Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
MacVidCards, Thanks for your insights.

... .

As far as why the AA is BOOSTING frame rates, I can only guess.

I have noticed inconsistencies using the Uningine benchmaks where AA doesn't seem to slow down the speeds, whereas it used to be a boat anchor that would sink fps.

Conceptually it feels like the part of core that handles this is unused unless you turn this function on.

Also, how do you have the global settings for FSAA etc?

The Nvidia control panel can trump anything set in renderer or let program choose.

Here is a pic of my Control Panel AA settings.
 

Attachments

  • NvdiaControlPanelTopCapture.JPG
    NvdiaControlPanelTopCapture.JPG
    45.8 KB · Views: 212
That's unbeleivable, it's as if the second half of the card does almost nothing when in SLI mode!

All this talk of how much better GPUs run in Windows has me doubting whether I'll bother with OSX for the video/AE/C4D workstation for much longer... (

I'm in the midst of making that move with respect to some of my visual content creation applications and, so far, I have no regrets. But if one doesn't want to bother with having more control over how his/her systems perform, what those systems can do, and what he/she can do with them, he/she shouldn't take the leap. Like the debate involving hackintoshes vs. true Macs, Windows doesn't float everyone's boat. Nor does it float all of my boats. I have hundreds of mechanical tools in my tool chests. Like those wrenches, saws, screw drivers, hammers, etc. in my tool chests, a computer system, to me, is just a more sophisticated, electronic tool. I still use my Ataris and Commodores for the things that they do best (music and 2d animation/effects). I still use my Linux systems for servers and special heavy duty jobs. My Macs remain my systems of choice for some productivity applications and my Macs help to keep my music riding the waves. No sad face - but rather smile like this :) from now on. Also, it doesn't have to be a complete and final divorce from Macs. Who knows, you might become a system polygamist just as I am.;) Love shouldn't be bestowed on inanimate objects; nor can they, nor will their corporate creators, love you back. But I recognize that, as to you, I'm just preaching to the choir. You know to whom you should give your eternal love and I congratulate both of you in that upcoming union and will always wish the best for you two +++.
 
Last edited:
Thanks and let me know if there's any test you have in mind for Premiere that I can run to compare to your system. I may have some control in the BIOS over the CPUs that is unavailable on a real mac pro.

The forever question of "more cores or more GHz?" still exists. In the Hardware section of the Adobe Premiere Pro forums, I asked this question to try and get some insight from the other hardware builders there.

Primarily, my Premiere Pro editing consists of absorbing large, spanned AVCHD clips from my 2 cameras, flipping one of the horizontally (it's a mirror), cropping it, scaling it 50%, syncing the audio from the 2 cameras up with a WAVE file, and then exporting it to 720p MP4. All of the effects I work with in Premiere are GPU-bound, so they benefit from fast video cards.

The question is: should I throw 2 10-core Xeons at it, or a single Core i7 overclocked? As Bill points out in that thread: previous versions of Premiere didn't scale as well as newer versions with lots of threads.

Harm Millard is quoted in that thread, and he's considered the hardware expert. I don't always agree with his opinions on things (and have told him as much... in my oh-so-subtle way... ;-)) but he does produce good information and a lot of testing. What he's found up to and including version CS6 is that: IF you involve the GPU, then more GHz, not cores, are needed. In other words: you need a faster CPU to get the data to and from the GPU(s).

That sorta makes sense. AVCHD, it seems, isn't all about raw number crunching. Meaning a Geekbench winner might not be the right solution for that kind of footage.

Do you have a camera that can produce 1080p AVCHD footage? Set it to record 20 minutes of... something. Even if it's just the wall in your office. Inhale that into Premiere Pro, add a GPU effect or two (horizontally flip it, for instance...). Export it via AME to a 720p h.264 MPEG4 file at around 8Mbps or so. Before doing so, delete the CUDA text file so that you're 100% certain your card or cards are being used by both Premiere and AME.

Do the same test after disabling cores. See if it takes significantly longer, or if there's any change at all.

Doing this with Premiere CC might be a better test than with CS6, for what it's worth. Specially since CC can use as many CUDA or OpenCL able cards as it can get its hands on.

I don't imagine this test taking a short amount of time. So if you have a "real job" ... focus on that, of course. :)
 
The forever question of "more cores or more GHz?" still exists.

Hi Jason, I may not be able to get my hands on footage directly from an AVCHD camera and after a quick search I don't think you can encode -to- this format. If you like I'll test some footage if you shoot a test clip, upload it and send me a link via PM. Let's say 5 minutes exactly to give a nice even sample for the benchmark - with thought to how many times I may repeat the benchmark ;)

Now I propose the following for the test:
I can disable a CPU in my machine no worries. I can then overclock this single 6c CPU to 4.0GHz and run the test. Then the second CPU can be enabled and 12 cores downclocked to 2.0GHz for the comparison. I will disable turbo boost to minimize any bias - a 6c 4GHz overclock with turboboost doesn't really match up to any processors on the market ;)

I know you didn't like the concept of comparing cores x GHz but as far as I can tell this would be the most ideal way to pit the two setups against each other - both matched at 6 x 4.0GHz = 24, 12 x 2.0GHz = 24.

I only have a GTX 570 at the moment - but I'll be buying a Titan soon which would definitely help out if you say Premiere takes great advantage of the GPU.

While I'm at it I will also run the test for a variety of codecs with and without effects as appropriate.
ProRes (Premiere CS6, Premiere CC)
DNxHD (Avid MC6)
Sony XDCAM
Native Canon 5D (Premiere CS6, Premiere CC)
R3D RAW (RedCineX, DaVinci Resolve)
Sony F65 RAW

Now the only problem at this point is I may not have time over the next few weeks to put all of this into action - as Tutor has alluded I've got a pretty big (if not the biggest) commitment coming up which I'm busy organising and preparing for. Then there's the typically associated vacation afterwards so I'll need to factor in time away for that too :)
 
Let's say 5 minutes exactly to give a nice even sample for the benchmark - with thought to how many times I may repeat the benchmark

Actually, one of the challenges with AVCHD footage is that: when it gets too long, it spans multiple files (file size limitation on SD cards with Fat32 formatting). When it spans, the ingestion of said creates a whole new set of problems that increases workload. Sometimes significantly. That's why I proposed a 20 minute clip. Shooting a 5 minute one and uploading it for you won't really help because the workload (the type, not the amount) will change.

Don't worry about it. Focus on your other commitment. :)
 
Could you guys give me some recommendations for the best/reliable PCI express SSD storage cards out there?
I work mostly on huge Photoshop files and also I would like to put my system/app folder files on this same drive.

Thanks
 
The Cheap Bad Man is still cheap,

See http://store.apple.com/us/buy-mac/mac-pro. Ouch!

Apple's 2013 Mac Pros start at $2,999. "The entry-level version includes a 3.7 Ghz quad-core Intel Xeon E5 processor, dual AMD FirePro D300 GPUs, 12 GB of memory, and 256 GB of PCIe-based flash storage. The higher-end 3.5 Ghz 6-core version with 16 GB of memory will starts at $3,999, and additional configure-to-order options offer 8-core or 12-core Xeon E5 processors, AMD FirePro D700 GPUs, up to 64 GB of memory and up to 1 TB of flash storage." [ https://www.macrumors.com/2013/10/2...-red-mac-pro-for-product-red-charity-auction/ ] I see no compelling reason for me to downgrade. "Aloha, new Mac Pro." I'm sticking with my ancients and my current and future self-builds. Nevertheless, I hope that the new Mac Pros are snatched up like hotcakes - I need the dividend distributions and appreciation.
 
Last edited:
Could you guys give me some recommendations for the best/reliable PCI express SSD storage cards out there?
I work mostly on huge Photoshop files and also I would like to put my system/app folder files on this same drive.

Thanks

I get all of mine from http://eshop.macsales.com/shop/SSD/PCIe/OWC/Mercury_Accelsior/RAID , running some on Macs and others on PCs. I'm delighted with their performance and reliability. But you might want to see what Google has to offer.
 
Patience isn't only a virtue, it can give you solid options.

Assuming that the CPUs in the 2013 Mac Pro are easily swappable by the users, and if I wanted to buy one with the intention of upgrading the CPU, then patience would be my partner. In the past, Apple has sold refurbished Mac Pros at a discount and don't forget that it'll take some time for those refurbs to start being sold. Moreover, it's been impossible to upgrade beyond a CPU family. By that I mean, e.g., Nehalem and Westmere CPUs are in the same family, just as the Sandy and Ivy Bridge CPUs are in the same family, but the Westmeres and Ivy Bridges are in different families. Some creative soul can figure out how to go from the first family member to the next, but not how to go from one family to another. It was done by allowing the move from Nehalems to Westmeres, but because Apple sat out on the Sandy Bridges it hasn't been an issue [at least until now because if my predicate assumption is true, i.e., the 2013 Mac Pro CPUs are user swappable, then one could buy a bottom of the line (and better yet - refurb) 2013 Mac Pro and install top of the line Sandy Bridge and Ivy Bridge Xeons in it because the ability to run the last family member subsumes the ability to run the first]. For example, you could upgrade a four core Ivy to a reasonably priced eight core Sandy if cash is tight (sell the Ivy to defray much of the cost of a Sandy that you buy from someone who wants to move from Sandy to Ivy - think about a Sandy E5-2687W and that there are enthusiasts and other owners who must (?) make the move to the Ivy - V2 of that chip. But for maximum upgradeability, why not wait until the Haswell Xeon Mac Pro refurbs hit the Apple store. Then you'd have the potential to upgrade even higher because you'd be at the beginning of the swing of the pendulum - giving you the option of going to V2 of that family + you'd give others and Apple more time to iron out any wrinkles with the Cylinder and peripherals and allow some of those prices to drop + you'd give others time to gain experience to assure you that the CPU swap can be easily done by you and show you how to do it properly. Look before you leap. Just my 2 cents - from the tweaker in me to the tweaker in you.
 
Last edited:
I get all of mine from http://eshop.macsales.com/shop/SSD/PCIe/OWC/Mercury_Accelsior/RAID , running some on Macs and others on PCs. I'm delighted with their performance and reliability. But you might want to see what Google has to offer.

Thanks. I heard some people complaining about OWC because they use Sandforce controllers and that seems to have a higher failure rate. That's why I was wondering about other vendors.
I have been using OWC for years and they seem very legit. I purchased couple years from them back a SSD and so far so good. But I have never had experience their PCI express SSDs.
 
Thanks. I heard some people complaining about OWC because they use Sandforce controllers and that seems to have a higher failure rate. That's why I was wondering about other vendors.
I have been using OWC for years and they seem very legit. I purchased couple years from them back a SSD and so far so good. But I have never had experience their PCI express SSDs.

There are not really any other options compatible with mac as far as direct PCIe storage. There are a couple of PCIe cards which allow connection of 6Gb/s SATA SSDs e.g. Apricorn and Sonnet Tempo
 
Building a top OctaneRender farm just got cheaper by using Titanators

Big Bang For Your Bucks!

On November 7, 2013, Nvidia will release the GTX 780 Ti (I call it, "the Titanator.") [ http://videocardz.com/47400/nvidia-geforce-gtx-780-ti-released-november-7th-699 ]. The Titanator will cost $699. The Titanator will out perform a Titan for most uses that most Titan purchasers need. The Titanator has the same number of cores as a Titan. Titanator's cores are clocked higher (Titanator - 900 / 950 GHz or 950 / 1000 GHz vs. 863 / 900 - Titan) and its memory is clocked higher (1750 MHz - Titanator vs. 1502 MHz - Titan ) than those of a Titan at factory settings [ http://videocardz.com/47287/nvidia-geforce-gtx-780-ti-3dmark-score-exposed ]. Titanator crushes Titan and the AMD Radeon R9 290X in 3d Mark [NVIDIA GeForce GTX 780 TI: X5204 (100%); NVIDIA GeForce GTX TITAN: X4924 (94.6%) and AMD Radeon R9 290X: X4664 (89.6%)]. Titanator can have various aftermarket cooling solutions [ http://videocardz.com/nvidia/geforce-700/geforce-gtx-780-ti ]. Titanator pushes the price of the GTX 780 to $499 and the GTX 770 to $399. Titanator is intended to compete against the AMD Radeon R9 290 ($449) and 290X ($549).

With the Titanator, the costs of building a Tyan based eight GPU OctaneRender monster has dropped dramatically ($699 * 8
= $5,592 vs $999 * 8 = $7992; difference = $2,400). That $2,400 can easily cover the cost of CPUs and remember because OctaneRender is totally GPU based whether you drop the least powerful pair of E5-2600s V1s or V2s in this:

[ https://www.superbiiz.com/detail.php?name=TS-B759F2T - Mfr Part Number: B7059F77AV6R-2T
Motherboard: S7059GM2NR-2T
Processor: Dual LGA2011
Support Intel Xeon E5-2600 Series Processor
QPI up to 8.0 GT/s
Support up to 150W TDP
Chipset: Intel C602 PCH
Memory: 24x DDR3-1600/ 1333/ 1066 DIMM Slots, Quad Channel, ECC, Supports up to 384GB REG Memory or up to 192GB Unbuffered Memory or up to 768GB Load Reduced Memory
Slots: 8x PCI-Express 3.0 x16 Slots, 2x PCI-Express 3.0 x8 Slots (one for Mezzanine Card), 3x PCI-Express 2.0 x1 Slots, 1x PCI Slot (32-bit)
SATA: 2x SATA3 Ports, 4x SATA2 Ports, Support RAID 0, 1, 5, 10
SAS: 4x Mini-SAS Ports, Support RAID 0, 1, 5, 10
Video: Aspeed AST2300 Graphics
LAN: Intel X540-AT2 10-Gigabit Ethernet Controller
Drive Bays: 6x 2.5" or 4x 3.5" SATA3/SATA2 HDDs
Ports: 2x USB Ports, 1x VGA Port, 1x RJ45 LAN Port, 1x RJ45 Dedicated IPMI LAN Port
Form Factor: 4U Rackmount
Power Supply: 2400W 80 PLUS Platinum Certified Hot-swap (2+1) Redundant Power Supply; AC Input - 100-127V / 200-240V
System Cooling: 6x 120mm Hot-swap Fans
Front Panel:
Buttons: RST, ID, PWR w/ LED
LEDs: HDD, 2x LAN, ID, IPMI/Warning
Dimensions (WxDxH): 17.24 x 30.31 x 6.93 inch / 438.0 x 770.0 x 176.0 mm
Gross Weight: 35.0 kg
RoHS Compliant]
for Price: $4,649.99

or

the least powerful pair of X5500s or 5600s in this

[ https://www.superbiiz.com/detail.php?name=TS-B715V2R Mfr Part Number: B7015F72V2R
Motherboard: S7015GM4NR
Processor: Dual LGA1366
Supports Intel Xeon 5500/ 5600 Series Processors
QPI up to 6.4 GT/s
Support Up to 130W TDP
Chipset: North Bridge Intel 5520 & South Bridge Intel ICH10R
Memory: 18x 240pin DDR3-1333/1066/800 DIMM Slots, Triple Channel, REG/Unbuffered, Max Capacity of 144GB
Slots: 8x PCI-Express 2.0 x16 Slots, 2x PCI-Express 2.0 x1 Slots, 2x PCI-Express x16 Slots (run at x4), 1x 32-bit PCI Slot
Video: Aspeed AST2050 Graphics
LAN: Intel 82574L Quad-Port Gigabit Ethernet Controllers
Drive Bays: 2x 2.5" Hot-Swap SATA2 HDD bays
Ports: 2x USB 2.0 Ports, 1x VGA Port; 4x RJ45 LAN Ports
Form Factor: 4U Rackmount
Power Supply: 2400W (2+1) Redundant Power Supply, AC 100~240V, w/ PFC
System Cooling: 3x 12cm fans
Front Panel:
Buttons: RST, NMI, ID, PWR w/ LED
LEDs: PWR, HDD, Warning
Dimensions (WxDxH): 17.24 x 27.96 x 6.93 inch / 438.0 x 710.0 x 176.0 mm
Weight: 27.0 kg
RoHS Compliant ]
for Price: $3,616.99


won't make any difference.

So, you can get a barebones LGA 2011 renderfarm - all in one chassis for $10,241.99 (US) or a barebones LGA 1366 renderfarm - all in one chassis for $9,208.99 (US) - without taking into account shipping/taxes. Just add from $1K to $1.2K for ram and another $600 to $5.3K for CPUs (that depends totally on which system you get and on what CPUs you chose for other purposes since the least powerful ones would even work best for OctaneRender only use - less power draw for CPUs => more power available for GPUs), an OctaneRender Combo License (right now - 493.95 US Dollar), and whatever appropriate Windows OS ($100 for OEM Windows 8 and Server/Workstation - $700+) and storage you want ($600 - $1k+). Each Titanator has the 3d rendering performance of 10 E5-2687W V1; so eight Titanators has the rendering performance of 80 E5-2687W V1. I'm talking about (pre-shipping and taxes) spending about $12.5K (or more, depending on what others uses you need for the Tyan) for a system that'll out 3d render a fully loaded nMP by a factor of greater than 60X */ (not even accounting for your ability to overclock the GPUs for even better performance in Windows with EVGA Precision X) for a smaller or even cash outlay. That's big bang for the buck!

*/ 1) Top of line nMP may have a single E5-2697; 2) a single E5-2697 = 2.7 GHz * 12 cores = 32.4 GHz; 3) a single E5 2687W V1 = 3.1 GHz * 8 = 24.8 GHz; 4) 24.8 GHz (E5-2687W)/ 32.4 Ghz (E5-2697) = 0.77; 5) a single Tesla K20X has the rendering potential of ten E5-2687Ws [ http://www.nvidia.com/content/tesla/pdf/Tesla-KSeries-Overview-LR.pdf ]; 6) a Titan is faster than a Tesla K20X at 3d rendering; 7) 10 x 8 Titans > 80 E5-2687Ws; and 8) .77 * 80 = 61.6 E5-2697s.
 
Last edited:
Is this an actual Mac? Or did one of Tutor's beasts get loose again?

Image

That's not one of mine. It's a Yeti - now you see him, but soon you won't - because the Geekbench mods will soon vanquish that fake.

----------

I've coined a new measure. I call it "TE." "TE" stands for "Titan Equivalency." Since I use my Titans most for 3d rendering, I've defined TE in relation to OctaneRender's current Bench Mark Scene (in seconds), using Barefeats test found at [ http://www.barefeats.com/gputitan.html ] as a base. Here's how I'll be using TE:

Since the Titan that Barefeats used took 95 sec. to render that benchmark scene, (1) one GTX 680 that takes 190 sec. to render that scene gets a TE of .5, or (2) 2 GTX 680s that together render that scene in 95 seconds get a total TE of 1, or (3) 8 Titans that each render that scene in 95 sec., each get a single TE of 1, but combined they have a TE of 8, or (4) as in the case of my GTX 690 (tweaked) that renders that scene in 79 sec., it gets a TE of 1.20 [ 95 / 79 = 1.20253164556962]. Remember to keep the following in mind as number of GPUs rendering together exceeds two. In OctaneRender, your performance will scale linearly when using the same model GPUs with the same setting. How this works with only two GPU is easy - if one renders the scene in 200 sec., then the both of them will render the scene in 100 sec. But here's were the little tricky part starts: If you want to render the scene in 50 sec., it'll take twice the number of GPUs that it took to render the scene in 100 sec. That means it'll take 4. If you want to render the scene in 25 sec., it'll take twice the number of GPUs that it took to render the scene in 50 sec. That means it'll take 8. So if you keep this math in mind and just mull it over a little, it'll begin making perfect sense.
 
Santa cometh early this year to those with PCIe slots that need filling.

The GTX 780 Ti has 960 FP64 CUDA cores; 2880 FP32 CUDA cores (more cores of each type than factory GTX Titan and GTX 780); 5 graphics processing clusters; 15 streaming multiprocessors; 240 texture mapping units; at 993 MHz - geometry = 5 billion triangles/s and texture fillrate of 238 GTexel/s; 48 raster operating units; 3GB GDDR5 memory clocked at 7 GHz on a 384-bit interface = 336 GB/s bandwidth; TDP of 250 watts - the same number as the GTX TITAN and GTX 780; Base Clock of 875 MHz/Boost Clock of 928 MHz; power balancing technology to improve overclocking capabilities; single precision floating peak performance of 5.720 TFLOPS; two power connectors - 8+6 Pin; and at a price of $699. [ http://videocardz.com/47587/nvidia-geforce-gtx-780-ti-official-gaming-performance ]

One more thing: EVGA is preparing a 6 gig model [ http://videocardz.com/47611/evga-preparing-geforce-gtx-780-6gb-ram ].

I'm sticking with calling this new card - "The Titanator." Now I must plan for the GTX card pushdown and getting a few more OctaneRender licenses.
 
Last edited:
Funny that Swampus mentioned that Geekbench-beast!

I'm currently setting up a small but capable (for my needs) and reasonable :)D) priced PC and was browsing the 4930K geekbench scores. Was about to ask Tutor how that underclocking works exactly and if a noob can do that too. A score of 98.000 sure looks way more attractive than 18.000 or whatnot with the latest nMP 6core model.

Anyway, wasn't sure where to ask some specific questions work-related builds since those PC forums I found discuss not much more than the latest gaming rig config.

----------

The GTX 780 Ti has 960 FP64 CUDA cores; 2880 FP32 CUDA cores (more cores of each type than factory GTX Titan and GTX 780); 5 graphics processing clusters; 15 streaming multiprocessors; 240 texture mapping units; at 993 MHz - geometry = 5 billion triangles/s and texture fillrate of 238 GTexel/s; 48 raster operating units; 3GB GDDR5 memory clocked at 7 GHz on a 384-bit interface = 336 GB/s bandwidth; TDP of 250 watts - the same number as the GTX TITAN and GTX 780; Base Clock of 875 MHz/Boost Clock of 928 MHz; power balancing technology to improve overclocking capabilities; single precision floating peak performance of 5.720 TFLOPS; two power connectors - 8+6 Pin; and at a price of $699. [ http://videocardz.com/47587/nvidia-geforce-gtx-780-ti-official-gaming-performance ]

One more thing: EVGA is preparing a 6 gig model [ http://videocardz.com/47611/evga-preparing-geforce-gtx-780-6gb-ram ].

I'm sticking with calling this new card - "The Titanator." Now I must plan for the GTX card pushdown and getting a few more OctaneRender licenses.

Interesting. I'm not really on the fence to buy such a beast (using Maxwell, think I wouldn't really benefit from OpenCL as well as CUDA) but they sure are sexy to look at - spec wise. :D
 
This is beyond sexy!

Funny that Swampus mentioned that Geekbench-beast!

I'm currently setting up a small but capable (for my needs) and reasonable :)D) priced PC and was browsing the 4930K geekbench scores. Was about to ask Tutor how that underclocking works exactly and if a noob can do that too. A score of 98.000 sure looks way more attractive than 18.000 or whatnot with the latest nMP 6core model.

Anyway, wasn't sure where to ask some specific questions work-related builds since those PC forums I found discuss not much more than the latest gaming rig config.

----------



Interesting. I'm not really on the fence to buy such a beast (using Maxwell, think I wouldn't really benefit from OpenCL as well as CUDA) but they sure are sexy to look at - spec wise. :D

Would love to get a bevy of iChill GeForce GTX 780 Ti HerculeZ X3 Ultras:
 

Attachments

  • Of-Inno3D-d-s-Three-GeForce-GTX-780-Ti-Cards-Two-Are-Overclocked-and-Custom-Cooled-398244-6.jpg
    Of-Inno3D-d-s-Three-GeForce-GTX-780-Ti-Cards-Two-Are-Overclocked-and-Custom-Cooled-398244-6.jpg
    214.2 KB · Views: 315
Now you nMP owners can get a 780 Ti.

For about $4,400, you can have a 4-core nMP with room for storage & three PCIe slots. Then add $700+ to cover the cost of the 780Ti and you're done.

Magma has launched three Thunderbolt to PCI-e expansion chassis.

The one that retails for $1,399 has a pair of 5.25-inch bays, and room for three full-length, full-height PCI-e slots. Another one that retails for $1899 has the three PCI-e slots and eight 2.5-inch SATA drive bays. This functionality is installed in a rack-mountable aluminum chassis that has a 540-Watt power supply. That's more power than it appears that your nMP will be able to supply.
 
Update - Overclocking 101

Motherboards (MB) made by Gigabyte run multiple OSes best. You can overclock Xeons and i7s as follows:

Overclocking i7s
Nehalems and Westmeres (133 MHz base clock): yes - if motherboard (MB) bios allows it; if MB allows it, generally up to 30+% w/proper cooling = safe range (unless you start dropping the multiplier) for most chips, but mileage may vary (MMV) per chip.

Sandy and Ivy Bridges (100 MHz base clock): yes if MB bios allows it and you have a "K" series chip, but straps to hold down other DMI related functions, must work correctly to achieve significant overclock unless you rely predominately on increasing multiplier. Otherwise, you can overclock a non-K series by generally up to 7.5% (generally, cooling isn't an issue) with the right MB bios. But MMV per chip. Since Sandy Bridge intro, Intel locked the non-K chips down in virtually ever other way possible and tied together, with one clock, a host of functions to DMI, including things like PCIe. At about 8% and greater, other things, like your SATA drives, start to drop out because that increased clock rate is too great for them and your system won't boot.


Overclocking Xeons
Nehalems and Westmeres - Same as i7 Nehalems and Westmeres; except the good, dual+ CPU MBs w/tweaking options, other than the EVGA SR-2 MB, are few.

Sandy Bridges and Ivy Bridges (100 MHz base clock): Same as for non-k series i7 Sandy and Ivy Bridges. Note that there are no K-series Xeons or the equivalent. The good, dual+ CPU MBs w/tweaking options, other than the Supermicro DAX MB series, are few.

Observation
Inversely, I wouldn't buy a used i7 K series, or a used Nehalem or used Westmere w/o a good warranty because they could have been overclocked in dangerous ways/degrees. The non-K series i7s and the Sandy and Ivy Bridge Xeons cannot be overclocked to a level that necessitates special cooling or that overstresses the cores. So buying one of them used isn't as risky.
 
For the bakery chefs out there -

Here's a holiday recipe with some icing that carries a punch:

Take

1) a CPU like the $2,495 Intel Xeon Twelve-Core Processor E5-2697 v2 2.7GHz 8.0GT/s 30MB LGA 2011 CPU, Retail [ http://www.superbiiz.com/detail.php?name=E52697V2BX ];
2) a motherboard like the $229 GIGABYTE GA-X79-UP4 LGA2011/ Intel X79/ DDR3/ 4-Way CrossFireX&4-Way SLI/ SATA3&USB3.0/A&GbE/ ATX Motherboard [ http://www.superbiiz.com/detail.php?name=MB-X79-UP4 ];
3) a computer case like the $150 NZXT SWITCH 810 No Power Supply ATX Hybrid Full Tower Case (Gunmetal) [ http://www.superbiiz.com/detail.php?name=CA-SW810G ];
4) a PSU like the $284 LEPA G Series G1600-MA 1600W 80 PLUS Gold ATX12V v2.3 & EPS12V v2.92 Power Supply [ http://www.superbiiz.com/detail.php?name=PS-G1600MA ];
5) a CPU cooler like the $116 NZXT KRAKEN X60 280mm All-In-One Liquid CPU Cooler for Intel LGA775/1150/1155/1156/1366/2011 and AMD Socket AM2/AM2+/AM3/AM3+/FM1/FM2 [ http://www.superbiiz.com/detail.php?name=FAN-X60 ];
6) a set of memory like the $900 CORSAIR Dominator Platinum 64GB (8 x 8GB) 240-Pin DDR3 SDRAM DDR3 2133 Desktop Memory Model CMD64GX3M8A2133C9 [ http://www.newegg.com/Product/Product.aspx?Item=N82E16820233363 ];
7) a GPU, or two or three or four, like the $730 EVGA 03G-P4-2883-KR GeForce GTX 780 Ti Superclocked 3GB 384-bit GDDR5 PCI Express 3.0 SLI Support Video Card [ http://www.newegg.com/Product/Product.aspx?Item=N82E16814487001 ];
8) a couple of SSDs like the $517 Crucial M500 2.5 inch 960GB SATA3 Internal Solid State Drive (MLC) [ https://www.superbiiz.com/detail.php?name=960GM5SSD1 ];
and
9) your OS(es) of choice

and spend a Saturday mixing these together.

For icing, use the features of your motherboard's bios to overclock that memory and that Xeon by about 1.075 to realize a base speed of 2.90 GHz (rather than 2.70 GHz), turbo stage 1 = 6 to 12 active cores at 3.23 GHz (rather than 3.0 GHz), turbo stage 2 = 5 cores at 3.33 GHz (rather than 3.1 GHz), turbo stage 3 = 4 cores at 3.44 GHz (rather than 3.2 GHz), turbo stage 4 = 3 cores at 3.55 GHz (rather than 3.3 GHz), turbo stage 5 = 2 cores at 3.66 GHz (rather than 3.4 GHz), and turbo stage 6 = 1 core at 3.76 GHz (rather than 3.5 GHz).

$2495 + $229 + $150 + $284 + $116 + $900 + $730 + $730 + 517 + 517 = $6,668 w/o the OS(se), shipping and taxes, if any. Um, um good!

If you're on a CPU cost reduction diet, just substitute the E5-2697 v2 with a less costly one from the list at the bottom of this page: [ http://www.cpu-world.com/CPUs/Xeon/Intel-Xeon E5-2637 v2.html ] and you can always mix and match with other ingredients too.
 
Thanks for the ***** menu! (No insult intended! :D)

Question: I'm about to build my first small rig myself and do wonder about the CPUs at most. Using Maxwell Renderer paired with Rhino 3d for modelling and the usual Adobe PS/Illustrator combo - not for professional work as one may call it since I don't earn any money from it, but it is directly connected with my professional work, fwiw. Throw in CAD and a quick game here and there, that's about my usage scenario.

Problem is, I really have no idea whether a 4930k is better, or maybe a E5 2643 v2 on a dual board which I can upgrade in a year or so to a 2x6core..and on and on..
I really can't find a lot of info about the hyperthreading. How does a E5-2650 v2 with 2.6 base clock and a turbo freq of 3.4 perform over a longer period (say 3-4 hours) in single threaded progs? Permanently?

Part of my problem is that most sites/forums I visit/found talk about BF4 optimized rigs or are far above my level.

Can you help me out, Tutor? :D Hope I could explain my problem somehow understandable..
 
I thank you for sharing this, my wallet does not ;) Really struggling to decide between the 2643 and 2687, whether the 2687 is worth 50% more £.

Did you find any benchmarks regarding scaling? I looked at the same two CPUs as well but they are so darn expensive..if the prospect of prices coming down quite a bit at some point a second 2643 later on could be much cheaper than a single 2687 upfront..
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.