Need CPU & Chipset advice for building a "HackPro"workstation

[frankiee]

Yeah, my current one also supports 50GB disks, but has some (minor) problems I described above. So, which model would you recommend?

[frankiee]

I also have a specific question about the R4E. Rampagedev you said that you also control your temps with pwm fans. But how do you do it? I stumbled over this interesting thread, and I am a bit confused, bc they say (and seem also to prove) that the R4E fans are NOT pwm controlled (except one) but rather voltage controlled.

 

See here http://www.overclock.net/t/1401311/asus-rampage-extreme-iv-pwm-problem

 

Think that would make quite a difference, but is it true? The manual says otherwise, so I am really confused.

[Rampage Dev]

I also have a specific question about the R4E. Rampagedev you said that you also control your temps with pwm fans. But how do you do it? I stumbled over this interesting thread, and I am a bit confused, bc they say (and seem also to prove) that the R4E fans are NOT pwm controlled (except one) but rather voltage controlled.

 

See here http://www.overclock.net/t/1401311/asus-rampage-extreme-iv-pwm-problem

 

Think that would make quite a difference, but is it true? The manual says otherwise, so I am really confused.

 

Every last header is a PWM header. You need PWM Fans to make the PWM features work. All but 3 of my fans are PWM and the 3 that are not are Temp controlled fans. If you read what the end user was BSing about he was using a splitter on a PWM header which can actually cause a short on the board. Seen many people have issues with PWM splitters. 

Wrapped up testing on the E WS board today. All is well. 

[frankiee]

Thanks for clarification. Not so easy for me to decide what's BS and what's not. Just new to all of these things. But I never planned to use a splitter anyway, just attach each fan to it's own header, and - if possible - control them all independently by temperature. If I actually go for the R4E, I think I might even install an additional temp sensor (depending on the case, maybe near the HDDs for example) - and if only just for the fun of it

[frankiee]

Wrapped up testing on the E WS board today. All is well. 

 

Great news! So now, the final shooutout will be between the P9X79-E WS and the R4E.

 

What I like about the WS is more SATA slots and of course the firewire header on the board, saving me an additional card. And, this FW card - if going for the R4E - should have an internal header, are there actually existing some which are also OS X compatible?

 

On the other hand, temp / sensor management seems to be way better on the R4E, the WS has only 4 non CPU fan headers (2 less than the R4E), and if I would choose the FT02, they all will be already occupied then. Plus, while they can also be PWM controlled, the only temp source really seems to be only the CPU, and I think this sucks, not what I imagined at least. So in this case, I might end with an additional separate fan controller instead. And, the WS has not much USB headers onboard. (think just 1)

 

To bad you can't get everything ....

[joe75]

frankiee, for easy reference, PWM headers or fans will be 4 pin

[frankiee]

frankiee, for easy reference, PWM headers or fans will be 4 pin

Yeah, at least this is one of the things I already know But I have to admit I am still struggling quite a bit with the optimal cooling solution.

 

Also I am not sure if a R4E will really save me an additional fan controller. So my favorite case would be still the Silverstone FT02 (but also the Nanoxia DS and the Fractal Design XL R2 are on my list). This case uses a total of 4 fans (except CPU fans), and three of them are huge 180mm fans. I'd like to upgrade these to the Slverstopne AP182 fans, which are better.

 

Now, we have 4 fans, but only 3 "opt" fan headers on the mobo (which can be temperature controlled). So I might still need a splitter after all? Or is it possible with the R4E to use another header, lets say "chassis", but still use the sensor reading of an "opt" header as input at the same time?  I was under the impression that each "opt" can be only connected to it's own sensor, but since I don't have an actual BIOS to play with, I am not sure. Or are there additional sensors on the Mobo I could use? At least controlling a chassis fan directly via CPU temps still seems quite weird to me.

 

And apart from the fact that this fan also draws too much power (and needs am additional molex adapter) it is unfortunately only a voltage controlled fan. Read at least some reviews where using a PWM controller gave problems, and was not usable.

 

So, can I actually switch RPM control on the rampage IV extreme also to voltage control, or is it really PWM only?

 

I am also thinking about CPU cooling. While rampagedev recommended a h80i, I am not sure if I would do actually better with air cooling. At least a good tower cooler (Noctua NH-D14 for example) should give me both better cooling and way less noise. At least with the h80i the first thing I'd do is to replace the stock fans, which are rather loud. And I am also not _too_ keen about flowing liquids in my computer. Main drawback I see, however are the sheer size of these things, and the resulting clearance problems which may occur. I checked the RAM, and while it should fit, it may require removing the cooler I wanted to change something there. But cooling / noise ratio is way better than especially with smaller AIO coolers that seems to be a fact.

 

PS: and the WS-E fan control is really pathetic, seems the minimum duty cycle is a whopping 60% (compared to "0" on the R4E) .... this board would need a fan controller anyway I guess, also temps can be _only_ controlled by the CPU at least guessng from the manual - useless in my eyes. So the R4E might have a real advantage here, _if_ I can get my desired setup to work with it. And I am not sure about that yet, so any clarifications would be really helpful.

[joe75]

I personally don't know anything about the board. Using PWM or any controllers is strictly for noise control, some might claim to want added power management too but the draw from fans in very small. For actual cooling performance you'd want these fans full speed and ones that push the greatest amount of air which would also lead to increased noise. Any case fans would normally be left at full speed and there would be no problem using molex or splitters for them and this would free up headers for fans you might want controlled off PWM.

 

If your goals are for silence your best options with air are PWM on hardware fans and fan controllers or lower voltage fans for everything else.

[frankiee]

Well, the plan ist to have a completely automated, temperature controlled cooling with the actual controller being rather "low level", ie not controlled by an ordinary app or driver (would be difficult in OS X anyways) but rather by the BIOS or even via a dedicated controller. And I also like to have some sensors spread around, I think if the only temp input would be only the CPU, I can imagine it will sound like with the G5, where you only had to stress your CPU a little bit, and ALL fans were going on turbo immediatly, so that the machine sounded like a vaccum cleaner. The fans sometimes even ramped up when you were just scrolling a web page Oo. So certainly it should work better than that. Things like this can kill my whole experience ... So, proper and SILENT cooling would be very important for me.

 

Maybe rampagedev can help me a bit on this R4E specific topic? A bit diffcult to asses all this if you don't have a "real" BIOS to play with. Or is there such a thing like a BIOS simulator where I could play around a bit?

[tj5]

 My point was that no one needs "boot camp" on a PC because you can have a normal native install.

If you have a normal native install and you use parallels you can have it use the native windows installation as a boot camp vm. It will use the native ntfs drivers so you get a better disk i/o as well as native disk space so all installed programs are available for running native or thru parallels. I have my system setup that way and it works great. I rarely if ever boot windows natively anymore.

[frankiee]

If you have a normal native install and you use parallels you can have it use the native windows installation as a boot camp vm.

 

That is exactly what I do. I partitioned my System drive in two partitions, with a native Windows 7 and native OS X install. And this Windows 7 partition also doubles as a parallels installation. So it would be important for this to work exactly this way also on a hackintosh. But I am not sure if it will be more hassle with the partitioning, since I also want a recovery partition. So an option would be to use just a separate drive for windows. That's one of the open questions I cannot answer right now.

 

I rarely if ever boot windows natively anymore.

 

In many cases I do that as well. As I mainly use VMs for software testing, it is quite common that I have 4-5 VMs open at the same time. I would also like to open more, but this is the point where my Mac Pro begins to bog down with only 16 Gigs of RAM (also CPU load is quite noticeable, with a total load - ie all 8 cores - of up to 50% even when the VMs are more or less idling).

 

But: as I wrote before, there are cases where I just need or want a native install:

 

- I need native booting for some software I use that demands a more direct access to hardware. In my case, drivers for an USB microscope as well as the windows phone emulator.

- I want native booting for gaming. Tried some games (and I don't play _that_ demanding games, no FPS for example) in paralells, and at least with my setup there's a noticeable difference, esp in Graphics

[frankiee]

so, finally I have my buildlist together, and this is how it looks like now:

 

System
CPU: INTEL 4930K
GPU: NVIDIA Titan
Mobo: Rampage IV Extreme
RAM: 8 x 8GB G.Skill RipjawsZ PC3-17000 CL11
Power Supply: CORSAIR AX1200i

Storage
1x Samsung 840 EVO 1TB
1x Samsung 840 EVO 500GB
1x Samsung 840 EVO 256GB
2x WD Red WD30EFRX 3TB
1x Panasonic Slim Internal Blu-ray Burner UJ-265

Case & Cooling
Case:  Silverstone SST-FT02B
Case Fans: 3x Silverstone AP182 & 1x Noctua NF-S12A
CPU Fan: Noctua NH-U12S
Fan Controller: aquaero6 system controller

Addons

EX-16450 - PCI-Express FireWire 1394A
SilverStone FP58B - Slot loading Drive Adapter
Silverstone FP55B 3,5" Adapter
Silverstone FP35 multi-function Front panel
Silverstone SST-CP05-SAS Hot Swap Adapters x4
 

The only minor gripe I have with this setup is CPU cooling. I like to stay away from H2O and regarding air coolers, the NH-U14S - which I would prefer - would block the first PCI slot. Don't know if I also could install my 1st GPU in slot 2, but if I would install a 2nd one I get problems with clearing anyways. But I am still positive the NH-U12S can handle it.

 

So if anybody is actually interested how this all works out, I can post something like a build log - just let me know

[shilohhh]

Case & Cooling

Case:  Silverstone SST-FT02B

Case Fans: 3x Silverstone AP182 & 1x Noctua NF-S12A

CPU Fan: Noctua NH-U12S

Fan Controller: aquaero6 system controller

 

 

The only minor gripe I have with this setup is CPU cooling. I like to stay away from H2O and regarding air coolers, the NH-U14S - which I would prefer - would block the first PCI slot. Don't know if I also could install my 1st GPU in slot 2, but if I would install a 2nd one I get problems with clearing anyways. But I am still positive the NH-U12S can handle it.

 

So if anybody is actually interested how this all works out, I can post something like a build log - just let me know

I have the FT02B (Love it!), R4E, 3930K. I use Noctua NH-D14 and it doesn't block the 1st PCI slot. It's super quiet and efficient.

[frankiee]

So I have at least a basic build running right now, including a working, but maybe still not perfect OS X installation. Yeah, the case is quite nice, but also a bit tight to work with, but I do not regret buying it until now at least. Mavericks seems to run quite smooth - no KPs or anything - also very fast boot times (around 6 secs without kernel cache). Still needs quite some tweaking though.

 

As for the cooler, I just settled for the NH-U12S (the NH-D14 would block the RAM slots, whereas it would actually be the NH-U14S that would block my 1st PCI slot). And CPU temps seem to be quite low. Maybe not as low as you can with H20 that`s for sure, but since I am not planning to do any extreme OC it should be enough. Also has enough room in the case and I have access to all RAM slots. BUT I have a problem with this CPU fan running at 100% all the time no matter what I do in the BIOS. Still checking with Rampagedev to find out what's the cause - but since the CPU fan is working OK on the "CPU_OPT" header I fear that something with the mobo is going wrong. Looks like "CPU_FAN" does not transmit any PWM information

 

Geekbench scores seem to be quite low for this CPU, though (17000+ compared to 13000+ on my 5 year old Mac Pro) but still all running at stock speeds. Hope this will get better.

[shilohhh]

So I have at least a basic build running right now, including a working, but maybe still not perfect OS X installation. Yeah, the case is quite nice, but also a bit tight to work with, but I do not regret buying it until now at least. Mavericks seems to run quite smooth - no KPs or anything - also very fast boot times (around 6 secs without kernel cache). Still needs quite some tweaking though.

 

As for the cooler, I just settled for the NH-U12S (the NH-D14 would block the RAM slots, whereas it would actually be the NH-U14S that would block my 1st PCI slot). And CPU temps seem to be quite low. Maybe not as low as you can with H20 that`s for sure, but since I am not planning to do any extreme OC it should be enough. Also has enough room in the case and I have access to all RAM slots. BUT I have a problem with this CPU fan running at 100% all the time no matter what I do in the BIOS. Still checking with Rampagedev to find out what's the cause - but since the CPU fan is working OK on the "CPU_OPT" header I fear that something with the mobo is going wrong. Looks like "CPU_FAN" does not transmit any PWM information

 

Geekbench scores seem to be quite low for this CPU, though (17000+ compared to 13000+ on my 5 year old Mac Pro) but still all running at stock speeds. Hope this will get better.

I get about 22k at 4.2ghz and 24k at 4.5ghz in geekbench 3 64bit. You'll be limited to CPU strap or blck OC until asus fix their IvyB-E BIOSs to allow for proper multiplier OC. You could try shamino's 0024 bios for the R4E found at the rog forums but I haven't tested it as I have SandyB-E and can use bios 4206.

 

My noctua outperforms most budget closed loop liquid coolers

[frankiee]

I get about 22k at 4.2ghz and 24k at 4.5ghz in geekbench 3 64bit. You'll be limited to CPU strap or blck OC until asus fix their IvyB-E BIOSs to allow for proper multiplier OC.

I heard that CPU strap isn't too bad, so just downclock memory, and up the strap. bclk would be a no no for me I guess. But of course I am also a newb with OC so still learning Plus I want a really stable machine since this is still a workstation and I have also lots of RAM. But well a bit more perf wouldn't be bad, especially if I compare this:

 

http://browser.primatelabs.com/geekbench3/199856 (new hackPro) to this: http://browser.primatelabs.com/geekbench3/197861 (old Mac Pro, nearly 6 years old!)

 

The new machine only really dominates the old one when it comes to memory performance. But I actually think some "real tests" would be more interesting

 

My noctua outperforms most budget closed loop liquid coolers

 

I believe that! Plus, they are much more silent, so I am glad I went this route. Now, if I only could solve the fan problem with the R4E ...

[zzerg]

Hi guys, I am in a similar boat here - have to replace my 6-year old dual Xeon hack pro for scientific computation.  

XEON E5 v2 | ECC | REGISTERED are musts for me, so even workstation boards like GA-X79S-UP5-WIFI don't fit the bill, because they don't take registered dims (some do ECC unbuffered)

 

Please advice for a C60x server motherboard. I am looking at Supermicro X9SRA and Gigabyte GA-6PXSV4 (single CPU)... I could go dual CPU, but the second CPU is not immediately necesary.

 

RampageDev, you have my deepest respect for your work/blog; I saw there lots of X79 boards supported (DSDT). Any chance you'll help me with a Supermicro C60x DSDT?

 

Thanks

[frankiee]

Why not take a P9X79 WS or P9X79-E WS? Takes XEON CPUs and ECC RAM as well. Should run fine in OS X and is supported by RampageDev's SSDT method.

[zzerg]

frankiee, the asus boards take the XEON and ECC, but not REGISTERED memory (they take ECC/NON-ECC unbuffered memory)

 

http://www.tomshardware.com/forum/48545-30-difference-registered-unbuffered

[frankiee]

Ahh that's too bad. Well, out of the two you have mentioned I'd probably take the Gigabyte one. I think C60X chipsets should work, but especially supermicro boards seem to be used not very often for running a hackintosh. Maybe it would help to find someone who has gotten OS X to run on one of these boards, or else you will play lottery a bit. But it should work .... in theory. If it will work 100% that's another question. For example, power management still does not work 100% with certain CPUs.

[zzerg]

Maybe I should wait a bit for Apple to release the 2013-Mac-Pro, then many things will clear up. I bet the new mac pro uses C60x chipset, supports registered ram, and hopefully (fingers crossed) brings OSX power management on C60x/X79 boards.

 

I am torn between Gigabyte - easier BIOS, more mackintosh support; and Supermicro - more reliable. An IT friend told me Gigabyte boards are a bit of a lottery. especially on electrical circuitry. I will also look into Asus C60x server boards.

[frankiee]

I think any C60X boards should work, however the devil could be in the details ....

 

Power Management is expected(!) to be working with all X79/C0X chipsets as soon as an updated OS X build for the new Mac Pro is out.

 

BIOS wise, I personally like the ASUS more than the Gigabyte ones, so it might be worth a look.

[Rampage Dev]

Maybe I should wait a bit for Apple to release the 2013-Mac-Pro, then many things will clear up. I bet the new mac pro uses C60x chipset, supports registered ram, and hopefully (fingers crossed) brings OSX power management on C60x/X79 boards.

 

I am torn between Gigabyte - easier BIOS, more mackintosh support; and Supermicro - more reliable. An IT friend told me Gigabyte boards are a bit of a lottery. especially on electrical circuitry. I will also look into Asus C60x server boards.

 

Why do you believe lies about Gigabyte? They are no different the any other manufacture when it comes to installing Mac OS X. 

 

X79 is just a C600 chipset with the SAS part removed. Nothing more. ECC has nothing to do with the Chipset but the CPU installed. Power management is already here but is limited unit the new Mac Pro comes out. 

 

Please read and become educated and not a mindless person who is misinformed and believes lies when it comes to Hacks.