{"video_id":"mWp5HjLGkjY","title":"How Important Is VRAM Bandwidth?","channel":"Techquickie","show":"Techquickie","published_at":"2023-05-05T14:58:16Z","duration_s":309,"segments":[{"start_s":0.0,"end_s":5.16,"text":"One of the things people got outright angry about when the now-canceled RTX 408012 gig","speaker":null,"is_sponsor":0},{"start_s":5.16,"end_s":12.16,"text":"from NVIDIA was announced was the memory bandwidth. A spec you might also see described as bus width, or something similar.","speaker":null,"is_sponsor":0},{"start_s":12.16,"end_s":19.08,"text":"But why exactly is it so important? Is it some ultra-wide wheeled vehicle that carries your data to and fro, assuming they","speaker":null,"is_sponsor":0},{"start_s":19.08,"end_s":23.64,"text":"have a permission slip? Well, actually, that questionable analogy isn't far from the truth.","speaker":null,"is_sponsor":0},{"start_s":23.64,"end_s":28.76,"text":"You see, a bus in computing is a general term for a data link between two components.","speaker":null,"is_sponsor":0},{"start_s":28.76,"end_s":33.92,"text":"For example, USB stands for universal serial bus, because it's supposed to be a do-everything","speaker":null,"is_sponsor":0},{"start_s":33.92,"end_s":41.2,"text":"connection for transmitting data. But in graphics card land, the memory bus connects the GPU itself to your card's video","speaker":null,"is_sponsor":0},{"start_s":41.2,"end_s":47.52,"text":"memory or VRAM. The basic way a graphics card works is that the GPU, which you can see here, acts upon","speaker":null,"is_sponsor":0},{"start_s":47.52,"end_s":52.48,"text":"data stored in the VRAM, which are these little modules, to generate the images you see on","speaker":null,"is_sponsor":0},{"start_s":52.48,"end_s":57.68,"text":"your screen. Among other things, VRAM contains the textures that make the objects on your screen look like","speaker":null,"is_sponsor":0},{"start_s":57.72,"end_s":61.16,"text":"they're supposed to, as well as data used to create visual effects.","speaker":null,"is_sponsor":0},{"start_s":61.16,"end_s":65.96,"text":"The higher the resolution, texture quality settings, or VFX settings, the more VRAM","speaker":null,"is_sponsor":0},{"start_s":65.96,"end_s":70.8,"text":"your GPU will use, which is why you sometimes see people criticize NVIDIA or AMD when","speaker":null,"is_sponsor":0},{"start_s":70.8,"end_s":75.92,"text":"they pair a powerful GPU that could otherwise run games on high settings with an insufficient","speaker":null,"is_sponsor":0},{"start_s":75.92,"end_s":83.64,"text":"amount of VRAM. Unacceptable! And the more data that sits in VRAM, the wider memory bus you need to get that data to your","speaker":null,"is_sponsor":0},{"start_s":83.64,"end_s":88.88,"text":"GPU. But how do you read a spec sheet that gives a number of bits, or gigabytes per second","speaker":null,"is_sponsor":0},{"start_s":88.88,"end_s":92.64,"text":"for your memory bus, and how do you know if that number is high enough?","speaker":null,"is_sponsor":0},{"start_s":92.64,"end_s":97.4,"text":"So the bit width is just telling you how many bits the bus can move for each cycle of the","speaker":null,"is_sponsor":0},{"start_s":97.4,"end_s":102.32,"text":"memory clock. Basically, how much data is being moved per one hertz.","speaker":null,"is_sponsor":0},{"start_s":102.32,"end_s":106.56,"text":"Of course, seeing as how memory speeds on modern cards are typically well over a thousand","speaker":null,"is_sponsor":0},{"start_s":106.56,"end_s":109.8,"text":"megahertz, this adds up to quite a lot.","speaker":null,"is_sponsor":0},{"start_s":109.8,"end_s":116.0,"text":"For example, the top-end RTX 4090 has a bit width of 384 bits, but this translates into","speaker":null,"is_sponsor":0},{"start_s":116.0,"end_s":119.84,"text":"around 1 terabyte per second of total memory bandwidth.","speaker":null,"is_sponsor":0},{"start_s":119.84,"end_s":123.64,"text":"Now most of you aren't going to be buying a daddy warbucks-level graphics card, so","speaker":null,"is_sponsor":0},{"start_s":123.64,"end_s":127.88,"text":"how much does bus width matter as you go down the product stack?","speaker":null,"is_sponsor":0},{"start_s":127.88,"end_s":132.64,"text":"The answer is, not a whole heck of a lot, without additional context.","speaker":null,"is_sponsor":0},{"start_s":132.64,"end_s":136.96,"text":"Just like it's unwise to compare raw clock speeds between different generations of CPUs","speaker":null,"is_sponsor":0},{"start_s":136.96,"end_s":141.08,"text":"because their underlying architectures are so different, you can't simply look at the","speaker":null,"is_sponsor":0},{"start_s":141.08,"end_s":146.36,"text":"raw bus width for your VRAM and decide that a product is over or underpowered.","speaker":null,"is_sponsor":0},{"start_s":146.36,"end_s":152.0,"text":"It's not that simple. A good recent instance of this was the reaction to the 12 Gigabyte version of the NVIDIA","speaker":null,"is_sponsor":0},{"start_s":152.0,"end_s":157.04,"text":"RTX 4080 prior to Team Green pulling the plug on the card completely.","speaker":null,"is_sponsor":0},{"start_s":157.04,"end_s":160.76,"text":"Although the criticisms of the card being positioned as a 4080, despite the fact that","speaker":null,"is_sponsor":0},{"start_s":160.76,"end_s":165.04,"text":"it had lower-tier specs, were completely warranted, some of the anger at the memory","speaker":null,"is_sponsor":0},{"start_s":165.04,"end_s":173.52,"text":"bus specs had us raising an eyebrow. That first it seemed silly that the 408012 Gigabyte version was shipping with a 192-bit","speaker":null,"is_sponsor":0},{"start_s":173.52,"end_s":177.48,"text":"memory bus, narrower than the previous gen RTX 3070.","speaker":null,"is_sponsor":0},{"start_s":177.48,"end_s":183.04,"text":"However, with the higher power draws and thermal output of the Ada Lovelace GPU series, the","speaker":null,"is_sponsor":0},{"start_s":183.04,"end_s":188.6,"text":"408012 gig was to get a massive 48 megabytes of level 2 cache, which is extremely fast","speaker":null,"is_sponsor":0},{"start_s":188.6,"end_s":193.4,"text":"memory that sits on the GPU die itself and is separate from the main VRAM.","speaker":null,"is_sponsor":0},{"start_s":193.4,"end_s":197.52,"text":"This might not sound like much, but it's actually a huge leap from the 6 megabytes","speaker":null,"is_sponsor":0},{"start_s":197.52,"end_s":200.52,"text":"that the previous flagship 3090 Ti came with.","speaker":null,"is_sponsor":0},{"start_s":200.52,"end_s":205.52,"text":"The idea is that substituting VRAM bandwidth for cache is easier on your card's power","speaker":null,"is_sponsor":0},{"start_s":205.52,"end_s":210.04,"text":"and thermals, and we might see this become more common in the future as we try and squeeze","speaker":null,"is_sponsor":0},{"start_s":210.04,"end_s":214.52,"text":"more and more performance out of the GPUs themselves. Okay, so that was a lot.","speaker":null,"is_sponsor":0},{"start_s":214.52,"end_s":219.16,"text":"Here's our point. A wider bus and more memory bandwidth is better.","speaker":null,"is_sponsor":0},{"start_s":219.16,"end_s":224.88,"text":"That's true. However, as always, you can't rely just on the spec numbers when buying your next graphics","speaker":null,"is_sponsor":0},{"start_s":224.88,"end_s":230.04,"text":"card. Instead, read reviews and benchmarks and specifically look to see if there are reported problems","speaker":null,"is_sponsor":0},{"start_s":230.04,"end_s":234.2,"text":"with textures or visual glitches at higher frame rates and quality settings on cards","speaker":null,"is_sponsor":0},{"start_s":234.2,"end_s":241.88,"text":"that have relatively narrow memory buses. Prioritize a strong GPU and having enough VRAM and worry about memory bandwidth only","speaker":null,"is_sponsor":0},{"start_s":241.88,"end_s":248.16,"text":"if something seems off when looking at test results. Besides, there are already enough specs to keep track of anyway, like if the card you","speaker":null,"is_sponsor":0},{"start_s":248.16,"end_s":252.84,"text":"want will even fit in your case. Thanks, NVIDIA.","speaker":null,"is_sponsor":0},{"start_s":252.84,"end_s":259.0,"text":"And thank you for watching this video, earnestly. Like the video if you liked it, dislike it if you disliked it, check out our other videos,","speaker":null,"is_sponsor":0},{"start_s":259.0,"end_s":262.32,"text":"comment below with video suggestions, and there's one more thing I'm gonna ask you","speaker":null,"is_sponsor":0},{"start_s":262.32,"end_s":265.68,"text":"to do, okay? Can you subscribe and follow to TechWiki?","speaker":null,"is_sponsor":0},{"start_s":265.68,"end_s":268.96,"text":"Just this one time, okay? Just a little bit. Can we see if we can do that?","speaker":null,"is_sponsor":0},{"start_s":268.96,"end_s":269.96,"text":"Okay, thanks.","speaker":null,"is_sponsor":0}],"full_text":"One of the things people got outright angry about when the now-canceled RTX 408012 gig from NVIDIA was announced was the memory bandwidth. A spec you might also see described as bus width, or something similar. But why exactly is it so important? Is it some ultra-wide wheeled vehicle that carries your data to and fro, assuming they have a permission slip? Well, actually, that questionable analogy isn't far from the truth. You see, a bus in computing is a general term for a data link between two components. For example, USB stands for universal serial bus, because it's supposed to be a do-everything connection for transmitting data. But in graphics card land, the memory bus connects the GPU itself to your card's video memory or VRAM. The basic way a graphics card works is that the GPU, which you can see here, acts upon data stored in the VRAM, which are these little modules, to generate the images you see on your screen. Among other things, VRAM contains the textures that make the objects on your screen look like they're supposed to, as well as data used to create visual effects. The higher the resolution, texture quality settings, or VFX settings, the more VRAM your GPU will use, which is why you sometimes see people criticize NVIDIA or AMD when they pair a powerful GPU that could otherwise run games on high settings with an insufficient amount of VRAM. Unacceptable! And the more data that sits in VRAM, the wider memory bus you need to get that data to your GPU. But how do you read a spec sheet that gives a number of bits, or gigabytes per second for your memory bus, and how do you know if that number is high enough? So the bit width is just telling you how many bits the bus can move for each cycle of the memory clock. Basically, how much data is being moved per one hertz. Of course, seeing as how memory speeds on modern cards are typically well over a thousand megahertz, this adds up to quite a lot. For example, the top-end RTX 4090 has a bit width of 384 bits, but this translates into around 1 terabyte per second of total memory bandwidth. Now most of you aren't going to be buying a daddy warbucks-level graphics card, so how much does bus width matter as you go down the product stack? The answer is, not a whole heck of a lot, without additional context. Just like it's unwise to compare raw clock speeds between different generations of CPUs because their underlying architectures are so different, you can't simply look at the raw bus width for your VRAM and decide that a product is over or underpowered. It's not that simple. A good recent instance of this was the reaction to the 12 Gigabyte version of the NVIDIA RTX 4080 prior to Team Green pulling the plug on the card completely. Although the criticisms of the card being positioned as a 4080, despite the fact that it had lower-tier specs, were completely warranted, some of the anger at the memory bus specs had us raising an eyebrow. That first it seemed silly that the 408012 Gigabyte version was shipping with a 192-bit memory bus, narrower than the previous gen RTX 3070. However, with the higher power draws and thermal output of the Ada Lovelace GPU series, the 408012 gig was to get a massive 48 megabytes of level 2 cache, which is extremely fast memory that sits on the GPU die itself and is separate from the main VRAM. This might not sound like much, but it's actually a huge leap from the 6 megabytes that the previous flagship 3090 Ti came with. The idea is that substituting VRAM bandwidth for cache is easier on your card's power and thermals, and we might see this become more common in the future as we try and squeeze more and more performance out of the GPUs themselves. Okay, so that was a lot. Here's our point. A wider bus and more memory bandwidth is better. That's true. However, as always, you can't rely just on the spec numbers when buying your next graphics card. Instead, read reviews and benchmarks and specifically look to see if there are reported problems with textures or visual glitches at higher frame rates and quality settings on cards that have relatively narrow memory buses. Prioritize a strong GPU and having enough VRAM and worry about memory bandwidth only if something seems off when looking at test results. Besides, there are already enough specs to keep track of anyway, like if the card you want will even fit in your case. Thanks, NVIDIA. And thank you for watching this video, earnestly. Like the video if you liked it, dislike it if you disliked it, check out our other videos, comment below with video suggestions, and there's one more thing I'm gonna ask you to do, okay? Can you subscribe and follow to TechWiki? Just this one time, okay? Just a little bit. Can we see if we can do that? Okay, thanks."}