at 3860 i'm hitting 64 C according to occt
going to be getting a new cooler soon. need suggestions. not going back to water as long as im in the dorms.
Printable View
at 3860 i'm hitting 64 C according to occt
going to be getting a new cooler soon. need suggestions. not going back to water as long as im in the dorms.
been thinking some and may go back to water but keep the computer semi-hidden.
70c
source
Naturally I say water cooling and instantly have it out of my room and in the cleaning process... Just need tubing, a pump mount, and rad screws and it will be ready to put in.
air cooling? cheap air cooler id say go with xigmatek gaia and add a second fan for push pull
http://www.newegg.com/Product/Produc...82E16835233082
dont dabble in water cooling setups but i was curious about maybe getting an all in one system
the new NZXT Kraken X60 looks promising
http://www.amazon.com/gp/product/B00...=ATVPDKIKX0DER
imo if you do water do it right and do it custom. it really isnt as hard as it looks. just start with a simple cpu loop.
I will only have a swiftech 3x120, 355 with xpsc restop, and a gtz because i have no reason to upgrade.
If i could justify it i would add a 140 mm rad at the back.
Are you reading the Core temps? Under stress testing it's possible to get in the mid-60C range depending on what you have for a HSF, the ambient temp., Vcore and CPU-NB voltage. As far as the official Core max operating temp for the FX-83xx series CPUs - this is an unknown as AMD has not published any thermal data on these CPUs yet.Quote:
Originally Posted by stangracin3
70C is reported to be the TjMax according to Core Temp. 60C is reported to be the 24/7 max temp for the FX-8350 but no one ever runs their 8-core CPU under 100% load for anything other than stress testing because there isn't any software that can load all 8 cores to 100% other than stress testing software.
FWIW, unlike with the Deneb and Thuban CPUs, 60C does not seem to induce errors in the FX processors when stress testing, according to my testing and that of numerous other reviewers.
Stress testing of an AMD 8-core CPU is far more load than it will ever see in actual use with normal software which is not going to load all 8 cores to 100% probably ever, thus the Core temps are not likely to get out of the 50C range on a hot day.
If you are looking for a better HSF there are a number to chose from that perform very well, cost less than a closed loop cooler, are quieter than a CLC and are thermally more efficient than a CLC. IMO the only reason to ever consider a CLC is if you do not have room for a proper high end HSF. If you are serious about liquid cooling and want to spend $175.+ then go with a proper open loop system. Otherwise a high end HSF will do a better job and cost less than a CLC.
In addition to the HSFs being superior in all normal metrics that consumers typically use to determine a cooling solution for their CPU, CLCs can and do leak causing expensive PC hardware damage, lost data, RMAs, etc. This liability exists with all liquid cooled system but seems more common in CLCs. Thus I can't recommend a CLC to anyone who has room for a quality HSF as it can in fact cool better than CLCs as testing has shown.
What you buy is up to you. The link below has an excellent database of the popular HSFs and the Corsair H100, H80, etc. As you will see the CLCs are not what most people believe they are. Based on personal experience I would recommend the single fan Xigmatek Aegir SD128264 dual heat-pipe HSF which performs extremely well, is quiet and often on sale for <$50. and it doesn't ever leak water. ;)
http://www.frostytech.com/articlevie...id=2665&page=6
http://www.worldcommunitygrid.org/ :yepp:Quote:
Originally Posted by AMDforME
read one of the above posts and you will see i already have 90% of whats needed. maybe need to invest another $50 to get it back up and goingQuote:
Originally Posted by AMDforME
That isn't conventional software... ;) It's meant to use every single CPU cycle available. That's why it's use for stress testing.Quote:
Originally Posted by Evantaur
If you intend to Prime 24/7 then you need to look at all the parameters including power consumption, which can be very high and cause you electric bill to be significantly higher. You can actually adjust P95 to use less than 100% of available resources and/or adjust active cores, etc.
As I said people should buy what makes them happy. For ~$50. you could have a Xigmatek Aegir if you want a quality HSF. If you want water then that is what you should use.Quote:
Originally Posted by stangracin3
BeepBeep2-Quote:
Originally Posted by BeepBeep2
Your technical ignorance is showing in regards to the proper scientific testing methodology for CPU coolers.
Frosty Tech uses a controlled thermal load, CPU sized by brand (both Intel and AMD), heat source, which does not vary like CPUs and benchmarking software. Frosty Tech provides accurate thermal capacity data for a large variety of coolers using three different controlled thermal loads of 85w, 125w (for AMD) and 150w. They will soon be adding 200w data. I have yet to find any website providing better or more accurate CPU cooler thermal capacity data. They also include fan noise measurements and have listings for ultra-quiet HSF systems for HTPC or similar.
Just because you don't like the objective, scientific data does not make it incorrect. It's vary common to have reviewers test a HSF in a PC case. While that may appear to the uninitiated to be a sound test methodology, it isn't as the case design, fans used, hardware in the case and it's location, GPU card(s), etc. all impact the results of the data in addition to the aforementioned variations in thermal output from the CPUs and inconsistent loads. And to make matters worse the test data derived from this type of testing only applies to that particular case, fans, CPU, GPU, etc. Ten people could have the same PC case with different fans, HDD, GPU etc and get much different results than the reviewer.
Frosty Tech's data is properly generated thermal capacity data that can be duplicated by anyone willing to invest the time and money to build a similar lab grade CPU cooler test set-up. The same can not be said for someone trying to duplicate the test data from a review site that tests the cooler in a PC case.
I have no association with Frosty Tech but I do understand that their test data is accurate and valuable for those who want quality information and understand the test procedures and why they are the proper test methodolog to determine the coolers true thermal capacity.
i agreeQuote:
Originally Posted by stangracin3
however i dont want to invest into a proper watercooling setup
the all in one systems like the nzxt kraken x60 are perfect for my skill level
im assuming your 8320 is using the stock cooler @ 3.8ghz for 64c in OCCT?
yeah didn't want to take it any higher even though i know the cooler can take it because its a nice heatpipe cooler.Quote:
Originally Posted by tbone8ty
AMD always does nice with there stock coolers for the high end
In theory it is a great scientifically controlled testing method. The question is whether the dummy load accurately portrays the way real CPUs distribute heat from their surfaces in a manner that may not be uniform or in conformity with the dummy load? I don't pretend to know the answer to the question, but it is the thing that always nags about their methodology.Quote:
Originally Posted by AMDforME
That's why a heat spreader is used on a CPU. The thermal signature doesn't impact the total thermal output which is spread across the heatsink by the heat spreader.Quote:
Originally Posted by Momonishiki
Frosty Tech is the only site that I know of that tests with both an AMD and Intel sized heat source to distinguish the slight difference in HSF performance that actually results from a different sized heat spreader. In addition the dummy load is far more accurate and controlled than any CPU would be and that is why Frosty Tech's cooler database is an excellent source of accurate HSF/cooler thermal capacity data, unlike the in-case HSF/cooler tests which are only representative of the specific and exact hardware the HSF/cooler is installed in.
Believe what you want ;)
In theory the methodology is excellent. The actual results however do not make sense, are not ranked logically.
Either some samples have defects or they are not consistent between testings...
Look at the list yourself, some technically worse heatsinks have better results than their counterparts, some near-duplicates of models have been tested that score way better than the other.
Insult me all you want, their rankings can not be accurate based on what they say.
By the way, what exactly was I ignorant about? There's no reason to argue a point and expand an off-topic with a 5 paragraph essay in the middle of a thread.
^^^
BeepBeep2-
I didn't insult you, you're the one that insulted me in post #12 and disparaged Frosty Tech because you don't understand the technical subject and you disagree with accurate thermal data when I recommend Frosty Tech's test database because it is very useful and far more accurate than testing a HSF in a PC case with all sorts of variables.
You should educate yourself instead of insulting me or anyone else with your comments about Xigmatek, Frosty Tech, etc. People who want to learn do their homework so they can intelligently contribute to the forum discussions. If you don't understand the subject, then the onus is on you to do the research instead of making inappropriate comments that are a disservice to other forum members who might desire accurate thermal test data.
The FT test data is not just a scientific experiment, it's the proper means to evaluate the thermal efficiency of a CPU cooler by only changing one variable, by having a known thermal quantity controlled heat source and without the influences of a specific model PC case and internal hardware. Frosty Tech has invested a lot of time, knowledge and money to do it properly. In addition with FT's data you get (3) thermal load tests and results specific to AMD and Intel model CPUs. No other site provides this much information, let alone this kind of accurate CPU thermal information.
Their results in fact are ranked logically based on actual performance - which is the whole point. The results might not rank the coolers they way you think they should rank or some other website ranks them based on an improper testing methodology, but based on actual thermal efficiency under properly controlled scientific testing conditions where the only variable is the CPU cooler itself, the test results are accurate and that's what matters.
In addition the top ~8-10 or more CPU coolers typically test with 1-2 degrees C in Frosty Tech's testing. The only way anyone would know *exactly* how those top ~8-10 CPU coolers perform in their PC case, with their CPU, with their case fans and with their GPU, etc. is to test everyone of those CPU coolers. In the end any of the ~8-10 top CPU coolers is going to do an excellent job and may still show only 1-2 degree C difference between all the coolers, so all of them may work essentially the same.
Frosty Tech is providing accurate CPU cooler thermal data for those who want to make an informed decision. People are free to use whatever information is available to them. It is however ignorant of you to disparage a website that is one of the very few who knows how to properly test CPU coolers and provides this valuable information for free to anyone who wants it.
You will find that a few of the technically competent review sites that test CPU coolers in an actual PC case (because they don't have a lab grade testbed), will have a disclaimer that the test results ONLY apply to the actual hardware tested. They make this statement because they understand the pitfals of testing a CPU cooler in a PC case vs. the more accurate testing that Frosty Tech does in controlled lab conditions. I'm sure this disclaimer is ignored by many, but at least a few websites know to post this disclaimer.
BTW here's a clue for you: Just because a company claims that a newer product is superior to the previous model, does not make it true. You are being duped by marketing hype many times, if you blindly believe this. The test lab separates fact from B.S. and that is what all engineering and scientific communities rely on for results that are repeatable and that can be duplicated by anyone willing to invest the time and effort to properly test the product in question.
It's what you don't understand that leads you to false conclusions.
Hopefully this additional CPU cooler info. will be useful to the OP and other's interested in proper CPU cooler test data specific to the AMD or Intel CPU model they are using - as the coolers do not perform exactly the same on both model CPUs, based on heat spreader area and thermal load.
got everything installed
got it at 4.365 GHz at 1.425 and it seems to be overvolting by about .025-.04 so i can prolly lower the voltage a bit but instead I will probably raise the clock.
229x19
NB and HT are at 10x
ram about 2136. haven't tweaked yet. pretty much running rated speed only 3 mhz above...
in Large FFT's according to amd overdrive it is loading at 54c and cores at 39-41ish. so that puts the cores around 55...
Looks good stang, what parts exactly again?
Hope you can get 4.5-4.6+
sabertooth.
8320 cooled by an old gtz
2x4GB Gskill 2133 9-11-10-28
5850
right now i have occt running at 4.71 235x20
4.71 235 x 20 or 230 x 20.5 failed about 3 minutes in. So I backed it back down to 4.6 (230 x 20) and raised the CPU-NB to 2530. OCCT ran for about an hour before I stopped it. loaded at 55C that time. How high will the CPU-NB normally go?
what is max safe voltage?
I am using the Kraken X60 on my 8320 and I am still impressed. No idea what the real temps are though, Coretemps says 17C with a roomtemp of 21C :rolleyes:Quote:
Originally Posted by tbone8ty
According to the ASRock software, idle is 33C @ 4200 MHz, 1.425V with both fans at a quiet 900 rpm. For a hasslefree setup and quet operation, I can fully recommend it. Depending how lazy I am in the upcomming weeks, I will put the whole thing which has been sitting on my desk for the last 2 years into a case with a maxed out water cooling setup, so I can compare temps (also under load of course) :)
Will edit my thread after I maxed temps out with core damage, in case it still works. Back in the days it gave me the most reproducable and most accurate temps.
Edit: I almost forgot how hard it is to even move the mouse with core damage running at full blast :D
Kraken Control Fan Setting: Silent, which means 900-950 rpm
Idle: Liquid temp: 31C, CPU temp: 33C
At Extreme Fan Preset, the 2 fans rev at 1600 rpm and you can add another 2 fans or replace the existing ones without any hussle. NZXT uses normal radiator screws and the fans are not factory-attached in some weird way. If I wouldnt have bought the whole water cooling gear I need years ago, then I dont think I would replace the X60 anytime soon. I would rather get another one, mod the mounting and put it on my GPU ^ ^ Next time I remove the heatsink/pump, I will add a external temp probe, I somewhat think the temp readings from the tools that seem to work are whacky. Are there any new tools out for AMD/AM3/990FX for accurate temp readings, that I am unaware of? Apart from the ASRock software, I cant change the bus speed with any tools, Overdrive (also Beta version) aint working. What happened to setFSB and the lot? :(
Load: Liquid Temp: 36C, CPU Temp: 51C
what do you have the CPU-NB at?
1.3V for CPU-NB, memory running at 933 MHz 9-9-9-24 (2x 4GB) with 1.665V, NB at 2400 MHz with 1.210V, HT at 1800 MHz with 1.24V, Vcore is set to 1.425V in BIOS, but CPU-Z readings show 1,360V, so I have to measure it once I find out where my voltmeter is (I just moved).
I didnt tweak or finetune any settings, was pretty much my first try and I encoded many videos with those settings, so I guess there is some more room for more OC :)
I updated my previous post with temp readings
use amd overdrive.
Currently im trying to get mine stable at 4.8...
cant get it stable enough for OCCT or p95 so here's a few benchies
cinebench at 4.85
http://i39.photobucket.com/albums/e1...ps351db1d3.jpg
cinebench at 4.96
http://s39.beta.photobucket.com/user...ba130.jpg.html
and AIDA64
http://i39.photobucket.com/albums/e1...ps83706e81.jpg
The 4.96 image wont open theres a link above
hey Fr3ak
try using hwmonitor...i highlighted the most important cpu temp
http://i1183.photobucket.com/albums/...psec996e53.jpg
the cores #0-7 temps are not as accurate, more like an algorithm of actual temp...they are usually about ~10C below cpu temp when cores are loaded.....correct me if im wrong guys
the most accurate cpu temp for the chip is highlighted yellow... at least thats my understanding
Got 1249pgt fx-8350 today will test tonight!
nice, dont forgot at max validation with 1.55-1.575V. How good CPU could be...Im curious, I dont saw yet so new batch of Vishera :)
CPU Temp = the CPU SOCKET temp, not CPU cores. The CPUT/CPU socket temps typically run 15-20C higher than the CPU cores, because the socket has no cooling other than free air and the mobo.Quote:
Originally Posted by tbone8ty
All that counts are the CPU core temps - under max load. AMD says the temps from 40C upwards are accurate on the FX processors. The temps below 40C are not accurate, which is obvious when the CPU temp is below ambient.
Socket temp doesnt need "active" cooling, so I dont think your 15-20C are very accurate and just a fairly okayish guess in the best case. Socket temp directly depends on CPU temp and if heat is "drawn away" from the CPU by the CPU heat sink, socket temp goes down as well. The only and main difference always was and always will be between core and socket temp is socket temp reacting a lot slower in idle/load changes.
You misunderstand, I didn't say the socket needs active cooling. I said the socket temp is not what you should be using to determine the CPU "temp". The socket temp means nothing and does vary with CPU temp and ambient temp. It typically runs 15-20C above the cores which have the heatsink for cooling.
It simply does not matter what the socket temp is. All that matters are the core temps under max load and they only matter for stability and longevity. Nothing else really matters and using the incorrect temp sensor can lead to false conclusions about the CPU's thermal condition. Using the socket temp as a means to determine if the CPU core temps are reasonable under max load, is incorrect as AMD's CPU temp specs are for the cores, not for the socket.