Aces High Bulletin Board
General Forums => Hardware and Software => Topic started by: humble on February 06, 2005, 04:03:45 PM
-
Curious what normal temp under a full load is for AMD 64....
Bumped my FSB to 220 and its running ~45C going up to 54C under a full load...didnt know whats considered normal for the chip.
-
Thats ok.
Standard my 3500 with stock cooler would go as high as 65 under AH loads.
Now overclocked it rarely goes above 50 with my current cpu cooler.
-
Originally posted by humble
Curious what normal temp under a full load is for AMD 64....
Bumped my FSB to 220 and its running ~45C going up to 54C under a full load...didnt know whats considered normal for the chip.
my AMD64 is 38 idle 43 load but im water cooled.
your case temp will effect your cpu temp if u air
cool. u might need to pull in and out more air to
get your case temp down. 45c idle tells me your case temp
is pretty high with hot air hanging around. also
look and make sure you have your cables and wires
tiddy as they can restrict air flow.but 54 is not all that high, just keep it under 60.
-
I'm getting some lockups at 220 with VC reset to "stock"....temp is fine ~54/55 C. It dumped me to desktop in AH once and froze on me once. Curious about cause....from what I understand I'm well within limits of chip...memory or other things I should check out???
-
Under 60C is "normal", but if you're getting lockups at lower temps you might consider adding .05 - .1 volts to the cpu to see if that will stabilize it. If better cooling is an option, give that a try too. 50-55C shouldn't be cause for concern unless you're seeing instability, and even then you'll have to determine if it's the memory or the cpu. That may not be too difficult since the athlon 64 cpus are only multiplier locked upwards. You can run any athlon 64 at a lower multipler without problems, so to check your memory, reduce the cpu multiplier a few increments (from 10 to 8, etc) and see if you still get instability. If it keeps crashing, it's not your cpu. If it becomes stable when running a lower cpu multiplier, you either need better cpu cooling or a *slight* boost in voltage.
-
Thanks....so far it seems to be AH related. I can run benchmarks fine and play Call of Duty with no problem....
-
try to go from 220 to 218
-
I've found that AH stresses my computer more than the benchmarks do, so an otherwise stable overclock will often result in a blue-screen crash when running AH. Dunno why, but that's what happens. I'm having to completely re-run my overclocking stability tests to determine whether it's my cpu or ram.
-
Eagl,
If memory serves you I think you have the Neo 2 platinum.
If so this is the same board I have, perhaps we could compare notes. I am running a amd64 3000+ @ 2.4ghz. I only played for about 45 minutes last nite with no crashes. I am however getting a broken cockpit pillar issue when I pan my view. I am more inclined to think it vid card.
-
I only played for about 45 minutes last nite with no crashes.
I bet you were alt hugging! I get usually shot down several times in that time. :p
-
Originally posted by eagl
I've found that AH stresses my computer more than the benchmarks do, so an otherwise stable overclock will often result in a blue-screen crash when running AH. Dunno why, but that's what happens. I'm having to completely re-run my overclocking stability tests to determine whether it's my cpu or ram.
Easy. Flight modeling requires an enormous amount of floating-point calculations. There is not an FPS game on the planet which comes close to what is required of a good flight simulator in terms of floating-point calculations required per frame.
Floating-point calculations are what drives the CPU temperatures up the most.
-
Originally posted by humble
Thanks....so far it seems to be AH related. I can run benchmarks fine and play Call of Duty with no problem....
united offensive???.....whats your sig. and when u on?:D
-
Originally posted by buzkill
united offensive???.....whats your sig. and when u on?:D
Actually just bought it....used to play DoD some....havent been online with call of duty yet....do have united offensive also....will probably use snaphook as game ID
-
The torture tests I use are the 3 prime95 tests. One stresses primarly the FPU, one produces max CPU heat, and the last one loads up 100% of available physical ram as well to load the entire memory and cpu system. Before I fiddled with some settings, I could run stable at 220 FSB but not at 230 FSB. That may have been cpu related though, so I backed off the cpu multiplier, loosened up the mem timings, and tried again.
Right now, here are my parts and settings:
Athlon 64 3200 winchester
2x512 meg crucial ballistix pc3200
thermalright XP90 heatsink with 92 mm fan @ 3100 rpm
GeForce 6800GT at 400/1100 gpu/mem
HTT mult - 4x
FSB - 230
mem speed - 230
mem timings - 2.5 3 3 10
mem voltage - auto
cpu mult - 9 (2070 mhz)
cpu voltage - 1.45v
AGP/PCI bus locked at 66/33
I'm most worried about cpu temp since the mobo is reporting around 55C at max load during torture tests. The heatsink paste hasn't fully settled in so I expect that to drop a bit, but adding more airflow through the heatsink didn't seem to do anything for cpu temps so I just might be at the max heat load for that heatsink, dunno. Or it's not seating properly even after removing it, reapplying the paste using a different method, and re-seating the heatsink.
Using the same fsb but 10 on the cpu mult and 2.5 3 3 8 on the ram timings, prime95 died after 6 hours and AH would bluescreen after about 45 min of play. With the current settings, prime95 ran 10 hours without any problems but since I'm still messing with the settings I haven't done any real testing with AH yet. I also haven't done any real benchmarking yet, just using prime95 and memtest86 for stability testing.
-
55c is a tad high for 2100ish mhz, but its still fine
myself i have a slient boost cooler which is made for quiet, not performance. I have had my 2000mhz wincester running at 2300mhz and 1.6volts (extreme) most of winter and never toped 50c
now running at default cos i think the 4 months at 1.6volts has done some damage...... :rolleyes: but thats the risk you take
-
I know many use pastes etc. when seating Heatsync to CPU.
I do not. I clean very thoroughly the bottom of heatsync with alcohol. And dampen a Q-Tip with alcohol and clean point or points where heatsync will seat with CPU.
Never had a heat related issue with any machines I have built.
I know of instances where to much paste has been applied, and appled unevenly causing a lousy heat transfer to the heatsync.
In my case a good cleaning before installing Heatsync has been my choice.
Also....even if heatsync and fan are doing their job........airflow must be good in the box. My mini server tower has a fan on top blowing air out of case. Two 80mm fans bringing air in, and a Bay Freezer which is a device that installs in a drive bay that has twin 60 or 80mm fans one takes air in.....one exhausts it. Newer Vid cards have fans on them also...so moving the air in the case in past couple of years has become more important. Everything running at higher speeds....including video cards, using more power and generally spreading heat throughout the case.
By the way....playing AH with Athlon64 3400 temp solid 50.
Other applications 42. Mobo stays at 35.
-
Originally posted by Hajo
I know many use pastes etc. when seating Heatsync to CPU.
I do not. I clean very thoroughly the bottom of heatsync with alcohol. And dampen a Q-Tip with alcohol and clean point or points where heatsync will seat with CPU.
Never had a heat related issue with any machines I have built.
I know of instances where to much paste has been applied, and appled unevenly causing a lousy heat transfer to the heatsync.
In my case a good cleaning before installing Heatsync has been my choice.
Also....even if heatsync and fan are doing their job........airflow must be good in the box. My mini server tower has a fan on top blowing air out of case. Two 80mm fans bringing air in, and a Bay Freezer which is a device that installs in a drive bay that has twin 60 or 80mm fans one takes air in.....one exhausts it. Newer Vid cards have fans on them also...so moving the air in the case in past couple of years has become more important. Everything running at higher speeds....including video cards, using more power and generally spreading heat throughout the case.
By the way....playing AH with Athlon64 3400 temp solid 50.
Other applications 42. Mobo stays at 35.
.
WHAT? you dont use paste or ANYTHING? jesus how havnt you had a failure? when i did this once (as a test ages ago) the pc would reboot within 5 minutes due to a temp over 70c.
I suppose what you do will work with the CPU die exposed, and a heatsink lapped with 2000 grit paper, but not with a CPU die under a cap, and a default boxed cooler.
try doing what you do on a prescott CPU and it will never even make windows let allone work
-
Overlag....done it with all AMD CPUs'.....and as you know they run hotter then Intel generally.
Arctic Silver etc. is not needed if you have a tight fit. Remember no matter how well the fit...using paste or not heat syncs and fans are effected also by ambient temps within the case. If the air is cooler, heat will be absorbed and disappated at a faster rate allowing lower cpu temps.
-
The heatsink paste I'm using is sorta old so maybe I'll buy a new tube of AS5 and give it a shot. The problem is that shipping just KILLS me. I'll end up paying $20 shipping on a $5 tube of paste. The only other things I'd like right now are a couple of thermally controlled 80mm case fans, but that would still be $20+ shipping on about $20 worth of stuff.
I'll probably just wait a bit longer until I have something else to buy, and tack it onto the order.
-
Originally posted by Hajo
Overlag....done it with all AMD CPUs'.....and as you know they run hotter then Intel generally.
Arctic Silver etc. is not needed if you have a tight fit. Remember no matter how well the fit...using paste or not heat syncs and fans are effected also by ambient temps within the case. If the air is cooler, heat will be absorbed and disappated at a faster rate allowing lower cpu temps.
cant belive you never had a failure of some sort. i guess your using well made coolers with very very good base finish on them? the likes of swifttech bases are like what i mean.
when i hand lapped a heatsink to 1200 grit, and used it on a P3 copper mine core 99% of the thermal paste ended up around the edges of the core because the contact was so good. But on the new A64s/P4s that cap on top of them is alittle rough, as is most standard heatsink bases.
smoother the contact patches the better but you still need paste of somesort even if it is a very very small ammount
-
Overlag....tend to agree. If using paste only a tiny dab spread with a razor blade EVENLY will do the trick.
Had a friend of mine assemble his first Computer. I hoped to call him before installing his cpu and heatsync. He told me he was done and installing his pci cards. Asked him if he used Arctic silver.....he said yes.....I asked 'How much?" He said the whole tube. LOLOLOL I told him to get the heatsync off before he burns everything up LOL
True story!
-
Originally posted by Hajo
Overlag....tend to agree. If using paste only a tiny dab spread with a razor blade EVENLY will do the trick.
Had a friend of mine assemble his first Computer. I hoped to call him before installing his cpu and heatsync. He told me he was done and installing his pci cards. Asked him if he used Arctic silver.....he said yes.....I asked 'How much?" He said the whole tube. LOLOLOL I told him to get the heatsync off before he burns everything up LOL
True story!
yup seen that before :eek: fools, cant they read instructions/?! lol
-
Originally posted by eagl
The heatsink paste I'm using is sorta old so maybe I'll buy a new tube of AS5 and give it a shot. The problem is that shipping just KILLS me. I'll end up paying $20 shipping on a $5 tube of paste. The only other things I'd like right now are a couple of thermally controlled 80mm case fans, but that would still be $20+ shipping on about $20 worth of stuff.
I'll probably just wait a bit longer until I have something else to buy, and tack it onto the order.
Where you at....awful high shipping cost?
-
I'm in merry olde England. Shipping to my house via fedex draws an 18% import tariff. My P.O. box on base doesn't accept anything but US Postal Service, but since that's unusual for most online retailers, the USPS shipping fees tend to start high. Newegg charges about $18 for the first item and then about $2 for each item after that. It's not too bad when you're ordering a few hundred bucks worth of stuff, but it makes it a losing proposition when ordering just a few inexpensive parts.
Buying them locally isn't great either due to the exchange rate. Do I pay newegg to ship parts and pay in dollars, or pay literally twice the amount at a local shop due to the exchange rate? The tradeoff occurs at, ironically, about $20 in merchandise. More than that and it's cheaper to have it shipped. Less than that, and it's cheaper to pay the high exchange rate.
-
I've got a small tube or two of paste, not sure what brand. They came came with one of my higher end cooling fans. Happy to dump one in the mail for you....post or email me PO box....
-
I have a tube of generic paste that came with my HSF, but from past experience that stuff tends to dry out after a while so bumping the case may cause a heavy heatsink to shift enough to ruin the thermal interface. That's why I've been using the artic silver stuff, because it works really well and never dries out.
I suppose I'm probably too picky, but thanks for the offer though :)
-
Originally posted by Hajo
Overlag....tend to agree. If using paste only a tiny dab spread with a razor blade EVENLY will do the trick.
Had a friend of mine assemble his first Computer. I hoped to call him before installing his cpu and heatsync. He told me he was done and installing his pci cards. Asked him if he used Arctic silver.....he said yes.....I asked 'How much?" He said the whole tube. LOLOLOL I told him to get the heatsync off before he burns everything up LOL
True story!
Hey Hajo, I'm glad you didnt Doc-ument who that was :)
shamus
-
Originally posted by eagl
I have a tube of generic paste that came with my HSF, but from past experience that stuff tends to dry out after a while so bumping the case may cause a heavy heatsink to shift enough to ruin the thermal interface. That's why I've been using the artic silver stuff, because it works really well and never dries out.
I suppose I'm probably too picky, but thanks for the offer though :)
Hehe....
Exactly why mine is lying around...ignored the stuff they gave me and bought a tube of artic silver. I've still got bout 1/2 the tube but its probably as old as yours (15 months or so). Still debating if I'm gonna upgrade the cooler on my 3200+...dialed it back to stock since I dont need the horsepower...its running at 41C again at idle so I dont think I'm gonna mess with anything for now...no clue how long the stuff lasts once you crack the tube open....happy to send it your way if you think itds better than what you've got...
-
The problem is that AH2 stresses the CPU and the Graphic board, so the temp in the case will rise above what you experience with Prime95.
Prime tests CPU but no graphics.
-
Quite right. A high end video card will probably heat up about 20-30F while running AH, depending on the HSF on the video card.
I know my X800XT gained about 28F when running AH and is much happier since I switched out the default HSF for an Artic Cooler 4.
Plus exhausting the heat out of the case has had other side benefits. The CPU and Northbridge areas are much cooler. Northbridge is running 40F cooler now, when under load.
Also added a low speed 92mm fan above the NB/RAM, which is just a 'dead air' fan. It just moves a low volume amount of air to make sure there are no deadspots.
I'll have to post a pic one day, when I get rich and famous and can afford a digital camera. :)
-
or get those lottery numbers nailed down.....lol
-
I agree on the importance of getting the case nice and warm during testing. One thing I do is run a little windowed graphics thingy called rithdrbl that puts a ton of stress on the vid card. I'll run that and prime95 for an hour to get the case nice and hot, then halt rithdrbl to give the prime95 process 100% of the cpu to see if it gets any errors. Every hour or so, I'll re-run rithdrbl to heat up the case again.
I suspect I'm having heatsink/cpu mating problems. Temps are just too warm for the cpu and they didnt' drop during the artic silver burn-in process like they should have. And the heatsink remains completely cold to the touch, when it should be pretty darn hot. So something is going wrong.
-
humble, my temps were about the same as yours with the stock cooler, Ive read in a few places that the Nforce 3 boards were having trouble with the temp readings so i took the temps i was getting with a grain of salt . I have since put a thermaltake venus 12 on and my temps are 36c idle , and 50c at full load with a probe . this is at a fsb of 215. Im having trouble going over that,while still keeping my ram at CL2 if i go 2.5 latency i can go to 220 without issue , and my temps really dont change much. if i go anything over 220 windows will do the infinate boot thingie just keeps rebootin itself. fwiw
38
btw have any of you guys had problems removing the cpu heatsink, everytime i do it pulls the cpu right out of the socket no matter how much i wiggle and try to be gentle. seems like the artic silver creates such a bond i have to slide it off the heatsink just to break it the suction
-
I had mine running at 220, AH would run for 30-40 minutes till it dumpred froze etc....temp didnt seem to get over 55-56 C so dont think it was a heat dump since system didnt lock.
-
I just removed and reattached my heatsink with less heatsink paste, and temps are down about 6C. I was up to around 56C at max load, and right now I'm around 50C at load. Plus this time, I can feel the heatsink get warm.
I'm pretty sure I just had too much paste on the cpu. I applied it this time using the "drop of paste in the center" method recommended for cpus with heat spreaders, and it seems to be working fine. Hopefully as the paste settles and spreads I'll get another couple of degrees drop in temps. This is with the cpu at 2070. After I verify it works ok and remains cool, I'll probably bump the cpu multipler to run it at 10x instead of 9x, which would get me 2300mhz instead of 2070.
If the cpu crashes at 56C, then try increasing the voltage .05 or .1 v. That can help. Otherwise, get a better heatsink and try again.
-
Hey eagl, long time no speak. I hear you are still flying, and that's great news. Is Roamer still in the Pigs? Do you know his current email address? Thanks, Squid. :)
edit - sorry for the hijack, humble.
-
Squid as in eric? If so, heya!
Shoot me an email. eagle at flyingpigs dot com. I have roamers email addy but don't want to post it in a forum.