TechSpot

Watercooling kit hardware help (CPU)

By H3llion
Jul 25, 2013
Post New Reply
  1. H3llion

    H3llion TechSpot Paladin Topic Starter Posts: 973   +99

    Ah I see, fair enough. That said my PC did not shut down when reaching 100C (not all cores were on 100) but it instantly throttled down (AI Suit 3 I guess).
     
  2. GhostRyder

    GhostRyder TS Evangelist Posts: 2,262   +541

    The chip wwill auto throttle itself after hitting its temp limit which will save it from any major damage. Although temps this high are not helping it, I would just be a little cautious until you have everything set up, but Haswell runs warm to begin with.
     
  3. H3llion

    H3llion TechSpot Paladin Topic Starter Posts: 973   +99

    I hope my chip isn't shitty. If I can get 4.5 stable with the wcling then im laughing (although 4.7 or even 4.8/9 would be fabulous).

    Could I stick the PC case in a fridge? :D As long as no condensation/icicles/frost forms then it should be fine ^^
     
  4. GhostRyder

    GhostRyder TS Evangelist Posts: 2,262   +541

    4.5 is not a stretch on the cooling system listed, that's probably going to be a snap. 4.7 will be the limit from what ive been hearing or reading, I have not seen almost any yet hit above there (ive seen one at 4.9). But you will just have to experiement with your chip, its all in the luck of the draw.

    Your chip is fine, I doubt you got a bad chip, its such a rare thing these days. If your still having issues with your new water cooling at stock, then we will see if there is an iseue.
     
  5. red1776

    red1776 Omnipotent Ruler of the Universe Posts: 5,906   +90

    intel_haswell_i74770k_ipc_overclocking_review/6
    Make what you will of this. Kyle over @ [H] came up with this stat regarding OC-ability of the avg of the shelf 4770K

    From [H] OCP Intel 4770 OC Review
    http://www.hardocp.com/article/2013/06/01/
     
  6. dividebyzero

    dividebyzero trainee n00b Posts: 4,904   +715

    The full link: http://www.hardocp.com/article/2013/06/01/intel_haswell_i74770k_ipc_overclocking_review/6
    It's actually from Asus's ROG program. They do preliminary binning as part of their QA for the boards and includes qualification (engineering) samples and initial production. Asus issue their findings as part of the press kit ( Here's the Sandy Bridge numbers. Pretty low binning numbers considering what late production run D2 silicon was/is capable of)
     
  7. red1776

    red1776 Omnipotent Ruler of the Universe Posts: 5,906   +90

    Yeah I read that, but these days I pin the info to the site that I found it on or repeated it. Seems so unlikely that those numbers are not by design for that process and that the move to the 22nm FinFET process was that problematic. I can't figure how Intel is gearing up for 10nm process when transistor density is already getting dodgy at 22nm. There are at least three scenarios or plans of action at this point and they all have a serious downside.
    I have also seen more burned CPU's in the last 18 mo than I can remember than the last 15 years. Although that is next to impossible to quantify/qualify because the ramp up of OC'ing being THE main selling point, and UEFI BIOSs making OC'ing more accessible and comfortable.
     
  8. dividebyzero

    dividebyzero trainee n00b Posts: 4,904   +715

    Intel probably aren't helping themselves by moving away from soldered heatspreaders in the OC bell curve. Add in the fact that the CPUs themselves aren't using appreciably less power than their predecessors, but the heat generation is more localised as the die package is decreasing markedly.
    22nm for Intel is obviously a learning curve, and they will bear the brunt of the issues while everyone else is a node (or more) behind in high power applications. From what I've read, Intel are bracing themselves for the yield/cost per mm^2 that EUV is going to swat everyone with. The tools simply aren't ready, and Intel is investing a LOT of cash in the area. Easy to see why Intel have taken their foot off the gas when it comes to their internal schedule.

    As for the transistor density argument...basically everyone's in the same boat - just that Intel (and possibly IBM) will be bearing the brunt of charging into that particular wall, and is obviously why Intel have standardized clock speed at 3-3.5GHz over the last few generations for desktop, and enterprise tops out at the lower figure for the most part.
     
  9. red1776

    red1776 Omnipotent Ruler of the Universe Posts: 5,906   +90

    Most definitely everyone is in the same boat...but AMD has something that intel did not have, someone else going first. It will be interesting to see how well they took notes.
    and it may not help. The proprties of electricity and electrons does not change...unless we get into QM and Schrödinger's cat type discussion. :)
     
  10. dividebyzero

    dividebyzero trainee n00b Posts: 4,904   +715

    Does AMD even enter into the discussion? Last time I heard, AMD's stake in GlobalFoundries was sitting at 0%.
    AMD's problems aren't about process tech, since they at the whims of others and regardless of what designs they come up...need to be fabbed elsewhere, but more about (1) making a decisive call in what they want to be, (2) mindshare, and (3) getting squeezed out by Qualcomm/Calxeda/ AppliedMicro at one end and Intel at the other.
    If you're pinning your AMD hopes on GloFo, I might point out they are either going to be playing follow the leader with Intel (likely), or -with the aid of magic unicorns and fairy dust- take a process lead (unlikely) and suffer the pitfalls of being the industry leader. I'd also note that GloFo's record on adhering to a timetable, and having decent yields are abysmal. Of course, AMD could hit up Samsung or UMC...or even go all-in with TSMC (now there's a spine chilling thought).

    Anyhow, we seem to be drifting off the subject of 4770K's under water. If we get any further off topic I'd expect a visit from the captain.
     
  11. red1776

    red1776 Omnipotent Ruler of the Universe Posts: 5,906   +90

    Whoa nelly!. way to micro. I am big picturing the the convergence of mores law, tran density, and overcoming the challenges of getting below the process nodes we are at now and there continuing to be tangible performance increases.

    ...where is Cranky anywho?
     
     
  12. dividebyzero

    dividebyzero trainee n00b Posts: 4,904   +715

    Keep up "mores law" references and I'm sure he'll turn up to slap your wrist for lacking spelling and capitalization
    It sounds like Moore's Law is still viable at 5nm based upon known technology (Germanium and Indium-Gallium-Arsenide transistors with finFET and the outside possibility of TFET aka your Schrodinger's Cat) - that takes in around the next decade of process tech.
    How it plays out in silicon performance is anyone's guess. It's still more theoretical than practical, although I'd note that we still appear to be lacking in software enviroments to take advantage of what we presently have ( multi-core optimization, less overhead in ISA etc.)- I'm picking that it will take the best part of that decade just to start utilising AVX-512.
     
  13. H3llion

    H3llion TechSpot Paladin Topic Starter Posts: 973   +99

    You guys, hijacking my thread!!!! :D Just one question in terms of reliability of pumps. How reliable are they? Can they go on without breaking for a year or two or is it more of a hit and miss?
     
  14. dividebyzero

    dividebyzero trainee n00b Posts: 4,904   +715

    I knew someone would notice!
    I don't think I've ever had a pump fail in under that timeframe, although I've heard/seen instances where they haven't met the usually quoted 5 year lifespan.
    Since you're going with a D5 I'll concentrate on that. To get the most out of the pump life it's a good idea to disassemble the pump once a year or so to clean it out. Depending upon how thoroughly you pre-flushed the system, how often you replace the coolant, if you use dyes, and what kind of biocide you use it's quite common to get a film of gunk between the impeller and the impeller chamber which will eventually take a toll on the pumps performance.
    Basically, remove the impeller and wipe out the chamber with a very soft lint free cloth and a little isopropyl alcohol, then rinse thoroughly with distilled water.
    The impeller is held in place by a reasonably strong magnetic field. Be very careful when removing and replacing the impeller to the impeller chamber- damaging the ceramic "ball" in the chamber is very bad news- second only to running the pump dry.
    [​IMG]

    It's also a very good idea to replace the silicon O-ring at this time also, since the original one would be compressed and may not offer the best sealing when you reassemble the pump. You can buy the O-rings individually, or in a bulk pack.
    Most D4/D5 pumps seem to last many years. You might get the occasional pump that is DOA (usually from bad handling by the postal service), but generally, the pumps are solid and usually go the distance. A decent regime of changing coolant will help, and if you're using straight distilled water and silver as a biocide it will be better still.
     
  15. H3llion

    H3llion TechSpot Paladin Topic Starter Posts: 973   +99

    Sweet, thanks that has 100% reassured me. Will be ordering either on Friday or during weekend. Had few doubts about pump failures.

    PS: Maxing GRID 2 pushes 780 windforce 3x to 65C and my 4770k to high 80s (89s even). I hope I have a half decent chip... won't know until the cooling arrives I guess :3

    I ordered my 780 through Dabs, £525 or so + tuesday morning delivery @ £535. My 460 had a DVI adapter, does Gigabyte no longer provide DVI adapters? (I am sort of derailing my own thread heh)
     
  16. dividebyzero

    dividebyzero trainee n00b Posts: 4,904   +715

    According to HardwareLUXX's review, the card ships with what seems to be the standard no-frills Nvidia package- a DVI-VGA adapter, a couple of PCI-E power adapters (2x6pin to 8-pin) and the usual worthless paper setup guide and outdated driver CD.
     
  17. H3llion

    H3llion TechSpot Paladin Topic Starter Posts: 973   +99

    Odd, I had no DVI-VGA adapters, PCI-E/Molex cables x2 and the disc+manual.
     
  18. GhostRyder

    GhostRyder TS Evangelist Posts: 2,262   +541

    Well DVI to VGA adapters, I have a bag of like 40 of those, would you like one ;)
    I had heard only what ive been seeing posted by other users, had not seen the numbers or percentages yet, did not expect them to drop so low so fast. Seems 4.7 though is still common enough.

    H3llion
    If you have a pump fail within a year, its just a bad one, ive never had one fail on the last 4 liquid machines ive made and the last few my friends have built including one of those fish tank comps, so I think your good.

    The Shrinking of the die is for Intels race to be as power conservative as possible in the mobile market. I fear that we will hit a point where the die shrinks will go to far and will harm the desktop market for overclocking and clock speeds in general in a year or so. Hopefully both companies can keep up with the tech and we wont get too much lower.
     


Add New Comment

TechSpot Members
Login or sign up for free,
it takes about 30 seconds.
You may also...


Get complete access to the TechSpot community. Join thousands of technology enthusiasts that contribute and share knowledge in our forum. Get a private inbox, upload your own photo gallery and more.