Oldsmoke, yes this is the 3930K .

At full load 100% max temp is at 85c fan speed at only 4400-4700rpm, As I say above, as cpu temp is getting hotter, the fan will spin faster to keep the temp in check. But this is at full load at 4.7Ghz . Realtemp Thermal Status stay OK thru out the test.

Here is an example, I rendering a 35 minutes footage on Vegas timeline right now. CPU Usage at 42-57% and the realtemp showing 62-73c and the fan speed at 3000-3400rpm

Cpu load at different level , depends on what plug ins you are using. The same test footage I post above but with BCC7 color correction. CPU usage only at 42-57% . Newblu fx color fixer color correction, cpu usage 98% and much faster.

How much faster is it rendering at the overclock? I tried overclocking mine years ago when I got it. It would be stable but the % of decrease in rendering time was small enough where I said I'd rather have the CPU run a tad cooler then get an extra couple seconds shaved off.

I was about to say the same. I know I can get mine stable and fairly cool, 75C at 4.6GHz but the decrease in render time beyond 4.3GHz wasn't significant anymore. I suspect that at higher speeds the other components become bottle necks including the RAM. The 3930K doesn't like RAM speeds higher then 1600MHz. I tried 2133 but again the performance improvement wasn't significant.

Thehappyfriar

How much gain in rendering from 4.6-4.7Ghz. Since you ask, I am doing the test righ now t. I will report the finding later on. I am not concern with increase heat issue at all. My Delta fans can handle the heats and eat it away. With the noctua stock fan, I felt warm air coming out. at the exhaust fan. With Delta fans, I don't feel any warm air at all. I am not kidding you. The Delta fan I placed at the back of the full tower removed any warm air quicky. So there is no warm air building up in side the tower.

Sure wish i could give Vegas a test run on one of the servers at work. It's only a 2.1GHz Opteron 6172 box (yeah, about 3 years old), but it's got 48 (yes, forty-eight!) processor cores in it. We've also got a measly little 128GB RAM and 10TB RAID 1 drive space. We built it for AT&T for a database project they were having trouble with. They couldn't get their queries down under 90 seconds on their best hardware but we were able to get them to run in under 5 seconds on this beast. Unfortunately they eventually dropped the project, but we got to keep the box!

Too bad it's running Linux instead of Windows, or i'd give it a try.

I suppose at $49,000, it'll be a while before i build one for myself.

I think you need a shroud going to the CPU fan. I might well be that the rear exhaust is so strong that it takes away the air the CPU fan needs for cooling.
I changed to water-cooled last year for exactly that reason. There is a time where can't get enough cool air into casing anymore. I can't wait to get my two GTX570 water-cooled too.
Also keep in mind that the TCASE is only 66.8C for the 3930K according to Intel's website.

Chienworks

Well, you are comparing a $3500 system to a $49,000 system is like compared apple and orange.


Oldsmoke,

I don't need any shroud for the fan. The 3 Delta are equal in speed and cfm. The exhaust fan will not over power the other. All 3 work in pwm mode. I got the 2 Delta fan sandwich the heatsink. As warm air heat up the case, the exhaust will remove it quickly.


As you know Intel temp recommandation is on a conservative side and well as their cpu voltage. for example they recommand not to go over 1.35v and I don't remember seeing any body burning up their cpu under 1.5v on the 3930K. I have mine up to 1.4v

ok the results are in.

18 miniute avchd on the timeline with Newblu fx color fixer color correction. Render strictly on CPU ONLY MainConcept Avc 1920x1080 30P

4.6Ghz render time 21:17 minutes
4.7Ghz render time 20:36 minutes

For what its worth, definately a gain and free computing power. And I am for it.

>>>>I will be keeping it at 4.7Ghz, I love this cpu<<<<

So the new Mac pro is a game changer, yet you have been running a PC at almost a full 1GHz faster for how long now?

What will mac drones say now?

Bruce,

Which TIM did you use?


..

skeeter123,

TIM, you means thermal paste? I used Noctua NT-H1 thermal compound.. For any one who has 3930k/4930k with 4.2-4.5ghz. If you don't mind, can you drop exactly 18 minute avchd footage onto vegas timeline and use Newblu color fixer for color correction. Render that in MainConcept AVC 1920x1080 30P CPU ONLY.

Turn off gpu in perferences and restart your computer. See what you got in rendering times. Or perhaps my system is slow for what it is.



PS.

I understand some of you guys may not have Newblu FX on your system. So, I am re- render the 18 minutes avchd witout Newblu FX.

This is for the 4.7Ghz results, CPU ONLY. Turned off gpu in preferences and restart the computer to take effect.

18 minutes avchd render MC AVC 1920x1080 30P No Newblu FX.

The results. 16:38 minutes that is about 4 minutes drop in rendering time without FX. Now for anyone saying between 4.2Ghz and up don't make much of a different. Well, I am at 4.7Ghz, you can test for youself on your 4.2-4.5Ghz see how much of a different basic on what i have show here. 18 minutes avchd straight from camera and render with out any FX. See what you can come up with. I appreciate your time if you can do this simple test. just for comparision between 4.2-4.5 to my 4.7Ghz. .

deusx, I don't follow MAC Computer developments.. So, I am not going to comment on that.

That test doesn't make sense in my case; I will never render CPU only. With GPU on, the difference is so minor that it don't have to go to higher frequencies. Certainly, if you render CPU only I don't dispute that there is an improvement.

Anyway: HXR-NX5U AVCHD footage 1080-60i rendered to MC AVC 1080-30p, CPU OFF everywhere: 25:02min.
Same with GPU ON everywhere: 8:35min.
CPU: 3930K@4.3GHz

I don't have 30p footage but that would be faster for sure.

That is why I was specifically set out for CPU ONLY for testing purpose to see how much of a improvement going from 4.2Ghz to 4.7Ghz. Heck, if I chose to render with GPU and of course my render time would be much much less.


PS. My test footage is also 1080 60i from one of my old canon camcorder. But Thanks for the test. I do appreciate your time.


PS.S. In regards to GPU render, it seems that your GTX570 and my AMD 6970 are pretty close in performance based on the Vegas benchmark that Hulk putting together. Example, In XDCAM render, the AMD 6970 edge out the GTX570 by 10s and in MainConcept AVC render your GTX570 edge out the the AMD6970 by 10s. 33s vs 43 and 41 vs 51 sec

But what is the difference with GPU on and a higher frequency?

I promise, I will try to run the test with GPU on some time later on and will post report the good, the bad and the ugly.

I know that MC AVC is the fastest with CUDA on my system as it uses both cards almost equally. I think the difference in load might be coming from the difference in PCIe slots; x16 and x8. I need to "slim" down my GTX570s as they are 3slots wide; once watercooled they only occupy two slots each and then I can get both into x16 slots.

It might actually be that two 6970 with Crossfire are faster if both are utilized under Vegas.

@Chienworks

I also have several such boxes (40 real cores, 80 with hyper) at disposal but when I wanted to use them to test Vegas I found a problem... which version of 7 can you install on such servers ? I was told 7 only supports a limited number of cores... then, usually these servers have very basic, no-gpu, video cards...

Hi, Bruce!

Yeah. Thermal paste....I tried Arctic Silver 5 for my 4930k this time....so far, so good...

Any chance of you posting your 18 minute footage to Dropbox or similar? Would love to use the same source for testing...I have a bunch of space available on Google Drive if needed...

...

18 minutes avchd into drop box? h e l l no. Beside, I don't have a drop box account anyway. Look, this is just a quick and dirty comparision. It may not be supper accurate. Can you just drop in 18 minutes any avchd footage 1080 30P or 1080 60i footage and let it rips. What ever the results and go from there. At the end of the day, we are all be using GPU rendering any way.

PS. I have tried a tube of Arctic Silver 5 and it did not made a different in temp. So I went back to Noctua paste. Are you enable turbo stepspeed on your ? if you do, then of course your temp will be cooler. I disable turbo on mine, so I am always right at 4.7Ghz even in idling.

Disabled SS. Sits at 4.55Ghz all the time...I haven't noticed any diff with Artic Fox either...I'm a retired software/hardware engineer, so I do like to experiment... but I keep it at Stock voltages..I don't change the voltages because there is a point that WILL shorten the life of the silicon. Now, will it do it by the time the next uber-nice chip makes it way down the pipeline? Maybe, maybe not.. YMMV...

I was offering my Drive account if you wanted get all sciencey about it...I have lot's of GoPro and RX100 MK II 1080p footage. I'll stick to the RX100 stuff since the GoPro stuff is a bit higher bitrate than the RX100 (15-26Mbps vs 30Mbps). What bit rate is your source?

I usually cut Mercalli Prodrenalin footage. That stuff is in the 150-250Mbps range....ugh...I can put some Prodrenalin source material in my Google Drive if you want to have at it....PM me if interested...

I'll also test it against the R9 290 that should arrive this week. Gonna be fun in nerd-land....

..

Oldsmoke,

As promised, I rendered the 18 minutes footage with GPU. 7:48 minutes. GPU load at 92% solid and 15% cpu. MC AVC 1920x1080 30P