Page 2 of 3 FirstFirst 123 LastLast
Results 11 to 20 of 30
  1. Collapse Details
    #11
    Senior Member
    Join Date
    Jul 2013
    Posts
    518
    Default
    Quote Originally Posted by filmguy123 View Post



    Yes, Sep 7 by Adobe - currently in beta, part of the next version hopefully but you could access now via beta (they are still working out the kinks).

    Nvidia Thread
    https://community.adobe.com/t5/premi...1415455?page=1
    AMD Thread
    https://community.adobe.com/t5/premi...1415460?page=1

    .
    Thanks!


    Reply With Quote
     

  2. Collapse Details
    #12
    Senior Member Samuel H's Avatar
    Join Date
    Jun 2010
    Location
    Madrid, Spain
    Posts
    8,156
    Default
    I have a threadripper workstation (3970X: 32 cores, 64 threads, 128GB of RAM), but I got it for work, completely unrelated to video. I also use it for video (Resolve) but it's absolutely overkill. If you can wait for the new models, get the new CPU with 16 cores instead: it will be cheaper, and for most stuff it will be faster. The exception will be 3D rendering, if you do much of that.

    With such a massive number of cores, each task doesn't take much less than it used to when I had an 8-core Ryzen, but I don't have to ever worry about all the other things the computer may be doing at any point in time. I can have 5 matlab sessions running, using 25 cores and pulling 400W off the wall socket with the GPU basically on idle, and I can still open resolve, edit colors and export a 4k video without even noticing the computer is doing anything else. But if I close everything else and just open Resolve, it's not noticeably faster than my previous build.


    Reply With Quote
     

  3. Collapse Details
    #13
    Senior Member puredrifting's Avatar
    Join Date
    Nov 2004
    Location
    Los Angeles, Ca.
    Posts
    11,183
    Default
    As a total PC novice (been hardcore Mac for the past 20 years), is Ryzen taking over from Intel as far as PC build dominance? I just built a new live streaming system
    using a Aorus 299 MB, NVidea Quadro RTX5000, and the Intel 10990 i9 CPU. I am sure this will be fine for my needs (vMix Live and Call), no overclocking, but I am curious
    as all I hear about on the boards is Ryzen, Threadripper, Ryzen?

    My live streaming partner, who is an engineer, dislikes AMD and specifically wanted me to stick with Intel. What say you?
    It's a business first and a creative outlet second.
    G.A.S. destroys lives. Stop buying gear that doesn't make you money.


    Reply With Quote
     

  4. Collapse Details
    #14
    Senior Member Teddy_Dem's Avatar
    Join Date
    Aug 2013
    Posts
    233
    Default
    Quote Originally Posted by puredrifting View Post
    As a total PC novice (been hardcore Mac for the past 20 years), is Ryzen taking over from Intel as far as PC build dominance? I just built a new live streaming system
    using a Aorus 299 MB, NVidea Quadro RTX5000, and the Intel 10990 i9 CPU. I am sure this will be fine for my needs (vMix Live and Call), no overclocking, but I am curious
    as all I hear about on the boards is Ryzen, Threadripper, Ryzen?

    My live streaming partner, who is an engineer, dislikes AMD and specifically wanted me to stick with Intel. What say you?
    AMD has been gaining a lot of ground in the past few years or so with Ryzen and Threadripper as they are offering more cores at lesser prices. This article is good breakdown on how it happened over the past 6 years mainly due to Intel's manufacturing woes:

    https://www.pcworld.com/article/3563...struggles.html


    Reply With Quote
     

  5. Collapse Details
    #15
    Senior Member
    Join Date
    Apr 2008
    Posts
    4,000
    Default
    Intel had the crown and a large mote. A lot of the software written had special instructions for Intel processors and they just worked better than AMD. Then Intel got lazy and also fell on their face in trying to move to 10nm in their production process. They have been stuck in a low gear for six years. AMD made some bold moves to meet and pass Intel where they predicted them to be, and now AMD has the attention. Weather or not it is a good to choice to leave Intel for video editing is unknown, but AMD is giving more performance than Intel right now. 16 cores with a boost speed of 4.9GHz is pretty nice. All of this running at a low power rating is where everybody wants to go. I will be watching the space. Reliability is very important and Intel systems do work. (I see a lot of similarities between Canon and Intel). Great to have competition to spur development.


    Reply With Quote
     

  6. Collapse Details
    #16
    Senior Member
    Join Date
    Oct 2014
    Posts
    7,130
    Default
    The general consensus over the last few years has been that Intel still has an edge in a single-core performance but AMD delivers a better bang for the buck. But, as Samuel pointed out above, it also depends on what one is going because one could build out a total monster with either - heck, one could build a supercomputer with either chip, if he had enough funds - and end up with a surplus to the requirements.

    What has changed over the last week with the AMD announcement is that AMD is now even or even slightly ahead of Intel in a single thread performance. But the new CPU's are now higher priced as well, almost on par with Intel.

    Given the full roster of chips, the key isn't a brand per se but the right budget for the needs. Or it'll be like driving your kids to preschool in a Dodge Challenger SRT Demon Wide Body - it'll get them there but you still have to drive over the speed bumps at 20 MPH.


    Reply With Quote
     

  7. Collapse Details
    #17
    Senior Member Samuel H's Avatar
    Join Date
    Jun 2010
    Location
    Madrid, Spain
    Posts
    8,156
    Default
    Right now, AMD holds the crown, by a nice margin. Not only is it faster "at any given price" (the fastest $100 CPU is AMD, at $250 it's AMD, at $500 it's AMD, at $1000 it's AMD, check the graphs here, they have price on the label) but also "at any price", which has often not been the case. Right now, even if you have $2000 or $4000 to spend on a CPU, the fastest one is still AMD. The $2000 one is the one I got, performance tests here.


    Reply With Quote
     

  8. Collapse Details
    #18
    Senior Member
    Join Date
    Jan 2007
    Location
    Portland, OR
    Posts
    4,760
    Default
    There's also another key point of consideration: not only does AMD now hold both the single-core (by a slight margin) and multi-core (by a large margin) performance crown, they ALSO have platforms with PCI-E 4.0 and Intel does not, making it even more difficult to justify building an Intel rig today if you care about system connectivity.

    For example, with an AMD X570 (not HEDT system) I get 40 lanes of PCI-E 4.0, which is the equivalent of having 80 lanes of PCI-E 3.0. That's a huge deal. It means I can run 4 pci4 m.2 NVME SSD cards and an RTX 3080 in my system with zero bottlenecking - all 4 drives cranking 7000mb/s simultaneously - on a non HEDT platform. It also means for many many years, I won't have any peripheral compatibility issues and will be more future-proofed.

    In addition, all of this comes at a lower TDP cost. Intel has really stretched their current 14nm chips to try and compete with AMD here by pushing the thermal envelop. So AMD is winning with a lower TDP which matters for system cooling and power efficiency (and possibly better long term reliability - heat and electronics aren't friends, and maxing out silicone in such an aggressive way is unproven in the long run on those intel chips).

    There is nothing wrong with that Intel 10900 chip, it was a good buy and it will be great for many years. But for those looking at a new system AMD just makes more sense right now for most people. I am sure Intel will be back with a vengeance when they finally get their 10nm chips out the door on a PCI-E 4.0 platform, but for desktop, that is currently looking like 2022 - an eternity to wait if you need a new system right now, and who knows where AMD will be by then.

    As for industry support of AMD vs Intel, this is a rather moot point - they are both x86 platforms and the historical Intel AVX2/AVX512 (previously) exclusive instruction sets just aren't an advantage today at all, it was interesting for a while for some niche tasks years ago. AMD could implement them today but hasn't bothered as they are basically dead in the water. In any case, AMD's market share means they can't be ignored from a developer POV - all of the new game consoles are built entirely on AMD hardware, so if anything, AMD will have the advantage in gaming (which tends to drive the enthusiast PC market). And In the professional realm, there has been such a massive shift to AMD Threadripper and Ryzen platforms by small shops and enthusiasts (which will only continue to amplify in the coming 2 years) that one would be very hard pressed to make any sort of argument for Intel having an edge there. These are two competitors essentially working on making the best version possible of a shared architecture, and Intel is a few years behind. Developers are hugely incentive to develop and test equally on both platforms - not that it is needed too much given how similar they are.


    1 out of 1 members found this post helpful.
    Reply With Quote
     

  9. Collapse Details
    #19
    Senior Member
    Join Date
    Aug 2008
    Location
    Ottawa , Canada
    Posts
    723
    Default
    I also think AMD has the crown at the moment. Current NLE are moving to GPU taking most of the load so the task for the CPU is feeding the GPU ( s ). That means fast storage and connections where PCIe 4.0 has the advantage for AMD too. Direct connection between CPU, GPU and storage will be important where AMD has this covered, even more so with the Threadrippers with 60 PCIe 4.0. The new 16 core Ryzen motherboards have just enough lanes to provide direct connection between CPU, GPU and NVME 4.0 devices and with 4 PCIe 4.0 lanes for the chip set for everything else. That is still equivalent to half the full complement of PCIe 3.0 lanes for competing Intel CPU's. The RYZEN's have 24 PCIe 4.0 lanes and Intel 16 PCIe 3.0 lanes for parts like the i9 10900K. Depending on X570 motherboards the AMD CPU has direct connection to the first GPU slot and one of the NVME so could provide full transfer to these location at PCIe 4.0. Other storage could be provided by the chip set. So a big advantage over Intel.

    Even on my older 1920 Threadripper I can see the encoding ( NVENC h265 encoding in Resolve by my 1080Ti , so old now too !) difference in time between source/destination being on hard drive, SSD or NVME. As expected the difference is considerable between hard drive and NVME. So essentially seeing the difference between SATA hard drive speed, SATA SSD speed and NVME. I expect the bottleneck is still the NVME speed on my system in that the CPU and GPU could still go faster ?


    Reply With Quote
     

  10. Collapse Details
    #20
    Senior Member
    Join Date
    Jan 2007
    Location
    Portland, OR
    Posts
    4,760
    Default
    Ron, this is interesting. I'd love your feedback on my post here on how to layout my new system with storage: http://www.dvxuser.com/V6/showthread...post1986845237

    And FWIW - Since I didn't exactly spell it out in that post, and because it's relevant to the topic here -- here is how I'm planning to configure my lanes as such:

    CPU (24, with 20 available (4 dedicated to chipset interconnect)):

    8x RTX 2080 (only need 8x thanks to PCIe4.0, it's equiv to PCIe3.0 16x and have confirmed this)
    8x M.2 Expander Card with two m.2 nvme SSDs (this will be in the 2nd 16x slot connected to the CPU, running at 8x bandwidth - that is, a full 4x per nvme ssd).
    4x M.2 Slot with one m.2 nvme SSD (this is the motherboard slot connected to the CPU)

    Chipset (16 lanes available, via a PCIe4.0 4x lane back to CPU)

    - M.2 Slot with one m.2 nvme SSD (this should be a minimal bottleneck, if anyone, even with all other chipset devices running full speed thanks to pcie4.0)


    Related to the discussion here... hopefully it helps paint a picture of how powerful and cost effective PCIe4.0 for many of us. This kind of internal storage interconnect at this price point and TDP is game changing. It's like having an 80 lane PCIe3.0 workstation setup. There is no external solution on market that can achieve this kind of throughput - 10gbe will net you 1250 mb/s, USB 3.2 Gen 2 ~2400mb/s, and newest thunerbolt around 4800mb/s. With all of these internal PCIe4.0 ports you can run multiple nvme drives simultaneously at 8000mb/s throughput and incredibly fast direct connect to the CPU, no latency.


    Reply With Quote
     

Page 2 of 3 FirstFirst 123 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •