

[GeezerMan]

Originally posted by: kenton
Sorry for double posting again. I think it's this old ass version of Norton 2003 that messes up my javascripts. Old Norton + Updates > Freeware, in my opinion. When you have 2GB of RAM you don't have to worry about the 30MB Norton eats... Plus, out of all the Norton naysayers I've *NEVER* had *ONE* problem. Not one, ever!

Anyways, overclocking using the 4 pipe heatsink is wonderful at least on single core processors. I had an Arctic Cooling Alpine64 (it's a super large Al block with a 92mm fan on top, I mean the size of the heatsink dwarfs anything OEM and it was $6 at newegg) and it didn't let me Venice go past 2.6GHz without me getting worried about heat. It was getting pretty close to 55C under load and gaming and this was at 1.58-1.6V (love those Venices!!!!)

Anyways, I slapped on the four piper and I was able to overclock another 167MHz without passing 55C. There's several Venice/754 loyals on Hardforum that say 1.6v and 55C is the limit on air so I reached it happily.





Also, I'm having a goofy problem out of this Opteron.

- Prime both cores at 2600MHz -- Stable for 24 hours
- Prime ONLY CORE0 at 2600MHz -- Stable for 5 hours
- Prime both cores again -- Stable > 5 hours (we'll see what happens, if it's another 24 hour prime)


Tempted to part out and sell this machine on eBay and sell some other hardware that's collecting dust (brand new motherboards, couple of Clawhammers, some low-megabyte DDR RAM, etc) and get a CORE 2 DUO...

Funny, I can relate. One can get these deals, like the MSI-7125, the Opteron 165, etc. One day, you wake up, and you have brand new parts laying around doing nothing for you but collecting dust. Sell it all, and you can have one hot machine.

 

[jhill1977]

Your dilemna with your Plextor drive sounds VERY familiar to me. I had a PX-716A do the same thing to me after about 16 months. Mine would read and write DVDs..but no luck reading or writing CDs or CD-Rs.

Originally posted by: Lurker1
Got the board, got it installed and flashed (what a pita - no floppy and a quirky Plextor that won't recognize CDR??? WTF?) and I'm up and running. Looks good so far, still loading drivers. Looks like a smokin deal though.

Oh, and Geeks definitely could use a lesson in packing these.... Geez, I've had boards bought on ebay sent better packaged than this.

 

[kenton]

http://www.kenton.org/2700.JPG

Decided to switch to the oldschool dual instances of Prime95, setting each one to Core0 and Core1, respectively.

Anyways, Core0 would fail after half a second with stock voltage @ 2700MHz (300HTT x 9). Upped voltage in the BIOS to "1.420" which actually results in 1.38-1.41v.

As we can see, stable for 24 hours. I'll do 2800MHz next because that's my final goal and then I'll work myself down on the voltage, if I can.

 

[GeezerMan]

Originally posted by: kenton
http://www.kenton.org/2700.JPG

Decided to switch to the oldschool dual instances of Prime95, setting each one to Core0 and Core1, respectively.

Anyways, Core0 would fail after half a second with stock voltage @ 2700MHz (300HTT x 9). Upped voltage in the BIOS to "1.420" which actually results in 1.38-1.41v.

As we can see, stable for 24 hours. I'll do 2800MHz next because that's my final goal and then I'll work myself down on the voltage, if I can.

How can you tell that if you set the voltage to 1.420 it is actually 1.38 - 1.41 ?
How do you know which program to believe?
Thanks

 

[kenton]

Using CPU-Z and SpeedFan (which has a graph option) I can see voltages over time. Unfortunately, SpeedFan only -seems- to report the last 13 minutes in the graph. Bit of an arbitrary number, indeed.

Of course without a digital multimeter I don't know how to accurately intercept actual voltage, and even if I owned one I wouldn't know what to attach it to to get CPU voltage. Not keen on placing the points directly on my motherboard when it's powered on.

I don't think I'm going to mess with it much today, I'm dead tired... After flying to Atlanta -- getting your heartbroken on a STUPID attempted punt return from your own 5 (LET THE FSCKING BALL GO INTO THE END ZONE) and after flying back home (Little Rock) all in the span of like 30 hours I'm beat.

 

[kenton]

Well, I lied. I'm really bored today (NFL = yuck) and I played around with it some more.

Upping the HTT from 300 to 311 results in 2800MHz and a perfect DDR400 memory ratio. This was my ultimate goal. At 1.39-1.41 (1.420 BIOS) voltage Core0 failed immediately, much like it did at 2700MHz. Reboot and set voltage at 1.428 (resulted in the same reported voltage as 1.420) -- failed. Reset voltage to 1.44 in the BIOS (actual was 1.42-1.43) and it failed again. Reset voltage to 1.47 (wow!, reported 1.44-1.45) and it still failed.

I don't have the cooling needed (i.e. water or removed IHS) to up the voltage too much more than 1.420 so I'm perfectly happy at 2700MHz.

Settings in the BIOS:

Core Voltage: 1.375V + 3.3% = 1.420
HTT: 300
HTT Multiplier: 3x (Actual HTT speed is 1800MHz out of 2000MHz, there will be zero performance loss)
DDR Divider: 133MHz (Actual is 192MHz or DDR384 -- Not too shabby)


Overall I'm very happy out of this $200 motherboard/CPU combo (laugh.. 25% of that was this great board!) My final review on this motherboard is at least a 9 out of 10. I love the layout including the floppy drive connector (don't use it, but if I ever had to...) and IDE locations.

I don't understand why we can't put those useless PCI-E slots on the bottom of the board, as having 4 PCI slots is nice as hell (most single PCI-E 16x mobos have 2-3) but it just sucks that two will be blocked by any normal PCI-E graphics card and I'd like to put an exhaust blower below the graphics card leaving 1 free PCI slot which is taken by my AuzenTech X-Plosion sound card (I need the DTS encoding, I have a real HT)

My girlfriend is getting me the BFG GeForce 7950 GT OC Edition at Newegg for Christmas so this machine is about to be rocking.

For anyone curious because I did a lot of stability testing on this board, this is my system specifications:

Athlon 64 Venice 3400+ & Opteron 165 cooled with Arctic Cooling Freezer 64 Pro w/ Stock MX-1 paste
MSI K8N Neo4 Motherboard (duh!)
Ultra 2x1GB DDR400 RAM (in the slots closest to the FRONT of the case)
ATi Radeon X600 SE 128MB PCI-E 16x (cheapest eBay card!)
AuzenTech X-Plosion DDLive/DTS PCI sound card
Two Optical Drives
One IDE Hard Drive (soon to be replaced by a huge SATA)
Ultra X-Finity 500w, replaced by an Antec TruePower Trio 650

Once I drop in the fat SATA drive and the graphics card this machine will umm... seriously own.

 

[GeezerMan]

Originally posted by: kenton
Well, I lied. I'm really bored today (NFL = yuck) and I played around with it some more.

Upping the HTT from 300 to 311 results in 2800MHz and a perfect DDR400 memory ratio. This was my ultimate goal. At 1.39-1.41 (1.420 BIOS) voltage Core0 failed immediately, much like it did at 2700MHz. Reboot and set voltage at 1.428 (resulted in the same reported voltage as 1.420) -- failed. Reset voltage to 1.44 in the BIOS (actual was 1.42-1.43) and it failed again. Reset voltage to 1.47 (wow!, reported 1.44-1.45) and it still failed.

I don't have the cooling needed (i.e. water or removed IHS) to up the voltage too much more than 1.420 so I'm perfectly happy at 2700MHz.

Settings in the BIOS:

Core Voltage: 1.375V + 3.3% = 1.420
HTT: 300
HTT Multiplier: 3x (Actual HTT speed is 1800MHz out of 2000MHz, there will be zero performance loss)
DDR Divider: 133MHz (Actual is 192MHz or DDR384 -- Not too shabby)


Overall I'm very happy out of this $200 motherboard/CPU combo (laugh.. 25% of that was this great board!) My final review on this motherboard is at least a 9 out of 10. I love the layout including the floppy drive connector (don't use it, but if I ever had to...) and IDE locations.

I don't understand why we can't put those useless PCI-E slots on the bottom of the board, as having 4 PCI slots is nice as hell (most single PCI-E 16x mobos have 2-3) but it just sucks that two will be blocked by any normal PCI-E graphics card and I'd like to put an exhaust blower below the graphics card leaving 1 free PCI slot which is taken by my AuzenTech X-Plosion sound card (I need the DTS encoding, I have a real HT)

My girlfriend is getting me the BFG GeForce 7950 GT OC Edition at Newegg for Christmas so this machine is about to be rocking.

For anyone curious because I did a lot of stability testing on this board, this is my system specifications:

Athlon 64 Venice 3400+ & Opteron 165 cooled with Arctic Cooling Freezer 64 Pro w/ Stock MX-1 paste
MSI K8N Neo4 Motherboard (duh!)
Ultra 2x1GB DDR400 RAM (in the slots closest to the FRONT of the case)
ATi Radeon X600 SE 128MB PCI-E 16x (cheapest eBay card!)
AuzenTech X-Plosion DDLive/DTS PCI sound card
Two Optical Drives
One IDE Hard Drive (soon to be replaced by a huge SATA)
Ultra X-Finity 500w, replaced by an Antec TruePower Trio 650

Once I drop in the fat SATA drive and the graphics card this machine will umm... seriously own.

All the great equipment, and a girl too. Maybe you have the girl because of your great equipment.:laugh:

Thanks for your postings. I will now try my first overclock....:shocked:

 

[LeonarD26]

My old 30 gig drive is going to be retired since my new 250gig Seagate SATA drive is arriving today (SATA is new for me . Do I need to do anything special when installing XP on this new system? Does this board support large hard drives out of the box?? I plan on going with a ~40gig C: and the rest on D:

BTW, I have the HTT at 260 and has been running stable all weekend. I plan on lowering this down for my XP install, but so far I haven't had any problems overclocking...

 

[GeezerMan]

Originally posted by: LeonarD26
My old 30 gig drive is going to be retired since my new 250gig Seagate SATA drive is arriving today (SATA is new for me . Do I need to do anything special when installing XP on this new system? Does this board support large hard drives out of the box?? I plan on going with a ~40gig C: and the rest on D:

BTW, I have the HTT at 260 and has been running stable all weekend. I plan on lowering this down for my XP install, but so far I haven't had any problems overclocking...

Just plug it in. I never had any problems installing a SATA and loading windows. Now setting up a raid takes more work

 

[kenton]

I have not installed Windows on this machine using SerialATA. I do know that if you're installing an OS from a DVD (Like Windows Vista) your DVD-ROM/RW must be the secondary master, I had my CDRW as Sec. Master and DVD-ROM as Sec. Slave and Vista would not install.


As far as I know, though, Serial ATA is built into XP SP2 so if you're using an SP2 disc it shouldn't need drivers. The SerialATA comes from the chipset and not a 3rd-party (i.e. Promise, Silicon Image, etc)

 

[Lurker1]

Kenton:

I used a SATA drive and XP SP1. Worked flawlessly. BTW, my DVD+-RW is master on the primary IDE - no problems there either. Its the only device on IDE though, there's issues when you have IDE and SATA hard drives, or you're trying to boot off of IDE only - the boot HD needs to be the master on the primary. Also note that boot options allow you to select boot order and bootable devices. IDE boot can be disabled for an IDE/SATA HD split, but I don't know the specifics, as I haven't actually tried that combination myself.

I've only successfully OC'd to 2300 straight up at stock voltages (RAM running at 230). Anytime I change any of the other settings (so far) the MB refuses to POST. Evidently somewhere something no longer works the way I thought. I'm going to use your last couple of postings tonight to try again.

On the disk side, I have been playing with multiple partitions and Norton Ghost and Partition Magic. I'm trying to create a nice ready to go small clean image for games separate from my "working" install. With bootup times around 15s or less, I can afford to reboot. Right now I'm wishing that XP could not see multiple primary partitions... it was safer back then when you couldn't corrupt your alternate installations.

Turns out that Partition Magic appears to be the way to go - can copy a partition and everything just works. Don't know what they've done to Ghost, but it's not the Ghost my friends talk about. Can't create images anymore.

 

[Zap]

Originally posted by: Lurker1
I've only successfully OC'd to 2300 straight up at stock voltages (RAM running at 230). Anytime I change any of the other settings (so far) the MB refuses to POST.

The pertinent settings are dropping RAM to 166 and HT multiplier to 4X. When you change the RAM, you may also have to change the timings - manually set them to what they are supposed to be at DDR400 or higher.

 

[GeezerMan]

Yeah, I'm doing my first overclock tonight with the Opteron 165, and this motherboard. Everyone stand back.
I got it going with no boot or obvious problems at 2200MHz. Have not done any Prime95 testing yet. I would like to see if I can get 2600MHz at stock cpu voltages. I have 2GB of Corsair XMS PC3200. Corsair says I can set voltage up to but not including 2.9
Looks like the command rate is greyed out in the bios at 2. Can't choose 1
I figure I will set the ram at 2 3 3 6 since it's rated at that. Set ram to 166, HT multiplier to 4, HTT to 288

How does this look?

Thanks

 

[kenton]

I'd leave RAM at auto for now, or failing boot upon auto something safe like 3-3-3-8 2T, reason being -- why introduce an element of instability when you're stressing another component?

Since my CPU is already known stable for 24 hours dual core loaded at 2700MHz and since I'm only running at 192MHz/DDR384 I plan on going from 3-4-4-8 2T to 3-4-4-8 1T (1T offers the most significant gain) and running "memtest" until I get 500% coverage without error (this should take 4-5 hours with 2GB of RAM) and then I'll move to 3-3-3-8 and then 3-3-3-7.

Now, once you get your CPU stable at your desired speed, feel free to move faster with memory timings than I plan to, you seem to not have budget RAM and my machine is the definition of "Anandtech Hot Deals forum" (Every part, save for the case/opticals was influenced by this forum) you don't have as much to worry about as I do. I'd start at 3-3-3-7 and once you know for sure the CPU is good to go, hit up the RAM timings and run memtest and then some more Prime95. I know a lot of people think 6, 8, or even 12 hours Prime stable is really "stable." When it comes to the finer things in life -- including a fat overclock I prefer to know what I'm getting is actually what I'm getting. I want to know 100% that my computer is stable and won't embarrass/annoy/timeburglar me at any time unless it's due to an unforeseen component failure. Run it for 24 hours, most of us have several computers anyways so what do you have to lose?

Case in point, my Venice 3400+, which I've decided to put in my KN1 Lite (I've got 2x512MB Mushkin DDR400 and tons of IDE HDD/Opticals) and grab another Rosewill case on Newegg and my X-Finity just came back from Ultra so I've got basically another system right there. Video card coming from the X600SE that's being replaced by the 7950GT 512MB. I might even put it in the paper for $350 and see what happens. Oh I digress, my Venice failed Prime at 19 hours in with 1.58v. With 1.61 it was perfectly stable for 24 hours.

One thing, you need to know what memvoltage your RAM is stable with those timings, default DDR is 2.5v and this board maxes at 2.8v. This means nothing to me since I have budget RAM running under-spec, but it means a great deal to you since you mention Corsair says up to 2.9v.

That PC3200 oughta run up to about 240MHz with some relaxed timings, 3-3-3-8 probably, especially since it's HQ Corsair. My first stick of XMS was a 512MB stick of DDR333 that ran perfectly stable at DDR400 in a KT400 motherboard. The KT400 chipset didn't even officially support DDR400.

To hit 2600MHz dead on, you'll want 289, by the way.