|
HWiNFO64 (or x32) will show VRM temperatures. GPU-Z should also have a readout if it supports the card.
|
#
?
Jun 2, 2014 14:08
|
|
|
|
| # ? May 27, 2024 14:29 |
|
|
So, picked up a "cheap" radiator,  from one of these  also getting a H.A.F squarish case, I should have got a workbench case years ago, but there was always something else to get. HAF has a bunch of fans I can cannibalise for the rad, next, a cheap aquarium or whatever water pump, borosillicate glass and a blowtorch. staberind fucked around with this message at 14:52 on Jun 2, 2014 |
|
#
?
Jun 2, 2014 14:41
|
|
|
It's the late 90s again!
|
|
#
?
Jun 2, 2014 16:09
|
|
|
deimos posted:It's the late 90s again! In a few years we'll be hitting VapoChill and Prometia in their stride once again!
|
|
#
?
Jun 2, 2014 16:27
|
|
|
I could have sworn I read that 780tis didn't have a vrm thermometer, but I hope I'm just wrong about that. Only reason I didn't check myself is I'm out of town, so thanks for the info that's preferable to say the least
|
|
#
?
Jun 2, 2014 18:37
|
|
|
Yeah yeah, I am making a very 90's cooling system, I hope its offensive and brutal, on the other hand, I did pick up an intel clc, to use as a "control", if my frankencooler can maintain cooling of just the cpu, I am happy, but I am overbuilding the cooling to do memory and gpu as well.
|
|
#
?
Jun 2, 2014 21:09
|
|
|
staberind posted:Yeah yeah, I am making a very 90's cooling system, I hope its offensive and brutal, on the other hand, I did pick up an intel clc, to use as a "control", if my frankencooler can maintain cooling of just the cpu, I am happy, but I am overbuilding the cooling to do memory and gpu as well. Sigh...
|
|
#
?
Jun 2, 2014 21:30
|
|
|
I always wanted to use a tiny radiator or heater core or something. They can be almost essentially free and the capacity rivals even the largest (and expensive) radiators.
|
|
#
?
Jun 2, 2014 21:50
|
|
|
staberind posted:Yeah yeah, I am making a very 90's cooling system, I hope its offensive and brutal, on the other hand, I did pick up an intel clc, to use as a "control", if my frankencooler can maintain cooling of just the cpu, I am happy, but I am overbuilding the cooling to do memory and gpu as well. Watercooling memory...
|
|
#
?
Jun 2, 2014 21:53
|
|
|
So Intel's Devil's Canyon CPUs for overclockers are launching tomorrow, including the Core i7 4790K and the Core i5 4690K. Leaks ahead of embargo seem to confirm that rumors of 4.0Ghz base frequency and 4.4Ghz were true, and Kyle at HardOCP hinted we can look forward to 5Ghz overclocks (though people said that about Haswell too). Interestingly, photos of the back of the CPU show significant changes to the SMD power delivery component layout versus Haswell or even Haswell-Refresh processors, suggesting more changes than simple packaging improvements and more heavily binned CPUs. I wonder if there's a different silicon stepping too?
Alereon fucked around with this message at 05:24 on Jun 3, 2014 |
|
#
?
Jun 3, 2014 05:22
|
|
|
Alereon posted:So Intel's Devil's Canyon CPUs for overclockers are launching tomorrow, including the Core i7 4790K and the Core i5 4690K. Leaks ahead of embargo seem to confirm that rumors of 4.0Ghz base frequency and 4.4Ghz were true, and Kyle at HardOCP hinted we can look forward to 5Ghz overclocks (though people said that about Haswell too). Interestingly, photos of the back of the CPU show significant changes to the SMD power delivery component layout versus Haswell or even Haswell-Refresh processors, suggesting more changes than simple packaging improvements and more heavily binned CPUs. I wonder if there's a different silicon stepping too? This was the impression I got when I first read the clock speed leak. Based on the fact that a 4.4 turbo factory speed with a TDP just 4 watts over the 4770k, this might imply actual changes. There were some people who could barely get their 4770k stable at 4.4 ghz at any voltage. Then there were those who could get closer to 5.0 ghz. Would be nice if they took the "silicon lottery" part out of it, by whatever means. It would also make the i7 appealing for reasons other than hyperthreading. And if this was true, it would also mean my bank account is about to suffer some more.
|
|
#
?
Jun 3, 2014 05:27
|
|
|
Alereon posted:Interestingly, photos of the back of the CPU show significant changes to the SMD power delivery component layout versus Haswell or even Haswell-Refresh processors, suggesting more changes than simple packaging improvements and more heavily binned CPUs. I wonder if there's a different silicon stepping too?  From what I understand, Devil's Canyon chips are already cherry-picked by Intel to remove the lottery element. Alereon posted:I wonder if there's a different silicon stepping too? 
|
|
#
?
Jun 3, 2014 12:06
|
|
|
I'm sure those who care about this thread have seen the 5.5 GHz tweet already, and I'm getting terrible nasty thoughts about my own (stellar in its own right) 3770k in the back of my head, ew. Given decisions between motherboards on par with ROG fanciness, do the VRM configurations available on ATX size boards still make a max clock difference compared to smaller form factors? Or is the only difference nowadays between ATX and stuff like Mini ITX just case airflow? I ask because I was dumb enough to buy an mATX case for my mATX system. in other words do I want 12 or 16 or 20 phases for that leet stylin 6 ghz to satisfy my manchild Sidesaddle Cavalry fucked around with this message at 20:09 on Jun 4, 2014 |
|
#
?
Jun 4, 2014 20:07
|
|
|
Sidesaddle Cavalry posted:I'm sure those who care about this thread have seen the 5.5 GHz tweet already, and I'm getting terrible nasty thoughts about my own (stellar in its own right) 3770k in the back of my head, ew. matx cases can have just fine airflow. so can ITX.
|
|
#
?
Jun 4, 2014 21:06
|
|
|
Unless you're doing LN2 or phase-change benchmarking runs you don't need anything more than 8 to 12. I kind of want to buy one of the unlocked (!!) G3258 Pentiums when they come out. Should be interesting chips if they're loosely binned at all.
|
|
#
?
Jun 4, 2014 21:09
|
|
|
Don Lapre posted:Watercooling memory... The changes in voltage should not really be enough to cause an appreciable effect, its more for the "can this be done?" cheese factor. That's the same reason people climb Everest, fake moon landings and have breast implants.
|
|
#
?
Jun 5, 2014 02:17
|
|
|
staberind posted:The changes in voltage should not really be enough to cause an appreciable effect, its more for the "can this be done?" cheese factor. This always made me laugh a little 
|
|
#
?
Jun 5, 2014 03:23
|
|
|
Guh loving socket everything. socket the GPU, socket main memory and stick some traditional heatsinks or waterblocks on them.
|
|
#
?
Jun 5, 2014 03:46
|
|
|
Ignoarints posted:This always made me laugh a little Thats what I am talking about, tell me that's an April 1st image, please!? Alternately, why not just submerge everything in mineral oil or whatever, I have seen it done. Iirc, the only problem they had was oil siphoning out through the attached cabling.
|
|
#
?
Jun 5, 2014 06:25
|
|
|
 Must be in serious demand as they're all sold out from NewEgg.
|
|
#
?
Jun 5, 2014 09:24
|
|
|
staberind posted:Thats what I am talking about, tell me that's an April 1st image, please!? Its a nitrogen block, so it will be way colder but... obviously totally useless for actual computer use. Dyno queen edit: wow I just saw my renters insurance covers accidental damage to my PC up to $3000 with a $250 deductible. So if I burn out my GPU or CPU somehow I might accidentally throw my computer down the stairs  i know this is insurance fraud Ignoarints fucked around with this message at 15:16 on Jun 5, 2014 |
|
#
?
Jun 5, 2014 14:42
|
|
|
staberind posted:Alternately, why not just submerge everything in mineral oil or whatever, I have seen it done. https://www.youtube.com/watch?v=CIbnl3Pj15w
|
|
#
?
Jun 5, 2014 16:59
|
|
|
Welmu posted:Oil, pfft. Get with the times, nowadays all the Fixed that, have you seen at the price-per-gallon of that poo poo?
|
|
#
?
Jun 5, 2014 17:25
|
|
|
Why isn't Novec cooling liquid used in data centers? seems like it would save a lot of power and allow for smaller footprints? Whats the major draw back besides the cost of the liquid and change in design and maintenance? Don't the benefits outweigh the drawbacks? https://www.youtube.com/watch?v=a6ErbZtpL88 edit: answered my own question quote:I've used the original 3M solution. r0ck0 fucked around with this message at 17:29 on Jun 5, 2014 |
|
#
?
Jun 5, 2014 17:25
|
|
|
deimos posted:Fixed that, have you seen at the price-per-gallon of that poo poo? Only $330
|
|
#
?
Jun 5, 2014 17:26
|
|
|
r0ck0 posted:Why isn't Novec cooling liquid used in data centers? seems like it would save a lot of power and allow for smaller footprints? Whats the major draw back besides the cost of the liquid and change in design and maintenance? Don't the benefits outweigh the drawbacks? Yes but I'm sure its a matter of cost to implement it over existing systems. As far as I know it's so new that only one company is providing actual servers designed to use it, with something like 2 or 3 small time customers so far.
|
|
#
?
Jun 5, 2014 17:29
|
|
|
r0ck0 posted:Why isn't Novec cooling liquid used in data centers? seems like it would save a lot of power and allow for smaller footprints? Whats the major draw back besides the cost of the liquid and change in design and maintenance? Don't the benefits outweigh the drawbacks? Oil is waaaaaaay cheaper. (And it's being used already, TSUBAME is oil cooled.)
|
|
#
?
Jun 5, 2014 17:30
|
|
|
I don't have much experience with data centers, however, I've helped change rackmounts at Telehouse, its the London gateway between Europe and the US, so basically where all the uk isp's rent their server space. comparing the two, I can see the attraction of submersive cooling; despite the air getting changed roughly 12 times a minute, its full of microscopic swarf, it coats everything, even with filters.
|
|
#
?
Jun 6, 2014 00:20
|
|
|
Ignoarints posted:I'm starting to put together what I need to watercool my 780ti with the NZXT G10 bracket. I am concerned about the VRM's. From my quick look, they appear to have a separate metal heatsink as opposed to a bare board. Which is nice, if it's sufficient. this is funny to me because I just got a nzxt g10 for my 780, along with handheld flir cam through work and it owns
|
|
#
?
Jun 6, 2014 03:03
|
|
|
right arm posted:this is funny to me because I just got a nzxt g10 for my 780, along with handheld flir cam through work and it owns Jealous as hell. The pictures from people using those seem so informative. I'm just going to go ahead and put little heatsinks on everything though It's funny to me how I don't remember making that post at all. I was literally sleep typing or something.
|
|
#
?
Jun 6, 2014 04:54
|
|
|
Ignoarints posted:Jealous as hell. The pictures from people using those seem so informative. I'm just going to go ahead and put little heatsinks on everything though all I've used mine for so far is locating my cats at night and looking at footsteps it's real cool also, the g10 is great. perfectly quiet on my asus 780 and like 10° cooler under load and over clocked with zero fan noise. worthwhile investment imho
|
|
#
?
Jun 6, 2014 06:51
|
|
|
I am rebuilding the 290 hopefully on Saturday, so I should have potato pics for you Igno.
|
|
#
?
Jun 6, 2014 07:10
|
|
|
deimos posted:I am rebuilding the 290 hopefully on Saturday, so I should have potato pics for you Igno. sweet! right arm posted:all I've used mine for so far is locating my cats at night and looking at footsteps  I want some of that action
|
|
#
?
Jun 6, 2014 07:54
|
|
|
Ignoarints posted:
|
|
#
?
Jun 6, 2014 13:29
|
|
|
Josh Lyman posted:What's the hot area? VRMs? Yeah, with a synthetic load with the water cooling bracket. http://www.pugetsystems.com/labs/articles/NZXT-Kraken-G10-Review-527/ TL:DR picture version: Stock Titan full game load-  G10 bracket full game load-  The watercooler actually reduces the temperature of the whole board, which is awesome. The heat from the GPU isn't soaking into the rest of the components (I'm guessing, but that's rad) However things takes a turn Stock Titan Furmark-  G10 bracket Furmark  The VRM's heat overcomes the bracket's ability to remove it and gets kind of way too hot for comfort. That is for Furmark which is unrealistic, however the whole point of the bracket (for me) is to overclock it. Hopefully heatsinks will delay that threshold. But there is no real way for me to know I think :/ I'm also pretty surprised the memory gets that hot but I don't know how hot they are able to get either. edit: Well, I didn't think about how wide the VRM heat zone is, maybe a cheapo IR thermometer will give me suitable numbers to at least know if im in the danger zone Ignoarints fucked around with this message at 15:03 on Jun 6, 2014 |
|
#
?
Jun 6, 2014 14:49
|
|
|
What beautiful photos, showing the problem with water cooling. Sure the core is cool as a cucumber, but without airflow over the rest of the components, look at how hot the board gets.  I bet a low RPM 120mm fan would give the card enough airflow to help a lot.
|
|
#
?
Jun 6, 2014 16:07
|
|
|
SocketSeven posted:What beautiful photos, showing the problem with water cooling. Yeah it was exactly what I was looking for. Keep in mind though the GPU dropped by something like 40 degrees C and realistically the whole card really was cooler under a game load. But I can't assume they'll all be like that, especially when overclocked. If anything its a real case for a waterblock but that's a serious expense... but I'm sure thermal pictures of it would be mouth watering. It's just a bit uncomfortable knowing that the heat can runaway like that. Hopefully a IR thermometer does the trick for me
|
|
#
?
Jun 6, 2014 16:12
|
|
|
Does an epoxied mini heatsink really do anything compared to just a fan? How about when there's always a fan and bare ICs vs mini heatsinks?
|
|
#
?
Jun 6, 2014 16:31
|
|
|
Shaocaholica posted:Does an epoxied mini heatsink really do anything compared to just a fan? How about when there's always a fan and bare ICs vs mini heatsinks? Yes. I've done several things to cool the VRM banks on my 290. Running OCCT (with framerate unlocked) my time to 105°C is as follows (with ambients around 32-34°C): Bare chips: 30-35 seconds Custom milled copper bar with Aluminum fins epoxied: 120-140 seconds Bottom bracket of the stock cooler with the vapor chamber removed: ~180 seconds (I didn't test this more than once because I was only testing for clearance to the G10) Bottom bracket of the stock cooler with the vapor chamber removed and Aluminum fins soldered: more minutes than I'd care to test. All of the tests were done using 1mm thick thermal pad (to electrically insulate the VRMs from the heatsinks). Keep in mind that the VRAM chips that were cooled with independent copper heatsinks on the first 2 tests are cooled with the bottom bracket on the last 2. deimos fucked around with this message at 17:04 on Jun 6, 2014 |
|
#
?
Jun 6, 2014 17:00
|
|
|
|
| # ? May 27, 2024 14:29 |
|
|
deimos posted:Yes. Nice man. I'm hoping off the shelf copper heat sinks will cut it for me. I wish I could go that far though. Any recommendations on thermal pads because I've read quite a few reports of them just falling off with what comes with them. Although they are often countered by people saying they didn't do it right. Last thing I'd want is one falling off without me knowing and I actually was running too hot since I don't have a way to monitor vrm temp
|
|
#
?
Jun 6, 2014 19:08
|
|