Is it just me, or does a 32" screen feel smaller with every passing day? Honestly, those standard 23 or 24-inch monitors I used to rely on now feel about as tiny as a postage stamp.
That’s honestly pretty rough. Most of these components feel like they’re straight from the bottom tier of the market. If you could save a few hundred bucks by picking up a cheaper monitor, wouldn't it make more sense to redirect that extra cash toward...
I just stumbled upon some fresh data regarding the 6800 XT in Resolve:
https://www.computerbase.de/2020-11/...800-xt-test/5/ (check out the results under the standard Candle benchmark)
That Puget Systems benchmark—which feels like a bit of mumbo-jumbo since it’s so heavily dependent on your CPU, drive speeds, and file types—isn't really the best way to measure pure GPU performance, is it? According to these numbers, the RX 6800 XT is either slightly ahead or holding its own, but it absolutely crushes the RTX 3090 when it comes to TNR. Of course, we have to remember this was done at HD resolution; could things shift once we start looking at UHD? Still, when you consider that the RX 6000 series sits around $1,000, which is roughly half the price of the RTX 3090, the fact that the performance isn't even close to being half as bad is pretty impressive. (I can't seem to pin down the exact standalone price for the RX 6800 XT because everywhere I look, they're just bundling it into pre-built systems rather than selling it individually.)
I’m not seeing massive leaps from the RTX 3060 when it comes to running da Vinci, as it’ll likely hover around the same performance level we saw with the 2060 Super. If you look at the current 3060 Ti, it sits pretty much right in that sweet spot between the 2060 Super and the 3070... though, of course, all of that depends on whether the price actually makes sense for you, and having that 12 GB of VRAM certainly doesn't hurt.
@Timothy Kim60/">@@Timothy Kim60: The real headache is that all the GTX cards are capped at just 6 GB of RAM, though I have been weighing the idea of running an AMD and Nvidia combo specifically using OpenCL mode. While everyone talks about how RT in Resolve is a game changer, or how Tensor cores are supposed to supercharge things like Optical flow and Temporal Noise Reduction x1, does that actually hold up in the real world? It often feels like it doesn't quite live up to the hype.
In most tasks, the RTX 2 xxx and 3xxx series generally handle Navi just fine, but it gets interesting when you look closer; in Temporal Noise Reduction x1 and Optical flow—areas where the Tensor cores should theoretically dominate—Navi holds its own quite impressively. In fact, looking at the performance of the RX 6900 XT versus the RTX 3090, they are remarkably neck-and-neck, and in some instances, the Navi card actually manages to outpace the 3090!
It leads you to wonder, isn't it? Such bizarre results...
@wiredotter14/">@@wiredotter14: Since I work as a freelancer, I don't have a formal company setup to deal with... though honestly, it’s becoming such a struggle to find a used RX 580 8GB at a decent price, isn't it? Given how old they are and the sheer amount of new GPUs hitting the market lately (at least on paper), if the price difference is only a few bucks, I’d much rather just buy something brand new. I was scrolling through eBay again, and the selection of decent used cards is looking pretty slim, maybe $500+ options worth checking out. This one right here actually claims to be NEW in the title, yet when you read the description, it clearly says it's used...
Are we talking about DaVinci Resolve here? If so, you really need a more powerful graphics card and at least 16 GB of RAM to get by. But looking at that CPU—which honestly performs about the same as an Intel Core i7-4770K—it might be smarter to just aim for a GTX 1060 or a GTX 1060 Ti given how crazy card prices are right now. That processor should handle decoding a single 4K h265 stream or maybe two h264 streams, but after that, you're relying almost entirely on the GPU.
Most of the heavy lifting in After Effects is handled by the CPU, though they did move some specific effects over from New York City once CC became a thing, such as Motion Blur and Lumetri Color.
Thinking about building a new rig? Honestly, don't even bother with anything less than 6GB of VRAM; just skip it. If you grab an RTX 2060 Super with 8GB, you'll be set for most tasks for quite a while, including DaVinci Resolve. Though, I suppose we can all hope that prices will finally drop early next year! You'll want a processor with solid single-threaded performance to keep things smooth in After Effects and the Fusion tab in DaVinci, but for most video content creation, decent multi-threading is usually plenty fast enough.
wiredotter14 said:What exactly are you using it for?
Well, I mainly use it for DaVinci Resolve, specifically for real-time playback with certain effects—though I'm not really talking about color grading here. My goal is simply to make sure the director isn't breathing down my neck because they have to wait for a full render just to see if a cut feels right or if the pacing works... but honestly, what can I say? We all deal with the same pressures, so I won't dwell on it too much. 😬 In Resolve, graphics cards don't actually use Crossfire; instead, each card handles its own specific frame, which means if one card is significantly weaker than the others, the more powerful ones end up just sitting there waiting for it to catch up. In Resolve, almost everything is handled by the GPU.
My buddy Christian Torres3 helped me figure out that the RX 5500 XT is actually quite a bit slower—sometimes by as much as 0-13 fps depending on the specific task and how many effects you throw at it—compared to the RX 580. It seems pretty clear that it was designed strictly with gaming in mind, so there isn't much way around those limitations. That’s why I’ve shifted my focus toward finding an RX 570 instead; it should be only negligibly slower than the RX 580 when it comes to compute tasks, and it generally has a smaller physical footprint. I found this Asus model:
Since I already have an Asus motherboard, using this card offers some nice compatibility perks, plus this specific model features a dedicated fan connector where I can plug in a small side fan to help cool both cards if things start getting too hot.
I realize I might be overthinking this and bothering everyone with so many details, but I'm approaching the finish line of a project where I'm really going to need that extra GPU power.
So, a new plan is starting to take shape: I'm thinking about swapping out my current card for an RX 5500 XT instead of going with another RX 580. On paper, the RX 5500 XT pulls about 50 W less power than the RX 580, and while prices for both are hovering around the same level right now due to how the market is looking, who knows? Maybe a good sale will pop up soon. With my existing RX 580, I can actually dial the power down to 100W in the software, though in reality, it's probably pulling closer to 130 W. Moving to the newer card should theoretically keep everything running much cooler, shouldn't it? My main hesitation, however, is that while they seem comparable in raw performance, the online data regarding how the RX 5500 XT handles content creation and compute tasks is all over the place, and that 128-bit bus is definitely making me a bit nervous.
If anyone happens to own an RX 5500 XT and wouldn't mind lending a hand, I'd really appreciate it; I can easily walk you through installing DaVinci Resolve and a benchmark so you could send me the results.
Oops, I can't quite remember if I had actually applied the undervolt when I was taking those measurements earlier, and since my meter is sitting right there on the plasma screen, I really don't feel like getting up to move it again. I suppose I could try walking through it step by step—maybe grab the graph first, then attempt an undervolt combined with an underclock down to what is typically considered standard for polaris (though I should probably measure exactly how much of an FPS hit I take first, even if it shouldn't be more than one). If the system keeps shutting down regardless, then I guess it's time to look into a new power supply.
Just to clarify, the power consumption was measured while running an undervolted GPU—as I mentioned before—but without any underclocking (staying at 1380 MHz), all while the CPU (R1600X) was overclocked to 4GHz. Dropping the GPU clock by 100-200 MHz doesn't actually make much of a difference, though you might lose a few FPS here and there; the more critical factor is keeping the VRAM at 2250 MHz. As for Compute, it doesn't seem to offer any real advantage, likely because the overhead from the GPU-CPU/RAM transfer for every single frame basically cancels out any gains from faster compute speeds. Honestly, I’d probably just grab an RX 570 instead, since it's about 5% slower and if the price is the same, why bother? :coffee:
I used a power meter to check exactly how many watts the whole rig was pulling from the wall while running a combined stress test on both the CPU and GPU, and it clocked in somewhere between 380 and 400 W. When I ran Unigine Heaven, it pulled about 300 W, though the CPU wasn't being pushed to its absolute limit there.
The most aggressive undervolt I managed to pull off without causing the drivers to crash was dropping from 1.150 V down to 1.075 V.
Timothy Kim60 said:So, that reading shows 120W, but don't you think the information provided by MSI Afterburner or similar tools doesn't actually represent the total power draw of the GPU? 😁 So, looking at this setup, the power supply definitely needs to be swapped out for something else entirely. Where the old unit sits, there’s just not enough wattage to handle the load, and honestly, the build quality isn't exactly anything you'd call reliable or serious. Wouldn't it be better to start with a solid foundation?
It seems like you might have hit "send" a little too early! Did you have a specific post or some text you wanted me to rework for you? Just paste it whenever you're ready, and I'll get to work transforming it into that smooth, conversational American style we discussed. $167 You really need something that strikes a balance between solid build quality and raw power. Wouldn't it be better to aim for something a bit more substantial? Perhaps the best move would be to go for... Is there anything else you need me to work on?
I've been checking my power consumption through the AMD adrenalin software lately, though I can't help but wonder if everyone else is just seeing the exact same numbers?
I almost forgot—I was actually looking at this one: I was browsing through some hardware listings online recently and couldn't help but wonder about the current state of the used market. It really makes you stop and think, doesn't it? When you see all these different components floating around on sites like eBay, you have to ask yourself if the value proposition actually holds up for the average builder today. Is it truly worth hunting down older gear, or are we better off just saving up for something modern? It’s an interesting puzzle to solve when you're trying to balance a budget against performance.
It’s a bit strange when you look at the prices on international retail sites, isn't it? It seems like every single RX 580 out there is actually more expensive than what we're seeing here in the States.
I've also seen some listings popping up on eBay for what's supposedly brand new stock, though one has to wonder about the actual origin of those items? $417So, I was wondering, are there any hidden pitfalls or specific things to watch out for when looking at that XFX card?
I've been sitting here weighing the pros and cons of picking up another RX 580 to pair with my current AsRock Phantom gaming card, mostly because the market for new GPUs is just wild right now; if I wanted to actually double my speed in Resolve, I’d have to shell out at least three times the cash for a single modern card. My main concerns, however, are power and heat. Right now, I'm running a Chieftek GPS 550 AB-A which provides two 6-pin PCI-E connectors:
My existing RX 580 pulls about 120W at peak load in Resolve when I've got it undervolted, so running two would push that to 240W, and once you factor in the 100-125W from the CPU and chipset, I'm looking at something that exceeds the total 360W rated on my 12V rails. If you think this setup is pushing it too far, what kind of power supply would be a solid choice for this scenario while keeping things around the 400-$167 mark?
As for cooling—I'd have two RX 580 cards stacked right on top of each other. I currently have a 120mm fan in the front and an 80mm one in the back, though I could swap that rear one for a 92mm. I have a 120mm fan on the side, but it won't fit due to the CPU cooler, unless I mount it externally. Am I just inviting a total disaster here? 😬
What happened to the title of this thread? Wasn't it originally "entry into the world of Macos"? Now, suddenly, it feels like something that's been twisted into something semantically awkward and just plain wrong.
Andrew Miller102 said:Getting 4K at only 30Hz is pretty terrible, and that’s basically what you end up with when you're running an HDMI 1.4 standard motherboard+an APU and integrated graphics combo.Whether you're using a 4K TV or a monitor, it's all the same story—trying to deal with 30Hz is just going to give you a massive headache.🍿
If we are talking about something like The New York Times, HDMI 1.4 might or might not be an issue depending entirely on what you're actually watching. If you're settling in for a movie, 24Hz is perfectly fine—in fact, it's often preferred—because whether you have a 24Hz or a 60Hz display, the source material is still filmed at 24 fps, which can lead to that annoying pull-down judder on a 60Hz screen anyway. Since 30Hz isn't really a standard, whereas 50/60Hz is typically used for sports, news broadcasts, or certain series, it seems a bit unfair to claim 30Hz causes headaches while 24Hz doesn't. I mean, movies have been shot at 24 or 25 fps since the beginning of time, so why is it suddenly a problem now? Looking at your link, I think it's misrepresenting a few things; if you slow down 30 fps by 50%, you get 15 fps, which obviously stutters heavily, but otherwise, 30 fps shouldn't be that bad (just like how 60 fps slowed by 50% becomes 30 fps). Besides, almost all high-action movies are filmed at 24 fps, with the rare exceptions being The Lord of the Rings at 48 fps or some of the newer films by Ang Lee.
I guess we’re just waiting to see the answer to that one question—why choose Windows over everything else, unless there's some absolute necessity for software that only runs on one specific platform? To be honest, looking back at my experience tinkering with Hackintoshes, you might find that the exact same hardware actually feels a bit snappier running Windows. For instance, a Ryzen setup tends to pull better numbers in Cinebench, and an AMD GPU can feel noticeably faster when you're working in DaVinci Resolve. But does that really matter? It only makes sense if you managed to snag some budget hardware; otherwise, any slight performance gain isn't worth the massive price gap between a custom PC-Mac build and a standard machine. If someone is truly convinced that Windows is that much more exciting, they could always just grab a laptop without Windows pre-installed and try to set up a Hackintosh. Though, you have to be careful there—if I remember correctly, the last version of Windows to play nice with Nvidia was Mojave. If you're aiming for Catalina or Big Sur, you’d likely need a laptop equipped with a discrete AMD GPU... but there are no guarantees, and you'd have to spend a lot of time digging through forums to be sure. Still, it’s not exactly a trap you can't escape from; you can always just wipe it and install Windows.