HELP! NF7-S cannot overclock Athlon 2500+ to 3200+
First, thanks for clicking on this post. Your time and effort is greatly appreciated!
Components: (Mobo+RAM+Proc from newegg received 1/5/2004):
Abit NF7-S
2 x 256MB, PC3200 (Buffalo Tech using Winbond CH-5 chips)
AMD Athlon 2500+ (retail box) with the following codes:
AXDA2500KV4D KO22244l31038
PQZEA0332MPMW
Alpha PAL 8045 Cooled by a 80mm fan
Visiontek Geforce3 Ti 200 (using 41.09 WHQL drivers)
WD 160 GB Special Edition JBRTL
16x DVD Lite-ON IDE
24x CD-RW Yamaha IDE
---powered by an Antec TruePower 430 and enclosed in a CoolerMaster ATC-201
Proceedure:
Upon installing the NF7, I updated to the latest BIOS, overclocked the 2500+ by selecting the 3200+ (three selections are available: 2500+; 3200+, and User Defined)
WinXP Pro + SP1 installed without a hitch under the selected 3200+ (using default voltage). Then, I installed the 3.13 nForce drivers. I noticed that computer would freeze 3-10secs within playing an MP3. Hmm, not good!
Started playing around with BIOS settings, specifically memory timings and voltage settings. After reading in these forums, I upped the Vcore voltage (tried different settings from 1.675->1.8) in order to gain stability. I also increased the VDimm voltage and the chipset voltage in various combinations to no avail. As for the memory timings, I tried all the various settings, from “by SPD” to “Optimal” and “Aggressive.”
At this point, I was not sure where the problem lay: memory or motherboard? So, I downloaded memtest and another memory tester from Microsoft. Memory passed both series of tests (from each program) without error. At this point, I think that the problem is related to the onboard sound (because it would freeze when playing MP3s). So, I disabled the onboard sound and installed Wolfenstein Enemy Territory. Proc is still overclocked and it freezes a couple seconds after joining a server. Went to BIOS, enabled sound and set CPU back to 2500. Game does not freeze, MP3s + DVDs all play fine. I set the memory to asynchronous mode (mem 400 + Proc 333) and all still ran stable.
I have spent many hours changing settings in an attempt to reach around 2.2 ghz (3200+) stable (without games crashing and without getting errors from Prime95). I am disappointed because a friend of mine has a $60 Shuttle nforce2 that ran his 3200+ without a problem. Any advice is welcome.
Bios settings which remained consistent:
SoftMenu III (CPU Interface Enabled)
Advanced BIOS Features Setup (Virus Warning Disabled; APIC enabled)
Advanced Chipset Features (System BIOS Cache Disabled, Video RAM Cache Disabled, CPU Throttling Disabled, ALL Spread-Spectrum Disabled)
Integrated Peripherals (SATA Disabled , FireWire Disabled, Parallel Port Disabled, Serial#2 Disabled)..REST all DEFAULT (but PCI transfer enhanced Disabled)
Components: (Mobo+RAM+Proc from newegg received 1/5/2004):
Abit NF7-S
2 x 256MB, PC3200 (Buffalo Tech using Winbond CH-5 chips)
AMD Athlon 2500+ (retail box) with the following codes:
AXDA2500KV4D KO22244l31038
PQZEA0332MPMW
Alpha PAL 8045 Cooled by a 80mm fan
Visiontek Geforce3 Ti 200 (using 41.09 WHQL drivers)
WD 160 GB Special Edition JBRTL
16x DVD Lite-ON IDE
24x CD-RW Yamaha IDE
---powered by an Antec TruePower 430 and enclosed in a CoolerMaster ATC-201
Proceedure:
Upon installing the NF7, I updated to the latest BIOS, overclocked the 2500+ by selecting the 3200+ (three selections are available: 2500+; 3200+, and User Defined)
WinXP Pro + SP1 installed without a hitch under the selected 3200+ (using default voltage). Then, I installed the 3.13 nForce drivers. I noticed that computer would freeze 3-10secs within playing an MP3. Hmm, not good!
Started playing around with BIOS settings, specifically memory timings and voltage settings. After reading in these forums, I upped the Vcore voltage (tried different settings from 1.675->1.8) in order to gain stability. I also increased the VDimm voltage and the chipset voltage in various combinations to no avail. As for the memory timings, I tried all the various settings, from “by SPD” to “Optimal” and “Aggressive.”
At this point, I was not sure where the problem lay: memory or motherboard? So, I downloaded memtest and another memory tester from Microsoft. Memory passed both series of tests (from each program) without error. At this point, I think that the problem is related to the onboard sound (because it would freeze when playing MP3s). So, I disabled the onboard sound and installed Wolfenstein Enemy Territory. Proc is still overclocked and it freezes a couple seconds after joining a server. Went to BIOS, enabled sound and set CPU back to 2500. Game does not freeze, MP3s + DVDs all play fine. I set the memory to asynchronous mode (mem 400 + Proc 333) and all still ran stable.
I have spent many hours changing settings in an attempt to reach around 2.2 ghz (3200+) stable (without games crashing and without getting errors from Prime95). I am disappointed because a friend of mine has a $60 Shuttle nforce2 that ran his 3200+ without a problem. Any advice is welcome.
Bios settings which remained consistent:
SoftMenu III (CPU Interface Enabled)
Advanced BIOS Features Setup (Virus Warning Disabled; APIC enabled)
Advanced Chipset Features (System BIOS Cache Disabled, Video RAM Cache Disabled, CPU Throttling Disabled, ALL Spread-Spectrum Disabled)
Integrated Peripherals (SATA Disabled , FireWire Disabled, Parallel Port Disabled, Serial#2 Disabled)..REST all DEFAULT (but PCI transfer enhanced Disabled)
0
Comments
Necro might be on to something, try installing MBM.
BUT, my guess is that you'll find that your CH5 chips are holding you back. While the BH5's were amazing, some of the CH5's are having trouble running at stock speeds.
Also, you should probably use the "User Defined" option under BIOS and increase your FSB settings incrementally. (rather than jumping straight from 166->200)
Best of Luck,
PS: One piece of advice, I'd make sure the system is running smoothly at stock speeds (everything: CPU, Mem, Vid, etc) before you begin overclocking.
I played around with a few bioses yesterday. I usually have the first bios, but modded by EQuito, D10. DEAD stable at pretty high overclocks. (11.5X230 SYNC, DC) Never had a reboot or freeze at that setting. Tried bios 20 and Call Of Duty kicked back to desktop within 5 minutes. I'm not saying it has to be the bios that is faulty but it might be worth a check. The CH5 chips isn't that bad actually. The newer ones actually goes as high as BH5 with lots of voltage.
Set the timings to 11,3,2, cas 2,5 (we'll change it later) at 166 fsb in 6/6 (sync), disable the 2 cache options in the bios (video and one more, can't remember the name), set the VDDR (memory voltage) voltage to 2.8, vcvore to 1.85 (we'll lower it later on if possible), 1.6 on the chipset voltage, default on the other voltage.
A couple of questions: Is that cpu unlocked? Can you take a reading of the voltages in the bios? 12V, 5V and 3.3V is all you need to check.
We'll sort this out.
CPU DISCONNECT is enabled by default. DISABLE it if you haven't already.
this may be a dumb question, but why is disabling this better?
First, an update: I suspected that some problems may have been caused from installing windows on an overclocked system. So, in order to eliminate this possibility, I set everything to default speeds and reloaded WinXP + SP1 + drivers (40.72 WHQL'd for Video and the 2.xx nforce downloaded from Abit website). (BTW, my CPU temps at stock are 35-37C)
Now, i attempted to overclock! I followed Mackanz's advice (upped Vcore and Mem. Voltage, set RAM timings, etc...). I knew that my CPU and Mem were stable at default voltage (having run Prime95 on them), so I attempted to gradually OC.
I went from 166 to 173 to 177... it seemed fine until I started to reach the high 180s and 190s. At that point, I OC slowly (2-3 MHz increments). Although I have not been able to fully test each setting (aka >12hrs Prime95) I worked along the principle of finding *unstable* settings rather than proving stable ones. My method was to run Prime95 for about 10-15 mins. Often, this was more than adequate to detect errors. Sadly, at 199 MHz FSB, Prime95 would spit out an error in the first min of testing. The threshold of stability seems to be 195-198 MHz (with vcore increased to >1.8, temps hovered around 47C-51C loaded).
What piece of hardware is to blame? Is it the Mobo or the Proc? RAM maybe? I am fairly sure it is not RAM, as I have tested the module for errors numerous time. As for the Processor, I am left with more questions than answers. According to the CPU code and stepping (AQZEA0332MPMW) the "AQZEA" stepping is not the best nor the worst and should almost guarantee a 3200+ with added voltage. Moreover, having been manufactured the 32nd week of 2003 the multiplier *should* be unlocked--yet it is not!
Lastly, on a side note, how much voltage is too much voltage? Is a Vcore of 1.8v getting dangerous? How about the RAM? Does such an increase only raise the temps, or are their long-term consequences to consider. In other words, at what point should a trade-off be made in voltage to long-term stability (at least 2-3 years)?
Cordially,
Nathan
One more thing, if I am right, and if your HD's don't like 39.6 PCI bus, you may have already corrupted your nice fresh OS install. Aren't I just full of....good news?
I'll try and google this but I'm not sure how to word the search....
1.8 vcore should be fine for the processor, if your max loaded temp is 51C. There are many running 1.85 and higher, though over that you would want more advanced cooling.
Memory voltage (Vdimm) is a topic of some debate. I am of the school of thought that it must be almost impossible to kill memory with stock motherboard voltages. I always run my memory at max available on the systems I overclock, meaning 2.85-2.9v 24/7 for months and months on my 2 current rigs. This is not, however, long term testing and I do not know the answer over a 2-3 yr period. I do know of many overclockers who feed 3.1-3.3v to their memory via voltage modding their motherboards and have not seen a story of someone killing their RAM with voltage - tho I give the standard "YMMV" disclaimer.
In your shoes, I would not hesitate to go to 2.8v or even 2.9v with your memory, see if that helps, then work backwards down to stability. But at 200 fsb/PC3200 it's stock RAM setting and absolutely should run stable at stock volts/SPD timings. Voltage increase should be reserved for seeking tighter timings or overclocked speeds.
To TheSmJ: Yes, I update the BIOS to the latest one, though I kinda wonder if it may handicap OC in some way.
To Keto: Thanks for your suggestions. Concerning the PCI bus speeds, I have read that the PCI bus is locked on the nForce2 boards. (See: http://www.overclockers.com/tips00213/ )
This makes me rest a little easier, though I am still determined to figure out which component is hampering my OC. If I pump enough juice to the CPU (say 1.8-1.9) , should that not force it to hit 2.2 ghz unless it is POS?
BTW, could increasing the northbridge voltage help out (and if so, why)?
Thanks
The AGP frequency is locked on a NF7-S but the pci bus isn't, that is controlled by the divider which for 200+ fsb should be set to 6:6 or anything that gives a 1:1 ratio eg 4:4 or 1:1.
Regards
Jim
mtgoat, I think here its a case of suck it & see cos on mine it wont go above 200 on default voltage and will only get to 210 on max, 215-220 with a volt mod. My previous NF7-s would again had to be raised to get above 200 and on max voltage would hit 234fsb.
Mtgoat, err I think something has been lost in translation over the pond, if you're refering to my "suck it and see" comment its not meant to be rude or vulgar. All it means in the UK is that you need to try it to see what result you get, it has no rude connotations over here.... unless one's mind is in the gutter.
Therefore I apologise for any offense caused but I promise you none was intended, I dont do that sort of thing unless under extreme provocation.
Regards
Jim
Apology accepted as well as offering my own as I was the one who forgot the amount of water between us, thus my lack of latitude in the interpretation.
Larry, no problem mate, glad we we got it cleared up, wots that phrase about being seperated by a common language, well I think we just valiadated it.
Oopps sorry again, Thrax is right.
The divider just sets the ratio to lock the pci bus at 33MHz, correct??
That's still a pretty good overclock that you have. DO NOT use Bios 20. Unstable as grandma on rollerblades. Use Bios 17 if you want a newer bios, or the modded D10 if you want performance (and for me stability). What's the total cpu speed right now?
Weird that your cpu is locked. What happens if you lower the multiplier?
Rename the B1 and B4 to .ace.
The EQ bios is the one from EQuito, works very well. I'm currently testing the others. B1 and B4 have totally different menu structure and hidden timings setting which you CANNOT change. The "under 10 multi" bandwidth issue is solved with the B1 and B4 bios it seems. Try them at your own risk. They work for me though.
In BIOS:
vCore - 1.85
vDDR - 2.7
chipset - 1.7v
ABIT EQ:
vCore - 1.8
vDDR - 2.7
Haven't really tried to go any higher since it's stable at this speed.
John.
Well that certainly brought a smile to my face . Urm not quite but nearly John.
I certainly would'nt recommend replacing the word sip with suck in most situations when talking to us British as generally we do use "sip" in most of the same situations as you guys do.
However ,the phrase "suck it & see" has been around for "donkey's years" and has been used in everything, from advertising indegestion tablets to try before you buy offers. But essentianly Ageek's sentiment is correct.
I just, naively, assumed that as its in common language over here it would be the same, and mean the same for you guys.
Apologies for going off topic yet again.
Jim
Back on topic, I may be switching to Equito's D10 BIOS. I'm currently using his modified 20 with the SATA drivers and as far as I know it I don't have any stability issues. I'm currently running a 2500+ at 215 FSB on a 11x multiplier for 2365 Mhz in a Koolance system.