GPU Buying Guide - (Read first post)

analogtiger19 said:That’s a pretty wild generalization coming from you, especially since some GeForce GTX 970 cards can show a 5-10 dB difference under load 🙂

That just doesn't add up; you must have mixed something up. If that were true, it would mean certain GeForce GTX 970 cards are several times louder than others.

When you consider how the human perception of loudness works, a mere 10 dB jump is actually felt by our ears as a sound being twice as loud, isn't it? On the other hand, a tiny fluctuation of just 1 or 2 dB is practically nonexistent to the senses.

It all depends on the specific values you're looking at. A jump from 10dB to 20dB actually means it's twice as loud; moving from 20dB to 30dB makes it four times louder, and 30dB to 40dB hits you with eight times the volume, and so on. Given that these graphs represent load levels likely sitting somewhere in the 40dB range, there is simply no way one GeForce GTX 970 could be 5dB louder than another, let alone 10dB.

No, you’re conflating raw sound intensity with how we actually experience noise. To your ears, a 20 dB increase will feel twice as loud as a 10 dB jump, but if we are talking about actual power—the physical intensity of the sound itself—a 70 dB signal is ten times more powerful than a 60 dB one.

On the human perception of loudness. Scroll here to The human perception of loudness, Sound level Comparison Chart and The Ratio.

Exactly, it says precisely what I was getting at: an increase of 10 dB translates to a doubling of how loud it actually sounds to us. If the Sound level Comparison Chart and The Ratio indicate that a volume of 30 dB yields a perceived loudness of 8, while 40 dB jumps to 16, does that really mean a 40 dB sound is sixteen times louder than 30 dB? Of course not; it simply means it’s twice as loud.

How accurate are those noise-measuring apps on Android? Are they actually reliable enough for any kind of comparative measurements?

If you’re actually planning to use your smartphone for measurements, then the quality of the microphone itself becomes far more critical than any software you download, simply because the hardware is much more likely to fail you than the app is. Of course, anyone serious about this is probably using a dedicated dB meter anyway. The real issue is that all measurements need to be conducted under identical conditions—things like distance from the source and ambient noise levels—which, if you ask me, is rarely the case when someone decides to post a device's technical specifications online. 🙂

Well, if we're talking about temperature, I can't say I'm surprised when it comes to coolers.🕺

Frame rate remains the only truly reliable comparative metric we have!

Bradley Lewis59 said:Exactly, that’s precisely my point—it states there's a 10 dB increase, which translates to double the perceived loudness. If the chart shows a loudness value of 8 for 30 dB and 16 for 40 dB, it doesn't mean 40 dB is sixteen times louder than 30 dB; it means it's twice as loud.

Essentially, yes, according to the chart, every 10 dB increase doubles the perceived volume. My main point remains: you aren't going to find a GeForce GTX 970 that is actually twice as loud as another one.

David Sanders4 said:Well, if someone is comparing based on temperature... when it comes to coolers, that doesn't surprise me at all.🕺

Frame rate is the only truly reliable comparative metric we have!

If you're comparing noise levels, you have to ensure they were measured from the exact same source. Otherwise, the data is useless because who knows what the variables were? They could have used different distances, tested with an open vs. closed case, and so on...

And what kind of noise levels are we actually looking at for the 970s when they're running under a full load?

It might be around 40 (just off the top of my head). You'll really need to dig into the reviews for that specific model to be sure.

Timothy Kim60 said:No way, you've got it twisted. That would mean certain GTX 970 cards are several times louder than other GTX 970 models

My bad😵 ...I left out (A) because the values come from the graph db(A), not just db... mea culpa 🙈

Hi,

I'm currently running a 4870 1GB, and it's definitely time for an upgrade. I'm looking for a solid best buy—something significantly better than my current 🙂
.
Ideally, I want a card that runs cooler and has a smaller footprint. This current thing is a total tank; it takes up half my case like some kind of 🙂
monster.
So, what would you recommend as the best value purchase right now?

You didn't really give me much to go on... how much are you looking to drop on a new graphics card? Also, what kind of CPU are you running right now, and what's your power supply situation? (Please make sure to include those details)

Edward Miller33 said:I’m definitely looking for a card that doesn't run hot, isn't a massive beast, and won't take up half my entire case like this absolute elephant of a card 🙂

Honestly, those are probably the least important criteria you could have possibly listed when defining a "best buy" category. What we actually need to know is whether you intend to play absolutely every single game on maxed-out settings... or if you're fine with most of them running smoothly, perhaps making a few honorable exceptions where you'll happily settle for mid or high settings just to call it a win.

Bradley Lewis59 said:Those are probably the least relevant criteria you could have picked for a "best buy" category. What actually matters is whether you need to run absolutely every single game on max settings... or if you're fine with most of them, making a few honorable exceptions where you'll settle for mid/high and call it a day.

I'm running an i5 4460 with 8GB of RAM. I play games here and there—mostly just checking out the big hits—so I don't necessarily need everything on high. My power supply is just a generic 500W unit, but I can swap that out if needed. Money isn't the issue, but as I mentioned, I'm looking for the best value. My current setup is starting to fail, which is why I'm looking for an upgrade.

You've got a killer processor, so there’s really no need to swap it out—it'll keep up with whatever GPU you decide to throw in there. Personally, if you're looking for the absolute best bang for your buck right now, I'd say go with the R9 290. That said, I haven't heard much about that power supply you mentioned, so it might be smart to upgrade that too. Correct me if I'm wrong, but I wouldn't risk putting that specific card in with that PSU...

http://www.newegg.com/product_info.php?products_id=14136

http://www.bestbuy.com/product/graphics-card-example
http://www.newegg.com/product/power-supply-example

Or maybe this combo is cheaper:

http://www.bestbuy.com/product/graphics-card-alt-example
http://www.newegg.com/product/power-supply-alt-example

I guess if I went with the R9 390 paired with an Antec unit, I’d probably be set for 3 or 4 years of high settings without any issues.

Regarding the R9 290 this one, if you grabbed it with a 3% discount from PCPartPicker, it would land just under $833 (including shipping) + you'd obviously need a power supply around $167. I have a gut feeling, though, that he wasn't aiming for a GPU in that specific tier. Anyway, Edward Miller33, I'm curious—which exact power supply are you running? If it managed to keep a 4870 alive all these years, it probably isn't some off-brand junk. Regardless, here is a more budget-friendly suggestion: R9 270

And here is a slightly pricier Nvidia option, the Nvidia GTX 960—it comes with a 3-year warranty, though it's quite short if that's a dealbreaker for you 😁

At the end of the day, all these GPUs represent excellent value for your money.