Creative Communities of the World Forums

The peer to peer support community for media production professionals.

Activity Forums VEGAS Pro Check specs to build a new PC…

  • Dave Haynie

    December 10, 2010 at 5:44 am

    [i]No one knows where Vegas will go with more GPU acceleration or whether choosing a Quadro card over a GeForce card makes any difference. In the absence of any compelling evidence, I would personally choose the cheaper option.[/i]

    That’s precisely why I wouldn’t go overboard on a GPU right now. Like CPUs, there are new GPUs on a regular basis. I had high hopes for Vegas 10, given how many other NLEs are using edit-time GPU acceleration. And true, you may get some via plug-ins (BCC7 uses some OpenGL, which is hardware accelerated…. don’t think they do CUDA or OpenCL).

    But at this point, most folks don’t see a huge or even important performance boost from the CUDA support that is in Vegas 10’s Sony AVC emitter. I don’t have much other use for 3D on my system… which is why I’m running a serveral-years-old nVidia 8800GT — does all the playback acceleration I need, and, well, I don’t have a great guideline for where to go next.

    So maybe there are better benchmarks on Vegas + CUDA now. But unless you’re buying the best possible CPU, I claim money is probably better spent on CPU until the GPU proves useful. On playback, it’s dynamite… a fast PC can be eaten up pretty significantly by 1080/60p playback. With the bog standard Windows 7 AVC decoder, I see less than 10% CPU used on my system, with GPU acceleration. GPU is great, when used. But unless Vegas makes it pay, I’d wait on better GPU support. At about $200, the nVidia GeForce GTX460 is kind of on the edge… not crazy expensive. But cheap enough to want to replace in six months or a year (we can’t really know when/if Vegas gets better GPU support)?

    Other things… yeah, it’s 6GB (technically, 6 GiB). The reason the i7 is faster than the i5 or the high end AMDs (I have a 6-core AMD) is the memory interface. Most modern CPUs (AMD, i-anything from Intel) have on-chip memory bus interfaces, rather than external memory controllers on a shared main bus. Some laptop chips have a single 64-bit bus. Most PC CPUs have dual 64-bit buses. The i7 has three 64-bit buses. Memory is always made in power-of-two sized; thus, the system you quote has three 2GiB memory modules.

    And I echo what other have said… don’t bother with 32-bit XP. Sure, install it as a secondary boot, or (maybe better and easier, depending on why you still want it) download VirtualBox and install it as a virtual machine, so you can run it alongside Windows 7. I did the conversion of my Windows 7 machine a year ago last October, and didn’t run into anything of significance that didn’t still work in 64-bit Windows 7. Yeah, it’s technically possible, but unless you run lots of really old stuff, not too likely. And without 64-bit addressing, you only get to use about 3.2-3.4GiB of DRAM… and if you buy a video card with 1GB, less than 3GB, if you plan to see all of the card’s memory (if Windows even permits that).

    -Dave

  • Fabio Mereghetti

    December 10, 2010 at 8:16 am

    OK I’m on my shopping route. I’m ready to buy:

    Asus P6T deluxe V2 160€
    INTEL Core i7 950 3.06Ghz 8MB BOX 198€

    Looking for a GEforce gtx 460 or 470 based card: which brand would it be better? (asus, gainward, Gigabyte, MSI, Palit, etc)
    Which specs should I look more?

    And what kind and brand of memory modules?

    Ah, and what about a proc. cooler?

    Thanks.

  • Fabio Mereghetti

    December 10, 2010 at 10:55 am

    Talking about ram modules:

    which specs would be better?
    1066, 1333, 1600MHz = ?
    CL (7,8,9)= ?
    other = ?

    How much GB?

    Somewhere I read of a “2GB x core rule”: does that make sense?
    If yes, does that relate to “phisical or virtual cores”? So 8 or 16GB?
    Videoguys reccommend 12GB so what?

    And about W764: which one (lan with no domain, just workgroups)?

    Thanks.

    Fabio.

  • Marlen Mathers

    December 10, 2010 at 5:32 pm

    Danny – very nicely done! I think I recognize some of those backgrounds. I probably need to explain my gravitation towards Vegas for chroma key. I took this video production thing up as a hobby when I inherited some equipment. I never had any formal training with NLE, just self taught and a few books like “Instant Vegas” etc. I really have only scratched the surface of what some of these powerful programs can do. As such, I learn things as I need them to get the project done. Not a great way to get the most out of the programs, but I have a day gig that I need to pay the bills! (That’s why forums like this – and the people here are so valuable to me)

    As my needs for green screen have not been very sophisticated, I just got more comfortable with Vegas as the default. Perhaps when I move into some more advanced keying that will change….

    here’s the link for Keith’s article. It’s not the original (as I can’t find it), but it is the same content…

    Marlen

  • Marlen Mathers

    December 10, 2010 at 5:36 pm
  • Marlen Mathers

    December 10, 2010 at 5:47 pm

    Gary –

    one question on the DIY8 Configuration. If I wanted to go with the WD 10,000 RPM Boot drive – but use the Hitachi’s 1TB 7200 RPM for source material and another Hitachi 7200RPM for rendering, will I see any degradation in performance because of the differing speeds? I’d like go all 10,000 RPM, but would like to wait to see if prices come down and then swap out the source and rendering drives. I noted that overall performance is incumbent upon how all the parts work together, so I didn’t want to mess things up by being “too smart by half” in looking down the road for upgrading….
    Marlen

  • Gary Bettan

    December 10, 2010 at 6:04 pm

    i would suggest the following. Stripe the two 7200RPM drives in a RAID0 for performance and throughput. Put all projects & video on the RAID.

    When exporting files for blu-ray or web, set up a folder (or partition) on the 10K boot drive and export there.

    Gary

    https://www.videoguys.com 800 323-2325 | We are the video editing and production experts!

  • Marlen Mathers

    December 10, 2010 at 6:43 pm

    excellent! Thanks Gary.

  • Dave Haynie

    December 10, 2010 at 8:02 pm

    Drive speed is essentially never an issue … for a single stream of video, anyway. I managed to make my 6-core system hard drive bound by compositing two DNxHD files at the same time, from the same drive; I think these were at 145Mb/s.

    The big advantage of higher RPMs these days isn’t speed. Yeah, it’s a little faster, but the big win is on seek time. A 10K drive or SSD is a great choice for your system drive, largely because Windows deals in thousands of tiny files, and because everything you do gets a boost from a faster drive.

    I typically keep my main project files on a single 7200RPM SATA drive (one per large project, and a second one for smaller projects). Simpler stuff I just render back to the project drive, but if there are enough assets in the same project, I direct the output elsewhere (usually to the “desktop”, which is on the boot drive) and, if necessary, move some of the assets to a second input drive.

    Again, it’s the seeking that’s the issue. If you have a couple of inputs, your render should happen at full speed… CPU is at 100%. If you have too many large inputs, your drive will slow down enough from seeking that, as I pointed out, you may need a second drive.

    A RAID can help, but only so much. You can go RAID0 (stripeset), which parallels two drives. The advantage is that you have the same capacity as two drives, so straight-line performance is doubled. On the downside, the RAID is subject to the worst-case behavior from either drive… so your seeks are always the slower of the two, and the whole partition fails when either drive fails. With enough buffering, the additional performance will add more speed than the slightly slower seeking will take away (within a single file, you seek half as often thanks to the paralleled drive, but that only matters if you’re eating data in large chunks).

    RAID1 gives you the same read speedup, same minor seek issues, and added security — your partition only fails when both drives fail. But you spend twice as much for the partition (two drives, one unit of storage).

    -Dave

  • Dave Haynie

    December 10, 2010 at 10:38 pm

    Which specs would be better??
    1066, 1333, 1600MHz = ?

    That’s the effective clock speed. Like your CPU, a faster clock means faster memory, all else being equal.

    The first consideration: what does your motherboard support? DRAM speed is a potential, like the speed on your SDHC cards, or the voltage on a flashlight bulb. A faster SDHC card only helps if the device can use the faster speeds; a 6V flashlight doesn’t get brighter with a 12V bulb. A faster DRAM DIMM only helps your memory speed if the motherboard can use that speed.

    With that said, the clock speed (1066, 1333, 1600… all in MHz, based on the effective clock speed of the DIMM) is the most fundamental indicator of memory performance.

    CL (7,8,9)= ?
    CL = “CAS Latency”; CAS = “Column Address Strobe”. Simply put, a DRAM is accessed by your computer like a spreadsheet — first you lock in the row, then the column. However, that’s kind of slow, and modern processors generally talk to RAM in bursts. Rather than supply both row and column for each access, they lock in the row, and allow multiple columns accesses… much faster. CL is the number of clocks it takes for this page-mode access. For the same speed RAM, a smaller CL is better. But a larger CL number on a faster DIMM is likely faster overall.

    -Dave

Page 3 of 8

We use anonymous cookies to give you the best experience we can.
Our Privacy policy | GDPR Policy