Gadgetory

since the launch of nvidia quadro r-tx line of GPUs last fall we've seen the collection expanded four models total with the r-tx 5000 on deck today sitting third from the top or two if you consider that the main difference between the six and 8000 is the size of the frame buffer as a 5000 series card this Quadra r-tx acts as a high-end probe is option delivering great performance of both creative and gaming applications but doesn't cost as much as a high-end workstation PC on its own compared to the outgoing quadrille p5 thousand the neutering based r-tx 5000 gives us more cores 3072 in total placing its peak performance ahead of the gaming targeted geforce r-tx 28 ET i before that this video is brought to you by e BJ's a new audio sound card engineered by audio note EVGA a CEO knows high quality audio and has begun bringing sound cards back the new audio sound card is capable of delivering hair-raising audio superior to onboard sound the card includes a line in headphone line out and mic in and a Sony Philips digital interface new audio also leverages the EVGA ace PCB design experience has upgradeable op amps and uses AKM premium components for its DAC and ADC learn more at the link in the description below while the frame buffer size has remained the same from generation to generation NVIDIA has delivered more than a 50% boost to memory bandwidth thanks to the move from g5 x 2g ddr 6 while it's obvious that the latest and greatest is going to perform better than the same class card of yesteryear the performance proof seen with the RT X 5000 come at the expense of increased power draw but when it boils down to it the quicker your renders and encodes execute the less power you'll be using overall it's commonly asked why Quadros and Radeon pros cost so much more than the gaming counterparts and the answer isn't too difficult the simplest explanation is that Rd certifications and optimizations lead to a good chunk of the cost representing improved performance in industry software as well as improve stability speaking of error correction memory also plays some of the role in the cost of the higher end part of the stack for those with 10-bit color needs gaming cards are fine for DirectX applications but OpenGL software will require pro level cards something that applies to both Nvidia and AMD there are some exceptions to this rule but trying to find a list of supported versus non supported 10 bit applications on gaming cards is something that will be done in vain so if you want to guarantee your color you will want an appropriate graphics card we've covered before there is no one-size-fits-all where gravis cards for creative workloads is concerned one GPU may be fine for a certain task well another you'd expect to do well could fall flat on its face to that end the performance ahead covers a wide range of different workloads and even if your application is not covered directly there's likely some other performance gauge that's relevant we'll begin gauging the r-tx 5000's Worth with the help of some encoding applications with Adobe's popular Premiere Pro leading the pack while we're currently only able to offer encoding performance we do have plans to add playback performance at some point in the future after time can be dedicated to developing some useful and reliable tests this first chart is a bit of a funny one since the top half performs pretty much identically with this particular encode this is partially due to the fact that this is a simple transcode of one resolution and format to another whereas a normal Premiere Pro project like those produced right here at gamers Nexus will throw in a lot more variants the more complex your project the longer it's going to take to encode and the more likely the GPU will become better involved in the process there's a clear point of diminishing returns with this simple performance look but this chart still offers a good lesson of never going to low end on your GPU if you do you can potentially cripple your encoding performance with a meagre bump and model at the low end making a dramatic impact of performance how do things change up for a proper project with this 1080p youtube project we see more interesting scaling than before but it's clear that the GPU only has so much say in a project like this when you start to pile on effects specifically designed around the GPU you'll begin to appreciate having a bigger card more but if working inside a Premiere Pro is your primary and most important task you clearly don't need a top-end GPU to get the job done but you also don't want to skimp on the CPU to complement Premiere Pro now encoding results we have a magics Vegas Pro playback test as well as a couple of encodes with the performance scene in this chart we have some explaining to do many video editors could be called unoptimized even Premiere Pro sometimes but the unoptimized can really hit Vegas Pro hard to make a long story shorter in video performance in Vegas Pro isn't ideal right now although it largely depends on whether or not you have a GPU that is affected by unexpectedly poor performance surprisingly in perhaps conveniently the quadrille RTS 5000 is the only quadrant our entire lineup that managed to handle this test fine although again it doesn't makes sense given the other performance we're seeing what this ultimately means right now is that NVIDIA is simply not an ideal choice for Vegas Pro and we really hope that will change in the future a recent build of Vegas Pro was released since this performance was tabulated but follow-up testing didn't change any of our results fortunately for everyone involved in code performance is far more realistic than playback but it's still unpredictable in some ways you'll notice that some GPUs end up falling behind technically more powerful options Nvidia is turning our detector seems to give that company's cars a slight performance advantage over the last gen but it's still not enough to compete ideally against AMD and thus we yet again reach a conclusion that Vegas Pro is AMD territory Autodesk recently released the first GPU powered beta of its popular Arnold renderer so naturally we couldn't wait to dive in at the moment Arnold GPU is in no way considered a production renderer as it's still missing some functions you may be looking for so it currently acts more of a preview of what's to come and that time will indeed come when a render made with the GPU will look like a render made with the CPU that all said Arnold GPU is currently only available for CUDA based graphics cards though that's something that could change in time as a current trend seems to be more renderers opening up a bit more in time Arnold will also support Nvidia's r-tx specific features like the ray tracing cores some software can really earn its beta tag and Arnold GPU is one of them despite the GTX 1660 Ti being a capable GPU it suffered so hard during a render here that we had to give up and letting it finish we're talking waiting 10 minutes still seeing zero percent progress and then having to kill the process we eagerly await future versions of the plug-in to see how this situation improves the GTX 1660 Ti isn't the only GPU struggling even the RT X 5000 we're focusing on is having a bad time despite being technically superior to both the GeForce RT X 2070 in quadrille RT X 4000 it still falls behind with a different project we get a second opinion but it's not really a better one a single GPU failed the first render but two more have joined it here even including the more than capable GeForce r-tx 2060 and we still have some GPUs falling behind more powerful GPUs like the RT X 2080 taking twice as much time as the RT X 2070 AMD's Radeon Pro render may not be as popular as other renderers in this video but it's slowly growing in popularity as a continues to iterate on its capabilities instability where AMD has done really well with pro render is with its performance while at last check multi-gpu testing with nvidia is a bit wonky single GPU configurations across-the-board scale very well pretty much as expected the pack of radians performed extremely well against nvidia in our Vegas protests but Nvidia strikes back hard with AMD zone renderer with his turning architecture offering a serious advantage to performance we've gone from 214 seconds with the Titan XP one generation to 101 seconds with the Titan RT X the next generation that's a monstrous gain but you don't need to tighten our TX to enjoy seriously boosted performance the RT X 5000 itself exhibited strong performance placing itself well ahead of the RT X 4000 and right behind the gaming focused our TX 20 a DTI moving from one scene to another finer tune scaling can be seen but the middle of the chart shakes things up a bit Nvidia stop turning cards still lead the pack with a healthy advantage with the Titan RT X once again proving more than twice as fast for this work as the older Titan XP fortunately for those I in the RT X 5000 performance is looking really good right about now naturally we'd have to point out that the bottom rung cards offer pretty brutal performance the WX 7100 and RX 590 pretty much croak here so we can safely say you will ideally want nvidia turing 4 pro render work or at least AMD Vega but with the gtx 1660 TI keeping rate up to the Radeon Pro WX 8200 Nvidia has a definite lead over all the upcoming blender 2.8 brings a lot of great stuff to the table including an improved interface for robust animation tools a brand new renderer called Eevee and of course performance boost to the classic cycles renderer blender 2 point AIDS viewport is powered by the new Eevee renderer and with the change comes a brand new option called lock development or look dev for short compared to solid shading look to have loads of project shaders and textures to give a good interpretation of what a scene is going to look like when it renders you'll still likely use solid mode for most work using looked at whenever you want to quickly gauge changes without wanting to wait on an actual render for those designing their projects at 1080p you'll be happy to know that you don't need to go top n to get decent performance while 60 frames a second is always a great ballpark we're not dealing with fluid gameplay here so don't let the lower values scare you for this kind of work even 30 frames a second when bumping the resolution to 4k the going really gets tough in videos turning cards once again exhibit some seriously strong performance here with the top three RT X cards we have on hand ruling the top of the chart AMD's Radeon 7 deserves an honorable mention though it even manages that performed last gens Titan XP this chart yet again helps us prove that you never want to go with a really low end graphics card when you were actually trying to get good work done while 30 frames a second is suitable enough for luke dev cutting that in half is not going to bode well with all of the testing we've done with blender we can honestly say that the GPU is going to be much more important for your render than the CPU you'll never want to skimp on your CPU of course but blender is just one of those applications it doesn't require a big CPU in order for a big GPU to do good work alongside it Nvidia stirring architecture has been on a relative rampage so far and that continues on through to these rendering tests the same three cards sit at the top with the RTS 5000 sitting in third place ahead of the Titan XP with a different render this time an interior scene the scaling changes a little bit with the Radeon 7 gaining a couple of spots putting itself ahead of the RT X 5000 it's interesting that the Radeon 7 handles the more complex scene better than the simpler BMW 1 but it's a good thing to see for the red team there are some exceptions but overall the faster your GPU the faster your renders in blender are going to be there are just going to be better options at a given price point so hopefully these charts will help you make your decision easier we're tackling many different renderers here but aside from the upcoming octane bench all of them require a full software suite to benchmark Lux mark acts as a great at-home way to run a super quick test that allows you to compare your systems performance to ours funny enough Lux mark was heavily promoted by AMD at the launch of its Radeon 7 so the reasons why are not hard to find out when you look at this chart with the lux ball render the Radeon 7 simply dominates with the same top-flight NVIDIA GPU sitting behind that in reality the Radeon 7 doesn't have as strong an advantage over the rx Vegas 64 as it does in this particular test but it sure would be interesting if it did the Lux ball scene is simply reflective ball sitting in space whereas the hotel scene is much more complex thanks to a huge number of reflective surfaces that grace the lobby with this project the performance picture changes a little bit with the Radeon 7 falling behind the same 2 NVIDIA GPUs that dominated last time ultimately both AMD and NVIDIA perform extremely lexmark with turing once again showing some strong performance along with AMD's Vega based GPUs unlike Autodesk Arnold renderer chaos cubes v-ray supports both Nvidia and AMD GPUs however the AMD OpenCL support seems to be in its earlier stages or our chosen projects have at least refused to render without issue on radio and hardware so for now we're sticking to Nvidia but hope to add AMD into our v-ray tests in the future note that we're avoiding the standalone v-ray benchmark as his graphics test has always left a bit to be desired and ultimately it's based on the older v-ray 3 not v-ray 4 which was released last fall the first of 2 v-ray scenes we test with gives us an interesting result where the RTS 5000 falls behind the RT X 4000 this is one of those occasions where logic effectively goes out the window because subsequent testing delivered the exact same results if you were paying close attention to the Lux ball results from earlier you may have noticed that the quadrille P 5000 fell behind the P 4000 which was yet another repeatable result sometimes oddities happen but it's interesting that our two present examples involve the same series GPUs from different generations when we move to the more complex t sat scene the RT X 5000 recovers itself from embarrassment now placing an expected distance ahead of the RT X 4000 the RT X 2060 keeps close to the gtx 980ti while the quadrille RT x 4,000 enjoys it plays ahead of the titan XP redshift is another renderer that's going to eventually take advantage of the RT x-series to RT course but without those the performance scaling still looks good the Quadro RT x 5000 is one of the fastest of the lot again coming ahead to last gen stein XP and for a fun comparison well ahead of last gens $5000 pascal pays quadrille p6000 the better your NVIDIA GPU the better your redshift performance it really is that simple at least until the renderer begins supporting our t axis special features at which point Turing is likely to look even more attractive like redshift o toys octane render supports only CUDA based graphics cards but that's something that will be changing in the near future especially as the company would like to continue supporting Mac OS which currently only supports Radeon from generation to generation octane render gives us a great impression of gains with both the RT X 20 a DTI and Titan r-tx sitting well in the clear up top the quadrille RT X 5000 actually falls slightly behind the Titan XP here hinting that octane hasn't benefited quite as much from the turning architecture as some of the other renderers for a second octane opinion we've also tested the beta version of octane bench which not only includes support for Nvidia RT X but even gives us a before-and-after score with our TX off we see similar scaling as with the original benchmark but when RT X is turned on the gains are simply incredible effectively tripling the performance of any given GPU how this performance will carry over to the real world we're not able to attest quite yet but based on this benchmark things are looking amazing we took care of blender viewport performance earlier but to take care of a bunch more we enlist the help of spec view perf 13 this is a standalone benchmark so anyone can download and run it to compare against our results the entire suite uses application traces to reliably recreate how real applications would utilize the hardware it's not a perfect gauge of real-world performance in these applications but it's a close enough gauge that both AMD and NVIDIA promote its use we're going to focus on 4k viewport resolution here with the exception of this first test suspect view / doesn't have it available in 4k both 3ds Max and Maya are pretty neutral with regards to performance between gaming and workstation cards so the faster your hardware the better the fluidity of the viewport at least with this particular test AMD falls behind Nvidia with the lowly GTX 16 DTI managing to outperform the technically superior Radeon Pro WX 8200 meanwhile the RT X 5000 performs exceptionally well leaping some 20% ahead of the last gen quadrille p6000 moving on a 4k resolution in maya we see fairly similar scaling as we did with 3ds max but the quadrille r TX 5000 falls a bit further behind the GeForce r-tx 20 a DTI than it did before it still sits in third place however again ahead of the Titan XP in quadrille p6000 the gtx 1080 TI holds on for dear life refusing to give up its position to the quadrille r TX 4000 a good chunk of the software featured in this video works similarly on both the gaming were station GPUs but CATIA bucks the trend it clearly favors pro level cards not just on the nvidia side but AMD's as well though Nvidia does seem to have an advantage overall in the matchup between the Radeon Pro WX 8200 in quadrille RT X 4000 which both cost around the same amount Nvidia card pulls comfortably ahead on the AMD side the radio and seven once again offers admirable performance the Titan cards enjoy some great performance here as well as CATIA is one of the applications that benefits from special workstation optimizations on those cards with SolidWorks 2 so Systema stakes its preference for workstation cards and cranks it rate up the performance advantage for workstation cars is so strong here that'd be hard to imagine going with a gaming GPU for this kind of work if you check out the show's official GPU support page for SolidWorks you may be surprised to see no gaming GPUs listed at all an added benefit were station GPUs also enjoy with SolidWorks as the real View mode which spec view purse test doesn't take into account since it simply can't work on gaming GPUs well can't is a loaded word since it's clearly possible if not for the drivers deciding on it but we're talking about a software solution that cost more for an annual license than the Quadro r-tx 5000 does SolidWorks cues his performance heavily towards workstation graphics cards while PTC's creo only mildly does with its workstation optimizations in tow the last gen quadrille p6000 manages don't press here more than usual but at the same time the gaming targeted RT X xx atti performs even better than it does meanwhile the Titan RT X delivers a strong lead at the top and the Quattro RTS 5006 just behind that in the Battle of Nvidia versus AMD the green team definitely wins here the first Radeon to make an appearance is the Radeon Pro WX 8200 10 rows in if we add a WX 90 100 to test it likely plays ahead of the art TX 2060 we saved the best viewport chart for last and by best we mean the most interesting well in a solution like SolidWorks you can get by with a gaming GPU even though the software clearly prefers professional cards Siemens and X unapologetically treats gaming GPUs like second-class citizens where even the lowest end Radeon pros will outperform the highest end gaming GPU snx is another example of a software package so expensive that its users are not likely to gas but the higher cost of workstation hardware especially when reliability and optimizations are weighed so heavily moving away from applications we wanted to take a quick look at the RT x 5000 performance in a couple of extra categories including cryptography as seen here and also scientific and financial analysis the better graphics card is at math the faster your complex compute jobs are going to complete the nice thing about SCI software as Sandra is that it gives every architecture a fair shake if there's an optimization available the application is going to be taking advantage of it giving us the best possible scenario out of any GPU for AMD that bodes well as we're seeing the Radeon 7 lead every single other GPU in the list even the RX Vegas 64 is flexing its muscles here following the RT X xx atti from just a few feet behind the quadrille RT x 5000 ranks a bit lower than we'd expect to see here falling short of even the GTX 1080 TI's performance performance picks right back up for the RTS 5000 in the financial tests now giving us the same top three we've seen throughout many of these results the Radeon 7 dominated the crypto test but Falls a few pegs back in the financial test it still performs extremely well when compared to the rest of the stack on downward highlighting great gains over Vegas 64 at the low-end the cards struggle in a test like this which gives us weird results like the W X 4100 and W X 5100 performing almost identically for the most part card scale here as we'd expect with both turning in Vegas 7 offering big performance up lifts with the scientific analysis test scaling at the top changes a little bit but the same top 4 cards remain however the RTS 5000 this time Falls but behind the others still offering solid scaling versus the r-tx 4000 sibling you may remember that right after the Radeon 7 launched AMD issued a driver update than the LOC double precision performance to increase its effectiveness fourfold because nvidia locks double precision on every GPU that doesn't cost multiple thousands of dollars the Radeon seven will always win this battle being that the quadrille r-tx is aimed at the workstation crowd were going light in gaming tests but did want to get a couple of 3d mark and VR mark runs in for a basic gauge of whereas scales at 4k resolution with specs similar to a geforce r-tx 2070 the quadrille RTX 5000 nears the gaming performance of the GT X 1080 Ti and lingers just behind the last gen quadrille p6000 in this particular test the Radeon 7 has the right to gloat as it sits in between the Titan XP and 1080 Ti with the DirectX 12 times by test performance changes once again thanks in part to the same Turing optimizations we've talked about a few times before in VR mark Turing optimizations reveal themselves once again to boost performance on both the GTX and RT X cards nothing can touch Nvidia's top two GPUs but the quadrille RT X 5000 says behind those if nothing else the chart proves that future VR content is going to be gruelling for GPUs and at least in the case of VR Mark Nvidia strengths in VR are currently better than AMD's with all of that performance covered you're hopefully far better informed now than you were earlier about which card you should be pursuing next as covered earlier even if your particular application isn't covered here much of the performance scaling should still be relevant it's only when we get into certain higher in CAD sweets when performance optimizations begin to favor workstation cards as for the Quadro RTS 5000 it's a lot of GPU in the significant upgrade over the previous generation quadrille P 5000 in some of our tests the gains with the new card were largely expected but in others Turing's architectural enhancements gave us even greater gains in one example we saw the neutering based Titan r-tx literally double the performance of the Pascal based Titan XP sadly that performance scaling is not seen across the board but it'd be quite interesting if it was at around $2,300 the Quadro RTS 5000 carries a big premium over the RT X 4000 but it also delivers more goods such as big performance improvements in some cases as well as a doubling of the frame bar sighs 8-gigabyte is a standard size for workstation cards today but that'll change before long and moving from Pascal tutoring we saw the top-end Quadro double from 24 to 48 gigabytes so the uptake in memory demand is clear and speaking of it wasn't tackled explicitly earlier but the RTS 5000 does in fact include error correction memory whether the RTS 5,000 is right for you largely depends on your budget and what you expect to see as a return from your work Quadro is all about stability and optimization making it a natural choice over gaming GPUs but if your demands are not so critical a geforce card can often fit the bill but the biggest exceptions being highlighted in this video and we once again come to the end of a performance results tell yuge if you love this kind of content you can support gamers Nexus that store gamers Nexus net as well as on patreon at patreon.com slash gamers Nexus and if you can stomach more of me you can Scout more of this type of content out at the tech edge YouTube channel thank you for watching and catch you guys again soon