All We Know About Maximizing CPU Related Performance

[Tutor]

..., I want to respect Tutor's page here for his work on CPU (not GPU) performance...

Thanks PunkNugget. However, it's within the scope of my post #1: … . "By using the phrase 'CPU Related Performance' I intend that this thread also include information regarding other modifications such as, but not limited to, better cooling, faster storage, video card upgrades for OCL assist and memory upgrades." … . "For all who are interested in this thread, what's new in CPU/CPU-GPU assisted performance and what might it mean to maximizing CPU performance?"

 

[PunkNugget]

Thanks PunkNugget. However, it's within the scope of my post #1: … . "By using the phrase 'CPU Related Performance' I intend that this thread also include information regarding other modifications such as, but not limited to, better cooling, faster storage, video card upgrades for OCL assist and memory upgrades." … . "For all who are interested in this thread, what's new in CPU/CPU-GPU assisted performance and what might it mean to maximizing CPU performance?"

No pressure. So to that I say: BRING IT ON WITH YOUR GPU SELF !!!

I'm just excited that things are working with the UD7 setup that I have and 480 GTX. Now if I could only get it to work on my HACKINBEAST that would be even better...

 

[DJenkins]

Thanks PunkNugget, going to take a look at those links tonight when I'm home from work.

 

[Tutor]

Change of Mind

.... . So my days of purchasing ATI cards are coming to an end.

Not! For OGL performance in Cinebench 11.5, my overclocked 4890s score about forty (40) fps higher per test than my overclocked Galaxy 680s and about ten (10) fps higher per test than my overclocked EVGA GTX690. Thus, since it appears that Nvidia and ATI cards excel at different functions I'm going to put one of each of them in my WolfPacks, other than WolfPackPrime0.

 

[PunkNugget]

As deconstruct60 cogently points out here: https://forums.macrumors.com/threads/1422732/ - "[F]ocusing on the CPU is misguided going forward. Currently and going forward there will likely be much more computational "horsepower" in two 200W PCI-e cards than there will be in the CPU "chamber" of a workstation. It isn't primarily just about the CPUs anymore and that is one of the problems with the current Mac Pro case design." This is especially so because of the CUDA cards. One GTX 690 renders 2.4x faster than 4 E5-4650s.

Currently, my render farm has 13,356+ overclocked CUDA cores (29,088+ GFLOPS) and 17,520+ overclocked stream processing units (26,160+ GFLOPS). Currently, the CUDA cores are the most useful. So my days of purchasing ATI cards are coming to an end.

But I remember talking to you at length about this when it came to using apps like AP and AE. I specifically said that using my 480 GTX would be better than the 4890, but you kept going back to the compatibility issue end of things (which there was with the nVidia cards), and you also mentioned that you didn't have much knowledge of the nVidia series. Seems like you do have some of that knowledge or at least know where to find it, because look at what you posted. I would have picked up another 480 (or 580) instead of those 4890 cards knowing more of this information that you're now sharing...

 

[Tutor]

But I remember talking to you at length about this when it came to using apps like AP and AE. I specifically said that using my 480 GTX would be better than the 4890, but you kept going back to the compatibility issue end of things (which there was with the nVidia cards), ... .

It's true that we talked at length about selecting video cards for your EVGA-based Hackintosh for which you had already installed an Nvidia card and tried to get that card working without success. So I recommended to you, after you had complained many times about the problems you were having with the Nvidia card in a Hackintosh, a card that I had no problems with. Moreover, I still do not have any problems with any of my nine (9) 4890s.

... and you also mentioned that you didn't have much knowledge of the nVidia series.

It's also true that at that time that I did not have much knowledge of Nvidia cards and I made that clear to you. However, that I might have lacked knowledge about something in the past (like Nvidia cards and their capabilities) is not an impediment to my acquiring more knowledge in the future.

Seems like you do have some of that knowledge or at least know where to find it, because look at what you posted.

Thanks for the recognition that an old dog can learn new tricks, but note that my post to which you refer never mentions the word "Hackintosh" which relates to your system on which you had sought my assistance.

I would have picked up another 480 (or 580) instead of those 4890 cards knowing more of this information that you're now sharing...

Although I created this post to share information when it becomes known, I now regret that my recommendations to you to solve your system's problems plead to me in many PM's and described in posts #'s, inter alia, 52, 53 ("I'm not yet sufficiently familiar with the EVGA GTX 580 to have constructive thoughts about it."), 95, 97, 98 (" I'm strictly an ATI man. … ."), 99, 100 - 105, 117, 119-121, and 270 - 273, have, like d00d's, left you dissatisfied. I am glad that you were resourceful enough to get your Hackingtosh working fast and properly. Moreover, I pray that you never have any more issues with it or the need for my help. Happy Thanksgiving to you too.

 

[PunkNugget]

Although I created this post to share information when it becomes known, I now regret that my recommendations to you to solve your system's problems plead to me in many PM's and described in posts #'s, inter alia, 52, 53 ("I'm not yet sufficiently familiar with the EVGA GTX 580 to have constructive thoughts about it."), 95, 97, 98 ("I'm strictly an ATI man. … ."), 99, 100 - 105, 117, 119-121, and 270 - 273, have, like d00d's, left you dissatisfied. I am glad that you were resourceful enough to get your Hackingtosh working fast and properly. Moreover, I pray that you never have any more issues with it or the need for my help. Happy Thanksgiving to you too.

First, you and d00d are two completely (polar opposite) people; at least when it came to extending your help and how you work with others (at least with me). So by no means am I going to allow you to compare yourself to d00d. Even though I did PM him and apologize for some of my comments to him, I've yet to hear back from him and I know that he checks his PM's. So now that's on him, not me...

Guys like you and I can get frustrated and even have attitudes but correct ourselves, apologize and are able to move forward.

Second, you've gone WAY BEYOND the help that I could ever ask for. It's just that I'm continuing to learn as you've mentioned in the past:

"...milage may vary."

and:

Trust your own judgment in the end, for "In lumine tuo videbimus lumen."

I hope when it comes to the posts that have been provided here and on here:

https://forums.macrumors.com/threads/1485401/

when it comes to the nVidia 480, 580, etc. will help those that need it when it comes to taking advantage of using CUDA and Mercury support. What I do know is that if you're using 10.7.0 - 10.7.3 for UC'ing it can be very difficult to use your 480 or 580 GTX, but if you have 10.7.4 - 10.8.2 you can use your 480 and 580 GTX. I might want to try out 10.8.2 on my SR-2 on a separate SSD to find out how that will fair when it comes to rendering purposes as I mainly built this machine to take advantage of rendering power using CS6. I know UC'ing won't work if I 10.8.2, but I will try it out on it's stock config using the 580 GTX and go from there.

Another thing that I'm finding out about is Telsa (which I don't know much about and need to dig into a bit more when it comes to using the apps that I currently have with CS6 and if they work at all).

But thanks for all the help Tutor (now and in the future). BTW, I too hope you had a great time with your family this Thanksgiving as well. Later...

 

[PunkNugget]

Hi BrainDeadFool, hope all is going well for you. When you have a moment I had a question... Later...

 

[PunkNugget]

Thanks PunkNugget, going to take a look at those links tonight when I'm home from work.

Hi D, Just wanted to see (when you have a moment) I sent you a PM. I had a question for you. Hit me up... Later...

 

[Tutor]

Clock tweaking of CUDA video cards has its advantages

Clock tweaking of CUDA video cards can improve their double- and single-precision Float performance. Within the near future, I hope to explore how this performance increase can be magnified, if any, by clock tweaking system bios parameters that affect throughput, such as system memory, QPI and, of course, the CPU. GT640 performance data after increasing (1) GPU offset by +150 MHz, (2) Memory offset by +310 MHz (CUDA-z indicates Memory clock rate increase of 1.46x) and increasing voltage to 1075 Mv, fan to 60% -> idle GPU temp of 43 degrees (C) and load temp of 56 degrees (C), yielding (a) 1.53x FurMark Score, 1.27x Tessmark Score and 1.19x GiMark Score and (b) 4.6034 Gflops [or 16%] improvement in Double-precision Float and a 65 Gflops [or 16%] improvement in Single-precision Float.

 

[PunkNugget]

Clock tweaking of CUDA video cards can improve their double- and single-precision Float performance. Within the near future, I hope to explore how this performance increase can be magnified, if any, by clock tweaking system bios parameters that affect throughput, such as system memory, QPI and, of course, the CPU. See posts #4 and 5 at https://forums.macrumors.com/threads/1495160/ for an example of how video card tweaks impact CUDA performance.

Well, I just did updated my SR-2 to 10.8.2 and wanted you to check out my Extra folder; pretty stremlined huh? Also, had my EVGA 580 GTX 3GB GPU optimized to have CUDA, Mercury and Fermi work at their optimal level and tried that OpenGL app link that you provided. Let me know what you think... Later...

 

[Tutor]

Well, I just did updated my SR-2 to 10.8.2 and wanted you to check out my Extra folder; pretty stremlined huh?

Much so.

Also, had my EVGA 580 GTX 3GB GPU optimized to have CUDA, Mercury and Fermi work at their optimal level and tried that OpenGL app link that you provided. Let me know what you think... Later...

Excellent performance.

 

[PunkNugget]

Hey All (especially for all us SR-2 Hackintosh owners):

Take a look at what RampageDev hooked up for me as a result of us talking about the onboard audio issue:

http://rampagedev.wordpress.com/2012/12/12/how-to-create-a-virtual-audio-device-under-mac-os-x/

I haven't had the time to if this works for you but if it does (and it should) Merry Christmas. Oops! I think I might get arrested for saying the word Christmas...

Also, he was the one who helped me with streamlining my Extra Folder down to just a few files...

He's also helped me optimize my NEW setup that is just screamin' when it comes to rendering (I'll be posting comparison results within the next few weeks):

MacHakPro2:
- ASRock Extreme9 Mobo
- i7-3930K CPU (not OC'd yet)
- GSkill 64GB RAM - 2400MHz @ 1.65 (model #F3-19200CL10Q2-65GBZ)
- EVGA 580 GTX 3GB GPU
- Silverstone ST1500W PSU
- Corsair H80 (for now) I will be using all Koolance Water Cooling parts (with the new CPU-380I) with Tygon tubing (once again).
- Silverstone FT02 Case

 

[DJenkins]

Pretty clever tip! Although as stated most people would still probably like to get actual sound coming out of their speakers eventually

A $10 USB sound card did the trick for me to start with, now I've swapped to a 6in/6out Edirol firewire device for recording my band's demos

I'm a fan of rampagedev's work, he's very dedicated to helping people and he persisted with me for a while but unfortunately my problems ended up not being DSDT related.

 

[PunkNugget]

Pretty clever tip! Although as stated most people would still probably like to get actual sound coming out of their speakers eventually

A $10 USB sound card did the trick for me to start with, now I've swapped to a 6in/6out Edirol firewire device for recording my band's demos

I'm a fan of rampagedev's work, he's very dedicated to helping people and he persisted with me for a while but unfortunately my problems ended up not being DSDT related.

About the USB Audio Stick, I couldn't agree with you more. The BEST one out there is the Siig USB SoundWave 7.1 Pro (Optical S/PDIF):

http://www.newegg.com/Product/Product.aspx?Item=N82E16829270004

You could get their cheaper one:

http://www.newegg.com/Product/Product.aspx?Item=N82E16812191020

but it doesn't have Optical S/PDIF Cable capability. MAN, when I hooked that thing up to my Logitech Z-5500's - HOLY COW the sound was INSANELY CRISPER AND BASSIER !!! And the surround capability - FORGET IT, BETTER THAN A THEATER !!! It's totally worth spending the $30.00 on that USB Audio Stick and buy an Optical Cable (if your speakers can use one). When I say that using an Optical Cable is the better way to go I'm not kidding. Anyway, enough of that. Now onto RampageDev...

Rampage is a good guy. He's helped me install every type of Mobo, CPU, GPU & RAM config out there - lickity split !!! He's fast, reliable, dedicated and is becoming a good friend. I like him. With the exception of few guys like us, there's not too many out there like that.

 

[DJenkins]

I guess this is related to our ongoing discussions regarding maximizing performance… so here goes.

I am one who has always been interested in having an all-in-one do everything box.

Now with our SR-2 builds we have 7 PCIe slots, which at first I thought was incredible. However, getting a hackintosh up to full functionality requires a lot of these slots to be used for things usually found onboard the mobo, or case of a mac pro.

For example on my machine:
Slot 1 - GPU
Slot 2 - inactive - covered by GPU
Slot 3 - empty, tight fit with GPU fans
Slot 4 - RevoDrive SSD
Slot 5 - Gig-E NIC
Slot 6 - Areca RAID 1213-4i
Slot 7 - Firewire 800 & 400

I also know there are people out there who like to have wireless and bluetooth cards installed.
Now I could possibly get a few PCIe extension ribbon cable things and use slots 2 & 3, but I think there's a more interesting option...

I remembered seeing a Mac Pro set up for Davinci Resolve colour grading with a Cubix box attached for multiple GPUs.
Doing a quick search I've also found this SR-2 machine with 8 (yes EIGHT) GTX 480 GPUs using another PCI expansion box manufacturer Magma.

Now they get pretty damn expensive if you want your expansion box to have 8x lane slots, but you could always move all your lower speed x4 devices externally and keep your SR-2 for the 8x slots.

So who's going to take the leap and try this?
Personally, on top of my current setup I'd go for a few GTX 580s, an AJA Kona video capture card, an OWC Accelsior PCI SSD etc. etc. haha

 

[Tutor]

I guess this is related to our ongoing discussions regarding maximizing performance… so here goes.

I am one who has always been interested in having an all-in-one do everything box.

Now with our SR-2 builds we have 7 PCIe slots, which at first I thought was incredible. However, getting a hackintosh up to full functionality requires a lot of these slots to be used for things usually found onboard the mobo, or case of a mac pro.

For example on my machine:
Slot 1 - GPU
Slot 2 - inactive - covered by GPU
Slot 3 - empty, tight fit with GPU fans
Slot 4 - RevoDrive SSD
Slot 5 - Gig-E NIC
Slot 6 - Areca RAID 1213-4i
Slot 7 - Firewire 800 & 400

I also know there are people out there who like to have wireless and bluetooth cards installed.
Now I could possibly get a few PCIe extension ribbon cable things and use slots 2 & 3, but I think there's a more interesting option...

I remembered seeing a Mac Pro set up for Davinci Resolve colour grading with a Cubix box attached for multiple GPUs.
Doing a quick search I've also found this SR-2 machine with 8 (yes EIGHT) GTX 480 GPUs using another PCI expansion box manufacturer Magma.

Now they get pretty damn expensive if you want your expansion box to have 8x lane slots, but you could always move all your lower speed x4 devices externally and keep your SR-2 for the 8x slots.

So who's going to take the leap and try this?
Personally, on top of my current setup I'd go for a few GTX 580s, an AJA Kona video capture card, an OWC Accelsior PCI SSD etc. etc. haha

Check out BHPhotoVideo to purchase a smaller/less expension external chassis for the single slot cards and use the other 3 dual slots for CUDA cards.

 

[PunkNugget]

Check out BHPhotoVideo to purchase a smaller/less expension external chassis for the single slot cards and use the other 3 dual slots for CUDA cards.

Here's some more info on the CUDA card thing:

GTX 580 has 512 CUDA cores, but it's not as straightforward. Kepler is a vastly different marchitecture to Fermi. Previously Nvidia's design philosophy was few high efficiency big shaders at high clocks (2 x core clocks). So, GTX 580's 512 shaders were 'hotclocked' at 1550 MHz. With Kepler Nvidia has adopted AMD's model - many small, dense not-as-efficient shaders at low clocks (AMD 7970 has 2048 shaders at 925 MHz). GTX 680 has 1536 shaders clocked at 1006 MHz. Kepler shaders are smaller and less complex, most notable FP64 has dropped to 1/24th rate versus 1/8th for Fermi. So, the numbers are not really comparable. However, what is up is FP32 and thus single precision FLOPS. Way up, at 90+%. FP16 texture fillrate is also up massively, though this has little relevance to CUDA. Memory bandwidth is identical, 192 GB/s, GTX 580 has a wider 384-bit memory bus. In more complex compute (CUDA/OpenCL/DirectCompute) applications GTX 580 is still faster. Perhaps Premiere Pro is not as complex, relying mostly on single precision brute force. Let's wait for benchmarks to see how well GTX 680 performs in Premiere Pro.

This bit of info came from a new found friend (RampageDev) as a result of a question that I had about the differences between the 580 GTX vs the Quadro 4000. Rampage also posted this (a Mac version that he created of MemTest):

http://rampagedev.wordpress.com/2012/12/14/how-to-run-memtest-in-mac/#more-1466

That's his online blog. Now Tutor may or may not have issues with this mac app that he put together since the PC version is more "reliable", but it from what he explains it works just as good. He's a smart guy and very generous with his time, BUT if anyone decides to use this man's brainpower skillz, then please be kind and pony up the contributory funds ($20 to $30) and he'll hook you up. He actually did me a solid and streamlined that Extra Folder that I posted for you guys and tweaked my 480 GTX and 580 GTX cards to work at their max performance to cut through the garbage of things that were blocking them from being utilized properly. If I haven't already mentioned this, well, I'm mentioning it now. He's really good at what he does and educates along the way. I think you'll like him Tutor (or anybody else for that matter) if you decide to reach out to him. He got some MAD SKILLZ...

PS - he NEVER demands the funds, but if a person with this kind of skill can get your system up and running properly (with the proper DSDT adjustments) within less than an hour or two isn't it worth it? Instead of having to go to TonyMacX86, InsanelyMac and whatnot to post your question of why your kernel panic happened when you installed your OS X or some new .kext file, and have to wait for days for the proper reply (if any)? He can pretty much decipher it (the kernel panic) very quickly and show you why it happened (teaching you how to read the code in that kernel panic) so you learn as you go (as he's done with me). You may not care to learn (as he will just fix it anyway), but it's an excellent way to catch on as he's helping you along the way, That's a big benefit for someone like me who's always wanting to learn. He's an excellent teacher!

 

[Tutor]

Alpha level - Splurge on the appetizers, but hold back on the full course.

Tweaking the GTX 580 GPU for Maximum, Safe Rendering Performance in Octane

Software Tools Used
(1) CUDA-Z-0.6.163 - used to measure CUDA performance and other GPU related parameters [ http://cuda-z.sourceforge.net/ ],
(2) GpuTest_GUI - used to benchmark certain GPU functions and ensure GPU stability [ http://www.geeks3d.com/20121113/gpu...gl-benchmark-furmark-lands-on-linux-and-os-x/ ],
(3) EVGA Precision X - used to over- and under-clock the GPU and affect a variety of the GPU's other settings [ http://www.evga.com/precision/ ],
(4) CINEBENCH 11.5 - Used to measure CPU and OGL performance and tune CPU and GPU for maximum performance [ http://www.maxon.net/products/cinebench/overview.html ],
(5) Geekbench 2 - Used to measure CPU and memory performance and tune CPU and memory for maximum performance [ http://www.primatelabs.com/geekbench/ ], and
(5) Octane Demo - used for GPU/CUDA based rendering [ http://render.otoy.com/downloads.php ]


Octane is owned by Otoy. Octane is a CUDA-based rendering solution that relies solely on the CUDA capabilities in Nvidia-based video cards. As such it is probably the fastest rendering solution now on the market. Since it uses only the GPU, the user can use his/her system's CPU(s) to performance any other tasks just as if Octane were not rendering away. But because Octane will max out one or more GPUs capabilities while its rendering, Otoy recommends that the user have another video card installed for interactivity and for doing other CPU bound tasks while Octane is doing its thing. At last, those GT 120's have a task made for their continued use. My MSI (aka WolfPack5) uses a GT 640 4-gig for interactivity and CPU bound tasks.

Octane has been in beta for a few years. Octane Render 1.0 Final was released last month. There are a variety of 3d packages for which some of the Octane forum members have created exporters. There are plug-ins in the works: there are betas for Maya and Poser and a finished version for 3d Max. For other 3d apps the plug-ins are in various pre-beta stages.

Rendering an animation in Otoy's Octane has two phases. The first phase involves perfecting the scene's look such as the lighting, materials/textures, camera(s) and object placement [ http://render.otoy.com/videos.php ]. This phase is much harder on the animator than it is on the system. The second phase involves the initiation of the renderer. This phase is very easy on the animator, but heavily taxes the GPU(s). During the first phase, I was able to overclock my two GTX 580s in my MSI 6-core Sandy Bridge 3930k system (total system cost, including 2 GTX 580s, under $2.25k), by about 6%, setting the the fans to run at 93% to keep the temperature at or below 50 degrees C. Since my CPU is overclocked to run at 4.13 GHz (non-turbo) and at 4.901 GHz (turbo), I set my GPUs' temperature max to 50 to avoid even the hint of any thermal issues. During the second phase, I underclocked my two GTX 580s by about 5%, setting the the fans to run at 70% to keep the temperature at or below 50 degrees C during render times of, at least, 3 hrs.

The first pic shows settings relevant to the first phase, related benchmarks and that all important small increase in CUDA capability to enhance interactivity. The second pic shows settings relevant to the second phase, related benchmarks and that all important small decrease in CUDA capability to keep everything cool. The third and fourth pics show CPU performance tests made while Octane was rendering. The last pic shows how the system benches under OSX.

 

[Rampage Dev]

Just wanted to stop in and say great work.

 

[PunkNugget]

Good job Tutor. My thing though, how was your GB score when OC'ing in Mac OS X 10.8.2? Also I see that you were able to to get the CPU@579MHz. If I'm not mistaken it appears that you were able to UC it, how? Especially since (from what you mentioned before) you cannot UC in 10.8.2...

 

[Tutor]

Good job Tutor. My thing though, how was your GB score when OC'ing in Mac OS X 10.8.2?

That 5th pic shows the GB score for ML 10.8.2.

Also I see that you were able to to get the CPU@579MHz.

Nope. Just how Geekbench 2 erroneously read it when it was overclocked to 4.13 GHz. I can still get it to turbo 5 bins, but only 5 bins.

If I'm not mistaken it appears that you were able to UC it, how? Especially since (from what you mentioned before) you cannot UC in 10.8.2...

I can underclock it. I just can't get it to increase the bins from idle to turbo. So - so far underclocking it just makes it slower. Underclocking won't be advantageous until I can figure out how to get it to turbo more than 5 bins above the underclock or MSI changes the bios to give me a working CPU strap higher than the current 1.67 CPU base strap that doesn't even work.

 

[PunkNugget]

That 5th pic shows the GB score for ML 10.8.2.



Nope. Just how Geekbench 2 erroneously read it when it was overclocked to 4.13 GHz. I can still get it to turbo 5 bins, but only 5 bins.



I can underclock it. I just can't get it to increase the bins from idle to turbo. So - so far underclocking it just makes it slower. Underclocking won't be advantageous until I can figure out how to get it to turbo more than 5 bins above the underclock or MSI changes the bios to give me a working CPU strap higher than the current 1.67 CPU base strap that doesn't even work.

You know, Both you and Rampage should work on this together as he has a TON of knowledge you should go to his site where his email is:

http://rampagedev.wordpress.com/premium-technical-support/

His email is on that page. Ask him about this issue that you're having to see if he can be of some help...

PS - Take a look at my new build I'm working on. Here's a pic of the largest Rad that I could find that fit PERFECTLY. It's a Magicool 540mm Radiator. The 120 mm radiator that's in the corner will give you some perspective on how big the radiator below it is. It takes up the whole bottom part of the case from end to end this you can see and takes three 180 mm fans. It's not like it hasn't been done before I'm just going to do mine a little bit differently that's going to create even more airflow that I'm looking for. I'll be posting this up pretty soon within the next week or so once I get all the parts in…

 

[PunkNugget]

I (again) just want to toss out some BIG-UPS to a great person who has not only taken the time to help (without asking for anything). This guy has done an amazing job helping me streamline my EVGA SR-2, ASRock Extreme9 & Gigabyte GA-X58A-UD7 mobo setup. He didn't ask for it but I contributed funds his way for each and every setup. TOTALLY WORTH EVERY PENNY !!! As I didn't have to search and wait for days or weeks (as I did in the past). He also provided a great DSDT Guide (that I'm providing for all to learn on their own). He's not only a good new found helper, but a great teacher and also a friend.

Again, if you don't want to wait DAYS or WEEKS to have others who post some kind of DSDT guide that DON'T have the desire to want to get back to you or help (and some of you may know who I'm talking about), then Rampage is a GREAT ALTERNATIVE; and again (like Tutor) also A GREAT TEACHER !!! Enjoy the guide as a gift during this Christmas season, later...

 

[Tutor]

... .a great DSDT Guide (that I'm providing for all to learn on their own). ... .

Rampage Dev and PunkNugget,
Thanks for the present to all of us. Happy holidays.

PunkNugget,
I'm disinclined to spend much more time tweaking WolfPack5 - the MSI - 3930k 6-core system. I do not expect that I can improve this system's performance by much more. WolfPacks 1 to 4 and 6 to 13, my Mac Pros (and yes, I'm hackintoshing [per netkas's forum] all of my old Mac Pros to give them new lives, but their hardware I'll leave in tack) and WolfPackPrime0, as well as my GPUs, will get the remainder of my free time these holidays. My Mac portables are next.