From: "Adam Mercado" Received: from firebrick.maple.relay.mailchannels.net ([23.83.214.59] verified) by media-motion.tv (CommuniGate Pro SMTP 6.1.0) with ESMTPS id 7244149 for AE-List@media-motion.tv; Wed, 13 Feb 2019 23:38:26 +0100 X-Sender-Id: dreamhost|x-authsender|adam@influxx.com Received: from relay.mailchannels.net (localhost [127.0.0.1]) by relay.mailchannels.net (Postfix) with ESMTP id 63BF543315 for ; Wed, 13 Feb 2019 22:48:42 +0000 (UTC) Received: from pdx1-sub0-mail-a30.g.dreamhost.com (unknown [100.96.19.78]) (Authenticated sender: dreamhost) by relay.mailchannels.net (Postfix) with ESMTPA id E0010434BB for ; Wed, 13 Feb 2019 22:48:41 +0000 (UTC) X-Sender-Id: dreamhost|x-authsender|adam@influxx.com Received: from pdx1-sub0-mail-a30.g.dreamhost.com (pop.dreamhost.com [64.90.62.162]) (using TLSv1.2 with cipher DHE-RSA-AES256-GCM-SHA384) by 0.0.0.0:2500 (trex/5.16.2); Wed, 13 Feb 2019 22:48:42 +0000 X-MC-Relay: Neutral X-MailChannels-SenderId: dreamhost|x-authsender|adam@influxx.com X-MailChannels-Auth-Id: dreamhost X-Stupid-Desert: 777042d70c8f7a16_1550098122311_3922972208 X-MC-Loop-Signature: 1550098122310:802165361 X-MC-Ingress-Time: 1550098122310 Received: from pdx1-sub0-mail-a30.g.dreamhost.com (localhost [127.0.0.1]) by pdx1-sub0-mail-a30.g.dreamhost.com (Postfix) with ESMTP id 6694680187 for ; Wed, 13 Feb 2019 14:48:41 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha1; c=relaxed; d=influxx.com; h=from :content-type:content-transfer-encoding:mime-version:date :subject:message-id:references:in-reply-to:to; s=influxx.com; bh=1zg1BNEjDtRFFOsjDEYYOqCpeA4=; b=QqHxrSFTEUmq3RcU/khsRRBJa00Y DmwgcF11Xz/NZMomuzI7o/J9rxlOTBZFkBAcVRlAbzbKOFYJfW/PV97OsZIXpSDD 2zillaM71l6RHhmQG/O9BCee2KQN41+eOMXg6LMFqRWPktmiWnJtXLRH+9vd/+Wz MM+YOvnwrQVAyu4= Received: from [192.168.43.2] (mobile-107-77-164-42.mobile.att.net [107.77.164.42]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) (Authenticated sender: adam@influxx.com) by pdx1-sub0-mail-a30.g.dreamhost.com (Postfix) with ESMTPSA id AD0FC80188 for ; Wed, 13 Feb 2019 14:48:40 -0800 (PST) X-DH-BACKEND: pdx1-sub0-mail-a30 Content-Type: multipart/alternative; boundary=Apple-Mail-110D81AF-B6FB-4A25-8AE3-04882D295529 Content-Transfer-Encoding: 7bit Mime-Version: 1.0 (1.0) Date: Wed, 13 Feb 2019 14:48:37 -0800 Subject: Re: [AE] Feedback on Workstation components Message-Id: <8394267B-98C9-485E-BADB-67E165199CA9@influxx.com> References: In-Reply-To: To: After Effects Mail List X-Mailer: iPad Mail (15G77) X-VR-OUT-STATUS: OK X-VR-OUT-SCORE: -100 X-VR-OUT-SPAMCAUSE: gggruggvucftvghtrhhoucdtuddrgedtledruddtgedgtdduucetufdoteggodetrfdotffvucfrrhhofhhilhgvmecuggftfghnshhusghstghrihgsvgdpffftgfetoffjqffuvfenuceurghilhhouhhtmecufedttdenucesvcftvggtihhpihgvnhhtshculddquddttddmnecujfgurhephfgtgfggfffukfhfjgfvofesrgejmherhhdtjeenucfhrhhomheptegurghmucfovghrtggrughouceorggurghmsehinhhflhhugiigrdgtohhmqeenucffohhmrghinhepshhhohhtghhunhhpohhsthdrtghomhdpphhughgvthhshihsthgvmhhsrdgtohhmpdhgohhoghhlvgdrtghomhdpihhmrghgvghstghivghntggvrdgtohhmrdgruhenucfkphepuddtjedrjeejrdduieegrdegvdenucfrrghrrghmpehmohguvgepshhmthhppdhhvghloheplgduledvrdduieekrdegfedrvdgnpdhinhgvthepuddtjedrjeejrdduieegrdegvddprhgvthhurhhnqdhprghthheptegurghmucfovghrtggrughouceorggurghmsehinhhflhhugiigrdgtohhmqedpmhgrihhlfhhrohhmpegruggrmhesihhnfhhluhiggidrtghomhdpnhhrtghpthhtoheptefgqdfnihhsthesmhgvughirgdqmhhothhiohhnrdhtvhenucevlhhushhtvghrufhiiigvpedt --Apple-Mail-110D81AF-B6FB-4A25-8AE3-04882D295529 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Thank you Carey for sharing your insight. That was a good read Adam Mercado Owner & Creative Idiot Influxx Media Production=20 714.928.9896 Moving Images. For Business > On Feb 11, 2019, at 6:27 AM, Carey Dissmore wrot= e: >=20 > A few comments:=20 >=20 > Puget Systems REALLY gets it when it comes to addressing the needs of our m= arket. They have a model where they do a ton of research and provide excelle= nt support for their tested, configured system designs. They share a lot of t= heir data and findings on their website, which benefits the community as a w= hole. They are uniquely positioned to do this type of testing and I=E2=80=99= m very thankful they share. For the vast majority of people who do NOT have i= nterest in building their own systems, but nevertheless are realizing that a= custom workstation is truly the best path forward for their needs, I can=E2= =80=99t think of a better choice.=20 >=20 > Now onto the geek stuff, (which Puget already knows): >=20 > A - Building PC: I=E2=80=99m inclined towards PC building too, and have bu= ilt many workstation machines. I built my first PC around 2001 mostly for ga= ming and offloaded encoding/rendering but I largely stuck with Mac for post p= roduction work until about 2012 when I made the complete jump to primary pos= t production on Windows workstations. Teddy and I had a good chat on the pho= ne back then just to gut check my system architecture before pulling the tri= gger on a $6k box build. I=E2=80=99ve built and rebuilt many since.=20 >=20 > B-Performance: It used to be that synthetic benchmarks were quite useful i= n determining performance as it applied to video post work. You pretty much a= lways needed the highest performance hardware you could get and there wasn=E2= =80=99t a whole lot of differentiation about tasks, as it all sort of lifted= with the general performance of the machine. You could look at relative per= formance and extrapolate your value or expected performance for the apps and= work you did and make a decision about value for money/ROI/Cost-benefit wit= h some simple math. =20 >=20 > Those days are over. That won=E2=80=99t work anymore. The trend for nearly= a decade is that PC performance gains have forked in multiple directions: C= PUs have added cores, and clock speed increases have crept higher but very s= lowly. Due to thermals, you are often left with a choice of more cores vs mo= re clock speed. GPUs have introduced massive parallelism tuned for specific c= ompute loads but data has to be moved from memory/fast CPU access over the b= us to the GPU and back so that is a factor in determining which types of =E2= =80=9Cloads=E2=80=9D are best suited to GPU compute. The problem is that syn= thetic benchmarks can be designed to torture test a specific type of computi= ng load that shows maximized performance of a particular component but it ha= s never been less relevant to apply those to your expectations in the apps y= ou ACTUALLY USE to do post production work.=20 >=20 > Therefore, the ONLY benchmarks that are relevant are the ones that are don= e by testing the apps you ACTUALLY USE to determine a particular component o= r configuration=E2=80=99s value to you. Here=E2=80=99s some apps that I use a= nd what I have found they need from a machine.=20 >=20 > For example here=E2=80=99s some differing compute loads and what types of P= C hardware/config will serve it well.=20 >=20 > Storage (applies to all): Everything wants fast storage and IO. Splitting u= p storage between media and boot and cache, each optimized is good. Storage/= cacheing has become an issue too and managing the flow of data and using hig= h performance caching devices separate from both system drive and media stor= age is key to optimizing. As you look at the applications below >=20 > AE: In daily use, largely still single threaded operations with dashes of m= ultithreading throughout, and a splash of GPU compute here and there, partic= ularly on plugins/effects. Today, this means favoring higher clock speeds ov= er core count on the CPU. Get as much RAM as you can justify because AE and R= AM previews will use it. Generally AE-only artists don=E2=80=99t need tons o= f data storage, but what they have should be fast. Often they are best serve= d by something like a 9900k system and a modest GPU from NVidia.=20 > Note that one of the challenges of optimizing AE for multithreading is tha= t not every type of compute task lends itself well to parallelism. Sometimes= you have to wait for the result of one computation to use the result of tha= t to start the next computation. This same performance optimization issue al= so affects other similar apps like Fusion.=20 >=20 > Premiere: A little more balanced between CPU cores and clock speed and GPU= . Having said that Premiere still favors clock speed over massive core count= , can take advantage of Quicksync on newer Intel CPUs BTW, and uses GPU but d= oes not rely heavily on it and primarily processes color and effects on GPU (= with some exceptions). Premiere doesn=E2=80=99t require quite as much RAM as= AE, but I guess it doesn=E2=80=99t hurt. Having said that Premiere seems to= need an awful lot of compute hardware thrown at it to smoothly handle media= at 4k. By comparison FCPX (which I don=E2=80=99t use because it=E2=80=99s M= ac only) is way more optimized and gets better performance from lesser hardw= are. I think there=E2=80=99s a lot of room for improvement for Premiere here= but it probably means a rewrite. >=20 > Encoding: h264 or now h265 encoding is massively parallel and really shows= off what a lot of cores can do. Currently CPU dominated we=E2=80=99re start= ing to see this moving to GPU options as well. If all you did all day was en= code stuff, go for core count and fast IO. However, for most of us, encoding= occupies only a few minutes per day compared to creative work. >=20 > Resolve: GPU dominated (for color page operations anyway). Here, CPU perfo= rmance matters but a mid-line CPU is probably adequate but what you want are= GPU compute units. The best units you can afford, plus more of them. 2 or 3= GPUs scale really well with Resolve, but in my testing I=E2=80=99ve found i= t best to keep them matched. I did some mismatched testing with different ge= n GPUs and it=E2=80=99s pretty variable and inefficient. Not that every litt= le bit doesn=E2=80=99t help=E2=80=A6but one card ends up waiting on the othe= r to finish a task. PCI-E lanes required for multi-GPU support push the user= to go with higher level platforms like X299.=20 >=20 > Cinema 4D: When rendering, massive CPU core count is your friend. But when= modeling, key framing, animating etc a lot fo that is more single threaded b= ut obviously previews can be run OpenGL (etc) on the GPU. Other renderers ar= e GPU-dominant. 3D is not my forte so I=E2=80=99m going to let others cover t= his. >=20 > But I said all of that as a preface to make this point: You can choose any= one activity or app and build a kick-butt machine tuned specifically to the= requirements of that activity, and walk away with maximized ROI. An AE-only= machine is going to look quite different than a Resolve-only machine, and a= gain quite different from a C4D optimized machine. But often we need to enga= ge in a mix of activities and the choice about whether to try to build the u= ltimate =E2=80=9Cdo it all=E2=80=9D workstation ($$$$) vs separating the tas= ks is something we all have to look at in light of our own mix of work. >=20 > Anyway, when I sat down I had no idea I=E2=80=99d write ll that but alas, t= he coffee was good and plentiful. Plus, when you really get into this stuff t= here are just a lot of considerations to get your head around. By no means d= id I write a definitive guide here (is there such a thing?) and understand w= hy Chris Zwar=E2=80=99s is over 10,000 words. I look forward to it. >=20 > Carey Dissmore >=20 >=20 >=20 >> On Feb 11, 2019, at 6:33 AM, Byron Nash wrote: >>=20 >> Puget really impressed me with the depth and detail of their hardware ana= lysis for very targeted demographics. It's nice to have someone really under= stand what the difference is between editing, motion graphics and GPU render= ing.=20 >>=20 >> On Sun, Feb 10, 2019 at 11:52 PM Teddy Gage wro= te: >>> Sure thing Chris, feel free to reach me at teddygage@gmail.com. Goes for= anyone- always happy to consult or chat hardware or software. Have also bee= n doing a fair amount of VFX supervising recently for a couple TV shows.=20 >>>=20 >>>> On Sun, Feb 10, 2019 at 11:14 PM Chris Zwar w= rote: >>>> Hi all, >>>>=20 >>>> I started researching performance last year, with a view to writing up a= n article and some thoughts for the ProVideo Coalition. Currently I'm about= 50% of the way through my plan and I'm over 10,000 words so it's a bit slow= going. Even so, I was wanting to reach out to list members who are especia= lly interested and knowledgable on performance related matters. >>>>=20 >>>> Teddy & Stephen were both on my list of people to approach for input an= d this thread just reinforces that! Are you two able let me know the best wa= y to contact you off-list please? >>>> And if there's anyone else who has strong knowledge or strong opinions o= n performance/ components then let me know too. >>>>=20 >>>> Cheers, >>>>=20 >>>> -Chris >>>>=20 >>>>=20 >>>>> On Sat, 9 Feb 2019 at 07:18, Byron Nash wrot= e: >>>>> Those are good thoughts Stephen and I'll keep that in mind going forwa= rd. I am planning to use a Blackmagic card for video output but the monitors= I have to use right now are 8bit HD. I'm starting to get into 4K VFX for fi= lms on my machine but most of the rest of the team is building HD promos/mog= raph content. My only pet peeve with the BM cards is when you have a 4K scre= en it won't downscale through the card and just goes black.=20 >>>>>=20 >>>>>> On Fri, Feb 8, 2019 at 11:10 AM Stephen van Vuuren wrote: >>>>>> > In regards to 10bit output an option not being discussed is using a= Blackmagic or Aja PCI card or device. These will give you the option of 10b= it YUV or RGB output (monitor dependent). 4K capable versions of these are a= vailable for as little as $200. >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> I mentioned external card but you are not budgeting for the 4K 10-bit= display that you will need to see the output. >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> I have a $199 BM MiniMonitor 4K and it=E2=80=99s great going out to a= 50=E2=80=9D HDR TV). But that does not get you 4K 10-bit on your desktop di= splay or apps that don=E2=80=99t support that card. You have to have a secon= d monitor where as Quadro allows you to have 4K 10-bit on your main desktop m= onitor. >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> If you only have room for on 4K 10-bit, a card plus additional monito= r is not an option. And even if you do, my setup I would rather always see 1= 0-bit as much as possible in all my work and catch issues. >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> As video moves to 10-bit and HDR in 4K, even for low-end, 10-bit is g= oing to be a bigger deal. >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> From: After Effects Mail List =20 >>>>>> Sent: Friday, February 8, 2019 12:25 AM >>>>>> To: After Effects Mail List >>>>>> Subject: Re: [AE] Feedback on Workstation components >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> If you=E2=80=99re talking about a budget system Quadro cards aren=E2=80= =99t even in the conversation (5k was mentioned). If one of the main uses of= the machine will be Redshift you=E2=80=99re better off buying the fastest G= TX or RTX card(s) you can afford. The RTX cards may end up being a better ch= oice in the near future, if Nvidia=E2=80=99s and Redshift=E2=80=99s recent a= nnouncements are to be believed. >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> In regards to 10bit output an option not being discussed is using a B= lackmagic or Aja PCI card or device. These will give you the option of 10bit= YUV or RGB output (monitor dependent). 4K capable versions of these are ava= ilable for as little as $200. >>>>>>=20 >>>>>>=20 >>>>>> Dennis Wilkins >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>>=20 >>>>>>=20 >>>>>>=20 >>>>>> On Feb 7, 2019, at 8:49 PM, Stephen van Vuuren wrote: >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> Steve, >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> That link is for 10-bit DirectX output which is for games and Direct X= apps =E2=80=93 not relevant to this discussion. Adobe and most other pro po= st apps don=E2=80=99t use or support DirectX. And just because a driver or m= onitor flags 10-bit does not mean there is actually 10-bits happening. >>>>>>=20 >>>>>>=20 >>>>>> Better info is here: https://imagescience.com.au/knowledge/10-bit-out= put-support >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> From: After Effects Mail List =20 >>>>>> Sent: Thursday, February 7, 2019 11:40 PM >>>>>> To: After Effects Mail List >>>>>> Subject: Re: [AE] Feedback on Workstation components >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> 10bit display support is all over the place on mac. MBP 2016 is drivi= ng a 24=E2=80=9D external HDMI monitor in 10bit will driving the native disp= lay in 8bit. recent model imacs with 5K displays running in 10bit.=20 >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> you can get nV cards on win to run in 10bit. here is how to run nV or= AMD on win10 in 10bit https://www.pugetsystems.com/labs/articles/Setting-G= raphics-Card-Software-to-Display-10-bit-Output-1221/ >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> S >>>>>>=20 >>>>>>=20 >>>>>>=20 >>>>>>=20 >>>>>>=20 >>>>>> On Feb 7, 2019, at 10:25 PM, Stephen van Vuuren wrote: >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> >>> Unless you NEED 10 bit color output (unlikely) do not get a Quadr= o. They are intended for industrial CAD / CAM workflows and are 1/10th as fa= st for rendering CUDA Redshift. They are intended for use cases where you ne= ed extreme floating point precision, like designing critical infrastructure.= Nobody is using Quadros for Redshift / Octane.=20 >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> I disagree that 10-bit is =E2=80=9Cunlikely=E2=80=9D. It=E2=80=99s mo= re of a ignorance of benefits that an unlikely scenario. I get a lot of pro c= olorist and post house delivering stuff with banding, too much shadow noise a= nd color/gamma problems that have obviously been graded in 8-bit on typical= 1000:1 panels. See it all the time.=20 >>>>>>=20 >>>>>>=20 >>>>>> Same as I see people mixing on headphones or uncalibrated speakers an= d charge pro prices for the service. The defense is always the same =E2=80=9C= looks/sounds great here!=E2=80=9D. >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> GeForce cards cannot display video in Premiere or images in Photoshop= in 10-bit, period or most any other pro app. You have to have a monitor car= d going to a separate, second monitor. That card or box + monitor for a 4K s= etup is not cheap. >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> Quadro cards offer native 10-bit OS level support with Quadros and Ma= c is now fully supported. >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> Most people think they do don=E2=80=99t need 10-bit output but end up= delivering work that looks bad in 10-bit, 12-bit and HDR environments or di= splays with wider than the 8-bit (and 6-bit+dithering panels out there). >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> DCP is 12-bit. Rec2020 is 10-bit as is all Netflix, Amazon etc. HDR> H= ell, YouTube HDR is 10-bit.https://support.google.com/youtube/answer/7126552= ?hl=3Den >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> If you=E2=80=99re delivering 10-bit and while it used to be rare, the= se days almost everything I deliver is 10 bit or higher. >>>>>>=20 >>>>>>=20 >>>>>> So I just don=E2=80=99t take people mastering final projects for prof= essional deliver very seriously that don=E2=80=99t have a fully 10-bit or hi= gher workflow. >>>>>>=20 >>>>>>=20 >>>>>> Quadro plus a good affordable HDR 4K 10-bit monitor is the same or le= ss money than a GeForce + 10-bit video output plus dedicated 4K display and o= ffers better drivers, more stability and more flexibility (like Photoshop wo= rk). >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> From: After Effects Mail List =20 >>>>>> Sent: Thursday, February 7, 2019 2:04 PM >>>>>> To: After Effects Mail List >>>>>> Subject: Re: [AE] Feedback on Workstation components >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> Thanks for the input. It has confirmed my intuition to stay away from= the Quadro and Xeon for this build. Looking at either 2x 2080's or 1x 2080t= i now. Don't have funds for 2 of those. :-(=20 >>>>>>=20 >>>>>> No need for 10bit output here. Mostly making HD TV promos on this mac= hine. >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> On Thu, Feb 7, 2019 at 10:58 AM Teddy Gage w= rote: >>>>>>=20 >>>>>> okay I'm seeing a lot of mixed advice. The problem here is that there= is a major real-world consequence of getting the wrong type of processor th= ese days, and it is REALLY important to buy the right system for YOUR workfl= ow >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> The big points -=20 >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> - Unless you NEED 10 bit color output (unlikely) do not get a quadro.= They are intended for industrial CAD / CAM workflows and are 1/10th as fast= for rendering CUDA Redshift. They are intended for use cases where you need= extreme floating point precision, like designing critical infrastructure. N= obody is using Quadros for Redshift / Octane.=20 >>>>>>=20 >>>>>> - The same goes for Xeons. Do not get a xeon system. Many builders tr= y to sell you on xeons because they can make more markup on it. They are ext= remely expensive and offer little to no performance benefits vs. i7 or i9. T= his is basically due to A) lack of overclocking ability and B) way more exp= ensive ECC RAM required (not necessary). The large number of cores means eac= h core is slower than the equivalent i7/i9. AE and most tasks in Cinema are s= till single-threaded. The exception is if you plan on doing a lot of CPU ren= dering in Arnold (which is going GPU soon anyway), or high end simulation wo= rk in houdini etc.=20 >>>>>>=20 >>>>>> - Basically - the more cores you have, the slower each core is, this i= s based on thermal and manufacturing limitations. AMD TR chips are good but t= he same applies - only necessary for pure multithreaded CPU tasks.=20 >>>>>>=20 >>>>>> - Currently, for future-proofing, the only GPU I'd look at right now i= s the GTX 2080ti - but try to find the "blower" models and not the "open fan= " designs, as those tend to just push hot air around inside your case.=20 >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> I'm not going into a specific parts list but from what you're saying I= 'd be looking at a system with an i9 chip with 8-12 cores, high megahertz bo= ost / OC abvility (4.2-5 Ghz) very good all in one liquid cooling (a 360mm r= adiator is essential), and one or two GTX 2080ti's. And yes, RAM prices have= blown up and stayed there due to manufacturing shortages, supposedly.=20 >>>>>>=20 >>>>>> =20 >>>>>>=20 >>>>>> I am currently rocking a machine I built with the core i7 7900X, 4x 1= 080ti, 128 GB RAM for similar workflows and it does great. I will say avoid A= sus motherboards for now, they can have issues >>>>>>=20 >>>>>> -Teddy=20 >>>>>>=20 >>>>>> =20 >>>>>>=20 >>> --=20 >>> _____________________________ >>> Teddy Gage >>> President >>> Shotgun Inc. >>> 360 VR | MOGRAPH | VFX >>> SHOTGUNPOST.COM >>> BKLYN NY 11215 >>>=20 >=20 --Apple-Mail-110D81AF-B6FB-4A25-8AE3-04882D295529 Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: quoted-printable Thank you Carey for sharing your insight. T= hat was a good read

Adam Mercado
Owner & Creative= Idiot
Influxx Media Production 
714.928.9896

Moving Images. For Busine= ss

On Feb 11, 2019, at 6:27 AM, Carey Dissmore &l= t;AE-List@media-motion.tv>= wrote:

A few comments: 

Puget Systems REALLY gets it w= hen it comes to addressing the needs of our market. They have a model where t= hey do a ton of research and provide excellent support for their tested, con= figured system designs. They share a lot of their data and findings on their= website, which benefits the community as a whole. They are uniquely positio= ned to do this type of testing and I=E2=80=99m very thankful they share. For= the vast majority of people who do NOT have interest in building their own s= ystems, but nevertheless are realizing that a custom workstation is truly th= e best path forward for their needs, I can=E2=80=99t think of a better choic= e. 

Now onto t= he geek stuff, (which Puget already knows):

A - Building PC: I=E2=80=99m inclin= ed towards PC building too, and have built many workstation machines. I buil= t my first PC around 2001 mostly for gaming and offloaded encoding/rendering= but I largely stuck with Mac for post production work until about 2012 when= I made the complete jump to primary post production on Windows workstations= . Teddy and I had a good chat on the phone back then just to gut check my sy= stem architecture before pulling the trigger on a $6k box build. I=E2=80=99v= e built and rebuilt many since. 

B-Performance: It used to be that synthetic benchmarks w= ere quite useful in determining performance as it applied to video post work= . You pretty much always needed the highest performance hardware you could g= et and there wasn=E2=80=99t a whole lot of differentiation about tasks, as i= t all sort of lifted with the general performance of the machine. You could l= ook at relative performance and extrapolate your value or expected performan= ce for the apps and work you did and make a decision about value for money/R= OI/Cost-benefit with some simple math.  

Those days are over. That won=E2=80=99t work any= more. The trend for nearly a decade is that PC performance gains have forked= in multiple directions: CPUs have added cores, and clock speed increases ha= ve crept higher but very slowly. Due to thermals, you are often left with a c= hoice of more cores vs more clock speed. GPUs have introduced massive parall= elism tuned for specific compute loads but data has to be moved from memory/= fast CPU access over the bus to the GPU and back so that is a factor in dete= rmining which types of =E2=80=9Cloads=E2=80=9D are best suited to GPU comput= e. The problem is that synthetic benchmarks can be designed to torture test a= specific type of computing load that shows maximized performance of a parti= cular component but it has never been less relevant to apply those to your e= xpectations in the apps you ACTUALLY USE to do post production work. 

Therefore, the ONLY= benchmarks that are relevant are the ones that are done by testing the apps= you ACTUALLY USE to determine a particular component or configuration=E2=80= =99s value to you. Here=E2=80=99s some apps that I use and what I have found= they need from a machine. 

<= div class=3D"">For example here=E2=80=99s some differing compute loads and w= hat types of PC hardware/config will serve it well. 

Storage (applies to all): Everything= wants fast storage and IO. Splitting up storage between media and boot and c= ache, each optimized is good. Storage/cacheing has become an issue too and m= anaging the flow of data and using high performance caching devices separate= from both system drive and media storage is key to optimizing. As you look a= t the applications below

AE: In daily use, largely still single threaded operations with dashe= s of multithreading throughout, and a splash of GPU compute here and there, p= articularly on plugins/effects. Today, this means favoring higher clock spee= ds over core count on the CPU. Get as much RAM as you can justify because AE= and RAM previews will use it. Generally AE-only artists don=E2=80=99t need t= ons of data storage, but what they have should be fast. Often they are best s= erved by something like a 9900k system and a modest GPU from NVidia. 
Note that one of the challenges of optimizing AE for mul= tithreading is that not every type of compute task lends itself well to para= llelism. Sometimes you have to wait for the result of one computation to use= the result of that to start the next computation. This same performance opt= imization issue also affects other similar apps like Fusion. 

Premiere: A little more bal= anced between CPU cores and clock speed and GPU. Having said that Premiere s= till favors clock speed over massive core count, can take advantage of Quick= sync on newer Intel CPUs BTW, and uses GPU but does not rely heavily on it a= nd primarily processes color and effects on GPU (with some exceptions). Prem= iere doesn=E2=80=99t require quite as much RAM as AE, but I guess it doesn=E2= =80=99t hurt. Having said that Premiere seems to need an awful lot of comput= e hardware thrown at it to smoothly handle media at 4k. By comparison FCPX (= which I don=E2=80=99t use because it=E2=80=99s Mac only) is way more optimiz= ed and gets better performance from lesser hardware. I think there=E2=80=99s= a lot of room for improvement for Premiere here but it probably means a rew= rite.

Encoding: h2= 64 or now h265 encoding is massively parallel and really shows off what a lo= t of cores can do. Currently CPU dominated we=E2=80=99re starting to see thi= s moving to GPU options as well. If all you did all day was encode stuff, go= for core count and fast IO. However, for most of us, encoding occupies only= a few minutes per day compared to creative work.

Resolve: GPU dominated (for color page opera= tions anyway). Here, CPU performance matters but a mid-line CPU is probably a= dequate but what you want are GPU compute units. The best units you can affo= rd, plus more of them. 2 or 3 GPUs scale really well with Resolve, but in my= testing I=E2=80=99ve found it best to keep them matched. I did some mismatc= hed testing with different gen GPUs and it=E2=80=99s pretty variable and ine= fficient. Not that every little bit doesn=E2=80=99t help=E2=80=A6but one car= d ends up waiting on the other to finish a task. PCI-E lanes required for mu= lti-GPU support push the user to go with higher level platforms like X299.&n= bsp;

Cinema 4D: Wh= en rendering, massive CPU core count is your friend. But when modeling, key f= raming, animating etc a lot fo that is more single threaded but obviously pr= eviews can be run OpenGL (etc) on the GPU. Other renderers are GPU-dominant.= 3D is not my forte so I=E2=80=99m going to let others cover this.

But I said all of that as a= preface to make this point: You can choose any one activity or app and buil= d a kick-butt machine tuned specifically to the requirements of that activit= y, and walk away with maximized ROI. An AE-only machine is going to look qui= te different than a Resolve-only machine, and again quite different from a C= 4D optimized machine. But often we need to engage in a mix of activities and= the choice about whether to try to build the ultimate =E2=80=9Cdo it all=E2= =80=9D workstation ($$$$) vs separating the tasks is something we all have t= o look at in light of our own mix of work.

Anyway, when I sat down I had no idea I=E2=80=99d w= rite ll that but alas, the coffee was good and plentiful. Plus, when you rea= lly get into this stuff there are just a lot of considerations to get your h= ead around. By no means did I write a definitive guide here (is there such a= thing?) and understand why Chris Zwar=E2=80=99s is over 10,000 words. I loo= k forward to it.

C= arey Dissmore



On Feb 11, 2019, at 6:33 AM, Byron Nash <AE-List@media-motion.tv> wrote:

Puget really impressed me with the depth and detai= l of their hardware analysis for very targeted demographics. It's nice to ha= ve someone really understand what the difference is between editing, motion g= raphics and GPU rendering. 

On Sun, Feb 10, 2019 at 11:52 PM T= eddy Gage <AE-List@= media-motion.tv> wrote:
Sure th= ing Chris, feel free to reach me at teddygage@gmail.com. Goes for anyone- always= happy to consult or chat hardware or software. Have also been doing a fair a= mount of VFX supervising recently for a couple TV shows. 

On Sun, Feb 10, 2019 at 11:14 PM Chris Zwar <AE-List@media-= motion.tv> wrote:
Hi all,

I started researching performance last year, w= ith a view to writing up an article and some thoughts for the ProVideo Coali= tion.  Currently I'm about 50% of the way through my plan and I'm over 1= 0,000 words so it's a bit slow going.  Even so, I was wanting to reach o= ut to list members who are especially interested and knowledgable on perform= ance related matters.

Teddy & Stephen were both on my list of people to approach for input a= nd this thread just reinforces that! Are you two able let me know the best w= ay to contact you off-list please?
And if there's anyon= e else who has strong knowledge or strong opinions on performance/ component= s then let me know too.

Cheers,

-Chr= is


On Sat, 9 Feb 2019 at= 07:18, Byron Nash <AE-List@media-motion.tv> wrote:
Those are good thoughts Stephen and I'll keep that in mind going for= ward. I am planning to use a Blackmagic card for video output but the monito= rs I have to use right now are 8bit HD. I'm starting to get into 4K VFX for f= ilms on my machine but most of the rest of the team is building HD promos/mo= graph content. My only pet peeve with the BM cards is when you have a 4K scr= een it won't downscale through the card and just goes black. 

O= n Fri, Feb 8, 2019 at 11:10 AM Stephen van Vuuren <AE-List@media-motion.tv> wrote:

> In regards to 10bit output an option not being discussed is using a= Blackmagic or Aja PCI card or device. These will give you the option of 10b= it YUV or RGB output (monitor dependent). 4K capable versions of these are a= vailable for as little as $200.

 

I me= ntioned external card but you are not budgeting for the 4K 10-bit display th= at you will need to see the output.

<= p class=3D"MsoNormal"> 

I have a $199 BM MiniMonitor 4K and it=E2=80=99s great going o= ut to a 50=E2=80=9D HDR TV). But that does not get you 4K 10-bit on your desktop display or= apps that don=E2=80=99t support that card. You have to have a second mo= nitor where as Quadro allows you to have 4K 10-bit on your main desktop moni= tor.

 

If you only have r= oom for on 4K 10-bit, a card plus additional monitor is not an option. And e= ven if you do, my setup I would rather always see 10-bit as much as possible= in all my work and catch issues.

 

As video moves to 10-bit and HDR in 4K, even for low-end, 10-bit= is going to be a bigger deal.

 

From: After Effects Mail List <AE-List@media-motio= n.tv>
Sent: Friday, February 8, 2019 12:25 AM
To: After Effects Mail List <AE-List@media-motion.tv>= ;
Subject: Re: [AE] Feedback on Workstation components

 

If you=E2=80=99re talking about a bud= get system Quadro cards aren=E2=80=99t even in the conversation (5k was ment= ioned). If one of the main uses of the machine will be Redshift you=E2=80=99= re better off buying the fastest GTX or RTX card(s) you can afford. The RTX cards may end up being a better choice in the near future, if Nvidia=E2= =80=99s and Redshift=E2=80=99s recent announcements are to be believed.

 

In regards to 10bit output an option n= ot being discussed is using a Blackmagic or Aja PCI card or device. These wi= ll give you the option of 10bit YUV or RGB output (monitor dependent). 4K ca= pable versions of these are available for as little as $200.


Dennis Wilkins

 



On Feb 7, 2019, at 8:49 PM, Stephen v= an Vuuren <AE-List@media-motion.tv> wrote:

 

Steve,

 

That link is for 10-bit DirectX outpu= t which is for games and Direct X apps =E2=80=93 not relevant to this discus= sion. Adobe and most other pro post apps don=E2=80=99t use or support Direct= X. And just because a driver or monitor flags 10-bit does not mean there is actually 10-bits happening.


Better info is here: https://imagescience.com.au/knowledge/10-bit-output-support

 

From: After Effe= cts Mail List <AE-List@media-motion.t= v> 
Sent: Thursday, February 7, 2019 11:40 PM
To: After Effects Mail List <AE-List@media-motion.tv>
Subject: Re: [AE] Feedback on Workstation components

 

10bit display support is all over the= place on mac. MBP 2016 is driving a 24=E2=80=9D external HDMI monitor in 10= bit will driving the native display in 8bit. recent model imacs with 5K disp= lays running in 10bit. 

 

you can get nV cards on win to run in= 10bit. here is how to run nV or AMD on win10 in 10bit  https://www.pugetsystems.com/labs/articles/Setting-Graphi= cs-Card-Software-to-Display-10-bit-Output-1221/=

 

 

S=




On Feb 7, 2019, at 10:25 PM, Stephen v= an Vuuren <AE-List@media-motion.tv> wrote:

 

>>> Unless you NEED 10 bit c= olor output (unlikely) do not get a Quadro. They are intended for industrial= CAD / CAM workflows and are 1/10th as fast for rendering CUDA Redshift. The= y are intended for use cases where you need extreme floating point precision, like designing critical infrastructure. Nobody is using Qu= adros for Redshift / Octane. 

 

I disagree that 10-bit is =E2=80=9Cun= likely=E2=80=9D. It=E2=80=99s more of a ignorance of benefits that an unlike= ly scenario. I get a lot of pro colorist and post house delivering stuff wit= h banding, too much shadow noise and color/gamma problems that have obviousl= y been graded in 8-bit on typical 1000:1 panels. See it all the time. 


Same as I see people mixing on headphones or uncalibrated speakers and charg= e pro prices for the service. The defense is always the same =E2=80=9Clooks/= sounds great here!=E2=80=9D.

 

GeForce cards cannot display video in= Premiere or images in Photoshop in 10-bit, period or most any other pro app= . You have to have a monitor card going to a separate, second monitor. That c= ard or box + monitor for a 4K setup is not cheap.

 

Quadro cards offer native 10-bit OS l= evel support with Quadros and Mac is now fully supported.<= u class=3D"">

 

Most people think they do don=E2=80=99= t need 10-bit output but end up delivering work that looks bad in 10-bit, 12= -bit and HDR environments or displays with wider than the 8-bit (and 6-bit+d= ithering panels out there).

 

DCP is 12-bit. Rec2020 is 10-bit as i= s all Netflix, Amazon etc. HDR> Hell, YouTube HDR is 10-bit.https://support.google.com= /youtube/answer/7126552?hl=3Den

 

If you=E2=80=99re delivering 10-bit a= nd while it used to be rare, these days almost everything I deliver is 10 bi= t or higher.


So I just don=E2=80=99t take people mastering final projects for professiona= l deliver very seriously that don=E2=80=99t have a fully 10-bit or higher wo= rkflow.


Quadro plus a good affordable HDR 4K 10-bit monitor is the same or less mone= y than a GeForce + 10-bit video output plus dedicated 4K display and offers b= etter drivers, more stability and more flexibility (like Photoshop work).

 

 

From: After Effe= cts Mail List <AE-List@media-motion.t= v> 
Sent: Thursday, February 7, 2019 2:04 PM
= To: After Effects Mail List <AE-List@media-motion.tv>
Subject: Re: [AE] Feedback on Workstation components

 

Thanks for the input. It has confirme= d my intuition to stay away from the Quadro and Xeon for this build. Looking= at either 2x 2080's or 1x 2080ti now. Don't have funds for 2 of those. :-(&= nbsp;

No need for 10bit output here. Mostly= making HD TV promos on this machine.

 

On Thu, Feb 7, 2019 at 10:58 AM Teddy= Gage <AE-List@media-motion.tv= > wrote:

okay I'm seeing a lot of mixed advice= . The problem here is that there is a major real-world consequence of gettin= g the wrong type of processor these days, and it is REALLY important to buy t= he right system for YOUR workflow

 

The big points - <= /u>

 

- Unless you NEED 10 bit color output= (unlikely) do not get a quadro. They are intended for industrial CAD / CAM w= orkflows and are 1/10th as fast for rendering CUDA Redshift. They are intend= ed for use cases where you need extreme floating point precision, like designing critical infrastructure. Nobody is using Qu= adros for Redshift / Octane. 

- The same goes for Xeons. Do not get= a xeon system. Many builders try to sell you on xeons because they can make= more markup on it. They are extremely expensive and offer little to no perf= ormance benefits vs. i7 or i9. This is basically due to A) lack of overclocking ability and B) way more expensive ECC RAM re= quired (not necessary). The large number of cores means each core is slower t= han the equivalent i7/i9. AE and most tasks in Cinema are still single-threa= ded. The exception is if you plan on doing a lot of CPU rendering in Arnold (which is going GPU soon any= way), or high end simulation work in houdini etc. 

- Basically - the more cores you have= , the slower each core is, this is based on thermal and manufacturing limita= tions. AMD TR chips are good but the same applies - only necessary for pure m= ultithreaded CPU tasks. 

- Currently, for future-proofing, the= only GPU I'd look at right now is the GTX 2080ti - but try to find the "blo= wer" models and not the "open fan" designs, as those tend to just push hot a= ir around inside your case. 

 

I'm not going into a specific parts l= ist but from what you're saying I'd be looking at a system with an i9 chip w= ith 8-12 cores, high megahertz boost / OC abvility (4.2-5 Ghz) very good all= in one liquid cooling (a 360mm radiator is essential), and one or two GTX 2080ti's. And yes, RAM prices have blown up and stayed t= here due to manufacturing shortages, supposedly. 

 

I am currently rocking a machine I bu= ilt with the core i7 7900X, 4x 1080ti, 128 GB RAM for similar workflows and i= t does great. I will say avoid Asus motherboards for now, they can have issu= es

-Teddy 

 

--
__________________= ___________
Teddy Gage
President
Shotgun Inc.=
360 VR | MOGRAPH | VFX
SHOTGUNPOST.COM
=
BKLYN NY 11215

=

= --Apple-Mail-110D81AF-B6FB-4A25-8AE3-04882D295529--