|
You gotta do at least 30min stress tests IMO. More if you're being stringent.
|
#
?
Aug 23, 2015 08:59
|
|
|
|
| # ? May 16, 2024 09:53 |
|
|
Reposting this here since the PC Building thread didn't have an answer and you guys seem to know a lot about airflow.Mutation posted:Okay, so I want to build a Silverstone SG13 using my 4690k and 970. I want to build it with a air cooler instead of the suggested Closed Loop Cooling solution.
|
|
#
?
Aug 24, 2015 05:50
|
|
|
Try the fan both ways, either way temps will be fine. I would also recommend a reference gpu so its blowing air out of the case rather than increasing case temps. BTW, get a cheaper board and a better video card if this is for gaming. Silly to get a $250 motherboard and a gtx970. Only other thing is you should also look at the Silverstone SG09. Its a microatx case, and has amazing cooling. It is a bit larger than what you got but fits great in silverstones bag. http://silverstonetek.com/product.php?pid=32&area=en
|
|
#
?
Aug 24, 2015 06:03
|
|
|
I got the motherboard because apparently the Asus Z97i Plus no longer exists (Not on Amazon, Newegg, and the Ebay vendor I tried buying it from lied about having it and didn't respond to my messages until a week later saying "whoops") and I got a MSI board before that but the CPU socket was too close to the PCI board so the CPU fan went over the PCI-E port, and that just made me extrememly paranoid to buy anymore boards whose CPU socket placement isn't as far away from the PCI-E port as Asus. So I found an open box discount for the Asus Maximus VII Impact on Newegg so I said "gently caress it all" and bought it. it didn't come with a wifi antenna Also I already have the parts, just not the case yet but I am dead set on it UNLESS this setup is not going to work or it's going to be very inefficient (like the CPU can't even go above 4.0ghz) . I just want to know if my cooling theory is fine and I'm not killing the PSU by having it suck up the CPU's exhaust. That bag is a good idea though. Automata 10 Pack fucked around with this message at 07:32 on Aug 24, 2015 |
|
#
?
Aug 24, 2015 06:06
|
|
|
I'm looking to OC an i5-3570k on an ASUS P8Z77-V because it's 2012 in my apartment or something. I've OC'd computers before, but it's been years and years since I've done it, so I'd like a reliable walk-through. I was rather surprised to find that I couldn't easily find a clear guide through googling, nor were any of the links in the OP particularly clear either. Are there any how-to's that come recommended, particularly for ASUS boards?
|
|
#
?
Aug 24, 2015 12:24
|
|
|
Oh hey its 2012 in my apartment too.  I can write you a more detailed step by step guide when I get home and can look at the BIOS, the quick and dirty guide is: 1) Set max turbo multiplier to 42 2) Don't touch anything else If you don't mind futzing with voltages and stability tests you can go higher but not by enough to be worth the trouble.
|
|
#
?
Aug 24, 2015 14:30
|
|
|
Parts Kit posted:Thanks. That was 72 degrees after about 5 mins so should I just not worry and give it a little more, maybe 4.5ghz? Normal processing temps have been much lower while playing games/using photoshop and it did auto throttle down when it hit 72 degrees.
|
|
#
?
Aug 24, 2015 16:55
|
|
|
cisco privilege posted:I wouldn't worry too much about 72degrees in Prime95, but I'd suggest testing for a longer period of time before moving up clocks. Also you should try to figure out how to use offset vcore settings rather than rely on the automatic voltage option, since alot of boards will pump vcore hard on auto if you go above ~4.2ghz on these chips. Better to have it raise the vcore to a max of ~1.38v when necessary instead of leaping to 1.45 or higher.
|
|
#
?
Aug 25, 2015 00:22
|
|
|
Alright, I've jammed everything together and tried it out. The Shuriken 2 Big Rev. B is a really tight fit, it goes right over the daughter board and there's virtually no space between it and the daughterboard, but I can still move the daughterboard a bit and see that there's no stress between the two. But anyways, it works! The maximum safe overclocked is 4.2 GHZ,, in which Prime 95 (before the Haswell fuckery update) brings the cores up to 74c and ran stable for several hours. If I dare to try and increase the voltage even by a little it, it slowly crawls to 80 but it stays at 80c from anywhere to 1.75 to 1.25. From my amateur guesswork it's that the PSU fan doesn't pull out the heat fast enough until a certain temperature threshold. But 4.2 ghz with a 4690k is pretty good considering the setup. The machine scores a 10559 in Firestrike 1.1. The Witcher 3 runs great on it, and while running a long session the card only maxes at 77c and the CPU only maxes at 65c. When running Prime 95 for a while I put my hand at the back of the PSU to feel the temperature of the exhaust. From a 1 to 10, 10 being "This is burning my flesh" the exhaust feels like a 6~7. Pretty warm, that's about it. I'm guessing that if the exhaust from both the CPU and PSU being stressed really hard is just "pretty warm" then it's a safe setup. Automata 10 Pack fucked around with this message at 05:35 on Aug 26, 2015 |
|
#
?
Aug 26, 2015 05:30
|
|
|
Hah, okay most of my heat issue was because the H80i GT wasn't getting regulated right. Partly because the motherboard was trying to run it in a quiet fan mode of some sort and partly because it needed the USB cable hooked up to let Corsair Link do the rest. Now in the same test at 4.1 ghz the hottest core just peaked at 57 degrees after 7 mins when it would have already hit 72 degrees prior. And as soon as the test was stopped temps dropped right the gently caress back down really fast. Still got to do harder stress tests and more bios tweaks obviously but that's a massive difference right there.
|
|
#
?
Aug 26, 2015 08:27
|
|
|
Just FYI you dont need corsair link. Just make sure the pump is running 100%.
|
|
#
?
Aug 26, 2015 14:58
|
|
|
use it to set your fans' custom curves then keep it turned off
|
|
#
?
Aug 26, 2015 15:03
|
|
|
Mutation posted:Alright, I've jammed everything together and tried it out. 4.2Ghz is a really low overclock you should push it way more easily, that 970 is cpu limited and physics score on 3D Mark is really bad. I did a firestrike on my own with my lovely 4 years old CPU but because my OC is a bit better I doubled your physics score http://www.3dmark.com/3dm/8346382
|
|
#
?
Aug 26, 2015 15:52
|
|
|
Don Lapre posted:Just FYI you dont need corsair link. Just make sure the pump is running 100%.
|
|
#
?
Aug 26, 2015 20:03
|
|
|
Hm, going back and looking for the CPU voltage settings it appears my Asus P8P67 LE doesn't let you directly set it. Some of their P8P67 boards definitely do, but quick googling suggests a few don't. What it does let me set a limit for is the max watts used in Turbo mode, which is currently set to auto. RealTemp is reporting under max load the processor is drawing 1.3811V at 92.7W. Presumably if I manually drop the max wattage (and the board doesn't do some kind of current drop) I should be able to force it to drop the voltage. Does anyone know for sure if that's right or wrong? A quick number crunch suggests 90W would drop me down to 1.35V.
|
|
#
?
Aug 27, 2015 00:48
|
|
|
I have an Asus P67LE as well, and I set an offset voltage from stock to limit voltage - seemed to work pretty well both with my old 2500K and my current 3770K. I got 4.3ghz on each proc using a Hyper212 Evo at ~1.28v. Didn't try for more - temps are/were good and I'm looking for reliability.
|
|
#
?
Aug 27, 2015 01:21
|
|
|
Ah okay, RealTemp is reporting VID, not VCore, so that explains some of the whack readings. Using the HW Monitor program is showing VCore max of 1.320V for the current offset settings while running under 100% load. Yay for rereading the OP. As for the watt limit, no that doesn't behave the way I was hoping. It fucks with both current and voltage and dropped my clock back down to 3.3ghz. JnnyThndrs posted:I have an Asus P67LE as well, and I set an offset voltage from stock to limit voltage - seemed to work pretty well both with my old 2500K and my current 3770K.
|
|
#
?
Aug 27, 2015 02:18
|
|
|
Sidesaddle Cavalry posted:Asus uefi havers trying to adjust adaptive voltage please go into digi power control settings and raise/lower CPU Load Line Calibration. Ok I finally got around to trying this, and yet again it did nothing about the voltage bump I get. The only setting voltages seem to obey is the offset one when the processor is either in Offset mode or Adaptive mode. Anything else doesn't change a thing. I'm a huge noob about overclock so it's quite possible that I'm missing something really big here. My mobo is a gryphon z87 by the way. The LLC setting on my mobo is divided in 8 levels, changing the level doesn't do a thing, the only effect is that changing it from "Auto" to whatever level also disable the "Enhanced Intel Speedstep Technology". Enabling and disabling the EIST also doesn't seem to do a thing. What I'm able to do is: - if I use manual mode, the voltage will remain the one I use (1.164v for 4.0GH) minus 0.002 for some reason (so it reads 1.162v from inside the OS) - if I use offset mode or adaptive mode, the offset setting seems to work like this: it stays at 1.192 + or - whatever I put in the offset setting, so if I put like - 0.060 it will stay at 1.132 while idle, which is great, but it will bump up to 1.225 under load. For some reason it's aways a +0.090 bump no matter what I do, no matter the LLC settings or speedstep settings. What I wanted to do is for it to stay at around 1.130v while idle and peak at 1.164v that is the maximum voltage I need at 40x, producing the least heat possible. Maybe I'm wanting the wrong thing? I don't get it, maybe it's because I changed some bios setting that is causing this and forgot about it? Reading about other people messing with adaptive mode on this very mobo seem to cause completely different results. I let my processor frequency fixed at 40x 100, should I change this somehow? program666 fucked around with this message at 04:31 on Aug 27, 2015 |
|
#
?
Aug 27, 2015 04:28
|
|
|
 Put my 2500K back at 4Ghz with auto settings and made sure XMP was off. No keyboard lag this time. 40 runs of IntelBurnTest on High. It claimed Linpack crashed right at the end but it got 40 correct results so uhh  I think it looks ok except for what might be an excessive Vcore for a measly 4Ghz. Off to read the guides and figure out how all the voltage settings interact.
Carecat fucked around with this message at 11:48 on Aug 27, 2015 |
|
#
?
Aug 27, 2015 11:43
|
|
|
program666 posted:I don't get it, maybe it's because I changed some bios setting that is causing this and forgot about it? Reading about other people messing with adaptive mode on this very mobo seem to cause completely different results. I let my processor frequency fixed at 40x 100, should I change this somehow? Is there some reason you want it at a fixed frequency? Most people just change the turbo target and raise the offset voltage to allow their cpu to hit that target.
|
|
#
?
Aug 27, 2015 13:42
|
|
|
I didn't make a lot of tests with variable frequency but when I tried it I got the impression that if I let the frequency vary it will do so according the voltages I use so at a 1.164 voltage it would never allow the processor to go all the way up to 4.0GHz anyway, and if I allow more voltage it will create too much heat and suffer throttling anyway. At 4.0GHz and 1.164v volts it already goes up to 85 C on the summer here, after a few minutes with prime 95, so that's what I use as a limit.
|
|
#
?
Aug 27, 2015 13:48
|
|
|
RAM question. Finally got my build finished (5820k on an X99-Pro) and I was thinking I'd have to set an XMP profile to run my RAM at the advertised 2666MHz but CPUZ seems to say it is already running at that speed with proper timings and everything. When I enable XMP it does its whole BCLK increase, cpu ratio decrease to get the 2666. Is the XMP actually needed? Is it doing something more I'm missing or do I not need to worry about it? If I'm planning to overclock the CPU anyway, is this preferred? I've only ever overclocked my old Q6600 and that was years ago and I never touched RAM or BCLK (always read it caused more trouble than anything).
|
|
#
?
Aug 28, 2015 18:29
|
|
|
You don't need to use the XMP. Just verify that all the timings and voltages are correct for the rated settings.
|
|
#
?
Aug 28, 2015 23:46
|
|
|
Sh4 posted:4.2Ghz is a really low overclock you should push it way more easily, that 970 is cpu limited and physics score on 3D Mark is really bad. I did a firestrike on my own with my lovely 4 years old CPU but because my OC is a bit better I doubled your physics score Uh, you have two more cores, as well, which will probably scale in physics tests. Also, first time I've heard Sandy Bridge-E called "lovely".
|
|
#
?
Aug 31, 2015 10:50
|
|
|
HalloKitty posted:Uh, you have two more cores, as well, which will probably scale in physics tests. Thats not the best part. Hard drive model 60 GB OCZ-AGILITY3 
|
|
#
?
Sep 1, 2015 20:00
|
|
|
go3 posted:Thats not the best part. Yeah, what's your point
|
|
#
?
Sep 1, 2015 23:15
|
|
|
Sh4 posted:Yeah, what's your point Your drive is bad and your emotions should be commensurate.
|
|
#
?
Sep 1, 2015 23:23
|
|
|
Sh4 posted:Yeah, what's your point
|
|
#
?
Sep 2, 2015 20:43
|
|
|
A couple weeks ago I saw a Craigslist ad for 2 120mm radiators for $20 each, 2 240lph pumps for $20 each, 6 Peltiers for $10 each, and a 120mm fan. . . Yesterday, I saw everything relisted for $40 for everything, so I jumped on it. I didn't really even want the Peltiers, but buying everything straight from Amazon would have been about $80 so I'm happy. My case, a Thermaltake V3 has holes in back to mount a radiator externally, I might even be able to wedge a second radiator inside at the top where a 120mm fan is currently mounted. . . The case has mounting points for 2 120mm fans at the top, the back fan is full size, but I had to use a slim (no airflow. . .) fan in the front to clear some junk on the motherboard. In other words, I have not attempted the radiator yet but I would be extremely surprised if it fit inside the case, I really would not like to mount it externally on top. I assume all I really need is some tubing and a water block, then I could use 1 pump and 1 radiator to cool my CPU, is there anything else absolutely necessary that I have overlooked? I see I can get a $20 universal block, or is there a cheap reliable block I should buy? I don't really want to overclock, I really just want silence and a chance to get my feet wet and upgrade components later, unless I decide water cooling sucks. . . To that end, I don't want to go with a $100 water block just yet. Adding a "simple" water cooling loop with 1 radiator and pump seems like the best way to start, or should I just dive in with dual radiators and dual pumps?
|
|
#
?
Sep 17, 2015 17:30
|
|
|
Well. My i5-5675c arrived from newegg. Coming from SandyBridge I have no idea how to overclock this thing. I remember it being stupid easy to do with SNB, but that was almost 4 years ago(Wow). Gonna keep tinkerin'.
|
|
#
?
Sep 19, 2015 18:01
|
|
|
Crotch Fruit posted:A couple weeks ago I saw a Craigslist ad for 2 120mm radiators for $20 each, 2 240lph pumps for $20 each, 6 Peltiers for $10 each, and a 120mm fan. . . Yesterday, I saw everything relisted for $40 for everything, so I jumped on it. I didn't really even want the Peltiers, but buying everything straight from Amazon would have been about $80 so I'm happy. You need a reservoir unless you want your loop to have air in it and make the same noise your stomach makes when you're hungry. Those pumps are terrible and I wouldn't use those for pc cooling, for watercooling you can cheap out on everything but not pumps, these will be noisy, unreliable and probably leak. Start with a D5 or an MCP, also dual pumps are useless, if you have a dozen of waterblocks and need more pressure get a better pump.
|
|
#
?
Sep 19, 2015 19:30
|
|
|
Sh4 posted:You need a reservoir unless you want your loop to have air in it and make the same noise your stomach makes when you're hungry. Those pumps are terrible and I wouldn't use those for pc cooling, for watercooling you can cheap out on everything but not pumps, these will be noisy, unreliable and probably leak. Start with a D5 or an MCP, also dual pumps are useless, if you have a dozen of waterblocks and need more pressure get a better pump. Thanks for your advice, I will admit the pump does look kinda crappy. . . I did finally test a pump with a 9V battery today, when out of water is loud as poo poo and vibrates a lot. I submerged it in my sink and it was quiet enough for a $10 pump but I could tell it was still vibrating a lot. That said, in looking for D5 or MCP style pumps I see they appear to cost approximately $90 and up. What is your opinion of something like a $60 Eheim pump? Years ago when I still visited bit-tech.net (2004ish) most people loved Eheim pumps for water cooling, but now all I see mentioned are D5s. Or what about a Laing DDC? You did not specifically mention the $20 universal block, in your opinion do you feel that is a good or bad idea?
|
|
#
?
Sep 20, 2015 16:50
|
|
|
Crotch Fruit posted:Thanks for your advice, I will admit the pump does look kinda crappy. . . I did finally test a pump with a 9V battery today, when out of water is loud as poo poo and vibrates a lot. I submerged it in my sink and it was quiet enough for a $10 pump but I could tell it was still vibrating a lot. Eheims were big in the early watercooling scene because they are top quality (made in EU!) and there wasn't really any good options around but they are obsolete for pc watercooling, iirc those were made for aquariums and such have pretty bad flow rate and pressure and require adapter fittings which add to the cost. My first watercooling used that same pump you linked and it was pretty bad. DDC's are pretty good pumps but I didn't mention them because they are a bit noisy and their price is so close to the D5 which is a superior pump qo they aren't worth considering anymore IMO. Waterblock looks fine, only difference between a top range block and this one will the material used and the maze inside so it will only be a difference in cooling, a couple of degrees top. Only advice I can give you is if those pictured fittings are included don't use them and get threaded pressure fittings. oh and I would advise against PWM pumps because I heard it increase bearing wear but pwm is a gimmick on a water pump anyway e.here's mine when it was new, maybe it give you some ideas, good luck 
Sh4 fucked around with this message at 00:58 on Sep 21, 2015 |
|
#
?
Sep 20, 2015 23:19
|
|
|
I always thought compression fittings where more for looks over barbs, I mean yeah they are more secure but I thought most people still considered barbs reliable enough. My cheap rear end radiator(s) have permanent barbs. I guess I have a few options, I can either a) use some super glue or silicone sealant as an extra measure (is this idea as dumb as it sounds?), b) mount the radiator outside the case with the barbs facing backwards and try to find a clever solution to make the hose do a 180 without collapsing, c) purchase a real radiator (but my Craigslist monies!!!) or d) suck it up and accept the dangers of water cooling.
|
|
#
?
Sep 21, 2015 02:11
|
|
|
Crotch Fruit posted:I always thought compression fittings where more for looks over barbs, I mean yeah they are more secure but I thought most people still considered barbs reliable enough. My cheap rear end radiator(s) have permanent barbs. I guess I have a few options, I can either a) use some super glue or silicone sealant as an extra measure (is this idea as dumb as it sounds?), b) mount the radiator outside the case with the barbs facing backwards and try to find a clever solution to make the hose do a 180 without collapsing, c) purchase a real radiator (but my Craigslist monies!!!) or d) suck it up and accept the dangers of water cooling. Barbs are okay, it's just that those pictured are really crappy. Don't use sealant or anything just hose clamps but careful to not damage your tubing.
|
|
#
?
Sep 21, 2015 02:50
|
|
|
I'm having issues with my i5 6600k overclocking on my ASUS ROG MAXIMUS VIII Ranger motherboard. From what i read everywhere I should use adaptive voltage and set the maximum voltage there but I tried to set the turbo voltage maximum on my ASUS ROG Maximus VIII Ranger to 1.28V but it doesn't work, it just blows past it to like 1.45V at load which is insane for 4.2ghz. Here are my settings:  Here's the actual voltage during stress test: 
Mayne fucked around with this message at 23:24 on Oct 1, 2015 |
|
#
?
Oct 1, 2015 22:55
|
|
|
I had something similar happen and I think people told me that CPU-Z is just displaying it wrong.
|
|
#
?
Oct 2, 2015 03:40
|
|
|
Your offset value is "Auto", change it
|
|
#
?
Oct 2, 2015 03:47
|
|
|
I did, changed offset to -0.001 to see if it works, also changed LLC to level5 instead of auto. The result  :  For now i'll keep using Offset mode with -0.1V set, during x264 testing i get 1.25Vcore and about 45C temps and 1.28Vcore/57C max on Prime95 28.7 small FTT test. Mayne fucked around with this message at 12:10 on Oct 2, 2015 |
|
#
?
Oct 2, 2015 12:02
|
|
|
|
| # ? May 16, 2024 09:53 |
|
|
LooKMaN posted:I did, changed offset to -0.001 to see if it works, also changed LLC to level5 instead of auto. Of course increasing LLC power will give you higher Vcore. You should start at lowest LLC (but not disabled as will get Vdrop) and increase if you notice instability and/or Vdrop under heavy load. edit : your memory voltage seems low, you might want to double check your memory settings or turn on XMP hth
|
|
#
?
Oct 2, 2015 15:55
|
|