|
I'm so very lost here. I have a i7 4770K an an ASUS z87-pro that I overclocked to 4.2 by setting the cores to 42 and the voltage to ADAPTIVE (I don't want to use a fixed voltage because I want my cpu to use less power when it's idle) Right now, It's been running prime95 for quite some time and hasn't gotten any errors, nor have I gotten any bluescreens. According to HWMonitor, my temps are around 82-83. However, this is where I'm totally confused. CPU-Z says my Core Voltage is 1.248. HWMonitor is reporting VID of 1.248V and IA Offset, GT Offset, LLC/Ring Offset, and System Agent Offset are 0. Asus AI Suite 3 is saying my Core Voltage is at 1.264. Which is right? What's a safe voltage for a 4.2 ghz overclock? Would adding more voltage improve temps, or only increase them?
|
#
?
Jan 10, 2014 02:43
|
|
|
|
| # ? Jun 2, 2024 22:14 |
|
|
More voltage increases temps. Power consumption is proportional to clock rate (more Hz = more watts), proportional to the square of voltage (more volts = many more watts), and also increases with temperature (more degrees C = a bit more watts). All in all, 1.25V Vcore (I would trust CPU-Z the most) is probably more than necessary for 4.2 GHz. I would try to lower that if possible - start at 1.20V and work your way up if it's unstable. If I'm Googling how Adaptive voltage is implemented from a good source (it's [H] forums), just enter your desired voltage in the "Adaptive Turbo Boost" field. That will be the voltage your CPU tries to hit when running over stock frequencies; below stock frequencies, it will use stock voltages.
|
|
#
?
Jan 10, 2014 02:58
|
|
|
Factory Factory posted:More voltage increases temps. Power consumption is proportional to clock rate (more Hz = more watts), proportional to the square of voltage (more volts = many more watts), and also increases with temperature (more degrees C = a bit more watts). Yeah, I tried to follow your advice, but apparently asus's adaptive voltage just means "we'll set it for whatever the gently caress we want to set it to," and no matter what numbers I put in for the Additional Voltage for turbo boost, it's still running at 1.25V. Has anyone else had experience with ASUS boards and this issue? Unfortunately, manual voltage just won't work for me. Maybe I could try an offset value.
|
|
#
?
Jan 10, 2014 04:17
|
|
|
UndyingShadow posted:Yeah, I tried to follow your advice, but apparently asus's adaptive voltage just means "we'll set it for whatever the gently caress we want to set it to," and no matter what numbers I put in for the Additional Voltage for turbo boost, it's still running at 1.25V. For overclocking haswell you first do manual voltage to figure out what speeds you can get stable at what voltage, then you set the adaptive voltage to approximate that voltage at the maximum afterwards. Adaptive voltage is a huge pain in the rear end on ASUS boards because it's very counter intuitive. I have the Z87-A but I imagine the bios settings are similar. I'll check how mine is set and get back to you. I was stable at 4.3ghz at 1.22v so I attempted to get that in my adaptive voltage settings but I think it goes slightly higher (though not enough to be worrying). Okay here is how mine is set (bear in mind this is the Z87-A and yours may be different). The trick for me was to get it to have a base voltage that made sense first and then set the offset. I had to try it a few times with different offsets to get it to land in the "stable" range.  Since it's all impossible to read the white fields are: CPU Core Voltage Offset 0.185 Additional Turbo Mode Auto Total Adaptive Mode CPU Core Voltage 0.185 (this is greyed out and not editable) and the bottom ones are both Auto. Now it seems like I've set it to go up to 1.4V there but in reality it sits right at about 1.22:  e: sorry that was huge, I resized the image Rexxed fucked around with this message at 05:35 on Jan 10, 2014 |
|
#
?
Jan 10, 2014 04:29
|
|
|
I was thinking about overclocking my processor slightly more and wanted to know if you guys have any suggestions. Right now I have a i5 4670k at 40x (100MHz base) 1.15v, and a Hyper 412 pwm heat sink, temperatures are hitting 75oC when running the prime95's small FFTs test. Also that's the lowest voltage it will stay stabilized. When I tried for a 41 multiplier the processor wouldn't stabilize in such a lower voltages but would go up to above 80oC before crashing so I gave up and let it at 40x. My questions are, what would be the temperature I should start worrying about, and if buying a second fan for the Hyper 412 would allow me to go for higher multipliers. I could also buy a pre-build liquid cooling if you think it will give me much more headroom for overclocking. Just one thing to note is that at first the processor would stabilize at 1.144v and in less than two weeks I already needed to increase it to 1.15v, and I'm not sure if this says anything about how fast it will degrade my processor, just letting you know. VVVV no, I just found out what it is now, seems a bit risky if you don't have proper equipment to do this (and I don't) program666 fucked around with this message at 14:44 on Jan 10, 2014 |
|
#
?
Jan 10, 2014 13:58
|
|
|
Did you delid?
|
|
#
?
Jan 10, 2014 14:35
|
|
|
program666 posted:VVVV no, I just found out what it is now, seems a bit risky if you don't have proper equipment to do this (and I don't) You mean a $3 pack of razors? $3 bottle of alcohol? $6 thermal paste?
|
|
#
?
Jan 10, 2014 15:52
|
|
|
Yeah I've been reading on the subject and it seems tempting. Well since I'm thinking about removing that clunky metal brick that is the Hyper 412 to try some deliding do you guys think it's worth changing the heat sink for something like the Seidon 120M or maybe the corsair CW-9060010-WW? The fact that they seem much less clunkier than a huge brick made of blades is probably enough for me but lowering temperatures would be a huge plus. But I'm also thinking that maybe having to increase the voltage in less than two weeks like I said before is a huge red flag and maybe I should not overclock my processor further. VVVV Gryphon z87, but why you ask? You think it's the reason I needed to increase the voltage so soon? My specs:  program666 fucked around with this message at 16:16 on Jan 10, 2014 |
|
#
?
Jan 10, 2014 16:03
|
|
|
What's your motherboard model? The quality of your VRMs may have a part in this. e: Yep that mobo's a keeper, just checking. Sidesaddle Cavalry fucked around with this message at 16:37 on Jan 10, 2014 |
|
#
?
Jan 10, 2014 16:10
|
|
|
program666 posted:Yeah I've been reading on the subject and it seems tempting. Well since I'm thinking about removing that clunky metal brick that is the Hyper 412 to try some deliding do you guys think it's worth changing the heat sink for something like the Seidon 120M or maybe the corsair CW-9060010-WW? The fact that they seem much less clunkier than a huge brick made of blades is probably enough for me but lowering temperatures would be a huge plus. Neither of those are really much of an upgrade. If you want closed-loop liquid cooling to be better than a heatpipe tower, you need to look at 140mm models at least, like the Corsair H90 or NZXT Kraken X40. Also your motherboard is really good, so it's not anything to do with your VRMs.
|
|
#
?
Jan 10, 2014 16:24
|
|
|
Thanks for all the tips, I'll get to it later. VVVV I actually gave up for a while. My case don't allow me to have a 140mm watercooler for example and the delid alone seem to take temperatures 9 - 11 degrees down, and that means one more multiplier only probably Besides I just made huge upgrade just now and I have no apparent bottlenecks so I'll mess with processors again next time I upgrade my GPU or something breaks. program666 fucked around with this message at 19:13 on Jan 10, 2014 |
|
#
?
Jan 10, 2014 16:41
|
|
|
Shaocaholica posted:You mean a $3 pack of razors? $3 bottle of alcohol? $6 thermal paste? Find a friend with a vise (or hopefully you own one yourself because you a cool guy). It was so incredibly easy.
|
|
#
?
Jan 10, 2014 17:17
|
|
|
Rexxed posted:For overclocking haswell you first do manual voltage to figure out what speeds you can get stable at what voltage, then you set the adaptive voltage to approximate that voltage at the maximum afterwards. Adaptive voltage is a huge pain in the rear end on ASUS boards because it's very counter intuitive. I have the Z87-A but I imagine the bios settings are similar. I'll check how mine is set and get back to you. I was stable at 4.3ghz at 1.22v so I attempted to get that in my adaptive voltage settings but I think it goes slightly higher (though not enough to be worrying). This is helpful, but mine is still spiking to 1.25 when I run prime95 (it stays where I want it when I'm just doing normal heavy-duty stuff like encoding video) The only way I've found to get it under control is to set a NEGATIVE voltage offset, but that's its own can of worms)
|
|
#
?
Jan 10, 2014 19:13
|
|
|
This is the n00best of questions, but after spending 15-20 minutes digging through my BIOS, I could not find where I adjust the clock multiplier on my Gryphon Z87. I gather it's in the Advanced Settings > AI options page, but where the gently caress in there it is has me at a loss.
|
|
#
?
Jan 10, 2014 19:33
|
|
|
It's the option "CPU Core Ratio", should be one of the firsts on the AI tweaker. You can set it to "auto" and leave it at that you may select "Sync All" or something (the one I use) that makes all cores use the same multiplier. Core voltages are halfway down after scrolling a bit, you can leave "CPU Input Voltage" as "Auto" and only change "CPU Core Voltage" and "CPU Core Ratio" to overclock your processor. If you leave "CPU Core Ratio" on "Auto" the motherboard will use too much voltage so I suggest you try to find the minimum to reduce heat. program666 fucked around with this message at 20:01 on Jan 10, 2014 |
|
#
?
Jan 10, 2014 19:59
|
|
|
Thanks, I saw the CPU Core Ratio option but didn't think it did anything but determine how each core was set.
|
|
#
?
Jan 10, 2014 20:19
|
|
|
From my understanding, my AsRock Extreme 4 Z87 board has some weird overclocking things going on, like requiring unexpected setting changes and defaulting to some weird things. Anyone with experience want to let me know how to properly OC my i5?
|
|
#
?
Jan 10, 2014 21:22
|
|
|
Alright. I delidded my i5-4670k before I had ever run it, cleaned it off really well, put a half-grain-of-rice size drop of AS5 on the die and clamped the IHS down onto the MB. I then did a line of AS5 down the middle of the IHS (as per AS instructions) and affixed the Hyper 212. I am getting higher temperatures than I am expecting when pushing the cores really hard with Intel Burn Test and to a slightly lesser extent Prime95 small FFT. Prime95 blend is pretty tame. I saw upper 80s C's peak on Intel Burn Test. Right now I am only running 4.2ghz at 1.22v, which seems totally stable, but potential for upper 80C temps seems bad. I've heard most of these should see 4.4ghz on air cooling, but maybe mine isn't great? Or maybe I didn't do the thermal interface that well? Or maybe Arctic Silver actually does have a break-in period and 1 day of running isn't enough? I know the Hyper 212 isn't exactly high-end air cooling, but I thought I'd do better than that.
|
|
#
?
Jan 11, 2014 17:27
|
|
|
Did you clear the heatspreader adhesive before re-mounting the chip? If not, then your delid didn't actually fix the gap problem. Use a credit card (or like object) to gently scrape away the glue that used to hold the heatspreader.
|
|
#
?
Jan 11, 2014 19:08
|
|
|
Factory Factory posted:Did you clear the heatspreader adhesive before re-mounting the chip? If not, then your delid didn't actually fix the gap problem. Use a credit card (or like object) to gently scrape away the glue that used to hold the heatspreader. Yessir I did.
|
|
#
?
Jan 11, 2014 19:30
|
|
|
I would say it's just the 212 EVO, then. The heat density of Haswell CPUs is really high, and the 212 EVO is not made to cool such things - it transfers heat mainly through surface area of the contact plate, not by rapid transfer on a small surface area.
|
|
#
?
Jan 11, 2014 20:17
|
|
|
Factory Factory posted:I would say it's just the 212 EVO, then. The heat density of Haswell CPUs is really high, and the 212 EVO is not made to cool such things - it transfers heat mainly through surface area of the contact plate, not by rapid transfer on a small surface area. What's insane to me is how FAST the cores claim temp changes. Dropping 40-50C in a few seconds when load drops from highest to 0.
|
|
#
?
Jan 11, 2014 20:24
|
|
|
ShaneB posted:What's insane to me is how FAST the cores claim temp changes. Dropping 40-50C in a few seconds when load drops from highest to 0. The core is pretty small, you remove the source of heat and it will dissipate pretty quick. It doesn't sound like the heatsink is going to handle much more of an overclock though.
|
|
#
?
Jan 11, 2014 21:02
|
|
|
I'm lookin' for a little voltage advice if any folks more experienced than me could help. If my temps are still good (70C's), how much should I worry about giving too much? Currently my setup is a 2500k and Gigabyte z77-d3h at 4.6ghz and 1.42 vcore using offset. It held at 4.7ghz around 1.47v, and I think it would do 4.8ghz around 1.5v. (Prime95 blend for up to 4 hours has been my go-to.) I know the OP says 1.38v for a 2500k, but I see people talking about hanging around 1.5v for higher clocks. I haven't found solid enough looking info on if running around 1.5v would mean the chip would die in 2 years, 8 years, or if it's a matter of finger crossing on how it takes it.
|
|
#
?
Jan 12, 2014 00:30
|
|
|
ShaneB posted:Alright. I delidded my i5-4670k before I had ever run it, cleaned it off really well, put a half-grain-of-rice size drop of AS5 on the die and clamped the IHS down onto the MB. I then did a line of AS5 down the middle of the IHS (as per AS instructions) and affixed the Hyper 212. I am getting higher temperatures than I am expecting when pushing the cores really hard with Intel Burn Test and to a slightly lesser extent Prime95 small FFT. Prime95 blend is pretty tame. I saw upper 80s C's peak on Intel Burn Test. At 1.22V and 4.25ghz my lidded i5-4670K gets to low to mid 90C range almost instantly with IBT with the Hyper 212. The thing to remember is that IBT is a good way to stress the cpu but it puts an unrealistic amount of stress on it. Even stress testing with Prime 95 brings them temps way higher than what I see in day to day use. IBT is a good way to make sure you've got a temp and multiplier that's not going to ever cause your chip to hit its thermal limit and also make sure your overclock is stable. Then once you set your adaptive voltage the machine will regain the capability to overheat because IBT allows it to push even more voltage in if you're using Adaptive at all. So, get your overclock stable and do your IBT testing and whatnot, then tweak things. Your temps delidded are lower than mine so it doesn't seem that bad to me. We were just unlucky to not get CPUs that magically overclock way better than others, which is random.
|
|
#
?
Jan 12, 2014 00:53
|
|
|
teh_Broseph posted:I'm lookin' for a little voltage advice if any folks more experienced than me could help. If my temps are still good (70C's), how much should I worry about giving too much? I would not run above 1.38V except for a suicide run. We're largely talking finger-crossing, but I would estimate that 3 to 4 years of 100% load time at 1.42V would kill it. It might die faster if things degrade the wrong way. If it's a weekend warrior machine that sees full load only a few hours a week, or you fully intend to replace the machine soon enough that longevity doesn't matter to you, then maybe a higher voltage would be acceptable to you. But I just can't recommend it.
|
|
#
?
Jan 12, 2014 01:23
|
|
|
If I wanted a watercooler for my i5 that was quieter/more effective than my Hyper 212 evo or whatever, what would be my best option(s)? I run a Fractal Midi R2.
|
|
#
?
Jan 20, 2014 01:33
|
|
|
I have a Fractal Arc Mini. Depends on how much you want to spend, but either a Corsair H100i or H110 would be easy to install and be a significant improvement in terms of effectiveness--your case can fit any kit with a 240mm-long radiator. If you manage to get your hands on a Switftech H220 I'm told those are even better. Quietness is also a thing to invest in. Usually aftermarket fans by Noctua tend to do nicely--they're occasionally called "tan fans" for their color though, if you're trying to color-coordinate for  . Should be the NF-F12 or -P12 model you're looking for for the correct size.e: Idly, I'm imagining that a 280mm-long radiator would be comparatively quieter with larger fans on it. I wonder if I can make enough space for one myself if I remove my optical bay?
|
|
#
?
Jan 20, 2014 18:03
|
|
|
The Swiftech H220 is better-than-280mm performance in a 240mm package because it uses copper radiator fins rather than aluminum. You can import it via NCIX, or Cooler Master's Eisberg 240L is essentially a rebadge. Probably what I'd go for on an mATX build rather than case modding.
|
|
#
?
Jan 20, 2014 20:19
|
|
|
Factory Factory posted:The Swiftech H220 is better-than-280mm performance in a 240mm package because it uses copper radiator fins rather than aluminum. You can import it via NCIX, or Cooler Master's Eisberg 240L is essentially a rebadge. Probably what I'd go for on an mATX build rather than case modding. After reading a lot on the subject, I feel like the Phanteks PH-TC14PE is what I wish I would have got out of the gate instead of this 212... I'll probably just live with it I guess?
|
|
#
?
Jan 20, 2014 20:24
|
|
|
Factory Factory posted:Cooler Master's Eisberg 240L is essentially a rebadge. Minor correction, but the Eisberg isn't the H220 rebadge - the Glacer 240L is. The Eisberg is an Alphacool product, and reviews indicates that pump noise is a pretty major downside, so the Glacer should be the better option (and cheaper too).
|
|
#
?
Jan 20, 2014 20:39
|
|
|
You could upgrade to it and sell the Hyper maybe. Some Goon would probably snap it up for close to retail (shipped).
|
|
#
?
Jan 20, 2014 20:39
|
|
|
Factory Factory posted:You could upgrade to it and sell the Hyper maybe. Some Goon would probably snap it up for close to retail (shipped). Would I sell that in Buy/Sell or is it kosher to post it in the parts building thread or something as a "hey new builders just get at me"
|
|
#
?
Jan 20, 2014 20:41
|
|
|
ShaneB posted:Would I sell that in Buy/Sell or is it kosher to post it in the parts building thread or something as a "hey new builders just get at me" Post an SA-Mart (buy/sell) thread, then post a "Hey folks, check my thread!" in the parts-picking thread.
|
|
#
?
Jan 20, 2014 20:51
|
|
|
GokieKS posted:Minor correction, but the Eisberg isn't the H220 rebadge - the Glacer 240L is. The Eisberg is an Alphacool product, and reviews indicates that pump noise is a pretty major downside, so the Glacer should be the better option (and cheaper too). Whoops, thanks for that.
|
|
#
?
Jan 20, 2014 20:57
|
|
|
What should I do with a H110 with a bum bottom plate? It was lapped ever so slightly off level and after a few days of burn in temps will gain about 15-20C from what I assume is the TIM migrating. Still works fine at stock speeds but whats the point. Anyway I've already replaced it but I've got this H110 with a good rad and pump just a bum plate. Anyone want it for parts?
|
|
#
?
Jan 21, 2014 04:05
|
|
|
What kind of benefit does push pull on a 120mm rad AIO have? On an h60 specifically. Would there be any point?
|
|
#
?
Jan 23, 2014 03:57
|
|
|
Push-pull is worth an up-to-5 C temperature difference, best case. This generally holds regardless of radiator type (including air towers).
|
|
#
?
Jan 23, 2014 04:41
|
|
|
Are there off-the-shelf closed-loop GPU coolers that are worth looking at? I'm not sure how different card specs impact the purchasing and affixing of the waterblocks to cards, etc. I haven't been able to easily find information on this. I'm currently debating the Kraken G10 adapter plate and getting a moderately priced compatible cooler. Seems like a solid way to go. The lack of VRAM cooling is concerning, though. ShaneB fucked around with this message at 17:54 on Jan 23, 2014 |
|
#
?
Jan 23, 2014 17:12
|
|
|
|
| # ? Jun 2, 2024 22:14 |
|
|
ShaneB posted:Are there off-the-shelf closed-loop GPU coolers that are worth looking at? I'm not sure how different card specs impact the purchasing and affixing of the waterblocks to cards, etc. I haven't been able to easily find information on this. Depends how off-the-shelf you mean. The Arctic Accelero Hybrid is the only full kit solution I know of, but it still requires all the effort of a Kraken G10 to install. In practice, they mostly vary in aesthetics. As for card details affecting these devices, they are overwhelmingly built to handle reference-model cards. Non-reference cards may still work, especially if nothing sticks up too far off the PCB, but everything kind of assumes that the GPU will be placed front-of-center on the card with particular mounting holes for the heatsink around it. If you have a weird card, you might have to resort to zip-ties. As for lack of VRAM cooling, I can confirm personally that it's not a big deal. I did The Mod with a G10-like bracket on my GeForce 680, and the VRAM is perfectly fine (verified by IR thermometer). Good case airflow and the rear heatsink together keep things plenty cool. It's not like they were terribly efficiently cooled in the stock setup - thick thermal tape interfacing to a flat metal plate.
|
|
#
?
Jan 23, 2014 18:57
|
|