EPTSimple Questions Simple Answers - Page 317
| Forum Index > Tech Support |
|
Belial88
United States5217 Posts
| ||
|
Belial88
United States5217 Posts
On April 24 2013 15:55 CuraOh wrote: how much power i need for a A8 3670k and a gtx 650 ti would be ok a antec earthwaths 380? maybe i overlock i litle the processor but only to 3.5 no more than that regards PD: the components where a gift and would be a pc for my girldfriend for play alot games in general but nothing too demanding The earthwatts 380 being good enough would all depend on the price you pay for it. Wattage means nothing, you need to abandon such a misguided way to judge power supplies. Quality > Quantity. A 700w Thermaltake TR-2 V1 will fare far worse than a Rosewill Capstone 450w on high loads. It's more like, how much quality do you think is sufficient? And an Antec Earthwatts 380 is a decent enough unit, but it's not really a high quality one, and I don't know of anywhere where it's sold at a decent price at the moment befitting it's quality. It's also a rather outdated unit these days. Right now the CX430/CX430m is a good, cheap bet, that will be sufficient for your build. It's not a great unit, but the support is outstanding for it and it'll be enough for your simple needs (and overclocking). I'd say the quality is slightly less than the earthwatts, but the earthwatts is overpriced at $35+. Unless your earthwatts is $25-35, there are much higher quality units for the similar price, and a similar quality unit at a lower price in the cx430/cx430m. and that green. i didn't realize how ugly it was until i got rid of my earthwatts. On April 25 2013 05:47 Sverigevader wrote: I have just now removed my H100i because the internal screw threads for one of the fan screw holes were broken/gone. Upon removing, I noticed something strange and I would like some information on this. This is a picture of the heat spreader on the cooler. + Show Spoiler + ![[image loading]](http://i35.tinypic.com/25iqi5l.jpg) This is a picture I made in paint to show you how the heat spreader on the CPU looked like when I removed the H100i + Show Spoiler + ![[image loading]](http://i34.tinypic.com/1e2gr4.png) As you can see. The heat spreaders have only made "contact" on the one side, leaving the other side untouched. My question: Who is to blame? I'm most likely going to get a new H100i so I would like to know if this could be my fault somehow. Is it possible that the h100i or even the CPUs heat spreader is uneven? Does it matter if they only make contact on half of the CPU heat spreader? Some info:+ Show Spoiler + I had no problems attaching the h100i pump unit to the CPU, I used a Philips head for the 4 thumbscrews to make sure they were equally tight and tighter than what my fingers could produce. I can't show you any temp values for obvious reasons but I remember some of them. I have a 3570k and with the H100i I got around 25-35 C when idle and like ~40 on load, not OCed. I ran a 4.5ghz OC 1.25 volts just to test my chip and remember the temps going up to 82 C with prime95 running. That was way too high for me so I lowered it to 4.4 GHz and around 1.235 volts with put it at 50-60C max. I did have some pretty uneven core temperatures. 0,1 and 3 were very similar, only different by a <5 C, but core 2 was always 10C higher, and I just assumed this was because of Intel being cheap... If you really have an issue, it would be very easy for you to just RMA your unit, no need to buy a new unit... You can also do an advanced RMA, where you just verify to Corsair you have the money on a card and they'll send you your replacement first. Corsair's support is, in my experience and most other people whom have had the misfortunate of RMAing just about everything from every company, the best in the industry (to make up for what they lack in product quality), so even if you were at fault here, Corsair would likely get you a replacement no questions asked (even if you bought your heatsink in a drug deal). However I really can't tell if there's an issue there. You picture is too low quality. Yes, all heatsink companies put way too much paste on the heatsink base, but I can't tell if the other half is just totally undisturbed, indicating that you mounted your heatsink incorrectly or there was a mount defect (which I doubt, frankly), or if it's smudged up and you are overreacting. You really need to make sure that the backside of the corsair mount, the one that goes behind your motherboard, that the screws go 'all the way in' into that x-mount plate. The fact you reached 82*C on the low overclock that you did, screams to me that you screwed up the mount, but it could be possible if it's in a warm room with low fans... maybe.... not really. For future reference, max practical temp for ivy when overclocking is really more like 90-95*C, you shouldn't be alarmed to hit 82*C (besides the fact you hit it on such a low overclock and voltage). It would be a colossal waste of money if you had an H100 and only did 4.6ghz, doing what a hyper 212+ could do for 1/6th the price. As long as your max peak temp stays below 95*C you are fine (and that's more for stability, 105*C is the thermal throttling point that's far below the chip's true max temp). | ||
|
Ropid
Germany3557 Posts
On April 25 2013 05:47 Sverigevader wrote: I have just now removed my H100i because the internal screw threads for one of the fan screw holes were broken/gone. Upon removing, I noticed something strange and I would like some information on this. This is a picture of the heat spreader on the cooler. + Show Spoiler + This is a picture I made in paint to show you how the heat spreader on the CPU looked like when I removed the H100i + Show Spoiler + As you can see. The heat spreaders have only made "contact" on the one side, leaving the other side untouched. My question: Who is to blame? I'm most likely going to get a new H100i so I would like to know if this could be my fault somehow. Is it possible that the h100i or even the CPUs heat spreader is uneven? Does it matter if they only make contact on half of the CPU heat spreader? Some info:+ Show Spoiler + I had no problems attaching the h100i pump unit to the CPU, I used a Philips head for the 4 thumbscrews to make sure they were equally tight and tighter than what my fingers could produce. I can't show you any temp values for obvious reasons but I remember some of them. I have a 3570k and with the H100i I got around 25-35 C when idle and like ~40 on load, not OCed. I ran a 4.5ghz OC 1.25 volts just to test my chip and remember the temps going up to 82 C with prime95 running. That was way too high for me so I lowered it to 4.4 GHz and around 1.235 volts with put it at 50-60C max. I did have some pretty uneven core temperatures. 0,1 and 3 were very similar, only different by a <5 C, but core 2 was always 10C higher, and I just assumed this was because of Intel being cheap... In this review of an ASRock Z77 Extreme4, the reviewer had serious problems installing his waterblock, thinking the cause was the ASRock using a very thin, easily flexing PCB: http://hardocp.com/article/2013/02/06/asrock_z77_extreme4_lga_1155_motherboard_review/6 The reviewer didn't see anything bad happen to the temperatures because of that. There are also (very good) coolers that are actually intentionally built with a curved convex base instead of perfectly flat, so it's perhaps not that important to have perfectly flat contact with the CPU's top. Example of a curved base: + Show Spoiler + ![[image loading]](http://www.frostytech.com/articleimages/201012/TRsilverarrow_base2.jpg) That said, I think your temperatures are too high and there's something off. The 10 C difference you see on that one core is normal. | ||
|
GolemMadness
Canada11044 Posts
On April 24 2013 23:21 Rannasha wrote: Not a direct answer to your question, but if you want a player with a basic interface, try Media Player Classic or VLC. 380W is plenty for that setup, even with some overclocking. I just want a simple library without useless crap taking up space. | ||
|
Sverigevader
Sweden388 Posts
On April 25 2013 07:56 Craton wrote: I'm not really seeing an issue here. There's nothing that says paste has to come off evenly when you remove a heatsink. If it stuck to the CPU on one side and the heatsink on another, so what? Do you mean that side of the CPU was totally pristine? That seems unlikely, since once you tighten the screws the pressure should automatically spread out the paste. Your temperatures looked fine. As I recall, you're only supposed to tighten that heatsink finger-tight, not screwdriver tight. Overtightening hurts, not helps (it causes it to flex, which impedes heat transfer). 82C under Prime95 is pretty low, frankly. Your max temp is 105C and you're basically never going to be running that high outside of a stress test. Mine is 89C under stress test, yet only 61C under HD streaming / gaming at 1440p. Cores varying in temp is normal. I'm tellin you it was completely untouched. I'll try only using my fingers the next time, though I don't think I over tightened it. On April 25 2013 13:58 Belial88 wrote: If you really have an issue, it would be very easy for you to just RMA your unit, no need to buy a new unit... However I really can't tell if there's an issue there. You picture is too low quality. Yes, all heatsink companies put way too much paste on the heatsink base, but I can't tell if the other half is just totally undisturbed, indicating that you mounted your heatsink incorrectly or there was a mount defect (which I doubt, frankly), or if it's smudged up and you are overreacting. You really need to make sure that the backside of the corsair mount, the one that goes behind your motherboard, that the screws go 'all the way in' into that x-mount plate. The fact you reached 82*C on the low overclock that you did, screams to me that you screwed up the mount, but it could be possible if it's in a warm room with low fans... maybe.... not really. For future reference, max practical temp for ivy when overclocking is really more like 90-95*C, you shouldn't be alarmed to hit 82*C (besides the fact you hit it on such a low overclock and voltage). It would be a colossal waste of money if you had an H100 and only did 4.6ghz, doing what a hyper 212+ could do for 1/6th the price. As long as your max peak temp stays below 95*C you are fine (and that's more for stability, 105*C is the thermal throttling point that's far below the chip's true max temp). Excuse my semantics, I meant that I would be getting a new one through RMA. Ambient temps are around 21-24C, 21C right now. Case is a Level 10 GT with the 200mm side intake angled upwards to the H100i heat sink. I don't think there was anything wrong with the mount or the pump unit when mounting them. Nothing felt wrong at least. I'll think about it when the new one arrives. You and I are different people, I care not for OCs in the +4.6GHz range. I care more about low temps and easy to clean interiors. A Hyper 212+ would be big and collect dust right in the middle whereas a H100i would not. And it is much more prettier  Higher res picture + Show Spoiler + http://imgur.com/t92Ff94 On April 25 2013 14:32 Ropid wrote: In this review of an ASRock Z77 Extreme4, the reviewer had serious problems installing his waterblock, thinking the cause was the ASRock using a very thin, easily flexing PCB: http://hardocp.com/article/2013/02/06/asrock_z77_extreme4_lga_1155_motherboard_review/6 The reviewer didn't see anything bad happen to the temperatures because of that. There are also (very good) coolers that are actually intentionally built with a curved convex base instead of perfectly flat, so it's perhaps not that important to have perfectly flat contact with the CPU's top. Example of a curved base: + Show Spoiler + That said, I think your temperatures are too high and there's something off. The 10 C difference you see on that one core is normal. I have a Asus P8Z77- Le Plus, I don't know if it could be because of that. I'll just try to be more... caring? next time | ||
|
Belial88
United States5217 Posts
I ran a 4.5ghz OC 1.25 volts just to test my chip and remember the temps going up to 82 C with prime95 running. That was way too high for me so I lowered it to 4.4 GHz and around 1.235 volts with put it at 50-60C max. Wait a second, you changed voltage but .015v and your temps rose 32*C? Something really doesn't sound right with your story. You need to use 5+ minutes of small fft prime95 for temperature testing, please (this is minutes 15-30 of blend). You need to use the maximum core temp, as in the highest temp reached by any of the cores (a variance between cores of even 15*C is perfectly normal). Changing frequency is responsible for a bit of temps, but 100mhz and .15v change in temps, that's literally a smaller difference in voltage than going a step up in LLC. Were these 2 tests taken in different countries? You and I are different people, I care not for OCs in the +4.6GHz range. I care more about low temps and easy to clean interiors. A Hyper 212+ would be big and collect dust right in the middle whereas a H100i would not. And it is much more prettier If you really care about low temps then you wouldn't use a water based cooler because they idle at higher temps than air coolers, particularly HDT coolers like the hyper 212. Also, a water cooler with a radiator like the h100 is going to get just as dusty as a hyper 212... you realize there's a big heatsink on it, right? The only difference in air and water cooling is the tubes are filled with treated water instead of a metallic liquid/powder and water has long and flexible tubes vs short, metallic tubes. 82*C is still a very 'low' temp. Bear in mind that the stock cooler on stock settings can very easily reach ~85*C. I don't know what you care about low temps for.... ivy bridge is extremely hardy. It's all about where the temp sensor is located too, it's like when people say they prefer AMD chips to intel because they run cooler. No, they run just as hot, just the thermal sensor is placed further from the cores so it reads at a lower temp (although AMD chips can't take as much heat either due to higher quality pcb i believe). If looks has anything to do with it, why didn't you just buy an h50 instead? Or hell an h40 or asetek 550lc would have been a better choice, $20. An h100 for 4.4ghz is a waste of money, and is not going to be much cooler than a hyper 212+ or h50 at such a low overclock. The h100, like all closed loop/water coolers, is only really going to be different than lower end and air coolers is when you really push it. You already are having higher temps than anyone gets with a hyper 212+. What you say doesn't make any logical sense. If you are concerned about power consumption then there are much more efficient ways to do that, like a higher quality PSU and motherboard. It's not that we are 'different' here, it's that you wasted money. The h100 is going to attract way more dust than a hyper 212 because it has 2 fans, it's going to clog with way more dust because of it's tighter FPI. It doesn't matter where your H100 is located, you have a case airflow and dust is going to be stuck in the middle whether your heatsink/fans are in the middle or not. And really, you could have a hyper 212+ or H50, put no fan on it, and it'll cool just as well and with less dust. It's just an absurd comment that makes no sense. Try remounting your heatsink. Don't worry about the tim, it'll spread properly as you have more than enough as it is on it. Make sure it sits evenly on the cpu, and make sure to check the back plate so that none of the screws aren't sitting flush against the board. Don't put back the other side panel until you got your h100 completely secured on. I mean it's like saying you prefer horse and buggy because it's lower temps than a jet engine. A jet engine is not very hot at only 1000*C, it's supposed to run at that. Buy what you want but you would have been way better off getting an i7 for your money instead. You spent a ton of extra money on a high quality motherboard, CPU, and heatsink, but you aren't properly utilizing them. It's like saying you don't want to put too much pressure on your mustang by driving 55 on the highway. It sounds like you have a blessed chip if you can actually do 4.5ghz@1.25v. That means you can very likely do 5ghz easier than most chips can do 4.6ghz, and you got the hardware capable of doing it. I see great potential being squandered is all, if you dont care about overclocking then whatever, but you got great hardware and i think if you wanted to, you could really have a lot of fun. It doesn't make sense to say "im not going to go to 4.6ghz, i want to stay safe!' when you got a cpu so good that it would be safer for you to do 5ghz than for many chips to do 4.5ghz. You don't even have to delid to reach a very high overclock on a chip that good with a cooler that high end. | ||
|
Sverigevader
Sweden388 Posts
On April 25 2013 16:09 Belial88 wrote: Then it's a waste of money, i don't know what you mean about 'low temps'. If you really care about low temps then you wouldn't use a water based cooler because they idle at higher temps than air coolers, particularly HDT coolers like the hyper 212. Also, a water cooler with a radiator like the h100 is going to get just as dusty as a hyper 212... you realize there's a big heatsink on it, right? The only difference in air and water cooling is the tubes are filled with treated water instead of a metallic liquid/powder and water has long and flexible tubes vs short, metallic tubes. 82*C is still a very 'low' temp. Bear in mind that the stock cooler on stock settings can very easily reach ~85*C. I don't know what you care about low temps for.... ivy bridge is extremely hardy. It's all about where the temp sensor is located too, it's like when people say they prefer AMD chips to intel because they run cooler. No, they run just as hot, just the thermal sensor is placed further from the cores so it reads at a lower temp (although AMD chips can't take as much heat either due to higher quality pcb i believe). As for prettier, that's totally subjective. IMO the larger and more offensive the air heatsink, the better it looks. I think a huge block of metal like the nh-d14 looks a lot better than the dead space of a closed loop. Not that I really care either way. If you wanted aesthetics of a closed loop then you would have fared much better with just getting some h40 or asetek 550lc for $20. An h100 for 4.4ghz is a waste of money, and is not going to be much cooler than a hyper 212+ or h50 at such a low overclock. The h100, like all closed loop/water coolers, is only really going to be different than lower end and air coolers is when you really push it. You already are having higher temps than anyone gets with a hyper 212+. What you say doesn't make any logical sense. If you are concerned about power consumption then there are much more efficient ways to do that, like a higher quality PSU and motherboard. It's not that we are 'different' here, it's that you wasted money. The h100 is going to attract way more dust than a hyper 212 because it has 2 fans, it's going to clog with way more dust because of it's tighter FPI. It doesn't matter where your H100 is located, you have a case airflow and dust is going to be stuck in the middle whether your heatsink/fans are in the middle or not. And really, you could have a hyper 212+ or H50, put no fan on it, and it'll cool just as well and with less dust. It's just an absurd comment that makes no sense. Try remounting your heatsink. Don't worry about the tim, it'll spread properly as you have more than enough as it is on it. Make sure it sits evenly on the cpu, and make sure to check the back plate so that none of the screws aren't sitting flush against the board. Don't put back the other side panel until you got your h100 completely secured on. I mean it's like saying you prefer horse and buggy because it's lower temps than a jet engine. A jet engine is not very hot at only 1000*C, it's supposed to run at that. Buy what you want but you would have been way better off getting an i7 for your money instead. You spent a ton of extra money on a high quality motherboard, CPU, and heatsink, but you aren't properly utilizing them. It's like saying you don't want to put too much pressure on your mustang by driving 55 on the highway. It sounds like you have a blessed chip if you can actually do 4.5ghz@1.25v. That means you can very likely do 5ghz easier than most chips can do 4.6ghz, and you got the hardware capable of doing it. I see great potential being squandered is all, if you dont care about overclocking then whatever, but you got great hardware and i think if you wanted to, you could really have a lot of fun. It doesn't make sense to say "im not going to go to 4.6ghz, i want to stay safe!' when you got a cpu so good that it would be safer for you to do 5ghz than for many chips to do 4.5ghz. You don't even have to delid to reach a very high overclock on a chip that good with a cooler that high end. You're right, it is subjective. I prefer a h100i to a hyper 212 because of how it looks and that the dust won't collect in a visible place. It will be much easier for me to clean a top mounted h100i than a hyper 212. I can flip open the top of the case which exposes the pull side of the h100i. I suppose I could have gone with a H55. When I checked the reviews on all of them the H100i looked like the better choice. I wouldn't save a lot of money in any case... I'll make sure to mount the new H100i, when I get it, correctly. I've never OCd before and when I saw the high temps at 1.25 V it didn't feel right so I lowered it, simple as that. Also, how longer would the CPU last with a lower OC/temps compared to higher? If a 4.6GHz at 1.3V and temps around 80C would result in a much lower life expectancy, it's not good for me. It's not like my life depends on that extra 100-200MHz. You make good points though. Thanks for the input. Maybe I'll try a higher OC  | ||
|
Ropid
Germany3557 Posts
Generally, I was under the impression that CPU's pretty much never die when looking at things like the cause of a server failing. It's always something else making someone have to work on or throw out a computer. I bet thinking about the motherboard failing earlier makes more sense than worrying about the CPU. I've seen a graph of someone overclocking an i7-3770k, measuring 50 W at 4 GHz, 100 W at 4.5 GHz, 150 W at 5 GHz, and the motherboard has to deal with that. | ||
|
Belial88
United States5217 Posts
You should try remounting your h100, i really doubt the product is flawed. I'm mounted over a dozen heatsinks and done 100 mounts, and I still screw up mounts sometimes. Try again. Take some pictures of your mobo cpu area, of the mounting brackets. I mean if something was off, you'd see it. A flat surface plus flat surface does not make a skewed contact point. Seriously, mount the h100 again and look behind your motherboard, on the cpu punch-out, make sure it's all straight back there because that's my guess on what's going on here. One of the screws for the X back plate is sort of slipping out a bit, that's why it's not sitting straight. Like you are just going to RMA your heatsink and the same issue will happen. Or, they could simply say nothing is wrong and send it back to you. You really need to be sure you aren't screwing anything up here. It's okay if you freaked at the temps you saw, better to be safe than sorry. There's a program called HWInfo, use it, go to the thread on overclock.net, the official hwinfo64 thread, it'll tell you how to make it so you can have your computer shutdown if you ever hit X temperature. So you can tell it to shutdown at 80*C, or 90*C (90*C as a max temp is honestly considered on the low end, most say 95, 100, 85 would be super conservative if you wanted to fly it like that). Coretemp also does this same function although it's UI is a bit easier to set up, frankly. It's a worse program though. Take your time, read up, and then do whatever. With overclocks, degradation, you are talking things like the cpu lasting only 20 years instead of 25 years. First off, we can't really tell you because ivy bridge has only been out for a year, it's possible that all overclockers just spontaneously combust in 666 days. But there are tons of overclockers, like myself, who actually torture our CPUs and really push them to the limit. And not a single person has shown any sign of 'degradation' on ivy bridge. Degradation means that you need higher voltage to be stable at the same overclock. Or, you start to see crashes on an overclock or setting that didn't occur before (and it's because of CPU degradation, and not some other issue obviously). Myself and a few others, like martinhal and valguar, we are pushing extreme voltages and constantly stress testing to see if any signs of degradation. So far, a large handful of us have been running 5, 5.1ghz at 1.5-1.6v 24/7 on ivy bridge with absolutely zero sign of degradation. The only person who had degradation on ivy bridge was a guy who did 1.9v on air, and he was trying to kill his chip. Sandy bridge, on the other hand, there was a real risk with overclocking. Going over 1.4v, within months of release there were lots of reports of degradation, people needing to bump up voltage to keep staying stable or had to reduce the frequency. CPU death is very rare, you have to be pushing extremes on liquid nitrogen to play around with voltages to kill, but only one ivy has died and there have been zero reports of degradation. It's a very hardy chip, if you ever see reports of anyone saying their chip degraded, then simply tune back your low overclock, haha. Of course, stay below 90-100*C, all of this is provided you stay below 90-100*C. I mean ivy bridge has a thermal limit in place where if you hit 105*C, it just throttles itself, so you can't actually damage the chip. I'm sure you could possibly degrade it if you run 24/7 compute workloads where the cpu was at like 5.1ghz@1.6v and 89*C for a year... but even then people are doing compute workloads on extreme overclocks and not having issues. It's just a chip that's built like a tank. So there is a risk, and no one can tell you exactly the risk, but there's no reason to believe that ivy will degrade in any appreciable or practical sense when overclocked. Just make sure you aren't the #1 most extreme overclock and you'll be fine lol. I mean maybe 4.6ghz@1.3v is half the life of stock, but that's like 20 years vs 40 years, you know. Or no one knows, really. And what about stock voltage? That's just an arbitrary value, why not be safer and downclock your CPU and voltage even further than intel's stock settings? Make the chip last longer, no? It's like a car, i mean you gotta use the thing, but driving on the highway for 70mph every day for a few hours isn't going to do anything to prevent the car from lasting a good 10+ years. If you want to drive it like a trucker, 24/7, well, these chips are basically built like trucks, not like cars, and they'll handle that for a good few years just fine. I understand you want to be safe, but there have been no reports of degradation on ivy. That's with millions of users. Now if this was sandy bridge, there's some danger, sure, but not with ivy bridge. It's really the chip to push. If you are really worried, there's the intel performance plan - for $20-35, you get unlimited overclocking coverage for a single chip. Meaning if you blow your chip out at 6ghz@1.9v on air, they'll give you a free replacement. Maybe this will help your peace of mind? Finally - if you use offset voltage and keep the power saving features turned on (as you should), like turbo, c1e, eist, your chip will downclock and downvolt most of the day. So take my system, i'm at 5ghz@1.5v, max temp 83*C. A pretty high overclock, many would say it's extreme. I would say it's just a high overclock, not extreme, but whatever. I'm perfectly okay with 5ghz@1.5v, because when I'm asleep, when im just on reddit or TL, my system run at 1.6ghz@.9v. Even during most of gaming, it just runs at 1.6ghz@.9v. So like 90% of the day, it's not at 5ghz@1.5v, it's at 1.6ghz@.9v. So you are doing way less wear and tear, because it's only on the highway for a little bit of the day anyways. | ||
|
Cyro
United Kingdom20285 Posts
![[image loading]](http://i.imgur.com/zmxrbTP.png) ![[image loading]](http://i.imgur.com/MZOmScH.png) ^Power usage scaling is almost entirely on voltage though, it's almost funny. A chip doing 5ghz at 1.285v will use like half as much power as one at 1.5v, wear and tear on motherboard if it even exists would be a lot lower, and you're essentially going from a "normal stock voltage" of say, 1.2 - to 1.285, and gaining 1.6ghz. Your CPU will run pretty much cooler and last longer than some dude doing 4.5/1.3 probably, to use an example My point from this, you can't really measure wear and tear on mobo, CPU etc, or even get any kind of idea, you cant say "this cooler can take 4.8ghz" it's more down to the voltages you are stable at. "This cooler can take 1.3v and 1.3v might get 4.4ghz or 5.2ghz depending on luck" is a more appropriate statement | ||
|
Sverigevader
Sweden388 Posts
On April 25 2013 17:23 Belial88 wrote: You should try remounting your h100, i really doubt the product is flawed. I'm mounted over a dozen heatsinks and done 100 mounts, and I still screw up mounts sometimes. Try again. Take some pictures of your mobo cpu area, of the mounting brackets. I mean if something was off, you'd see it. A flat surface plus flat surface does not make a skewed contact point. Seriously, mount the h100 again and look behind your motherboard, on the cpu punch-out, make sure it's all straight back there because that's my guess on what's going on here. One of the screws for the X back plate is sort of slipping out a bit, that's why it's not sitting straight. Like you are just going to RMA your heatsink and the same issue will happen. Or, they could simply say nothing is wrong and send it back to you. You really need to be sure you aren't screwing anything up here. Like I said in the first post, I'm returning it, not because of this, but because of a screw hole thread fail. I couldn't tighten one of the corners for one of the fans causing a lot of give between the heatsink and the fan which resulted in noise. When I held it down with my finger, the noise went away. But when the new one arrives, I'll make sure it is attached correctly. I've read that the thermal paste corsair supplies their new H100i with is pretty freaking good, so I'm reluctant to removing it to check if the problem persists. I have a tube of AS5 that I could use, but that's not as good as the pre-applied stuff right? | ||
|
Belial88
United States5217 Posts
You could also use cable ties or double sided 3M tape for the screw hole that failed, if you don't want to go and RMA. The stock paste corsair supplies with their h100 is decent. I didn't say remove the paste, I said you had enough paste left over still that you could still remount just fine. AS5 is terrible paste, it was good when it came out almost 15 years ago, but not anymore (a 15 year old CPU wouldn't be very good either). The pre-applied stuff isn't awesome, it's more that as5 is just terrible. | ||
|
Sverigevader
Sweden388 Posts
Can you really just remount a heatsink like that with old paste still attached? I always read that you should wipe it off and apply new paste on a fresh surface. | ||
|
skyR
Canada13817 Posts
| ||
|
mav451
United States1596 Posts
On April 25 2013 13:45 Belial88 wrote: Has anyone ever went from 2 sticks of RAM to 4, and had to reduce their overclock or could no longer do a certain speed? Core2 users usually had to do a combination of increased NB voltage and loosened timings/latencies to maintain current overclocks, once they filled all 4 DIMMs. I'm aware that Lynnfield/Bloomfield users had to raise QPI/VTT voltage as well. I know the move from a 4GB to 8GB kit definitely required additional testing on my part. Is there a reason you're going to 4 DIMMs? | ||
|
Cyro
United Kingdom20285 Posts
Chances are that if a poster is asking for help in the first place, that the difference in TIM will not be tangibly affecting their OCs. But it was something like 10c gap, IIRC, enough to be one of a chain of factors causing really excessive temperatures for a given voltage NH-D14: PK1: 49*C PK2: 48*C PK3: 45*C From Belial's paste blog, but those are 3 good pastes - and at a lot temp - where 4c is a much greater temperature gap than if it were 80 vs 84 for example, and iirc all three of those are at least decent/good paste Is there a reason you're going to 4 DIMMs? Belial bought 2x2gb modules, spent like 2 weeks overclocking them to 2400 cas8-11-8-28 or something and then decided 4gb RAM was not enough, so spent another month trying to get stable for more than 23 hours in prime with his next pair at 2200-2400mhz too =P I wonder if he'll be done stress testing before he sells the system to build Haswell or if he will have to say goodbyes with a half empty notebook and far too few primes throwing a whea error at 23 hours | ||
|
mav451
United States1596 Posts
 I guess my other point is this. I normally would agree with you, but my recent experience with my 750 has put a different tune on the combination of luck, regarding gains in going to a better heatsink. Consider this - my 750 did 3.6 through 3.8Ghz in fairly linear voltage increments. But around 3.9Ghz, the requisite increase to maintain stability became much steeper. With my 212+ (P12) it was doing 3.95Ghz 1.34vcore (long way from the 1.2v I gave for 3.6Ghz, and 1.1v for 3.3Ghz); anyway overnight P95 custom blend was putting me at 81+. I move to the D14. Drop about 12C to 70C for your basic 1792K custom blend. Problem is I still couldn't stabilize to an even 4Ghz, not even with a massive increase, say 10 increments to (1.4vcore). At that point my temps were in the 80's once again. So I would say yeah, to an extent it is one of many things. So really I bought this for Haswell anyway, but yeah not the best chip in the end. I've never been a fan of filling all four DIMMs. It wasn't even ideal before the IMC, and likewise I prefer not doing it in the current generation either. It'd be something else if he had the same week/year chips for his second pair. And yeah, I think it's funny cuz he gave a nice lecture on "how 4GB is enough" a while back. Every time I increased capacity, I've always preferred to sell off my previous kit. (E.g. 2x512 was sold to pay for my 2x1GB, and 2x1GB paid for my 2x2GB...and so forth) You see the Coolaler forum news on Haswell? Hitting 3K on the RAM, but then all the typical boilerplate benches are run at typical 1600 -_- + Show Spoiler + | ||
|
Belial88
United States5217 Posts
Can you really just remount a heatsink like that with old paste still attached? I always read that you should wipe it off and apply new paste on a fresh surface. You can. Absolute worst case scenario your temps are 3-5*C hotter. But in your case we are talking about an incorrect mount, most likely due to human error but it could be a faulty component (which I really doubt), so I really think you should just try and remount and get things right. The problem with remounting used applications of paste is extremely tiny air pockets, but pre-spread paste like on your h100 already has, will always have tiny air pockets so it's not going to be much worse to just remount without cleaning and reappliyng paste. For future reference, when applying paste, put a small, tiny ricegrain amount (as in a large ricegrain is way too much paste). You actually only want about 1/2 of the cpu covered in paste. Here's a bit of info on how to apply paste: http://www.overclock.net/t/1346069/belials-heatsink-tim-comparison-and-reviews-hyper-212-h50-nh-d14-pk-1-pk-2-pk-3/10 Lol that kind of hyperbole is a bit uncalled for on the AS5. Chances are that if a poster is asking for help in the first place, that the difference in TIM will not be tangibly affecting their OCs. Not at all. The same 'hyperbole' would be used if someone asked if the Pentium 2 would be a good CPU for Starcraft2. You do realize that AS5 is as old as the pentium 2 right? Thermal paste has advanced just as much as CPUs have in the last 11 years. There is little difference in modern thermal pastes. The problem is that AS5 is by no means a modern paste. Belial bought 2x2gb modules, spent like 2 weeks overclocking them to 2400 cas8-11-8-28 or something and then decided 4gb RAM was not enough, so spent another month trying to get stable for more than 23 hours in prime with his next pair at 2200-2400mhz too =P It was actually less than week for the initial 2x2 sticks, and it only took that long because gigabyte had released a new beta bios that I opted to join testing in, so half of the testing was actually in regards to the new F15R and F15Q bioses, not the RAM, and secondly because, as buggy as the F15 bios was/is, it has a ton of insanely intricate and minute timings that I saw fit to spend my time playing with. Slew rates, tXPDLL, REF_Panic_XM, refresh interval ix9s, some of which were very relevant to the particular RAM IC I was playing with (as in, a lot of people say with PSC just mess with the first 2 timings and slew rates and that's it). When I added 2x2gb of additional sticks, I've been having some odd problems because my IMC was not up to snuff. I only recently realized it was my IMC because no matter how much additional VTT/IMC voltage I added, had zero impact on stability (as in i'd crash sooner just as often as later). Even weirder, significantly reduced VTT/IMC voltages actually passed stability tests on 2200mhz CL8, if the IMC was suspect you'd think I'd at least need extra voltage to do 2200mhz when I can't do 2400mhz. And usb failure is an issue with PLL voltage being too low, so I addressed that too. I'll be done tommorow onsider this - my 750 did 3.6 through 3.8Ghz in fairly linear voltage increments. But around 3.9Ghz, the requisite increase to maintain stability was incredibly steep. With my 212+ (P12) it was doing 3.95Ghz 1.34vcore Ivy is fairly linear in it's voltage increase to increased overclock, it's not a sweet spot cpu like the 750 as you describe or amd chips. I've never been a fan of filling all four DIMMs. It wasn't even ideal before the IMC, and likewise I prefer not doing it in the current generation either. It'd be something else if he had the same week/year chips for his second pair. And yeah, I think it's funny cuz he gave a nice lecture on "how 4GB is enough" a while back. Every time I increased capacity, I've always preferred to sell off my previous kit. (E.g. 2x512 was sold to pay for my 2x1GB, and 2x1GB paid for my 2x2GB...and so forth) Perhaps, but PSC ram can be bought for $25 for 2x2gb and is basically guaranteed to do 2400mhz CL8 (both of my kits can do 2400 CL8). Now doing 4x2gb can't do 2400CL8, but 2200mhz CL8 is still a lot better than, say, the much more expensive 'budget overclocker' sticks like the crucial ballisticks LP that at best do 2200mhz CL11. Now RAM prices have changed obviously, and PSC is rarer to come by today than it was 2 months ago - even my recent purchase of PSC required PMing random people who said they had PSC 2 years ago on some forum and then asking to buy it from them - but if you have the time and savvy to figure out what 2GB density sticks are PSC XDZ or XDF, you can spend just as much as the cheapest 2x4gb kit while overclocking further than RAM three times as expensive. RAM overclocks generally don't mean anything, but that's assuming like 1333 to 1600 or 1600 to 2000. 1600 to 2200 is a huge boost, enough to see tangible results. It's not like I spent anything extra to get such speeds, just a lot of time, and since I enjoy doing this, it's time well spent on a rather significant performance increase. 4GB of RAM is enough, the difference is that I went to dual monitor and dual monitor tends to have you do stuff that consumes more than 4GB of RAM at once, my GPU is short on VRAM so it was spilling over onto my RAM load, and because I thoroughly enjoyed my first set of PSC that I had to get another. Frankly, I really don't need 8GB of RAM even on dual monitor, and my FPS or in-game performance certainly was not any lower when on 4GB, I just wanted to play with more RAM. Generally you are better off just selling your previous kit, yea, but the 2x2gb ram i had was so cheap and so high performance. Most knowledgeable enthusiasts and benchers actually go with 4x2gb PSC or BBSE RAM - it's only people who want the absolute best of the best and are okay spending $100-300 on RAM that actually buy 2x4gb, as PSC/BBSE will never reach over 2800mhz. There are a few kits at 2x4gb that overclock decently that aren't too expensive, but they won't overclock or bench or perform nearly as well as 4x2gb PSC and they cost slightly more. You'll find even a lot of the top HWbot scores are done with 4x2gb PSC. Not to mention that there's also a performance increase in going with 4 sticks of RAM called bank interleaving (kind of like dual channel vs single channel). But it's a lot easier to just buy samsung 2x4gb than it is to try and figure out what ram uses PSC ICs. | ||
|
mav451
United States1596 Posts
| ||
|
Belial88
United States5217 Posts
It doesn't look like haswell will be a big increase at all, although a bigger difference than sandy to ivy. | ||
| ||
