***Intel i7 4790K Owners thread***

[OhEsEcks]

my i7 920 is still running 1.4v to this day!..

45nm is an entirely different scenario than 32nm or 22nm. As die size shrinks, even minimal increases in voltage causes vastly greater heat generation. This is why you could safely put 1.5v or more through a Q6600 on air.

I'm pretty sure there was a thread on here when 32nm chips first came out warning people not to put more than 1.35v max through them because people where putting what they thought was safe from 45nm and 65nm through the chips and killing them. Edit: The thread is at the top of this sub forum actually.

However, if you're running a custom water loop, you can generally put much more through a chip because the loop does a better job of dissipating heat away from the chip.

 

[Deehjay]

45nm is an entirely different scenario than 32nm or 22nm. As die size shrinks, even minimal increases in voltage causes vastly greater heat generation. This is why you could safely put 1.5v or more through a Q6600 on air.

I'm pretty sure there was a thread on here when 32nm chips first came out warning people not to put more than 1.35v max through them because people where putting what they thought was safe from 45nm and 65nm through the chips and killing them. Edit: The thread is at the top of this sub forum actually.

However, if you're running a custom water loop, you can generally put much more through a chip because the loop does a better job of dissipating heat away from the chip.

So what would you personally recommend with a custom loop with a single 360 rad for 24/7 use? If it matters I also have a D5 Vario.

 

[stulid]

Is it worth flashing my UD3H-BK to the latest BIOS, or is it not worth the risk?

http://i.imgur.com/BhSBV0I.png these are the changes with each update. I'm thinking it will be worth it just for the proposed enhanced OC capacity.

Also, if I were to use my aftermarket air cooler first to test that the chip actually works properly before I get my custom loop set up, should I replace the thermal paste assuming that it's going to be on there for around 30 minutes to an hour?

I would update the BIOS, the really early ones used too much vcore.

 

[momo56]

I ran my i7 920 at 1.45V and 1.6V QPI for over 6 months.

That definitely did some damage to the chip. It used to be able to do 4.2 ghz but now it can barely do 3.8 at 1.4V. Apart from the reduced overclocking ability it still works at lower clocks.

Temperatures would normally be around 70 under load so I doubt that was the issue.

Anyway using a 4790K now. Just delidded and directly cooling the core. Pretty happy so far. In Aida 64 and BF4 the temp stays under 55-60 without the fans spinning.

 

[OhEsEcks]

So what would you personally recommend with a custom loop with a single 360 rad for 24/7 use? If it matters I also have a D5 Vario.

With a custom loop I'd be perfectly happy putting 1.35 - 1.4v through a 22nm chip. But that's about as high as I'd go. I got 4.8GHz on mine with just 1.25v on air. The higher the clock with the least amount of voltage is always the best option. But as I said before, the smaller the die, the more sensitive they are to increases in voltage. So a mild bump in voltage that might have had next to no effect on a 65nm chip, will have a much more pronounced effect on a 32 or 22nm chip in terms of heat. I don't switch my machine off either. So there's no heating up and going cold of the chip. Think of it like a filament light bulb. They always fail because they're being turned on and off all the time. But there's one in the US that has been burning for over 110 years now as it's rarely been turned off.

 

[Deehjay]

Just installed it today and did a fresh install of W8.1
When the clock speed goes up to the stock of 4ghz CPU-Z and the BIOS itself reads the core voltage as 1.4v, isn't this a bit high? And is the issue that stulid was talking about in the previous post with the early BIOS versions of the UD3H-BK using too much of it? If so, I'll get it updated ASAP!

 

[Deehjay]

Yep, all cleared up by simply updating the BIOS. The 1.4v on stock cooler was pushing the CPU to 100 degrees (it didn't sit there for long) and the temperatures were flaring all over the place, changing by amounts in the 10s every second. That's very dangerous... what are Gigabyte thinking not shipping all of these with the latest version?!

 

[McGraw]

Yep, all cleared up by simply updating the BIOS. The 1.4v on stock cooler was pushing the CPU to 100 degrees (it didn't sit there for long) and the temperatures were flaring all over the place, changing by amounts in the 10s every second. That's very dangerous... what are Gigabyte thinking not shipping all of these with the latest version?!

Happened to me...1.41v on the shipped bios. If I had bought my rig with a view to run at stock and learn overclocking later then I could have degraded my cpu quite a lot.

I would have been knocking on gigabyte's door for money for a new cpu.

 

[kimandsally]

Happened to me...1.41v on the shipped bios. If I had bought my rig with a view to run at stock and learn overclocking later then I could have degraded my cpu quite a lot.

I would have been knocking on gigabyte's door for money for a new cpu.

1.4v that's nothing my Gigabyte SOC Force was 1.5v until the BIOS was flashed shame on them they know the Intel specs it should not be down to the customer to even look at the voltage let alone flash the BIOS.

I have many friends who would have just run it and not looked and definitely not flashed the BIOS.

 

[Dave2150]

1.4v that's nothing my Gigabyte SOC Force was 1.5v until the BIOS was flashed shame on them they know the Intel specs it should not be down to the customer to even look at the voltage let alone flash the BIOS.

I have many friends who would have just run it and not looked and definitely not flashed the BIOS.

Those friends shouldn't be building their own PC's, if they are unable to do it properly (reading the manual, updating Bios etc). It's not exactly rocket science to read a manual, or to simply know to update the bios on a new motherboard for maximum performance/bug fixes.

 

[kimandsally]

Those friends shouldn't be building their own PC's, if they are unable to do it properly (reading the manual, updating Bios etc). It's not exactly rocket science to read a manual, or to simply know to update the bios on a new motherboard for maximum performance/bug fixes.

I was using that as an example, but nevertheless the manufacturers know the voltage specs so the ball is in their court.

Before you buy it as I did you research it and make adjustments to enhance the performance to suit what we require of the hardware, but the starting point should not be overvolting the CPU more than you would do it on a massive overclock.
I have decent water cooling but I can tell you I can't hold the temps down with the 1.5v that was on my BIOS as standard.

 

[setter]

Gigabyte boards in general seem to have very high voltage on early bios versions, Z97 boards in particular. No idea on other brands. My Asus is an older Z87, though vcore on bios 1504 for DC was high, but nowhere near as high as the gigabyte ones.

 

[PaulyD]

Got a 4790K on an Asus Z97 Pro Gamer mobo with 16GB (2x8GB Corsair Vengence Pro 2400mhz (11-13-13-31 1T).

CPU did 4.6ghz core, 4.4ghz uncore, 2400Mhz RAM @ 11-13-13-31 1T @ 1.19v load prime stable. 4.7ghz was @ 1.23v

Got 4.8ghz but with uncore at 4.6ghz stable, kept same RAM timings, finally got it stable (for me).

Prime custom (14GB RAM Usage):

Spoiler

Ver 26.6 so no AVX usage, always found this the killer to get stable (XTU, realbench etc pass but this fails)

Will round up with some final testing but happy with this at the voltage.

Need some help tweaking my memory down next, as won't be pushing any further with this chip, anyone help on that? I can post 1st, 2nd, 3rd timings if need be, majority on auto/def

 

[setter]

Nice one, looks like a pretty decent chip. AVX is indeed a real killer on theese, even at stock speed they hit crazy high temps unless you scalp it.

 

[PaulyD]

Cheers mate, might see you on the battlefield soon, been meaning to play some BF soon!

From what I read its a good clock at decent voltage. 4.4ghz uncore was so much easier but wanted higher to match the core

Wanted to get a 24/7 clock. I know 4.6ghz @ 1.19v is probably better (and cooler) but something about 4.8ghz makes me grin

 

[setter]

Yep, look forward to a few games again. Clocking wise, i havent really tried anything more on this one. Straight in at 4.7 on 1.300, voltage could go lower for all i know or chip could go higher. Too lazy to tweak it much more.

 

[BluD]

Has anyone got over 5ghz on these without 1.4+ volts? I thought these were suppose to get 5 ghz on air cooling, but i wouldnt fancy running over 1.4v on air.

 

[setter]

Havent seen many that will do 5g on decent voltage that you could use 24/7. In most cases ive seen they run hotter than the older 4770k too.

 

[Deehjay]

Got my 4790k under water now with a single 360 rad - temps are around 25 degrees idle and 50 degrees during RealBench. Any good? Time to overclock me thinks! Temps are flaring back and forth quite a bit but I assume that's due to the thermal compound not spreading properly yet.

 

[BluD]

If thats without delidding at about stock 1.2v thats a very good temp yeah. Dunno why your temps are jumping around, from the thermal past not been spread properly i normally find one or 2 cores is a lot hotter than the others, rather than jumping temps.