EPTComputer Build Resource Thread - Page 1315
| Forum Index > Tech Support |
When using this resource, please read FragKrag's opening post. The Tech Support forum regulars have helped create countless of desktop systems without any compensation. The least you can do is provide all of the information required for them to help you properly. | ||
Womwomwom
5930 Posts
| ||
|
Belial88
United States5217 Posts
On December 21 2012 23:16 Medrea wrote: 1.3v is pretty aggressive on an IB. You can get to 4.4ghz at around 1.05 volts or a bit higher depending on your luck. If you are looking for something like 1.4v for 4.8 or higher then you dont really know Ivy Bridge very well. 4.8 ghz can be done for like 1.2v. I would qualify 1.3v as "fire and forget" voltage to be bad and maybe even dangerous advice. Also consider you are adding about 6-10 watts per .05 volts. And the fact that you recommend upping the voltage if it doesnt work? I don't like that at all. Thats too cavalier. Better to do something like 4.6ghz at 1.1v Ivy Bridge gets temperamental past 4.6ghz anyway. 1.3v is not at all agressive on IB. It's well under the recommended 1.4v max voltage limit for 24/7 overclock on standard air cooling. Masterbreti is getting a closed loop, it's not water cooling but it'll compete with higher end air. People generally push past the max recommended voltage on closed loops and even higher end air. Even the most conservative 'max 24/7 safe voltage' of 1.35v is way above 1.3v. People say 1.35-1.4v max as the max safe 24/7 overclock voltage because it's still very conservative, meaning you can push right up to 1.35-1.4v and have no adverse affects. With an Antec 620 he should have plenty of cooling power to handle 1.3v. Yes, IB can overclock far on ~stock voltage, but why would you stick to ~stock voltage with an oveclock? 1.3v is NOT a bad or dangerous advice on voltage for an i5. I think your being alarmist and frankly it doesnt sound like you do much benching or overclocking to know exactly. I don't know why you'd caution that 1.3v is somehow high when 1.4v is listed as a very conservative max 24/7 voltage. You could push further than 1.4v and be perfectly fine. https://www.google.com/search?q=i5 3570k max recommended voltage&oq=i5 3570k max recommended voltage&aqs=chrome.0.57.5378&sugexp=chrome,mod=17&sourceid=chrome&ie=UTF-8 You would be hard pressed to reach 4.8ghz on 1.2v on an ib-i5. In fact no one at OCN has ever reached 4.8ghz stable on only 1.2v: http://www.overclock.net/t/1247869/official-the-ivy-bridge-stable-suicide-club-guides-voltages-temps-bios-templates-inc-spreadsheet/0_100 And 400 mhz is a huge difference, and will be noticeable. Good luck hitting even 4.5 on 1.1v, no one at OCN has ever reached that high an overclock on the 3570k on that low voltage either. IB i5 standard overclock is 4.5-5ghz, on voltages around 1.2 (for an absolute golden chip, only a single person has gotten 24 stable on 4.5@1.2v, and no one could push further than 4.5 on 1.2v or lower) to 1.48 (which is slightly above max recommended voltage of 1.4v, but should still be perfectly fine and safe on high end cooling). i5-3570k overclocks great, and doesnt require much voltage at all, but it does run hot, and it definitely appreciates more voltage. It's almost guaranteed to reach at least 4.5ghz, but your going to need around 1.3v on average for that. If you dont want to be anal about it all, just tuning 4.8ghz@1.3v should be a little above average 24 prime95 max ram tested stable, but should definitely be doable on most i5's if all your looking for is just gaming stable, and is well within max volt limits. The real max limit on voltage of an i5 is likely more around 1. Conclusion for air and watercooled IB PC's : look for the sweetspot and maybe it's not a bad idea to keep the voltages around 1.3ish Vcore max. But that's for 24/7 setups. Read more: http://www.madshrimps.be/articles/article/1000289/Intel-Ivy-Bridge-i5-3570K-i7-3770K-Review/9#ixzz2FlQFmWqz In fact intel recommends 1.55v as the safe max voltage. That's a HUGE range above 1.3v | ||
|
Myrmidon
United States9452 Posts
And 1.55V is clearly not the max safe voltage, no matter what kind of magical unicorn fairy tale figures anybody misinterpreted. That's even dumber than thinking SB i5 typical overclocks reach 150W CPU power draw easily. edit: 4.6 GHz on 1.1V is unrealistic for stability for most all samples. | ||
|
Womwomwom
5930 Posts
The last time they specified a max vcore was with Lynnfield and that was 1.5V. I have a hard time believing a smaller process with finfets is going to have a much higher maximum voltage. I understand you can't compare these things with different architectures but even AMD, who doesn't care about power draw with their high end products, specify 1.55V (iirc) as the max voltage of their horrific Bulldozer processors. 1.55V really? Can you post where Intel has specified this in their documentation? | ||
|
Myrmidon
United States9452 Posts
Anecdotes from people who haven't had the chips more than several months (the same kind of people that pushed SB too far and had to dial down later, and Lynnfield too far and had to dial down later, and Bloomfield too far and had to dial down later...) don't mean much. Most "enthusiasts" are pretty clueless overall and furthermore highly subject to wishful thinking and keeping up with the Joneses. Even if some setting turns out to be okay for long-term use, there's not too much justification that can be made now to say for certain it's safe, given what's possible to know at this point. Nobody knows for sure, especially not people outside Intel's labs (and what do they think? see link in previous post). At best, it's a good shot in the dark. | ||
|
Womwomwom
5930 Posts
But I'm still curious where exactly that 1.55v is mentioned. Logically, it is an absurd amount for an Intel processor and I cannot see any official Intel document state such a figure. | ||
|
Medrea
10003 Posts
On December 22 2012 16:37 Belial88 wrote: Good luck hitting even 4.5 on 1.1v, no one at OCN has ever reached that high an overclock on the 3570k on that low voltage either. ![[image loading]](https://dl.dropbox.com/u/34863088/vcore.jpg) Also this was what I was sourcing. http://www.anandtech.com/show/5763/undervolting-and-overclocking-on-ivy-bridge Those wishing for 4.8GHz at 1.4 volts (similar to Sandy Bridge) will run into a lot of issues if they think that 1.4 volts is appropriate for Ivy Bridge. In comparison, you may end up with something more reasonable like 4.6GHz at 1.1 volts, or 4.8GHz at 1.2 volts (as per some boards I have tested). Then it will be a case of deciding whether the small IPC gains that Ivy brings will be worth 200 MHz less on your CPU compared to Sandy Bridge. Im sure its just good values coming from a cherry picked review piece. But my sentiment remains the same. Overvolting is special to overclocking. Its paramount to find the lowest voltage you possibly can while remaining stable. And IB has issues once we get to 4.5 or 4.6 since it seems that the thermals start running away. Also 1.55v is really super high lol. Thats crazy. We generally want parts to last years. Those really high voltages might be stable at birth, but after some time there will be problems and youll be forced to clock it back down, even below where you could have been if you didnt try to attack the curve. Just dont go above 4.5ghz. IB's problems really start to ramp up there. Either way, on TL we dont recommend upping voltage to anyone. Because the next thing that happens is we get a thread where someone confused vcore with voltage stepping and ended up pushing voltage past 2.0v, ripping the chip apart, and then it kinda looks bad on us as well. If someone has to ask us for advice on voltage, they aren't qualified to actually do it anyway. | ||
|
Medrea
10003 Posts
I don't want to talk about it. Delidding an Ivy Bridge is like a necessity unfortunately if you want to keep your sanity at high clock speeds it seems. | ||
|
BamBam
745 Posts
| ||
|
Alryk
United States2718 Posts
On December 23 2012 01:48 Energizer wrote: Ey folks, I was wondering if anyone wouldn't mind sending me a few GPU recommendations, specifically those that would fit in a PCIe 2.0 slot as I have not yet upgraded my motherboard to have PCIe 3.0, and it will be awhile before I end up moving in that direction (I want to see the benchmark tests on the lastest AMD cpus before I make a decision on what motherboard/cpu to buy). Budget is ~$150, really dont want to spend much more than that as I expect ill be switching motherboards soon anyways and it would be a waste. Besides, anything would be an improvement over my HD 4830. Anything will fit in Pci.e 2.0, It's still the same slot, just with less bandwidth. It doesn't make a difference though, because iirc nothing uses up PCI 3.0 anyways (if even 2.0?) At 150$, get a 650Ti or a 7850 I think. 650 Ti being a bit lower than 150 at some times, 7850 being a bit higher. And when you switch motherboards, you can still reuse the GPU, just so you know. Every GPU out right now pretty much is PCI 3.0 compatible, and everything is backwards compatible anyway, so buying the GPU now can still be reused later. | ||
|
BamBam
745 Posts
On December 23 2012 02:22 Alryk wrote: Anything will fit in Pci.e 2.0, It's still the same slot, just with less bandwidth. It doesn't make a difference though, because iirc nothing uses up PCI 3.0 anyways (if even 2.0?) At 150$, get a 650Ti or a 7850 I think. 650 Ti being a bit lower than 150 at some times, 7850 being a bit higher. And when you switch motherboards, you can still reuse the GPU, just so you know. Every GPU out right now pretty much is PCI 3.0 compatible, and everything is backwards compatible anyway, so buying the GPU now can still be reused later. Ah, my understanding was that 3.0 wasn't backwards compatible, or if it was there would have been a performance hit. Thanks! | ||
|
Medrea
10003 Posts
On December 23 2012 08:32 Energizer wrote: Ah, my understanding was that 3.0 wasn't backwards compatible, or if it was there would have been a performance hit. Thanks! PCI-e 3.0 is fully backwards compatible with even 1.0, 1.1 and 1.2 devices as well as 2.1 The only compatibility issue I am aware with is if you have a 2.1 device and a 1.0 socket. Then sometimes it wont work. | ||
|
HaruHaru
United States988 Posts
![[image loading]](http://i.imgur.com/REvpc.png) I switched to a ram with a lower voltage and a non msi mobo and a more reliable psu. | ||
|
Medrea
10003 Posts
Also Case? I guess its covered? | ||
|
Myrmidon
United States9452 Posts
What was this computer for again? (CPU too strong relative to GPU?) Were you even planning to overclock? What's that site? As mentioned above, were you getting an aftermarket cooler, case, or optical drive? $50 for Corsair CX isn't really that good, by the way. It's not a bad pick, but at usual pricing I'd suggest something else. | ||
|
Belial88
United States5217 Posts
On December 22 2012 16:47 Myrmidon wrote: First of all, nobody's had any processors on Intel's 22nm process for many years yet, much less overclocked that long. And 1.55V is clearly not the max safe voltage, no matter what kind of magical unicorn fairy tale figures anybody misinterpreted. That's even dumber than thinking SB i5 typical overclocks reach 150W CPU power draw easily. edit: 4.6 GHz on 1.1V is unrealistic for stability for most all samples. http://www.intel.com/content/www/us/en/processors/core/3rd-gen-core-desktop-vol-1-datasheet.html Ah yes, sorry. It's 1.52. Please, there's no need for you to be so damn condescending. I have provided plenty of sources for everything I've said. The main argument here was someone saying that 1.3v is somehow too high for the i5, which is absolutely ridiculous. Most people say 1.35-1.4 is the max voltage for a 24/7 overclock on i5 but Intel has specified 1.52v in their data sheets. And, i5's can draw over 150w when overclocked. I've provided plenty of sources of that. The whole point of that discussion was how terrible the frozencpu benchmarks are, and even if CPUs do not draw more than 150w, a thermal plate that is inches wide does not accurately reflect CPU heat since CPUs primarily give heat off only on a very, very small, millimeters wide section in the middle of the IHS, not the entire IHS (hence why a tiny dab of paste not even covering the entire heatsink will perform just as well as glooping the whole thing, if not better in most cases). The max recommended voltages advised by Intel and AMD tend to be for long term, ie what is the max voltage you can run long term without inflicting damage. There's obviously a temperature limit and a shutdown and throttle temp wired into the CPU so there's no way you could just accidentally fry out the CPU, you'd have to intentionally fry it out. | ||
|
Belial88
United States5217 Posts
On December 23 2012 00:42 Medrea wrote: Also this was what I was sourcing. http://www.anandtech.com/show/5763/undervolting-and-overclocking-on-ivy-bridge Im sure its just good values coming from a cherry picked review piece. But my sentiment remains the same. Overvolting is special to overclocking. Its paramount to find the lowest voltage you possibly can while remaining stable. And IB has issues once we get to 4.5 or 4.6 since it seems that the thermals start running away. Also 1.55v is really super high lol. Thats crazy. We generally want parts to last years. Those really high voltages might be stable at birth, but after some time there will be problems and youll be forced to clock it back down, even below where you could have been if you didnt try to attack the curve. Just dont go above 4.5ghz. IB's problems really start to ramp up there. Either way, on TL we dont recommend upping voltage to anyone. Because the next thing that happens is we get a thread where someone confused vcore with voltage stepping and ended up pushing voltage past 2.0v, ripping the chip apart, and then it kinda looks bad on us as well. If someone has to ask us for advice on voltage, they aren't qualified to actually do it anyway. You can do anything if your not going to test it's stability. There's a world of difference in stable, meaning 24 hour prime95 custom blend max ram used test, and just being able to boot. IB does not have issues, it's simply a smaller chip. It's absolutely absurd that everyone keeps saying IB is 'hotter' or worse in overclocking than SB, it's utterly insane. IB is a cooler chip than the SB-i5, it's just on a smaller die so it gets hotter. As long as you have adequate cooling, and that guy's Antec Kuhler 620 (maybe with some pk3 or gelid) is definitely adequate cooling, you'll be fine. And, a chip can run much cooler if it doesn't require as much voltage, ie it's a golden chip. If the guy is lucky, he won't need more than 1.3 volts and can push a high end overclock, and thus wont need more than low end air cooling to get by. It's not that i5 has 'issues' at 4.5, it's that 4.5+ tends to be the limit of the chip on ambient cooling. Pushing an i5 to 4.5-5ghz up to 1.35volts is not going to do any damage to an i5 (obviously, cant say for certain since the chip hasn't been out for 10 years, but that can be said within reasonable expectations). You only get into damage when you start getting into benching, even 5ghz on 1.45v would most likely be just fine, it's just higher than the conservative 1.35-1.4 most people say just to be both on the safe side and because higher voltage than that on ambient cooling generally won't do anything fo ryou. I don't know how you can presume to speak for TL, but if you are overclocking, you are probably going to want to increase voltage. The voltage specified by the manufacturer is just a voltage set, if you want to overclock you should definitely give a small voltage boost. No one is going to get confused with vcore and voltage stepping, or push as far as 2v, your being alarmist. No motherboard is going to allow you to set 2.0vcore, it'll stop you well before then. Also, voltage supplied through a motherboard won't fry out a chip. What you have is voltages that are so high that will cause damage if run long term, like rougher water through a pipe, but won't do any damage if used just in short term, OR the motherboard will blow out (if it's low quality). And that's only on load too and maxing it out, not in general usage. Not to mention the parts he has will have overcurrent protections in place that'll simply stop too much voltage from being fed to the CPU or VRM. | ||
|
Belial88
United States5217 Posts
On December 23 2012 14:15 HaruHaru wrote: Hi all, back again. This is what I have switched the parts to: I switched to a ram with a lower voltage and a non msi mobo and a more reliable psu. cx430 for $49 is not a great deal. It is $25 on newegg right now, which is a great deal, and you can find the xfx pro550 for $39 right now which is also a great deal (id only get it if it was cheaper than the cx430 or you use SLI/crossfire or want to in the future). How much HDD space do you use? If you dont use more than 100GB, than you could just run off an SSD which is a reallyl, really big increase in performance. Instantly load up games, levels, booting up, restarting, etc, as opposed to waiting any number of seconds like you normally have to. With the build your going for I'd recommend at least looking into a low level ssd, if to just load games and the OS onto. I mean for $70 there's a couple 64-128gb SSDs out there. I really dislike the 650. You can find the GTX 460 for $60-80 (might have to find it refurb/used since it's slightly older). It'll outperform the 650 and has all the same features (or at the very least, all the same practical features). GPU companies are assholes and churn out rebrands of the same GPUs and label them as new generations and sell them for more money even though they suck. Case in the point, the 550, the 650, the 610, etc. All expensive GPUs that perform worse than the 460. You should really just try to find a nice 7950 or 660ti for your build, if you can afford to. The 650 is out of place. If your strapped for money, i'd recommend cutting a few corners, like go with a cheaper HDD and PSU and motherboard, for a way better GPU. If you can find one, I'd recommend you buy a P67 chipset motherboard instead of z67/77. You can find top of the line P67's for half the price of a shit z68/z77, like an asus sabertooth for almost half the price of the motherboard you chose. There aren't really any useful features of Z68/77 vs the p67, and most of the features that exist on z68/77 natively, exist on most p67's - usb 3.0, sata 6gb/s, pcie 3.0 (all of which aren't utilized by anything today, or if they are, they aren't even noticeably better, and are niche items, like they aren't cpu/gpu performance, but transferring large files on a usb drive, and not even noticeably better, and still exist on p67 anyways), lucid virtu is useless, quick sync isnt used by anything and you got a gpu, srt is good but not necessary if you use either an SSD OR HDD only and is a software implementation i believe... think that's it. Integrated graphics is useless, you use a gpu. | ||
|
Womwomwom
5930 Posts
If you are getting the values from the VID table, Myrmidon is right to be condescending, if he actually is. They even explain what the VID table is for and how its completely unrelated to safe operating voltages and overclocking. And everyone has really explained why the i5 3570k doesn't really draw 150W by itself when overclocked. Do we need to explain again why all of those links your posted of so-called 150W+ power draw doesn't actually suggest that the processor itself is drawing 150W? | ||
|
Belial88
United States5217 Posts
http://www.geeks.com/details.asp?invtid=P8P67-LE-PB-R&cat=MBB $49 Similar quality motherboard, almost 1/3rd the price. Any particular reason you need 8gb of ram instead of 4? I know it's only 20 in price difference, but that $20 in price difference means a MUCH cooler looking case, 1-2 very high quality fans that are ultraquiet (throw out your shitty stock fans), a better PSU (hey, now you have the option for SLI if you ever wanted to). Hell you could just get better quality RAM like the Corsair Vengeance. Just pull up task manager and see if you ever go above 3.5gb. im assuming haruharu has a case and heatsink already picked out. | ||
| ||
