|
Ignoarints posted:Hmmm I should do that. Especially when I get my second radiator for the GPU. That's a cool computer man!!
|
# ¿ Jun 28, 2014 20:05 |
|
|
# ¿ Apr 26, 2024 18:10 |
|
I accidentally banned my account with a thread tag and I don't have access to whatever email I used to register that account 6 years ago now. Which is the same thing that happened with my first account from like 2004.
|
# ¿ Jun 28, 2014 23:51 |
|
https://www.youtube.com/watch?v=GSllkFWVhJI&t=109s
|
# ¿ Jun 29, 2014 07:10 |
|
Don Lapre posted:If they are factory installed you should really leave them be
|
# ¿ Jun 30, 2014 14:38 |
|
Nice, once the "seal" is cracked you can just pull it off anyways by hand for ultra safety. Problem was mine was simply now going to come off that way there was an absurd amount of glue. For what it's worth the razor blade way was actually not all that hair raising or difficult, but I didn't want to have to do it.
|
# ¿ Jul 1, 2014 16:44 |
|
Looks like I would be delidding that cpu if I did get it. Someone got 5.5 ghz on air from a quick headline bit I read (dont know the circumstances) after delidding
|
# ¿ Jul 1, 2014 17:52 |
|
Torabi posted:I just bought some new ram today. 2x8 gb Corsair Vengeance LP at 1600 Mhz. My old ram was 3x2 gb Corsair XMS 3 at 1333Mhz. Both the new and old one are at 1.5V. Set that first setting to 1600 mhz. I don't know why but I can't see your picture now that I hit reply, but you can set your voltage too if it requires that as well to actually run 1.5v. Then you want to check your timings in Windows see if it took care of that on its own with CPUz or similar 1gnoirents fucked around with this message at 01:17 on Jul 2, 2014 |
# ¿ Jul 2, 2014 01:14 |
|
Don Lapre posted:Just delidded my 4790k. Dropped 14c using gelid extreme between die and ihs and on ihs and heatsink. 4.8 @ 1.263 and now a 14 degree drop ... jealous. What can I get you to do to push that to 1.35 and get 5.0 ghz :P
|
# ¿ Jul 2, 2014 03:16 |
|
Torabi posted:
Sorry in the bios where it says "DRAM Frequency" Don Lapre posted:I can't do more than a few ibts at 1.4v/5ghz. I'll have to figure out other settings to adjust. Haswell is complicated. Raise vrin to 2.0, cache voltage to 1.2, uncore multiplier to 36 and if its stable its one of those values you can start backing down from (or up from, as far as uncore multiplier, with turbo boost off). Or do it properly of course and raise each slowly lol 1gnoirents fucked around with this message at 14:41 on Jul 2, 2014 |
# ¿ Jul 2, 2014 14:38 |
|
Vrin can actually go higher and is a common roadblock, but 2.0 is a good moderate validation value to work from. It does increase temperature so its good to keep it as low as it will go eventually (and general lower voltage=better of course). Cache voltage has never really seemed to affect temperature for me but it does seem to improve stability especially combined with a fixed lower uncore. And uncore itself can make a whole setup completely unstable above a certain clock speed and there is very little benefit to having it 1:1 for Haswell overclocking, I actually like to start with that fixed to factory clock speed. LLC to "high" or "extreme" ... that's about all I can remember now though. I don't think C states matter. The ROG mobo you have (I think?) has a shitload more options that I don't know about too
|
# ¿ Jul 2, 2014 15:07 |
|
Hace posted:CoreTemp would bluescreen me 99% of the time with Win8.1 Wow, CoreTemp started loving with me in the last few weeks. I've had it running just fine on Windows 8 and 8.1 since last December or so, on probably at least 3 builds. Out of the blue it never works. Everytime it starts it essentially locks up explorer.exe after making it crawl for a minute. Then I can't see content of any windows (such as the task manager), and I'm not able to reboot. Although I am able to use other programs from icons. I've reinstalled it of course but never got much further than that. It was very but I haven't had a lot of computer time either so I just sort of forgot about it.
|
# ¿ Jul 6, 2014 03:01 |
|
Alereon posted:So it turns out my local shop has a single 4790K, and its an L4 batch that seems to be a great overclocker, with several people reporting 5Ghz on air. The available evidence seems to indicate that the brand new NZXT Kraken X61 AIO watercooler is now the best low noise cooler by a measurable margin, with both the pump and fans improving performance at lower noise levels. I was thinking of waiting for either news on a possible Samsung 850 Evo or another round of discounts on the 840 Evo, but I'm sorely tempted to just pull the trigger now. Nice. The x60 was pretty awesome on "silent" if I remember so pretty cool they made something even better. I just got done putting a Kraken X40 on my 780ti (with G10). I forgot to plug in the X40 3 pin(power cable? I'm not sure what to call it) into the GPU fan connector. So for the time being I plugged into a system fan header on the motherboard and figured I'd use the USB control to just run some settings manually. I was a little confused about how this was going to work really. The connector is 3 pins that runs the pump and radiator PWM fan, so I guess the USB connector must provide the signal to ramp up or down right? Unfortunately I don't get any fan speed report back with the NZXT software or any control (I do get pump speed and it varies based on liquid temperature I guess). It seems to be running at the lowest speed possible. I'm not sure why plugging it into the GPU pcb would make any difference but I'm hoping it will. I'm not sure how the MSI factory cooler worked with a 3 pin fan header. Does the board itself limit voltage to the fans? If that's how it works, would that cause a problem with the AIO? If the power input is expecting 12v at all times and being controlled via USB, would this be a problem? The instructions for either the X40 or G10 don't really say what is going on. I also bent the poo poo out of one corner of the bracket with the screws that mounted it to the GPU. I could barely feel any resistance - some cheap rear end metal. I'm pretty sure I put a slight bend in the card itself from when I looked at it on edge but just backing off a bit made it go back. I've seen these cards go through ridiculous bending without incident but it still annoyed the poo poo out of me. Despite all this and the radiator fan barely turning over I can't break 50*C edit: I'm blind I guess, it is a 4 pin after looking at a picture. So the correct way for me to do this is X40 4 pin connector -> GPU PCB. Then the integrated USB cable into the motherboard, and then plug the radiator fan into the integrated female fan headers. I guess I can plug in the G10 fan anywhere since I'd like it to run 100%. I dont know why it isn't working or reporting any information the way its setup now though Hopefully this 20+ degree drop will give me a bit more overclocking room. Plus I put some heatsinks on 1gnoirents fucked around with this message at 03:04 on Jul 13, 2014 |
# ¿ Jul 13, 2014 02:47 |
|
Alereon posted:A bit of an NZXT trip report: That's weird they didn't include instructions since the install sounds just like the x40 install, or any asetek. Which are pretty clear about those which standoffs and holes to use. I didn't know what cam software was but after looking at it that looks pretty badass - if it worked. Maybe it needs the kraken software workaround http://openhardwaremonitor.org/downloads/ running that and enabling .net framework through "Turn windows features on or off" menu edit: sucks about the fan though. I found out the hard way that finding better replacement fans for radiators is very difficult. Well I'm assuming in this case you'd have to exchange it but that doe simply suck. 1gnoirents fucked around with this message at 22:42 on Jul 19, 2014 |
# ¿ Jul 19, 2014 22:38 |
|
Null1fy posted:I guess I'd like some insight/advice/opinions. Essentially, I've been fringing on the idea of building a water cooled PC for some time for the benefit of my PC being able to tolerate 80-90 degree rooms in the summer and for my general enthusiast building purposes. I've had my eye on the Mountain Mods H2GO case, where I'd house either one or two radiators (front and/or back) to support processor and gpu cooling; I'm thinking something along the lines of a 4690K and a GTX770 (I can't justify the extra $200 for a 780). I've read that in a similar setup people can achieve 30-35C idle, versus 50 or higher in air cooling. Pretty significant. At the same time, I'm only ever going to run two 24" widescreen monitors off of a single GPU (I haven't seen any indicator of SLI being worth investing in to date) to mostly single-screen game with the 2nd monitor available for video streaming / web browsing. My other option I've considered is a Noctua processor fan and stick with air cooling. In either setup, I plan to fit the case with some 6 Noctua NF-F12 static pressure fans in positive pressure configuration for low RPM/noise and high volume cooling. My research indicates that the pump from the water-cooling should be the loudest component when the case is properly set up, or virtually silent in total air cooling. Someone else will have to answer fully, but in my opinion if I'm choosing between a custom water loop with a 770 vs saving $400 with air (or just AIO for the CPU) and putting that towards the GPU, there is no question there. I don't think it'd be worth paying for a waterblock for a 770 unless you already had one and wanted to SLI ... maybe. Which incidentally is a very good SLI combo (towards the SLI comment) not that I'd recommend it. You wouldn't get much more out of a 770 with a custom loop water block on it. I'd have a hard time with the cost of a AIO bracket + cheap cooler, although I would on one if I were to SLI. Buy the single most badass AIO for your CPU and get a single, much more powerful card with a proven cooler with the saved money.
|
# ¿ Jul 22, 2014 04:30 |
|
Welmu posted:
Yes. I'm just looking at the pros and cons of what you get for the money in this situation. A proven air cooled 770 isn't going to struggle or be loud even if the ambient temperature is 35* though. It's hard for me to see the value in putting a water block on that. You can push it farther for sure, especially if you unlock the bios and overvolt. But say instead you put that money into a 780ti, it won't come close. Is that hugely overkill for what will probably amount to gaming on one 24" 1080p monitor? Oh yes But just comparing cost benefit etc, etc. On the other hand if you just want to do it just to do it because its fun, I wont ever think thats dumb.
|
# ¿ Jul 24, 2014 14:46 |
|
Not an answer sorry, but does your ram voltage have to be lower than VIN or CPU input voltage (or equivalent) in this case? Overall that seems high as hell to me in general but voltage on ddr2 for AMD is something I know virtually nothing about.
|
# ¿ Jul 24, 2014 16:48 |
|
I mean vrin, not vcore. I was curious of AMD worked that way. Imo if your computer is incapable of running as it sits, and its not worth buying new parts for, and there is a free option that may or may not destroy it but enable it to work for a little while longer, I know what I'd do.
|
# ¿ Jul 24, 2014 17:13 |
|
If you really want a 770 then just get a AIO for your CPU and save a tooon of money. Or use that money to get a 780ti Or like I said if you just want something fun to do and badass by all means go for it, but if you're on the fence at all...
|
# ¿ Jul 24, 2014 21:29 |
|
Midnight City posted:
If that is stable that is full on conservative overclock and safe. Sometimes factory voltages are higher than that for boost. If anything looks "weird" its that the cache voltage is probably at a moderate level you'd more likely see when vcore is higher, or cache frequency is higher, or if it was being used to stabilize. It's not high at all though and no reason to even touch it imo, especially if its there for a reason. As long as your temperatures are good but even with the factory Intel cooler I 1.2 volts was always fine for real use temperatures in my experience
|
# ¿ Jul 25, 2014 16:18 |
|
Peanut3141 posted:Toms got to 4.8GHz with a Noctua U12S and 4.4GHz on stock with temps in the mid 80s. So, an EVO may get you a little more, but I'd be shocked if you got more than 200/300 MHz. Assuming you spent $100-120 on the CPU/MB, you'd be dropping 25% more $s for ~7% more MHz. You're well into diminishing returns. He is already at 4.5, although says no overvolt (I assume it must be autovolting). $35 more would put him at $135 for - assuming all is well with the current configuration - 4.7 or so with blend if I had to guess. Diminishing returns if we're talking value for the dollar here... don't start for a while. But literally so, for sure. Being dual core and the fact it uses raw speed to overcome not being hyperthreaded I personally think its worth at least an evo. Zero if you haven't updated your bios the cache voltage and vcore are locked together for some reason (maybe if you just try to set it manually, I dont know) until you get the new version. Also what is your vcore? Do you plan on manually setting that? I'd be pleased as hell though if it were actually stable at 4.5 as it sit. If it's truly stable at 4.6 hitting 90 degrees isn't going to kill it either. I'd make sure realistic loads were below 80 for sure. Play games on it. Its hilarious
|
# ¿ Jul 30, 2014 07:32 |
|
Yes it is literally diminishing returns by definition... but there is a point where it is an impractical way to measure worth or value imo compared to similar performance with other setups. Anyways though I'd be happy at 4.5ghz on the stock hsf. Frankly 4.6 at 90*C on the stock HSF is worth buying an Evo for just to see how far you could get with what is otherwise a budget cooler
|
# ¿ Jul 30, 2014 07:42 |
|
Peanut3141 posted:If you personally feel it's a good value proposition to drive a ways into diminishing returns, that's fine; but don't try to redefine terms to justify your opinion. I will call it "overall system value" or "reality"
|
# ¿ Jul 30, 2014 08:11 |
|
Zero VGS posted:I already updated the bios to that newest Anniversary CPU specific thingy. I then just manually set that one field to "46". So it is auto overvolting then? I don't think I should be overvolting manually or I'd be breaking 90c all the time probably. It's auto volting. All that means is that you can probably reduce the temperatures a bit by finding the voltage it can run at but since it is for someone else I'd also just leave it slightly conservative with auto everything. I've never seen blend-like temperatures (at least, after 15 minutes in) for gaming anyways Haha, a 290. It's going to be stupid good for the cost of parts edit: Well... I actually hate auto voltage so I wouldn't personally do that but practically speaking it will be easier and less time consuming that way for you 1gnoirents fucked around with this message at 16:46 on Jul 30, 2014 |
# ¿ Jul 30, 2014 16:44 |
|
Avalerion posted:My PC came with a build in overclocking tool (picture below). Looks like I can just increase the CPU ratio from the default x34 all the way to x41, should I? I'm somewhat concerned as the tool is reporting a voltage of 3.~V even without adjustments, when the OP states 1.38V as the "safe" value, though I have used it without issues for years. One of those voltages is for... the CMOS battery lol. I dont even know what the others are The voltage you're looking for is vcore or VID. If its not there, download CPU-Z Looks like you're already at 4200 mhz though x42
|
# ¿ Jul 30, 2014 17:46 |
|
I did see that but I assumed you were running at 4200 because of the above box. If its changing a lot that's due to load. CPU-Z is a more standard way to view it in Windows (I do not know what erazer is). Say if you auto tuned in the past and it settled on x42 as its overclock it won't necessarily tell you that in the overclock box since thats "manual". I guess
|
# ¿ Jul 30, 2014 19:49 |
|
To answer your original question, yes thats all you basically need to do. But slowly and one by one, and monitor temperature and voltage while you stress test it (simple fast test is IBT, this will make your computer get way hotter than normal so keep that in mind) If you dont feel like messing with voltages you dont have to but you wont get the most out of it (and sometimes you seriously really totally wont) and it can introduce instability depending on how good the auto voltage works. Also it can run hotter than it needs to. But its easier just to do the multiplier
|
# ¿ Jul 30, 2014 21:33 |
|
Avalerion posted:Cheers again, looks like I can pull off x35-x36 without any issue so I'll stick to that for now. Yes, and... does that have no fan
|
# ¿ Jul 31, 2014 20:07 |
|
Oh like actually inside and vertical haha. I was about to say you have the best chip on the planet for a passive cooler that small If you're running into a thermal wall for overclocking I suggest start messing with vcore. I'd set it at 1.20 to start and see how fast you can get your multiplier before it crashes, then raise the voltage 0.01 and so on until you run out of thermal headroom. That way you get a lot more speed out of the same voltage, or "effort" would be an analogy, that the CPU is doing Also if you do so I'd recommend just doing it in the bios as well 1gnoirents fucked around with this message at 20:24 on Jul 31, 2014 |
# ¿ Jul 31, 2014 20:21 |
|
Sh4 posted:I made the switch to AMD because poorness and got a FX-4350 which is fairly disappointing, I had an Athlon X6 1090T and I'm pretty sure it was performing better even at a lower frequency but anyway. As I'm sure you found out by now, its been downhill for a while. That being said I'm assuming returning poo poo isn't an option (although I'd seriously consider outright selling and getting something else). This is my anecdote. I had a lovely PSU go bad with an AMD 3 core-> 4 core unlock that I overclocked. I could not overclock with a new PSU. I thought I might have damaged the CPU, and I did, but I also damaged 16 gb of ram (every stick) and the motherboard. Which is something I didn't know until I actually tested out each and every single computer in a separate computer which took forever. It otherwise booted and ran despite all of that being damaged in some way. After a while I got blue screens, and poor performance in general, and another aspect was being unable to successfully overclock the same CPU despite having been overclocked before the bad PSU.
|
# ¿ Aug 3, 2014 18:21 |
|
There's no problem with getting the mobo you want and throwing a pentium in there for the time being simply because the price difference is so extreme. The only issue I have is its probably a poor choice if you plan on waiting too long (or just never upgrade... another serious possibility). I don't know. I can tell you now an i5 or i7 is simply better in every way (even single threaded - cache) but is it 5 times better? 7 times better? gently caress no. Not for games anyways, and very basic use. My fear with this is you'll be so happy with the pentium's performance that you won't feel the need to change. Is that really a bad thing? No, but, it seems weird to have that motherboard then. But then is even that really a bad thing? Are the advantages of that motherboard really any different between the penitum and any k processor? This pentium brings up some weird questions for me. But anyways, what Factory Factory said. Since you have a real deal overclocking motherboard you can go further with it too. Next things to tweak after vcore is vrin and cache voltage, but I doubt that will be important before you hit stock cooler limitations. Passed that it gets complicated and usually kind of pointless so I stop at that point. Before you overclock set all your motherboard power settings to "High". I wish I could be more specific but I am unfamiliar with the ROG bios. Things like LLC especially seem to matter at higher voltages, at least for me. But start with what Factory said and you'll likely be good to go until you get a cooler.
|
# ¿ Aug 6, 2014 15:41 |
|
I only have a singular experience with LLC settings so I'll defer to you for sure. By offset though do you mean auto-offset? In that case I always set that to 0. But I got to the point where I was at my highest self imposed thermal limit along with a vcore so on the tipping point 0.01-0.02 would cause it to crash, the best solution I came to was LLC "extreme" and zero offset
|
# ¿ Aug 6, 2014 16:38 |
|
I'm a little familiar with those for things other than CPU cooling. Be warned though, cheap ones are cheap for a reason, if you're going to use them for CPU's. One thing I thought of in that past was slapping one to the side of an endtank of a double-thick style asetek water cooler, if it were routed outside the case anyway. I dunno how practical that would be, but being non-critical you could use cheap rear end ones with thermal adhesive
|
# ¿ Aug 8, 2014 14:50 |
|
It isn't a very complicated mechanism. I wouldn't really worry about another one going bad to be honest
|
# ¿ Aug 10, 2014 20:36 |
|
CheeseSpawn posted:I hear DC is a crapshoot as well. Thanks for the feedback all, I guess I'll wait for the next decent refresh and chuck this thing to my brother at the time. I heard only the ones made in 2013 are. I forgot what the code is to look for though
|
# ¿ Aug 17, 2014 01:55 |
|
r0ck0 posted:my L4 does 4.7@1.25v. It needs too much voltage to do 4.8. it's not that great of a 300mhz overclock. meh at least it runs cool. 1.25 at 4.7 is good to very good, considering the architecture. 1.25 is basically achievable in stock conditions and is no way too much voltage. Also keep in mind its a little better than 300 mhz, since turbo boost at 4.4 on all cores doesn't happen. If it's cool just run it at 1.30 and see what you can get. 1.30 is, imo, when you're just starting to get into actual overclocking voltages (and a moderate one at that). There's no reason to but considering your numbers I'd personally see how far I could go just for validation on that chip since it sounds like a decent one. edit: kind of misread, what was the voltage requirement for 4.8
|
# ¿ Aug 17, 2014 19:02 |
|
A whole .10 for 100 mhz (especially at these voltages) can mean bottlenecks elsewhere, and that is almost always vrin (but also can be instability caused by 1:1 cache ratio, or voltage for that too). If you're interested it's a thing to try if you haven't. An easy test is put it back up to 1.35 volt and change vrin to 2.00 and see if it is stable. If so, then go back to your earlier numbers then work up
|
# ¿ Aug 19, 2014 17:20 |
|
r0ck0 posted:Ya I was thinking that the problem was with another value I just wasn't sure what other value would be the one to adjust. What is default vrin and do you think if I raise the vrin I can lower the vcore? Will raising vrin also increase the temp as drastically as vcore? Thanks yo. There is a chance the chip is just unstable past 4.7 at regular temperatures, but if its another value the first one to raise is vrin. The stock value is 1.80, putting that up to 2.00 is a good middleground value to test and see if that's the cause. It will increase temperature, but nothing like vcore (per increment anyways) and if it is the bottleneck you are allowing your vcore to be lower in this case. Not to say 1.35 vcore or higher for 4.8 ghz is uncommon, but it is pretty uncommon if you can run 4.7 at 1.25 vcore. In this case I would - Manually set the cache ratio (uncore) to 36 with turbo boost off - Set vrin to 2.00 and vcore to 1.35 and see if it works - If it works, I'd start back down to 1.26 and work my way up in vcore until I find stability which would hopefully be relatively low compared to 1.35, leaving vrin at 2.00 for now - Once I'm solid on a stable vcore, I'd lower the vrin until its unstable again. I like giving myself a healthy vrin overhead though - If all of it works out and I feel like I have the thermal headroom I might try for 4.9, etc Finally I'd raise the uncore frequency up until its unstable and lock that in as well. Cache voltage can be raised here as well, 1.20 is a nice easy number for this. If 2.00 vrin and 1.35v dont work initially, I'd set the cache voltage to 1.20 and test again. If that doesn't work I'd say you've done all the reasonable easy things at that point and just sit on 4.7
|
# ¿ Aug 19, 2014 17:49 |
|
Sorry man, kind of weird/rare. However unless you want this cpu to last 20 years, I wouldn't waste any more time lowering it from 1.235 lol
|
# ¿ Aug 20, 2014 16:25 |
|
After delidding, will be no problem. Typically wouldn't recommend a h55 for much but considering the cost and the fact it is literally for fitment in this case I dont see any problem with that either
|
# ¿ Aug 20, 2014 20:50 |
|
|
# ¿ Apr 26, 2024 18:10 |
|
program666 posted:These are somewhat new notions about temperatures for me. From everything I read I was under the impression I could let temperatures a bit beyond 90oC while running prime95. It's slightly conservative, and definitely the proper way to do things. However synthetics touching 90's isn't exactly a death omen either... but what exactly is your voltage. Even for a 4670k that sounds pretty high for voltages you typically need for 4.0 ghz. I'm pretty sure that's where I was on the stock cooler. There's kind of a lot of "dont buy aftermarket thermal paste" sentiment here in general, but in the context of this thread a few degrees matters and plus your stock paste or interface in general might have a minor problem. It's probably worth repasting at those temperatures (unless you're autovolting way too high or something else).
|
# ¿ Aug 25, 2014 14:24 |