Capacitor replacement on ASUS A8N32-SLI Deluxe...


 Share

Recommended Posts

Posted (edited)

Well I fixed that issue with system not being able to reboot when overclocked (as stock speed, reboots work fine (even with the lowered voltage)). in short... I had to increase CPU voltage a bit for rebooting to work in the 2.4Ghz overclocked state.

 

it seems it needs a little more juice in a overclocked state as at stock speed of 2.0Ghz I can run 1.300v on CPU and it reboots fine (and as I said any lower than 1.300v the system just won't boot and I gotta reset CMOS with jumper etc). but when I increase to 2.4Ghz and at the same 1.300v, during a cold boot it works fine, but any reboots (even adjusting settings in BIOS and saving and rebooting acts the same(but it does save the settings)) it does the reboot but hangs with HDD light on and no display (even pressing reset button, like I was saying, does something with DVD drives but ultimately don't do anything) and then I got to hold power button until it powers off and back on etc.

 

but I had to increase CPU voltage three steps in BIOS (i.e. from 1.300 to 1.3125 to 1.325 to 1.3375) before rebooting the PC (or even going to BIOS and saving settings and rebooting this way also) would work in general. but with the increase in CPU Voltage things now play out like this...

 

-VCore Voltage = 1.33-1.34v.

-Prime95 (room temp about 79.5f) = 62c peak, but usually around 59c. so temps went up a little here (vs when using 1.300v) but I am not too concerned as it takes a while for it to go over about 59c in general and I would rather have the rebooting functioning like normal for a minimum enough increase in CPU temps.

 

also, I noticed setting 'Q-Fan Control = ENABLED' (it's disabled by default) lowers fan speed and you can hear a noticeable decline in noise as CPU fan speed RPM drops quite a bit.

 

with Prime95 running CPU fan speed seems to top out around 5000rpm give or take 50rpm (at least within 20min of Prime95 which seems to be enough to peak CPU temp to my knowledge). at a idle it's around 2200rpm tops. NOTE: Linux 'watch sensors' command claims 'max' is 7200rpm. but assuming this is roughly accurate, my guess is the board won't start to really ramp up RPM's any further unless temps start to shoot up a fair amount beyond what I am experiencing(?).

 

I noticed once I stop Prime95, the fan speed RPM's still float at the 5000 or so for roughly 30sec-1min before a sharp drop in fan speed rpm (and noise) to about 2200rpm fairly quickly and then I noticed as it sits a while it slowly declines further as it's currently at about 1600 (it may go even low lower but I did not want to wait too long to see how far it will go). but just loading up the Neowin.net front page and scrolling a bit etc, since it puts a decent load on the CPU, you can see RPM's of fan creep up a bit to maybe 4000rpm or so but quickly go back down when CPU load drops shortly after.

 

p.s. I am actually using 1.1375 on 'Processor Voltage' in BIOS paired with the 'Over-Voltage CPU VCore = Enabled' which adds 200mw. hence, 1.3375v. which seems to be inline with that 'watch sensors' command of 1.33-1.34v.

---------------------------------------------------

 

Rubycon ZLJ 820uF 6.3V Low-ESR Impedance 105C radial capacitors caps 8mm... "8,000hrs at 105°C This rating is tested by the manufacturer at the maximum rated temperature and ripple current. The endurance of electrolytic capacitors roughly doubles for every 10°C reduction in temperature. At an operating temperature of 65°C, the estimated endurance would be about 128,000hrs."

 

128,100 hours is basically over 14.5 years ; so this computer will likely be ancient/dead by the time the caps act up in it again and it's not likely this computer will see 24/7 use. if it does, it probably won't be for any real length of time (it's possible I could be using this computer for a while if my main PC ever dies though (assuming it handles what I need it to handle well enough, which it might ;) ) so ill be good for the foreseeable future ;)

Edited by ThaCrip
Link to comment
Share on other sites

I was playing around with overclocking a bit further to 2.5GHz and tweaking RAM MHz and timings a bit and... 2.4Ghz is pretty much the sweet spot with 1.3375v to CPU along with 369MHz on RAM with 2.5(cas)-3(TRCD)-3(TRP)-7(TRAS) timings which is better than what the board was doing in 'auto' mode which RAM timings are a bit more relaxed. in 'auto' mode while it shows 266MHz for RAM on screen the real RAM speed was basically 300Mhz. so it appears when in 'auto' mode the board plays things a bit safer.

 

when I did 2.5Ghz, booted up and did Prime95 it errored shortly into the testing and upon reboot, it hanged and I increased CPU voltage up to 1.4v to see if that cured it and it still hanged and I do not really feel comfortable going beyond 1.4v to CPU. so as I mentioned above, I basically shifted back to 2.4Ghz/1.3375v and everything works well as this should give me a decent performance increase that's also safe (or minimally more risk vs stock 2.0Ghz).

 

from reading around online here is the gist of the formula to use for calculating RAM speed when overclocking on this A8N32-SLI board...

 

10 x 200 / 166 = 12.048 (and unless the number is perfectly even you always round up. so since it's not exactly 12 and is a bit higher, we round up to 13 here)

10 x 240 / 13 = 184.6 (double that number which basically comes out to 369.2, so basically 369MHz)

 

the first section is the default multiplier etc and the second section is my adjustment of the default 200 up to 240 to get the 2.4Ghz. also, that 166 number above assumes your using that in the BIOS to set RAM speed as if you set it to anything else you can adjust this stuff accordingly as the next higher option of 183 ends up putting RAM too high to 436Mhz as when I did that, it was not all that long after booting the OS that Linux froze solid. so then I dropped back to 166 and it seems to have cured it since speed is now 369MHz.

Link to comment
Share on other sites

Hello,


Have you noticed any difference in temperature reports from the overclocking?  These sounds like such minor increases in voltage (in practical terms) so I am curious as to what sort of increase they actually cause in the thermal loading for the machine.

Regards,

Aryeh Goretsky

Link to comment
Share on other sites

Posted (edited)
3 hours ago, goretsky said:

Have you noticed any difference in temperature reports from the overclocking?  These sounds like such minor increases in voltage (in practical terms) so I am curious as to what sort of increase they actually cause in the thermal loading for the machine.

 

Yeah, there is some difference. in short... about a 6c difference between 2.0GHz to 2.4GHz. NOTE: I just used the two examples (listed below) where 'rebooting' works like it should.

 

here is a bit more details... Ill say this (the A8N32-SLI BIOS defaults to 1.4v to CPU (the CPU apparently is officially rated for 1.35v according to... https://www.techpowerup.com/cpu-specs/athlon-64-x2-3600.c431 ) in regard to Prime95 SmallestFFT for 20 minutes on each test (room temp would be roughly 75-80f)...

 

-2.0Ghz (stock) @ 1.400v (i.e. BIOS defaults) (room temp about 75.2f) = 59c peak (maybe a little lower in general though)

-2.0Ghz (stock) @ 1.300v (room temp about 79.3f) = 56c peak, but usually around 53c give or take 1c. NOTE: 1.300v is the lowest I can go as any lower the system simply does not work at all (rebooting works here @ 1.300v).

-2.4Ghz (overclocked) @ 1.300v = 59c peak, but usually 56-57c give or take a degree. NOTE: while this works, 'rebooting' does not work.

-2.4Ghz (overclocked) @ 1.3375v (room temp about 79f) = 62c peak, but usually around 59c. NOTE: 1.3375v is the minimum voltage I can use when system is at 2.4GHz and have 'rebooting' work as expected.

 

basically like I mentioned in my previous post(s)... 2.4GHz is the most I can go "easily" (sounds roughly typical for CPU's of the type from what I can tell). when doing 2.5Ghz, the 'rebooting' problem returns at 1.3375v (Prime95 errored here to at 2.5GHz) and I tried going up to what the board defaults to which is 1.4v and still did not solve the reboot issue (so I did not attempt any further Prime95 tests at 2.5GHz beyond the 1.3375v test). I suspect I could go even higher voltage to CPU to potentially cure the 'reboot' issue at 2.5GHz and get a stable system, but it seems like it's going to take a solid increase in CPU voltage, which I imagine would be a solid increase in heat (i.e. higher risk), for minimal performance gain which is why I feel 2.4Ghz is probably the 'sweet spot' and should be pretty safe, since it's probably a minimal increase in risk vs 2.0Ghz @ 1.300v, and I am still supposedly running a lower voltage to CPU of 1.3375v vs what that TechPowerUp site said above with 1.35v for it's 'official' rating. so I feel safe enough even though according to the Linux 'watch sensors' command it claims '60c' is 'high' for this CPU and '95c' is 'critical' for this CPU. I figure even if we assume their assessment is correct with the 60c being a little risky, it's not likely the CPU would be 60c+ much at all in general when I am using that computer since to reach 60c+ it would likely have to have 100% CPU for some minutes as, off the top of my head, Prime95 takes it a little while before CPU temps peak/level off.

 

side note: who knows, maybe ill try running a Prime95 test in the future with the PC CASE OPEN and see if that fairs any better as I would not be surprised if there is some difference since the case that A8N32-SLI board is in is from a computer I got in 2001 and it's cooling is not exactly good as it's probably worse than most cases since there is no exhaust fan (besides what the PSU's 120mm pushes out) and only one roughly 80mm fan on side of case, which is a intake. hell, there is not even a place to put a exhaust fan on that case. so as you can see, without testing, it seems plausible opening case may help a little(?).

 

also, since I was using 250 x 10 to get that 2.5Ghz, I am still safe here as far as the board goes (i.e. "SB to NB Frequency" ) since I lowered the default 5x down to 4x (I am using the 4x option on all of my overclocking so far) which means it's 1000MHz (or 960MHz at 2.4GHz CPU), which is what the board does at it's defaults. but if I start going any higher than 2.5GHz I would then have to lower that 'SB to NB Frequency' from 4x to 3x as with 3x in place, even if I tried say 2.6Ghz (260 x 10), the board would be lower than 1000Mhz (i.e. 780MHz) and supposedly lowering this has negligible performance. NOTE: it appears one can go up from the 1000Mhz a bit (1000Mhz is the default) but it's safer to keep that to 1000-ish or less to help ensure this won't be a issue when overclocking from what I read about this A8N32-SLI board.

 

p.s. on my main PC's i5-3550 CPU... I seem to get about a 13c decrease in temps under load by lowering CPU by -0.130v (it goes by 0.010v increments), which is the most I can lower than and have a stable system as when I went to -0.140v, while it seems to work okay for a little while, usually within a day or two (off the top of my head) the system will just freeze solid on the image it's displaying on screen etc. but with the current -0.130v I have no system stability issues as I can leave computer running 30+ days without a reboot etc. but as you can see temp drops seem to have a noticeably bigger effect on my i5-3550 vs that A8N32-SLI boards AMD Athlon X2 3600+ CPU.

Edited by ThaCrip
Link to comment
Share on other sites

Ill blow budmans mind with this, but during lockdown I recapped an Asus P2B. People recap IBM 5160s. Just because something is old does not mean its not worth fixing.

  • Like 2
Link to comment
Share on other sites

Posted (edited)

In regards to 'side note' etc comment from my previous post...

 

I just did another Prime95 test for 20min with CASE OPEN (although door was still sort of pointed into case (but it's roughly a few inches away) and a little air might be flowing into it from 80mm case door fan) and with room temp at 81.1f the CPU peaked @ 62c but was usually 59-60c give or take 1c. so in this regard it's roughly the same (maybe a little better(?) because according to clock in room the temp is higher today than when I did the last test). under load 1.33-1.34v but at idle seems to be 1.36v.

 

even CPU fan speed might have dropped roughly 300rpm or so (4700rpm vs about 5000rpm).

 

also, at least based on a rough guess using my hand... it seems air flow out the back of the PSU is a little lower and air might be a little cooler (but I can't say this for sure though) and I noticed after stopping Prime95, not long after (which I feel more confident on the following claim), it seems airflow out the PSU slowed a bit more and was even cooler. but to speculate... this slowing down of airflow after stopping Prime95 could be because of standard stress the CPU load is putting on PSU in terms of wattage draw(?). so it's working harder as it's only 430watt (Seasonic), which ain't bad, but it's not like I got plenty of watts to spare from a guesstimate. EDIT: even a moment ago, after system has been idle a while, feeling air flow out the back of the case is noticeably lower than when the CPU was loaded up for any length of time as you can feel a small trickle of airflow when it's idle but under load there is noticeably more airflow out of it and even air temp seems noticeably warmer to.

 

also, after the PC was idling a while after boot up... I noticed it took roughly 3min (maybe a little less) of running Prime95 before my CPU reached the 60c mark. so this sort of goes inline with what I was saying in that even if 60c+ is pushing it a bit for the CPU temp (given the Linux 'watch sensors' claim of 'high' for '60c'), it likely won't see these temps much given one has to fully load the CPU for about 3min (or nearly 3min) before it reaches that point and prior to that point it, off the top of my head, it seems to float in the 55c range not long after starting Prime95. to give a pretty good feel of things in regards to temp. so I would imagine general usage, given it idles about 35-39c, would probably be about 45-55c.

 

also, while 62c is peak, I figured I would mention that's the hotter core of the two as the other core, runs I would say on average 2c, maybe 3c, cooler. even at a idle that seems to hold true.

Edited by ThaCrip
Link to comment
Share on other sites

Well I have not done a Prime95 test yet with the case door off (although I would expect similar results to the past tests off the top of my head) and the 80mm case fan disconnected (so there is no air being directed into the case)... but things seem to be okay when testing out a 1080p x264 video file and it must be using the GPU (Radeon 5670 512MB) as I can see that the GPU temps have increased vs when idle given what Linux says with the 'watch sensors' command (i.e. during playback it's showing 50.5c, which is WELL within safe range) and what's even more obvious it's using the GPU for video playback is CPU usage is floating between 8-10% which it would be WELL over that at 1080p if it was CPU only, especially given the age of the CPU (AMD Athlon X2 3600+) I am using as it would likely be at least triple that amount if not more (off the top of my head... probably 'at least' 30-50% CPU load or so) at which point the CPU fan would ramp up and noise levels from that case would increase noticeably. but 'as is', CPU fan speed is floating around 1600rpm, which makes the CPU fan much quieter vs when it ramps up and usually goes to about 3800-4000rpm or so briefly until CPU load decreases.

 

but it seems motherboard temp (which is passive cooling(i.e. no fan)) went up slightly (maybe about 3c) once I disconnected case fan on door, but probably nothing ill worry about. CPU temp is about 39-42c as it's playing back a video in Celluloid at 1080p using GPU and has, like I said already, typically 8-10% CPU load.

 

I had to tweak things in the 'Display' section of Linux Mint (which is simple enough to figure out after playing with it a bit so that things display properly using a 1600x900 res monitor paired with a standard 1080p(1920x1080) TV) and on the TV itself (I had to set the TV itself to 'just scan' so it displays properly and does not cut off any of the image etc). so then one has a nice dual-screen setup like how my main PC's is setup where I use the 24" monitor for general computing but if I want to watch a 1080p video etc ill typically drag it to the TV side of things and then play it full-screen.

 

with all of that said... should my main PC (i.e. i5-3550 / 1050 Ti 4GB / 16GB of RAM etc) ever die, it's possible I could 'get by' with this old A8N32-SLI setup for a while as, at least on the little use of it here and there browsing websites, it's not painfully slow (you can see my i5-3550 is noticeably snappier), especially because ill definitely put a SSD in it should that day ever come. but being this A8N32-SLI board only has 4GB of RAM (which is the max it supports) there will definitely be a good amount of swapping to the SSD when I load up a browser with some tabs etc since it will end up writing a fair amount more data to the SSD due to swap file usage. but I am not concerned because even if someone wrote 40GB of data EVERY SINGLE DAY to a typical SSD in recent-ish memory, for 10 years straight, that's still only 146TBW to the drive and any decent SSD should be able to pull at least that amount of data writing off before failure.

 

p.s. I installed 'mpv' (i.e. sudo apt install mpv ) on Linux Mint and then using it's default 'Celluloid' video player it uses the GPU by default, which like I mentioned above, cuts back on CPU load quite a bit and in turn helps keep CPU fan nice and low RPM (about 1600rpm) which in turn keeps computer much quieter ;) ; because it does not take much of a load before the CPU fan shoots up to about 3800-4000rpm, but it seems if the load is low enough and maybe brief touches at a moderate load, the CPU fan seems to stay nice and low RPM.

Link to comment
Share on other sites

Posted (edited)

Just to talk a bit about my two backup computers (A8N32-SLI desktop and HP2000 laptop, which I am sure are fairly outdated by modern standards). in regards to that same 1080p x264 video file, both using Linux Mint v20.1-Xfce (both on 5.11 kernel) using Celluloid (with MPV installed) here is a quick performance comparison with CPU load...

 

-AMD Athlon X2 3600+ (dual core) overclocked @ 2.4Ghz (paired with Radeon 5670 512MB GPU) = 8-10%

-AMD E-300 (dual core) 1.3GHz (which apparently has a Radeon HD 6310 integrated) = 26-30% (give or take a little).

 

so if that's a rough indication of CPU performance, assuming GPU does not matter (since I am sure the 1080p video file is being accelerated by each computers GPU given the lower CPU loads), the Athlon is roughly 3 times faster. but I am not surprised as browsing feels more sluggish on that E-300 CPU than my Athlon setup.

 

the AMD E-300 CPU is easily the weakest part of it as it originally had 2GB of RAM but I upgraded it to 8GB (2x 4GB) of RAM and currently has a Intel 545s 128GB SSD in it. so short of the CPU, it's respectable (it has UEFI and Secure Boot (I use UEFI on Mint but disabled Secure Boot) which the A8N32-SLI does not). but if that HP2000 ever dies, or I ever start using the A8N32-SLI on a regular basis, that Intel 545s would definitely be going into the A8N32-SLI setup (but then again if my main PC's motherboard was shot, I would likely transfer it's Samsung 850 EVO 250GB into the ASUS A8N32-SLI setup).

 

while I got the HP2000 used for little $ a while ago now, it appears it can't be older than Sep 2012 given that's the date on the battery, which I am assuming in the original battery. the CPU is the clear cut weak link of that setup, but it's passable for occasional light usage. but if I were to use it regularly, it would be too slow where as with the A8N32-SLI setup, putting gaming aside, I suspect it's possible it's still "good enough" for general browsing and the like, especially once I got a SSD in it. but should my main PC ever die, ill definitely be finding out if it's "good enough", or not since it's what I would shift to for a while until I debated to build another PC or use that setup for a while.

 

so at the end of the day... while the HP2000 has twice the RAM of my A8N32-SLI setup, I would rather use the A8N32-SLI if I had to use those on a regular basis especially given I suspect I can sort of compensate for lack of RAM on the A8N32-SLI by using a SSD (although I am sure writes to the SSD would increase noticeably(because of virtual memory writes), but like I say, a modern SSD can handle a lot of writes before it fails). about the only real positive of the HP2000 over the A8N32-SLI setup is it's clearly more power efficient, but at roughly a 3x performance hit, which is quite a bit when we are talking about CPU's that would definitely be on the slower side of CPU's in general (but not quite ancient level, at least not yet ;) ), ill take my chances on the A8N32-SLI overall between the two if I had to use one or the other on a regular basis.

Edited by ThaCrip
Link to comment
Share on other sites

Another small thing playing around with stuff a bit...

 

I noticed when using a USB powered 120mm fan (which is much quieter (nearly silent) than the Scythe S-FLEX SFF21F 120mm fan I tested) temporarily in the old A8N32-SLI board, the motherboards Nforce4 chip temp seem to drop quite a bit...

 

-Passive (no fan) = 49c (seems to have levelled off here after the OS has been running/idling for a while (it might go up a little under a bit of a load, but I did not really test this))
-120mm USB powered fan = 38c
-120mm Scythe S-FLEX SFF32F fan = 37c (so it's only maybe a touch better and airflow of this is definitely higher than the USB powered 120mm fan and this Scythe fan is definitely louder (clearly audible) as the USB fan is pretty close to silent operation)

 

NOTE: both 120mm fans (tested one at a time of course) in this test where in the same spot (basically roughly across from the 'ASUS ALife' copper square section), just inside the case, with the case door completely off.


NOTE: but I wonder if there might be a bit too much power draw from that USB powered 120mm fan(?) as I noticed when I put in a USB stick and tried to boot from it using F8 like usual I select the USB stick and normally it boots to it. but with that 120mm USB fan connected it boots into the Mint OS instead of the USB stick. but I figured maybe it was a rare glitch, so I tried again, and still the same result. then I realized it might be because of the USB 120mm fan and after removing that (for the record... the USB stick was not connected to the same port the fan was), sure enough, things worked as expected once again. all I had connected to the four total USB ports on that A8N32-SLI board was USB mouse/USB stick/120mm USB fan.

 

NOTE: just for kicks (with the passive cooling and things levelled off at the 49c board temp)... I even tried touching the copper pipe covering the boards Nforce4 chipset (mainly the 'ASUS Alife' area but I tried touching other areas a little to) and while it won't burn me, it's not exactly comfortable to the touch either, especially if it's not a light touch and I put a little more pressure with my finger and then try holding it there for seconds you can feel it start getting uncomfortable.

 

NOTE: a odd thing... I noticed even after powering system down with that USB 120mm fan connected that it keeps on spinning but everything else seems to be powered down (i.e. mouse/keyboard/ethernet port (since I don't see light on router (although I notice with Clonezilla if I power down through that when I run it occasionally, without doing a proper 'shut down' through the standard Mint OS like I usually do, the ethernet light is still lit up on router(but when shutting down through the Linux Mint OS, the ethernet light goes out as expected))) etc)

 

NOTE: even the couple of HDD's I got installed in it (250GB WD SATA(boot drive)/80GB Seagate IDE(small backup drive)) seems to have lowered temps a bit to.

 

NOTE: but in terms of that little fan that clips onto the heatpipe that comes with the A8N32-SLI board (which I never used even once so far) it specifically says, "For use in junction with Passive Cooler or Water Cooler ONLY." ; then in fine print it says, "installing the optional fan with an active CPU Cooler will interfere with internal CPU cooler airflow and endanger system stability'. I almost want to try it just to see what it does with motherboard temp and see if it does anything negative in regards to CPU temp. but at the same time... I figure as long as 50c or so is not that bad for a Nforce4 chip, then ill probably just leave things as passive cooling (since less noise/less stuff to potentially break). sure, if I installed that board into a better case with say some decent airflow I suspect that would help with board temps a bit (like if it was in my main PC's case, which is what the A8N32-SLI board was originally in back in March 2006), but I am not too concerned especially given it's not seeing plenty of use in general as if it was I might be a bit more concerned with getting a bit better cooling etc.

 

bottom line... I figure the A8N32-SLI board 'as is' (as I generally won't be using those 120mm fans in the test above), with no real intake/exhaust fans (besides the 120mm one on Seasonic PSU), seem to be okay assuming 50c or so is considered 'safe enough' for the boards Nforce4 chipset.

 

NOTE: I do have a 80mm fan built into the side of the case door but I prefer not to use it to keep noise levels minimal which noise of that computer is pretty low/comfortable as long as CPU fan does not ramp up (which 'Q-Fan Controll = ENABLED' helps quite a bit here by keeping fan RPM low when the system is largely idle-ish) which it does a bit once CPU load reaches a certain point for any length of time and then CPU fan is clearly noisy as just browsing a bit you can hear it come up here and there but then comes back down shortly after etc.

Link to comment
Share on other sites

  • 4 weeks later...
Posted (edited)

UPDATE: as I was playing around with some stuff with Windows 7 (yes, I know Microsoft has not supported it since Jan 2020 (but it's more of a minimal use sort of thing so not a big deal as I ain't got to worry about virus etc)) on another HDD on the computer along with some other things and I noticed rebooting was acting up again (like it seems to sometimes work but is not more steady reliable like it's supposed to be).

 

so what I tried and 'so far' seems to have fixed it is on my first try I switched from '2.4GHz @ 1.3375v' like I mentioned previously (which was giving me some (at least occasional) rebooting issues as I was saying) and dropped back to '2.3GHz @ 1.300v' and 'so far' things seem good once again. NOTE: when I was having reboot issues at the 2.4GHz @ 1.3375v, I went into BIOS and had to up it two increments in CPU voltage to 1.3625v (which is a bit higher than CPU's official 1.35v rating) and it seemed to have cured it but I would assume CPU heat would increase a fair amount so at this point I would rather play it a bit safer and opt for lowering GHz/voltage a bit...

 

...so I figure it's a minimal speed loss from 2.4GHz to 2.3GHz for what should be safer for CPU and heat in general since I am undervolting the CPU by about 0.050v vs it's official 1.35v rating and I figure if rebooting act up in the future (in regards to rebooting) I can either lower GHz a little more or increase voltage a bit as one of those two options will almost surely cure any reboot issues should they turn up in the future. but if it does act up in the future, ill probably try increasing voltage of CPU a bit first before the GHz drop since I am still a fair chunk under stock voltage.

 

p.s. plus, without testing, so who knows, maybe CPU fan speed could ramp up a little less, which if so, could be a small bonus since when that's low the computer is much quieter vs when it comes up a bit as case noise increases quite noticeably.

Link to comment
Share on other sites

On 30/06/2021 at 19:24, corrosive23 said:

Just because something is old does not mean its not worth fixing.

I never said that ;)  What I asked was what were you going to do with it

 

"What do you do with such a old board, that couldn't prob be done with a $30 pi? "

 

If you have some actual use for it...  Then it would make sense..   Other than you playing with it - seeing if it works.. What actual use does it have - are you going to actual use it for work or play.. Or just test that it still works..

 

example:

"I noticed when running Prime95 in regards to hammering the RAM etc, that the PC almost appeared to freeze temporarily due to it using swap and I forgot how SLOW regular hard drives are when they start doing that."

 

What have you actual done with it?  Other than test it?  Its a hobby of yours I get it - but I see no actual use case here.. Now on the other hand - if you restored say old cars - you could sell those for example... What you going to do with such an old computer after your done testing it, or have tested it to the point of it failing?  Is someone actually ever going to use it.. At least with old cars - people take them out for drives on a nice summer day, they show them off in shows, etc..  What exactly is going to happen to this old computer - other than sitting on a shelf collecting dust after you done playing with it, or move on to the next old computer to fix?

 

Is there some old computer show I am not aware of - where people get together and show off how old of machine they can still get to boot?

Link to comment
Share on other sites

On 02/08/2021 at 05:40, BudMan said:

example:

"I noticed when running Prime95 in regards to hammering the RAM etc, that the PC almost appeared to freeze temporarily due to it using swap and I forgot how SLOW regular hard drives are when they start doing that."

 

What have you actual done with it?  Other than test it?  Its a hobby of yours I get it - but I see no actual use case here.. Now on the other hand - if you restored say old cars - you could sell those for example... What you going to do with such an old computer after your done testing it, or have tested it to the point of it failing?  Is someone actually ever going to use it.. At least with old cars - people take them out for drives on a nice summer day, they show them off in shows, etc..  What exactly is going to happen to this old computer - other than sitting on a shelf collecting dust after you done playing with it, or move on to the next old computer to fix?

 

Like I have said before... it's pretty much my only backup desktop to my main PC. so while it does not see much use in general (ill turn it on occasionally to run updates on Mint etc), if my main PC ever dies (my main PC's motherboard is now 9 years and 3 months and counting, which is the longest I ever had a primary computer for(and has been on pretty much 24/7 short of occasional reboots/power downs)), there is a possibility that will become my main PC once again (at least temporarily, depending on cost to build another and what parts I can transfer over to another build etc) as it's still a passable computer for general internet and the like. it(the one in this topic) was my main PC from March 2006 til May 2012. considering it's age, it's performance ain't bad, at least for general internet and the like. because if it could not do at least basic web browsing well enough, I likely would not have bothered to do what I did in the first place.

 

plus, like I have said before, besides the 4-5 SATA ports, it's got some IDE ports which while I won't use those too much, it can be nice on the occasion one needs older hardware. I currently got my two old CD/DVD burners (Liteon 24102b(CD burner)/Liteon 1673s(DVD burner)) connected to the IDE ports.

 

p.s. even on that particular test in your example quote of me... when I said 'SLOW', that's typically not going to be a problem though in general, especially because if I started to use that on a semi-regular/regular basis, I almost surely would put a SSD in it, which I would likely remove the Intel 545s 128GB SSD from my HP2000 laptop if I needed one (since even that HP2000 laptop sees limited use). but of course, if my main PC's motherboard ever dies (or some show stopper level failure to where I can't fix it for minimal $), ill likely just remove the Samsung 850 EVO 250GB SSD and put it into the old motherboard in this topic etc.

Link to comment
Share on other sites

On 02/08/2021 at 02:40, BudMan said:

I never said that ;)  What I asked was what were you going to do with it

 

"What do you do with such a old board, that couldn't prob be done with a $30 pi? "

 

If you have some actual use for it...  Then it would make sense..   Other than you playing with it - seeing if it works.. What actual use does it have - are you going to actual use it for work or play.. Or just test that it still works..

 

example:

"I noticed when running Prime95 in regards to hammering the RAM etc, that the PC almost appeared to freeze temporarily due to it using swap and I forgot how SLOW regular hard drives are when they start doing that."

 

What have you actual done with it?  Other than test it?  Its a hobby of yours I get it - but I see no actual use case here.. Now on the other hand - if you restored say old cars - you could sell those for example... What you going to do with such an old computer after your done testing it, or have tested it to the point of it failing?  Is someone actually ever going to use it.. At least with old cars - people take them out for drives on a nice summer day, they show them off in shows, etc..  What exactly is going to happen to this old computer - other than sitting on a shelf collecting dust after you done playing with it, or move on to the next old computer to fix?

 

Is there some old computer show I am not aware of - where people get together and show off how old of machine they can still get to boot?

The pi cannot accurately replicate the windows 98/XP experience in terms of playing period correct games the way they were meant to be played at that time in computing history.  Far Cry for instance is almost unplayable with modern hardware, yes there are numerous "fixes" but they do not always work. Numerous dos/win95 games simply do not work correctly even in dosbox.

Link to comment
Share on other sites

  • 2 weeks later...

2.3GHz @ 1.35v (1.150v+0.200mv)

 

I settled on that lately as it's probably a bit safer for system stability I suspect and did not raise temp much (especially given it won't see any 100% load for any length of time, maybe once in a while but largely not) and gives it a bit more juice since I imagine when overclocking in general it's going to use more power, not less. plus, I am still at stock voltage so there should be no real risk to CPU from too much voltage (I might not mind going a little over stock as I suspect the risk should still be low enough if I were to do that even though I would rather avoid it).

 

so short of any issues cropping up, which I don't think there will be at this point(?), the 2.3GHz at 1.35v should do it given what I previously said with 2.4GHz at 1.3375v having a issue that turned up a while later and then I had to increase voltage two increments to 1.3625v for rebooting to stable out. so I am hoping with the 100MHz decrease in CPU speed (2.4GHz to 2.3GHz) will be enough to run at stock voltage of 1.35v.

 

either way, even if something acts up, I suspect I got to be very close at this point to where if I do have to adjust anything it will probably be one step up in voltage (i.e. 1.3625v) or a small decrease in MHz with it's current stock voltage of 1.35v.

 

p.s. but I guess sometimes it can take a while to truly get things rock stable because unless one really uses a computer a lot it can be potentially difficult to tell if it's truly stable when overclocking or messing around with CPU voltages. because unlike this old board from 2006, I can confidently say my main PC's i5-3550 being undervolted by -0.130v is stable since it's been on that for a rather long time now (several months at least) and it sees a lot of use (it's on 24/7 short of occasional power downs) as if there was any real system instability it would have shown by now as I know I am on the edge there as any further decrease system instability will definitely creep in.

Link to comment
Share on other sites

Well that is a trip down memory lane seeing the A8N32-SLI Deluxe randomly come up on Neowin, nice to hear that one of those motherboards is still going today! I believe this was one of the first, if not the first motherboard that had two x16 PCI-E slots and was capable of SLI.

 

My first job was as a systems integrator, essentially I'd sit at a workbench all day building bespoke computers customers had chose the spec of online.

 

Needless to say I built some very expensive systems at the time based around that motherboard. It was always interesting getting a build with dual 7800 GTX GPU's! Most of us we're 17-18 at the time and the minimum wage really wasn't a lot back then, especially as we were under 22. So I think we ended up with approx. £135 a week after tax, so it was kind of surreal to picking parts and building high end systems that cost what we earned in 2-3 months!

 

I always used Prime95 for stress testing overclocks, if that could run for 24 hours without crashing then it was a stable overclock. If it crashed then you had a problem.

Link to comment
Share on other sites

Posted (edited)

@InsaneNutter

 

Thanks for the little story from the old days ;)

 

speaking of GPU stuff... I noticed with the BIOS it has on it (currently v1405 BETA), GPU slot 1 basically does not work with my Radeon 5670 512MB (which I put in it back in 2010) but works with a Geforce 7900GT (which was originally in it when I built it back in March 2006) with newer BIOS's (although older ones work) as I made a thread on the issue in here back in Feb 2019... https://www.neowin.net/forum/topic/1379395-i-discovered-a-serious-bios-bug-no-post-or-display-on-asus-a8n32-sli-deluxe/ ; which was shortly after I revived the board from retirement (sitting in it's original box from May 2012 til Jan 2019 basically) given I got a PSU etc shortly before that point. but basically I just run the v1405 BETA bios with the Radeon 5670 512MB in GPU slot 2 as a work around.

 

so while I like the board, as it's a pretty high end board in it's day, I probably should have saved a bit of $ and went with something a bit cheaper since I never really planed on using SLI which was probably one of it's better selling points I suspect (along with overclocking) for many and I could have probably gotten a board that was close to that quality at a noticeably lower price point especially given the A8N32-SLI box, which I still have, says the word 'Gaming' on it which I am sure jacks up the price a bit to. still, it's a solid system for older tech at this point in time and respectable (for general use) considering it's age.

 

hell, at some point I might try seeing how it handles playing a couple of games (which I played on that system in the past) on it through Linux Mint/WINE to see how well it utilizes the Radeon 5670 512MB since there are no additional drivers for it besides whatever the Linux kernel uses by default. but even if that fails, I can still always use Win7 as a fall-back option which should work for sure since that's what I used to play the games back-in-the-day.

 

but yeah, I heard some places opt for the 24hr Prime95 test like you mentioned. but I suspect if there are any obvious problems it will probably show up quickly. that's kind of why I usually do 20min tests as during one of my overclock tests (@ 2.5GHz) a issue showed up quickly in Prime95 and usually heats up CPU enough to pretty much peak by around that 20min time frame. but I suspect there may be issues that can crop up that Prime95 might not catch, like some of the stuff I experienced in fairly recent memory with it which is why I have been occasionally updating my post in here tweaking GHz/CPU voltage as I am hoping my current 2.3GHz/1.35v (1.100v+0.200mw overvolt) finally reached the point the system is stable with random tasks under load and rebooting etc, which I 'think' it has, but like I say without using it more heavily it will be a bit more difficult to know with a high level of confidence like I have done on my main PC with undervolting it's CPU by -0.130v which is supposedly only 3.3GHz officially but when hitting it with Prime95 (and I would assume anything that puts a decent load on the CPU to) the CPU shows a constant 3.5GHz and the ASUS board it's in has no overclocking options either as it was nothing fancy when I bought it. but so far I like that boards reliability as it's been pretty much 24/7 use since I got it in May 2012 and only thing that failed on the motherboard there so far is the on-board sound which was a simple enough fix as I just disabled the on-board sound in the BIOS and then got one of those cheap USB sound cards with a 3.5mm jack on Ebay for $10 or so. but hopefully the board (ASUS P8H61-M LX Plus) will last for many years to come (since I likely won't need to upgrade for what I do for probably at least some odd years from now) as it's got all solid caps on it unlike the old A8N32-SLI board which only has those around CPU but the rest is the more usual kind of caps in that time period. but at least the ones I put in it, likely won't fail for a long time to the point I figure the board itself will be outright outdated before those fail again (at least the ones I replaced). but in regards to the A8N32-SLI setup... I wonder which will become more outdated first, it's CPU or 4GB of RAM(?). because right now both are still passable, especially for more of a general use computer.

Edited by ThaCrip
Link to comment
Share on other sites

I recently ran Memtest86+ v5.0.1 from a bootable Linux Mint v20.2-Cinnamon DVD+RW and it shows 191MHz, so DDR383 with custom 2.5-3-3-7 timings (as the board defaults to worse timings and lower speed when I got the CPU in a similar overclocked state when the board is on auto mode with RAM). the RAM is DDR400 with CL3 officially and I am running CL2.5 with, apparently a bit slower MHz (because of the 166 divider thing which I had to use otherwise RAM would be well over 400MHz due to CPU overclocking and fail to post etc).

 

I tried to go 2-2-2-7 RAM timings but the system would not post and I hit reset button like three times and then the board speaker itself output 5 short beeps and kept repeating that. so at this point I simply held power button until it shut off, waited some odd seconds, powered it back on and it recovered (thankfully, as I was hoping to avoid having to use jumper to reset everything) as it seems to temporarily lower CPU speed etc (based on what it reported on the BIOS menu you see shortly after powering it up) but once back in BIOS the same settings I used on RAM (along with all of my other custom settings etc) were still intact. so then I just lowered RAM timings back to their previously working "2.5-3-3-7", saved settings/reboot, and everything was good once again. I am probably not going to try any further attempts at tweaking RAM since it's close to it's stock MHz with tighter timings already as the timings I am using seem to be the standard ones for 333MHz RAM I think and, even from what I read, RAM speed has little to no difference in real world use (although I kind of liked to tweak it a little for good measure ;) ).

 

you can see in the attached picture it did 5 full passes (running over 10hrs straight) of Memtest86+ v5.0.1 without issue. I even tried Prime95's 'LargeFFT' (but custom set RAM use to I think 3.2GB or so (so I still left a little bit left for system)), which tests memory controller/RAM, for about 5-6hrs without issue (I read on some random sites it's a good idea to use both as Memtest is more strictly a RAM tester where as Prime95 helps a bit since it stresses both CPU and RAM a bit even though it appears it's RAM testing is not as good as Memtest86+ etc). but I noticed when doing that Prime95 stuff, while it completed without any errors, after I stopped it (CTRL+C) and then use system a little or go to a 'reboot' like usual it attempts the reboot and gets a ways into the process (as you can see it reloading a bunch of stuff back from HDD (250GB HDD) since it was unloaded from RAM due to the Prime95 using it basically (there was about 100-200MB swap during that test)) it eventually hangs very close to actually performing the reboot (like it's very close to triggering the actual reboot where system posts etc) and it did it twice when doing basically same Prime95 test a bit with 'LargeFFT' thing (so it appears it's a repeatable issue), bit I simply just press the 'reset' button and things are back to normal. but if I do more of a standard straight Prime95 CPU stress test (so it's not really using the system RAM at all) it's fine.

 

but off the top of my head it seems RAM usage on this system is probably somewhere around 500-900MB or so after booting up Mint v20.2-Cinnamon where as after stopping Prime95 (the LargeFFT custom 3.1-3.2GB of RAM test), if I recall correctly, I did a 'free -m' and based on that RAM usage (like 'used' RAM) was only roughly 300MB, if that matters. who knows, maybe it triggered some weird bug on Mint during a reboot which cases the hang issue(?). I guess if I wanted to I could try temporarily putting Windows 7 on this system and run Prime95 to see if it reacts similar or not (if I took a guess, this would probably work. so might just be a weird/rare bug on Linux in relation to this hardware(?)).

 

but I am not going to worry about that Prime95 triggered reboot issue since it only seems to be triggered by running Prime95 in LargeFFT mode where you burn up a high percentage of the RAM etc because when I just use the computer like normal, rebooting appears to be okay so far. so, short of something turning up in the future, as I mentioned in my previous post, 2.3GHz @ 1.35v seems to be basically stable.

 

p.s. who knows, for kicks, maybe someday ill play around underclocking the system just to see how low I can lower the CPU GHz/voltage and see if that keeps the CPU fan from speeding up as often (since I am hoping it generates noticeably less heat as I suspect the CPU fan ramping up is based on when it reaches a certain temp(?)) to help keep CPU fan noise down should I prefer a bit more quiet system. because currently, while if the CPU is mostly near a idle, the CPU fan speed stays low and keeps the computer much quieter, but if the CPU usage creeps up beyond a certain point for not all that long (which I suspect is tied to once the CPU reaches a certain temp(?)), the CPU fan speed ramps up.

A8N32SLI_Memtest.png

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

  • Recently Browsing   0 members

    No registered users viewing this page.