28 cores Dual socket build (Project Boobs) (Completed)

[alucasa]

The only problem I have at the moment is fan speed control. Being a server motherboard without IPMI and iKVM, the BIOS does allow me to control fan speed via 4 presets. (For info, if it's IPMI board, you have to set fan speed through IPMI viewer either remotely or on OS level)

The quietest preset sets fan speed to 5,000 RPM if possible. What that means is that I should not use 4 pin fans and all of the fans in the case is 4 pin, so they are all running at maximum RPM.

In other words, I will need to purchase 3 pin fans next month and even consider changing CPU heatsinks with swappable fans. Making it quiet will be my next goal.

 

[alucasa]

I have an update to make.

It seems my previous idle power consumption was wrong. Now that a day has passed and Windows did its own indexing and updating, idle power draw is down to 70ish watt which is much better than 130 I had a day ago.

Why is this important (to me at least)? On X99 platform, idle power consumption was always 100 ~ 120 watt. Lower idle power consumption is very important because I believe a computer stays idle or near idle most of times. This idle power consumption is what I likely get during web browsing. C612 chipset is certainly less power hungry than X99 chipset. A huge thump up for that (again for me at least).

This means it wastes less power on idle and does work twice faster when required: Bonus for me.

 

[biffzinker]

Seems impossible it dropped to 67 watts. Is Windows parking or shutting off the second Xeon?

 

[alucasa]

I was browsing TPU when I felt like checking the power usage. It is entirely possible that it's parking the 2nd CPU. Either way, I am impressed.

 

[Caring1]

......The quietest preset sets fan speed to 5,000 RPM if possible. What that means is that I should not use 4 pin fans and all of the fans in the case is 4 pin, so they are all running at maximum RPM.

In other words, I will need to purchase 3 pin fans next month ....... Making it quiet will be my next goal.

Try pulling the sensor wire from the 4 pin plug, it will function as a normal 3 pin fan then.

 

[alucasa]

Try pulling the sensor wire from the 4 pin plug, it will function as a normal 3 pin fan then.

Ohhh, yeah, that might work. Thanks.

 

[Kanan]

X99 has same or lower idle consumption, it's just that you maybe had (again) some background apps running eg. Windows Update or it's because your CPU was overclocked. Even X79 is under 100W.

http://techreport.com/review/27279/asus-x99-a-motherboard-reviewed/4

 

[Caring1]

X99 has same or lower idle consumption........ Even X79 is under 100W.

That's good to know, as i'm looking for a X79 Motherboard myself and power consumption is always on my mind.

 

[infrared]

Lower idle power consumption is very important because I believe a computer stays idle or near idle most of times.

Your computer should never be idle Get WCG up and running on that monster! lol

Seriously though, this is a pretty cool build

 

[alucasa]

X99 has same or lower idle consumption, it's just that you maybe had (again) some background apps running eg. Windows Update or it's because your CPU was overclocked. Even X79 is under 100W.

http://techreport.com/review/27279/asus-x99-a-motherboard-reviewed/4

Thanks but my X99 system never went below 100w. It was hovering around at 120. I did check numerous times.

 

[Kanan]

Thanks but my X99 system never went below 100w. It was hovering around at 120. I did check numerous times.

boards/setups/background apps etc. differ, those websites test non-overclocked, skeleton setups with skeleton windows only, so no wonder their numbers are lower. But this is true for any system, not only for X99. Though I wouldn't compare a new system (software and all) with a old, that's exactly what you did. That's again skeleton vs. old system full of apps (I guess).

 

[alucasa]

boards/setups/background apps etc. differ, those websites test non-overclocked, skeleton setups with skeleton windows only, so no wonder their numbers are lower. But this is true for any system, not only for X99. Though I wouldn't compare a new system (software and all) with a old, that's exactly what you did. That's again skeleton vs. old system full of apps (I guess).

My application library is exactly the same. I don't even have that many.

What I had and what I have now are exactly identical: Steam (same games since drive is transferred), LibreOffice, GIMP, Blender, Firefox. FTP and SSH clients. And, again, I checked power usage as a brand new system and after. And it never went below 100.

I will have another opportunity to test the system if I grab an cheap ES cpu.

 

[Kanan]

My application library is exactly the same. I don't even have that many.

What I had and what I have now are exactly identical: Steam (same games since drive is transferred), LibreOffice, GIMP, Blender, Firefox. FTP and SSH clients. And, again, I checked power usage as a brand new system and after. And it never went below 100.

Another factor I forgot, mainboards differ a lot =)

I guess after all you're right, that Supermicro board is very efficient. Makes sense, servers are supposed to be efficient.

 

[newtekie1]

Try pulling the sensor wire from the 4 pin plug, it will function as a normal 3 pin fan then.

You actually want to pull the PWM wire, not the sensor wire.

 

[Bo$$]

Looking good!

 

[alucasa]

You actually want to pull the PWM wire, not the sensor wire.

As for the noise, I think the CPU fans are the culprit. I did check by pulling out all fans and inserting back one by one.

But the main issue appears to be the case itself and how fans are aligned. I don't feel high enough noise with one by one. But when all four are together, it's a vacuum cleaner.

Edit: CPU fans are rated for 2,000 RPM max. Case fans are rated for 1,400 RPM.

 

[newtekie1]

Thanks but my X99 system never went below 100w. It was hovering around at 120. I did check numerous times.

The X99 and C612 are, AFAIK, the same silicon. They just have different features enabled/disabled. The power consumption could be a lot of things. The first thing that comes to mind is not just the chipset, but the rest of the stuff on the motherboard. Compared to your X99 Extreme4, the Supermicro board has less power hungry stuff on it. It doesn't have as many USB ports, it doesn't have a fancy high powered sound chip.

The other thing to consider is the PSU. I gather you aren't using the same PSU in each system. 150w is 20% load on a 750w PSU, you are idling below that. Once you start to get below that 20% load level, PSU efficiency starts to drop off pretty rapidly. Some PSUs are better than others in this area. This alone can make a 10-20w difference, especially if you had a bigger unit in your X99 rig.

 

[jaggerwild]

What's wrong with it? I like women's thighs.

But that's not a "WOMAN" a girl maybe........

 

[alucasa]

The X99 and C612 are, AFAIK, the same silicon. They just have different features enabled/disabled. The power consumption could be a lot of things. The first thing that comes to mind is not just the chipset, but the rest of the stuff on the motherboard. Compared to your X99 Extreme4, the Supermicro board has less power hungry stuff on it. It doesn't have as many USB ports, it doesn't have a fancy high powered sound chip.

The other thing to consider is the PSU. I gather you aren't using the same PSU in each system. 150w is 20% load on a 750w PSU, you are idling below that. Once you start to get below that 20% load level, PSU efficiency starts to drop off pretty rapidly. Some PSUs are better than others in this area. This alone can make a 10-20w difference, especially if you had a bigger unit in your X99 rig.

The X99 had FSP Hydro 650w.

But that's not a "WOMAN" a girl maybe........

They are Japanese. They look younger than how old they actually are. Actually, the photo is from a renowned cosplayer.

 

[alucasa]

Ordered two of these Zalman fan controllers to keep the noise down.... Man, I am really going over my budget this month...

 

[Cvrk]

Q: When was the last time most geeks touched a warm breast?

When my mother was breastfeeding me #vegan

 

[biffzinker]

Have you tried anything else that would benefit from the 28 cores?

 

[alucasa]

Have you tried anything else that would benefit from the 28 cores?

Nothing would really benefit more from higher # of cores than rendering.

With i7-4770/6700, I was looking at 3~4 hours per render. Mistakes and re-renders concerned, it can take days for final render.

With single e5-2683v3, each render went down to 1 ~ 1:30 hour. That was significant time saving. That's why I went for dual. With 28 cores, all of my render time is down to 30 minutes at decent quality setting.

Just the amount of time saved gives me less stress because mistakes cost me less.

 

[r9]

Nothing would really benefit more from higher # of cores than rendering.

With i7-4770/6700, I was looking at 3~4 hours per render. Mistakes and re-renders concerned, it can take days for final render.

With single e5-2683v3, each render went down to 1 ~ 1:30 hour. That was significant time saving. That's why I went for dual. With 28 cores, all of my render time is down to 30 minutes at decent quality setting.

Just the amount of time saved gives me less stress because mistakes cost me less.

Time is money, looks like you're saving quite a few bucks.
Just curious what do you render ?

 

[JunkBear]

What's the rig going to be used for?
I don't think 2 Xeon's will make your Pr0n collection go any faster

Démineur probably