Honestly, I would wait for the new C2 stepping to have been shipped for several weeks before buying any parts for LGA 2011. (The reason for such a large window is that there will be a period in which both C1 and C2 steppings are sold concurrently, and that the reseller does not specify the stepping at all.) The currently available parts are all of the C1 stepping, which has a few bugs.
If you must buy parts to build a new system right now, stick with the LGA 1155 i7-2600K or i7-2700K ($300 to $350) plus a Z68 motherboard ($150 to $200) and 16GB (4 x 4GB) of RAM (which should be available for less than $100). Although 32GB (4 x 8GB) kits are coming down in price, they still cost in the $300-ish range right now.
Is there a way to see on the box if the processor is C1 or C2 stepping?
Only the Intel S-Spec will reveal the stepping. The S-Spec is a five-letter code that starts with "SL". You try to match the S-Spec code with Intel's database. However, it is currently moot since all of the SB-E CPUs currently for sale are of the C1 stepping.
And many resellers will charge a relatively hefty restocking fee even for exchanges or not accept any returns whatsoever on CPUs even if the box is unopened.
this is the second time i have seen this what are these supposed bugs? we have not had any issues with the 3960 or 3930
you means the virtual nonsense which does not mean a hill a beans to video???
This is news to me, but I found this report after searching based on this thread:
I have a 3930K system being built right now, and would be interested in what the affect of this issue might be?
none at all
That bug affects only those people who run multiple hardware VMs. Video editing should not be affected by the bug.
So...why did you recommend against purchasing now, if it won't affect us?
Actually, Jim, I did not recommend against buying a 3930K or 3960X per se. It's just that I have some reservations about those CPUs (at least in their currently available stepping) that affect those people who run hardware-based VMs. And there are some people who also use their NLE systems to also run virtual machines.
Secondly, the current stepping of the X79 chipset has most of the goodies disabled by Intel at chip manufacturing level. As originally promised, the X79 chipset should have had PCI-e 3.0 support, Intel SRT SSD caching and 10 SATA 6.0 Gbps ports. But on all currently shipping boards with that chipset, the PCI-e 3.0 support and SRT are hard-disabled along with eight of the 10 SATA 6.0 Gbps ports, so that in the end the X79 offers little over the older X58 chipset besides the PCI-e x1 and x4 slots now running at full PCI-e 2.0 spec (the ICH10R used with X58 was only PCI-e 1.1 compliant).
"I would wait for the new C2 stepping to have been shipped for several weeks before buying any parts for LGA 2011"
That's a pretty clear recommendation against buying now. And that recommendation wasn't qualified to those running multiple VM's (which are probably a very, very small percentage of editors out there).
all currently shipping boards with that chipset, the PCI-e 3.0 support and SRT are hard-disabled
Not sure that's true. According to my reading, the PCI-e 3.0 support is there and functional, but simply not validated as there was no hardware to test it on before shipping. Intel left it up to the mobo makers to do the validating. And board makers are doing just that, as there are several models available using PCI-e 3.0.
I haven't looked into SRT as it's not a feature I'm especially interested in.
I was recommending "against" buying such a CPU at that point - until I did a bit more research.
Anyway, the first "real" results from i7-39xx series systems were posted on the PPBM5 site (prior to today, there were three entries of that CPU type - one from a system that's obviously hobbled by the first release of Premiere Pro CS5, version 5.0.0, while the other two were systems submitted by ADK itself). It turns out that even with 32GB of RAM, the i7-3930K needs to be heavily overclocked just to outperform an i7-2600K or i7-2700K (also with 32GB of RAM) at the same overclocked GHz overall in the PPBM5 benchmarks. At stock clocks, the i7-3930K isn't sufficiently faster than an i7-2700K to justify spending the extra $400 to $500 (including motherboard), particularly if a would-be user would be running no more than five disks (of which four of them are in two 2-disk aid0 arrays).
And yes, LGA 1155 (i7-2600K or i7-2700K) does perform better with 32GB than with 16GB, as far as CS5.5 is concerned. I only hope that the prices for 8GB DIMMs drop further: Right now, they still cost about 50 to 75 percent higher per GB than 4GB DIMMs.
Getting back to the virtual machine issue specifically, I am afraid I don't know what one of those is, but I am looking at Shadow Protect, and the video guide on their site makes constant refferences to virtual disk, and virtual machines. Is this the same thing and would a 3930k system have problems running Shadow Protect?
until I did a bit more research.
Ah. That clears it. Thanks.
At stock clocks, the i7-3930K isn't sufficiently faster than an i7-2700K to justify spending the extra $400 to $500
Interesting. You seen that in any tightly controlled tests? (Which PPBM is not.)
I have found that so-called "controlled" tests severely overexaggerated the differences between the CPUs (this means that the difference in controlled tests between the CPUs is far greater than it actually is in real-world performance). In these particular systems, the disk system and the GPU are the two biggest bottlenecks in MPE performance. And in this case, I was speaking "overall" performance, not performance in individual portions of the tests.
That said, the LGA 2011 systems seem to excel in the H.264 HD encoding tests but fail to distinguish themselves from the rest in the other PPBM5 tests. And in the case of the 223-second result by an i7-3960K overclocked to 4.1GHz, it is clearly hobbled (as indicated by the relatively slow MPEG-2 DVD encoding time) by the slow but very expensive Quadro 5000: It is based on an underclocked GeForce GTX 465 with a 320-bit memory bus and 2.5GB of RAM - and no amount or bandwidth of VRAM could compensate for the deficiencies of the GPU itself.
Just wait a bit for the new i73820 4cores 10 MB L3 cache (THE KILLER OF THE OLD GENERATION OF SANDYB). that in fak will cost the same as the i72700.:
Why you have to considerate to buy a X79 plataform:
1-Some are capable to hand 128GB RAM.
2- Quad chanell¨
3- Whit RAM cache and RAM disk cache as much RAM is on your board mor possibilities of improvising.
I guees you can even preorder the i73820.
I have found that so-called "controlled" tests severely overexaggerated the differences between the CPUs
I'll take that as a no, then.
In the near future Id like to change my MOB and CPU.Now I have I72600 (not oc.) with Asus P8P67 Deluxe with 16GB (4*4GB) 1600Mhz Dual ch ram.Im using for AE and PP CS5.5 to edit long avchd footages, with color effects, and After effects multiple layer graphic intros in hd 1080p or 1080i.
The possibilities what Id like to order (what this topic mentioned) :
- i7 2600K with the new Asus P8Z68 Deluxe /GEN3 or
- i7 3930K with Asus P9X79 Pro.
The new P8Z68 GEN3 support 32GB RAM in 4 slot, and PCIE 3.0.
I dont know there will be big performance difference or not?
But I see there is a price difference, which is not negligible.I read lots of comment ,and now I dont know who knows the truth about these cpu-s, cause one side says the new SB-E is the best peformance now, but others says there are lots of deficiencies.And the 2600k looks a very stable and favorite cpu, and with this MOB I can use with 22nm cpu-s later, when PCI-E 3.0 will reach the market.
So what do you think guys?