i7-4770k: Holy Christ This Is FAST

[peacedivision]

Try Prime95 in place large FFT for maximum heat load.



If you can get through without BSOD or the chip throttling its multiplier to stay alive you are good to go. Still ditch the stock cooler ASAP.

[meaty-btz]

Paste. Started with the stock paste from Intel, switched it out yesterday with something different. Not a brand I'm familiar with. I think it was called STAR-700 or something? It's 10% silver, at any rate.

All the fans are working. 2x120mm intake, 1x120 and 1x90 exhaust.

Temps are better today than before. I'm idling at low-mid 30's now, instead of the low-mid 50's from before the new paste and adding another fan. Got through wPrime 32M in 6.39 seconds at max 63 degrees. Still hit 86 during wPrime 1024M, though, and I cancelled it at 33% done.

We'll see how this does with rendering and other highly multithreaded loads ...

I still don't have a host for the file BTW.

As for the heat. Every should, but no one does, lap the heatsink and chip. But that really depends on how much you want to dink with it. Significant improvement in heat transfer. Thermal paste is more of a crutch to deal with the poor fit and heat transfer normally found. Stock coolers have always been sufficient in my experience except when over-clocking. Aftermarket is always better though, I'd still lap an aftermarket heatsink, same goes for a water cooler pack.

[Peter.Steele]

I'm pissed that I didn't jump the other night when TigerDirect had the Corsair H100 (not the 100i, unfortunately) on sale for $69.99. Ah, well ... deal was gone the next morning.

Anyway, a local store in Fresno has the Zalman LQ310 for ... well, relatively high at $79.99 plus tax, but they have it right now and I do occasionally like to support local businesses* ... I haven't seen any reports matching it up against Haswell, but it did seem to do well against a 3930k in one review.



*I'm hesitating with this store, though. The guy that I talked to on the phone about what cooling options they had in stock tried to tell me that a 4770k was Sandy Bridge, not Haswell.

[Fizz]

So it's fast ... but it also is running quite warm.

I'm using the stock cooler and the same cooling setup as on my previous 125W CPU. I'm idling right now at 50-54 degrees. I had one BSOD earlier due to temperature, and when I rebooted I recorded a max temp (from Core Temp 1.0 RC5) of 81 degrees.

Obviously I'd like to get water cooling, but here's the thing: if the stock air cooler wasn't going to be enough to run this chip at base frequency, then why would Intel use it?

They wouldn't.

80 degrees isn't really that hot to a CPU. Your machine restarted for some other reason.

I've built millions worth of Intel based PCs. The stock cooler has ALWAYS been adequate except for when you cram the computer into an enclosed space, overclock, let the HSF get so dusty you have an air dam, etc.

[Peter.Steele]

They wouldn't.

80 degrees isn't really that hot to a CPU. Your machine restarted for some other reason.

I've built millions worth of Intel based PCs. The stock cooler has ALWAYS been adequate except for when you cram the computer into an enclosed space, overclock, let the HSF get so dusty you have an air dam, etc.

Mmmm ... nope.


When I rebooted after the BSODs, the BIOS gave me a warning that the shutdown had been caused by high CPU temp. Every. Single. Time.

The temps have gone down significantly since I underclocked and scraped the stock Intel paste off the CPU and the fan and replaced it with a good solid layer of something else. Idle temps are ~20 degrees lower now. It still gets warm very quickly though - if I'm transcoding a video or doing any rendering, it'll easily get to 95+ as reported by Core Temp, and right now I've got it running slightly underclocked. I haven't taken it past 100 yet.

I think I'm going to probably just get an H100i at some point, but there's a couple of other upgrades I need to buy first.

Fortunately, this is all tax deductable!

[Fizz]

I'm not doubting a thermal event could have occurred, but it's not due to an inadequacy of the HSF.

I lost count of the number of times customer-built machines came to me with the HSF on 3 or less pins because they twisted the posts after mounting. If they were secured at all it was due to compound. Those resulted in similar reports of 'crappy stock intel fans'

Hint: If the arrows are pointing at the heatsink, they're in the 'unlocked' position.

If your board supports any kind of fan speed throttling you might have to adjust/disable.

[Peter.Steele]

I'm not doubting a thermal event could have occurred, but it's not due to an inadequacy of the HSF.

I lost count of the number of times customer-built machines came to me with the HSF on 3 or less pins because they twisted the posts after mounting. If they were secured at all it was due to compound. Those resulted in similar reports of 'crappy stock intel fans'

Hint: If the arrows are pointing at the heatsink, they're in the 'unlocked' position.

If your board supports any kind of fan speed throttling you might have to adjust/disable.

Yes, all 4 pins on the fan were connected. No, I didn't bend the pins on the motherboard. Yes, the arrows on the heat sink were pointing the right direction. Yes, the motherboard supports fan throttling. Yes, they were being controlled by the motherboard. No, the fans weren't shut off.

So, yeah, I'm not a complete retard.

[Fizz]

Yes, all 4 pins on the fan were connected. No, I didn't bend the pins on the motherboard. Yes, the arrows on the heat sink were pointing the right direction. Yes, the motherboard supports fan throttling. Yes, they were being controlled by the motherboard. No, the fans weren't shut off.

So, yeah, I'm not a complete retard.

Then I guess you found the first inadequate HSF I've ever heard of and I used to stress test machines in a 80- 95 degree fahrenheit room.

[Peter.Steele]

I would say it was only a partially inadequate HSF. Once I changed the thermal paste out it was more or less okay for basic web browsing and lightweight tasks* so long as I kept the voltage turned down. If I render it still jumps to 90+ almost instantly, but so long as I'm not doing anything requiring more than one thread, then hey, I seem to be okay. It's just probably not a great idea to do any, y'know, rendering or anything like that, until I get a new cooling solution.




*"Lightweight Tasks" in this case is defined as: free-to-play games, AAA titles from 2004, Go2Meeting, Minesweeper, that sort of thing.

[meaty-btz]

I would say it was only a partially inadequate HSF. Once I changed the thermal paste out it was more or less okay for basic web browsing and lightweight tasks* so long as I kept the voltage turned down. If I render it still jumps to 90+ almost instantly, but so long as I'm not doing anything requiring more than one thread, then hey, I seem to be okay. It's just probably not a great idea to do any, y'know, rendering or anything like that, until I get a new cooling solution.




*"Lightweight Tasks" in this case is defined as: free-to-play games, AAA titles from 2004, Go2Meeting, Minesweeper, that sort of thing.

just a thought popped in my head. There may be a defect in the chip between the chip and the heat spreader. It's rare as all hell but it can happen. If that is the case then you won't get good transfer from the chip to the spreader. No way to test for this as far as a I know though except "excessive heat related issues".

[dynamomark80]

I dont think it would be the heatsink at all. Im sure their are a ton or rigs out there with that same cpu/hs combo and intel would never put out a crap hs that couldnt at least handle the stock cpu at full load. It is possible the chip had a defect of that you ran it to hot and damaged it. It should have worked as shipped at regular settings. If nothing is wrong with the cpu I would take a look at the ran. Check hs fan rpm range and make sure case has proper ventilation.

On another side note, the need to oc a more then capable cpu to get miniscule performance gains is beyond me. I can understand oc on the older dual/quad core cpu on setups with a slow memory bus but not with todays current cutting edge tech.

[sd_shooter]

Post back in a couple years once Windoze has filled up your hard drive with useless restore points and other junk.

My WEI has remained constant since I bought my PC but my bootup time has gone from ~1min to ~5min.

[dynamomark80]

Post back in a couple years once Windoze has filled up your hard drive with useless restore points and other junk.

My WEI has remained constant since I bought my PC but my bootup time has gone from ~1min to ~5min.

I laugh at restore points. I should turn them off but for now I put thema nd my weekly backups to a separate drive just for backing up stuff. I got some drives cheap so Im up to 4 drive broken down to OS, media, games, and backups. Start-up time is still reasonable for a non-ssd equipped pc.

[yellowsulphur]

I dont think it would be the heatsink at all. Im sure their are a ton or rigs out there with that same cpu/hs combo and intel would never put out a crap hs that couldnt at least handle the stock cpu at full load. It is possible the chip had a defect of that you ran it to hot and damaged it. It should have worked as shipped at regular settings. If nothing is wrong with the cpu I would take a look at the ran. Check hs fan rpm range and make sure case has proper ventilation.

On another side note, the need to oc a more then capable cpu to get miniscule performance gains is beyond me. I can understand oc on the older dual/quad core cpu on setups with a slow memory bus but not with todays current cutting edge tech.

This is what a stock Intel heatsink/fan looks like that came with my 3770. Haswell and Ivy have the same size die so I would assume Intel's cooling solutions would be similar. These things are marginal and small fans can be quite annoying when they spin up. IIRC Intel recommends a larger cooler if the system is going to be operated for extended periods under heavy load.

[meaty-btz]

That is a very inadequate looking cooler.