Skip navigation
smallchanger
Currently Being Moderated

Is it too early for someone to reccomend a build with one of the new Sandy Bridge CPUs?

Jan 5, 2011 11:32 AM

Hi all I need to build a new PC and was told to wait until the new Sandy Bridge CPUs were out.

 

This will be my first build but I'll have some help from my brother putting it together.

 

I'm coming from working in CS3 on Vista 32 and want to build a PC to run Windows7 64 and CS5.

 

As for a monitor I think I'm going to get a NEC MultiSync EA231WMi 23". The reason being it is a cheap IPS. The only other one I was looking at is a ViewSonic VP2365wb but the NEC has better reviews. If anyone has any other recommendations for an IPS in that price range, I'd be grateful.

 

For a case I'm going to go with a Silverstone SST-FT02B Fortress or a FRACTAL DEFINE R3.

 

As for the guts, I'm a little lost.

 

I was going to get a 60gb SSD to stick the OS on but I see Harm Millaard reccomending a Velociraptor over on this thread. I didn't know what one was but I looked it up and discovered it was a 10,000rpm 300gb hard drive that costs about $280. Is that right? Are there cheaper and smaller versions?

 

As anyone any reccomendations on what other Hard drives I should get and what RAID I should use? My budget for the whole build is mid range I suppose.

 

So that brings me to the CPU, GPU and MOBO.

 

Does anyone know enough about the new Sandybridge CPUs to reccomend one and which motherboard I shoud get? Do I need a seperate GPU still. People are talking about it having an integrated GPU.

 

I suppose I would like to have 12gb of RAM with an option to upgrade to 24. I initially thought 8gb would suffice but people on here seem to be using 12 or 24.

 

Thanks for any advice.

 
Replies 1 2 3 Previous Next
  • Currently Being Moderated
    Jan 5, 2011 11:49 AM   in reply to smallchanger

    About the only comment I can make at this time is that the first Sandy Bridge options will only be dual channel memory.  Your options would be 8 or 16, not 12 or 24.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 5, 2011 11:56 AM   in reply to smallchanger

    Forget about Sandy Bridge for CS5. You are much better off with a i7-950/X58 mobo combo.

     

    Why?

     

    1. The 2600K is about as fast as an OC'ed i7-920 @ 3.8

    2. The memory architecture limits you to only 4 DIMM slots, so either 8 or 16 GB

    3. The P67 mobo only has 16 PCI-e lanes, so you can't install a much needed raid controller

    4. The intrinsic graphics do not support MPE, so you still need a GTX-4xx/5xx card

    5. The touted turbo mode works only marginally with multi-threaded programs, like CS5

    6. It does not have PCI-e 3.0

     

    If you want Sandy Bridge, at least wait for the Sandy Bridge E and the X68 mobo.

     

    Then you have:

     

    1. Octo cores

    2. Quad channel memory and 8 DIMM slots (16 or 32 GB)

    3. 40 PCI-e 3.0 lanes and the architecture for a good raid controller.

     

    If you need a new system now, entry level is an i7-950, 12 GB RAM on a X58 mobo and the rest.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 1:29 AM   in reply to smallchanger

    I would definitely recommend buying the sandy bridge and not the x58 system.

     

    All of the reviews and benchmarks I have read say clearly that sandy bridge, 2600k with 4GB of memory, outperforms an x58 i7 950 with 6GB of ram by about 20%.

     

    Simply google search 2600k benchmark and read the first link provided for all the information on the topic.

     

    http://benchmarkreviews.com/index.php?option=com_content&task=view&id= 622&Itemid=63

     

    Even though the x58 sounds better on paper, the new architecture of the sandy bridge is far more efficient, and is better suited for media encoding and decoding.  So I would say that CS5 users would get more benefit out of the sandy bridge than video gamers.

     

    I am going to be building a sandy bridge system later this month with 8GB of ram, only populating 2 dimms.  Later this year I will upgrade to 16GB when the higher capacity 8GB individual ddr3 dimms become available (Red Camera minimum system requirement is 8GB of memory)

     

    The sandy bridge is better suited for the future as well, due to the fact that it has a higher threshold for overclocking.  I'm sure that this platform is capable of 5ghz with air cooling, definitely with water cooling.  This setup should last you at least a couple years, before you would want to upgrade.

     

    My core system will consist of...

     

    ASUS P9P67 Deluxe

    Intel Core i7 2600k

    Zalman CNPS9700 LED

    EVGA GTX 570 SuperClocked

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 1:52 AM   in reply to omnifathom
    All of the reviews and benchmarks I have read say clearly that sandy bridge, 2600k with 4GB of memory, outperforms an x58 i7 950 with 6GB of ram by about 20%.

     

    You have missed the most important one for video editing: PPBM5 Benchmark

     

    The conclusion is simple:

     

    1. An i7-2600K with 16 GB, overclocked to 4.7 is 6% slower than an i7-920 with 12 GB overclocked to 3.7

     

    2. An i7-2600K with 16 GB, overclocked to 3.9 is 21% slower than an i7-920 with 12 GB overclocked to 3.7

     

    3. An i7-2600K with 8 GB, overclocked to 3.9 is 50% slower than an i7-920 with 12 GB overclocked to 3.7

     

    Synthetic benchmarks do not tell the full story. The only sensible approach to judge an editing rig is by using very practical benchmarks, that reflect typical workflows of an editor.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 2:22 AM   in reply to smallchanger

    All these Gigabytes are good. They differ in the specs, 5.1 or 7.1 sound, single or dual NICs, PATA or not. Just compare them and decide what is relevant for you.

     

    The i7-950 is the correct one. A nice video card would be the nVidia GTX-460/470/570.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 5:11 AM   in reply to smallchanger

    Dear smallchanger:

     

    I have with me these:

     

    UD3->I7930 4GHZ Air.

    UD5->I7975 4.4GHZ Water.

    UD7 well the XTREME-> i7950 4.25GHZ water.

     

    If you aske me wich of them are the best choice for the price to me. I will tell you  the ud3.

     

    TRY THESE: UD3-> i7 950 4.1GHZ AIR Prolimatech. Corsair dominator 6x4 9-9-9-20 1600mhz.

     

    and ask Harm for the hhd config.

     

    The  you will see that sandy B processor are the best processor for play games, and that the I7 920, 930 OR 950 are the best bang for the buck.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 5:16 AM   in reply to Crist OC/PC

    the pic.

    Attachments:
     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 5:21 AM   in reply to Crist OC/PC

    the pic.

    Attachments:
     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 5:23 AM   in reply to Crist OC/PC

    Gigabyte GA-X58A-UD3,5,Xtremen.jpg

     

    Finally. =)

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 6:23 AM   in reply to Harm Millaard

    absolutely wrong wrong wrong

     

    its time i posted numbers to prove that the sandy bridge beats everything other than the highest Xeon...

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 6:31 AM   in reply to Scott Chichelli

    Hard to refute ADK's own results. Look at the results you have submitted. They are slower on all counts in the comparison I gave above.

     

    Your statement is absolutely wrong, wrong, wrong, unless ADK submitted wrong test results.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 7:03 AM   in reply to Harm Millaard

    common sense Harm,

    our not oced Sandy bridge ranked 13th beating any stock processor. (for some reason you have it listed as OCed)

    remove the absurbly overpriced Xeons that makes it 8th

    remove every OCed processor its now #1.

    and its #1 without the absurd 8-12 drive raid arrays... which most people have no need for.

     

    for the average user your recommendation like mine was the 950 stock 950 which ranks 40th is severaly beat by the 2600 not oced..

     

    so your replacement recommendation should be as mine is, the 2600 over the 950 all day long..

     

    anything less you are arguing with your own bencmark?

     

    but just for giggles i am having Eric resubmit with an SSD OS and a 8 drive raid the OCed to 4.7GHz system.

     

     

    Scott

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 7:27 AM   in reply to Harm Millaard

    "An i7-2600K with 16 GB, overclocked to 4.7 is 6% slower than an i7-920 with 12 GB overclocked to 3.7"

     

    Did you mean the 920 @ 3.7 is 6% slower than the 2600K @ 4.7? The Sandy at 4.7 is the next one up above the 920 so I assume you meant it the other way around. BTW the 2600 at 3.9 with 16GB was 12% slower than your 920 if you are using the combined results.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 7:36 AM   in reply to ECBowen

    Eric. Take out 1  of your ram in the MB so you can make the ppbm in the same condition.

     

    3.9GHZ, 12GBRAM, and let see the result. OR you can make your own BM Public.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 7:44 AM   in reply to Crist OC/PC

    That would not be a correct configuration. 3 sticks would force the board into single channel mode and the benchmarks would be skewed.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 7:47 AM   in reply to ECBowen

    RIGHT. as is would not be a correct comparation the sandy with 16G OF RAM.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 7:57 AM   in reply to Crist OC/PC

    Sigh, that is different. You are comparing running the hardware in an incorrect configuration that reduces the overall speed of the system with all functions to running 2 ideal configurations with different amounts of ram. Just running 12GB of ram to 16GB of ram only effects projects that use greater than 12GB of ram which I have already stated did not show any difference with the 3 layer benchmark project we use. If you look on the PPBM5 the amount of ram required to run the project at ideal performance dropped as each Adobe update hit. This is far different than running the hardware in an incorrect configuration that purposefully slows the entire system down. In the case of the 1366 platform in tri-channel, the ram bandwith at DDR3 1600 is around 26GB/s  where as Sandy is 19GB/s. As you can see the 1366 plaform should require less ram to perform the same task since the data is coming and leaving out at a greater rate. So if you actually wanted to compare them correctly, you would do so at 12GB to 16GB because that would actually even the playing field between them with the greater ram bandwith on the 1366 platform. You are trying to compare Apples to Oranges here.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 7:58 AM   in reply to ECBowen

    These is very funny, these two proccesor are quad core whit H.T. one is a grate OC. THE SANDY. Let say 20% better O.C. than the i7920,930,950.

    The limitation is coming Mother board. You can be able to replay 20 question in 10 sec. BUT IF MAN ONLY MAKE YOU 10 question no matter how fast you can replay them. Im talking realy pro staff.

    Tell me. the sandy with SSD PCIX16X AND 32GB OF RAM 1333MHZ 9-9-9-27 @ 4.7GHZ will WORK fast that a i7950 with SSD PCIX16X AND 24GB 2000mhz 8-8-8-20

     

    Please argument your replay.

     

    THX.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 7:59 AM   in reply to Crist OC/PC

    there is no way to have the same amount of ram in both systems

    one is dual channel the other tri channel

    correct config for dual

    2,4,8,16,32

     

    correct for tri

    3,6,12,24

     

    anything other than that is not correct.

    lastly its rather immaterial.

    the 8 to 16 gig on the sandy bridge was barely noticeable...

     

     

    Scott

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:02 AM   in reply to Scott Chichelli

    Over your XEON CONFIG

    tudio North 4.3 14 GBAsusP6T7 SupercomputerD9       140.0D9          104.8Top      56.0D9        24.0D9        56.0Top       4.0D9         65.0Top      4.016.3 XInteli7-980X(OC) 4.31614

     

    THESE IS NOT A RIGHT TRIPLE CHANEL. even so.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:05 AM   in reply to Crist OC/PC

    Why would you run the 1366 ram at 2000 but the Sandy at 1333? You are purposefully skewing the results to support what you want to believe if you do that. Both platforms will run DDR3 1600. Why not run them both that way? Some of the boards support greater than 1600 on the Sandy platform. Why would you not run them that way? If you want to believe the 1366 is automatically superior in every way without evidence to Sandy then that is your choice. However the benchmarks do not agree with you in many circumstances. That is the case with many benchmarks both generic and video based. Audio is reflecting the same as well. There is far more evidence supporting the Sandy position than the 1366 position at the moment for many tasks. Argue with that till you are blue in the face, it does not change the results.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:10 AM   in reply to Crist OC/PC

    Did you not notice the version difference with Adobe on the results? 5.0.2 increased performance across every platform including the Xeons. The Xeon results were 5.0.1 which was the best at the time 5.0.1 was the only version available. Once again you are comparing apples to oranges. 5.0.2 was a big performance boost to the MPE and the results reflected it.

     

    BTW I will be updating the Dual Xeon with 5.0.3 soon since I have a Dual Xeon config coming through soon.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:11 AM   in reply to ECBowen

    Showme   dual chanell 4x8 with 1600mhz cl 8-8-8-20.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:15 AM   in reply to Crist OC/PC

    Cristobal,

     

    Do not be too harsh on Scott. His remark:

     

    the 8 to 16 gig on the sandy bridge was barely noticeable...

     

    only shows that he is the commercial guy. Eric does know the difference between a 141 and 210 RPI, 50% slower. Commercially that is hardly noticeable, technically it is. Well, if the price were 50 lower to reflect the performance loss, maybe Scott would find it more than ' barely noticeable'.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:15 AM   in reply to Harm Millaard

    I feel Im talking to the sale manager of mac, Dell or HP.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:21 AM   in reply to Harm Millaard

    Harm,

    real world benchmarks.....

     

    Premiere Pro CS5 Version 5.0.3 Testing

    I7 2600 3.4GHZ Turbo to 3.9GHz

    8GB Blackline 1600 CL 9

    460GTX

    4 WD 1Tb Sata 64 Meg Cache 600 Drives in 2 Raid 0 arrays

    Video material - AVCHD 1080P 24 Frame Each Cut to 30 minutes of material

    Export Codec - H264 HDTV 1080P 24 Preset Default

    4 Effects per Layer - Fast Color Corrector, Brightness & Contrast, Video Limiter, Sharpen

    Each Layer Scaled to 50% for 4 frame PinP view.

    3 Layer - 37:35

    4 Layer - 40:49

     

    16GB Blackline 1600 CL 9

    570GTX

    3 Layer - 36:17

    4 Layer - 40:05

     

    40.49 to 40.05.... WOW thats a real major difference....

    this was the benchmark he was talking about..

    now toss in AE, bridge and PS open heck yeah it will make a big diff.

     

    SCott

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:20 AM   in reply to Crist OC/PC

    i have a feeling  i am taking with an id 10 t error

    you havent a clue what you are taking about dude

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:25 AM   in reply to Crist OC/PC

    Ok so now your comparing 4GB sticks to 8GB sticks which are really hard to get unbuffered right now? Why would you compare 8GB sticks on the Sandy and 4GB sticks on the 1366? The comments about running both platforms at 1600 was based around the current available 4GB sticks. The sticks decide what speed and latency they can run including DDR3 1600  8 8 8 20. Not the platform. The IC's on the stick decide what latency the ram can handle. The board just reads those profiles. If I sound like I am a Dell or HP rep then they must also understand how the hardware works and where memory speed and latency settings are decided.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:25 AM   in reply to Scott Chichelli

    Tell me these:

     

    WHAT IS THE MAXIMUN  TRANSPORT DATA ON A PROCCESOR RUNNING @ 4GHZ.?

    WICH ARE THE LIMITATION?

    EXPLAINTME THE COMPLETT RELATION ON ALL COMPONENTS OF THE PC IS YOU KNOW THAT?

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:28 AM   in reply to Harm Millaard

    Harm, were your percentages based of the RPI versus the overall times?

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:33 AM   in reply to Crist OC/PC

    can anyone tell me if there is an ignore button on these forums like other forums?

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:41 AM   in reply to Scott Chichelli

    Scott,

     

    Without knowing your internal benchmark, I can only guess at what is happening here to show these small differences in contrast to the PPBM5 benchmark.

     

    With 3 or 4 tracks, all scaled to 50% (MPE assisted), with 4 effects (MPE assisted), it gives the impression of testing the efficiency of CUDA/MPE cards and to a lesser degree RAM/VRAM latency, more than the CPU. That might be the reason for the small differences you measure.

     

    The PPBM5 results show far greater differences, because it comprises more different formats (so more decoding/encoding by the CPU), more different effects (also non-MPE assited ones, also handled by the CPU) and different scaling throughout, which makes it more CPU intensive, not only CUDA instensive. Therefore I believe our test shows more of the CPU/memory capabilities than your internal test.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:43 AM   in reply to Scott Chichelli

    Is allowed to say "I don't know"

     

    I really respect your guys. you know how to make pc's for editing.

     

    B.R.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 8:46 AM   in reply to Harm Millaard

    yes i understand that Harm,

    on the other hand our internal tests are based on the workflow of at minium 75% of the market..

    the red4k test we use would probably show a slightly better performance gap.

     

    the reason for posting it was to put an end to Cristo rediculous posts.

     

    Scott

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 9:55 AM   in reply to Harm Millaard

    Remember the CPU has to handle the decoding and encoding as well as the memory allocation of the data being sent to the GPU for the MPE and back again  . That is why the CPU performance decides the amount of load and the performance of the MPE. This was meant to analyze the entire pipeline which now includes the GPU in the mix. If you watch the CPU usage during encoding on the testing, the usage is above 80% more than 90% of the time. That does not seem to point to the CPU as not being taxed or tested by the benchmark as much as the ram or GPU since the MPE currently involves all 3. The PPBM5 seems to point to storage differences more than anything else but we will have to see further once we test the PPBM5 on the 4.7 Sandy with better drive configurations including the OS.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 14, 2011 4:22 PM   in reply to ECBowen

    Marginal difference with SSD drives and 580GTX video card. The platform seems at it's ceiling or close to it with the PPBM5. Looks like the difference from here is Cache on the 1366 6 cores or Xeons and number of threads. I may also try updating our current benchmark projects with non accelerated Effects and see how much that changes the results between platforms since that should increase the caching and re-threading occurring.

     

    Eric

    ADK

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 22, 2011 9:33 PM   in reply to Harm Millaard

    So with these new SandyBridge MB's, there is no way to install a RAID card for an external array?

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 22, 2011 11:11 PM   in reply to Shaluda

    There is, just run your video card at half speed.

     
    |
    Mark as:
1 2 3 Previous Next
Actions

More Like This

  • Retrieving data ...

Bookmarked By (0)

Answers + Points = Status

  • 10 points awarded for Correct Answers
  • 5 points awarded for Helpful Answers
  • 10,000+ points
  • 1,001-10,000 points
  • 501-1,000 points
  • 5-500 points