Nvidia, you PROMISED!

[Emily Young]

Nvidia’s RTX 3080 is here, bringing Ampere to the masses and promising DOUBLE the performance of the RTX 2080! Could they really have been telling us the truth this whole time?

 

 

Buy an RTX 3080 (starting Sept 17th)
On Amazon (PAID LINK): TBD
On Newegg (PAID LINK): TBD
On B&H (PAID LINK): TBD

[Velcade]

[HenrySalayne]

Noise? Nvidia claimed it will be "10x" quieter than the previous generation (so probably 10 dB, SPL is not measured linear). 

[Fatih19]

So it was a bit over hyped but still a great product?

 

The title's tricking me.

[SavageNeo]

 

but still great performance

[Results45]

I'm hedging my bets on an overclocked RX 6800XT or 6900XT this time around. Not to mention 50% more VRAM........

Edited September 16, 2020 by Results45
Ryzen UP to the challenge of our rival.

[Spotty]

What happened to Linus' audio at the 9min mark when discussing Nvidia Reflex? Sounds like it was a recording from a skype call. Sticks out like a sore thumb from the audio recorded in the studio throughout the rest of the video.

 

@GabenJr Not sure if this is from the writers or the editors, but a few thoughts about the graphs in the video...

 

I find the graphs a little confusing having the minimum framerate aligned on the far left of the bar. I think it would look better if it was at the end of the blue bar where the value actually lies, in the same way that the other values (in this case average FPS) are shown.

 

An example of what I think makes more sense:

 

 

 

The Blender tests I think they should be shown as two different bars since the BMW and Pavilion test are independent of each other, so they shouldn't be displayed in the same bar. Not sure why the Blender test is displayed that way when other tests shown later in the video like the SpecviewPerf results with separate tests are each given their own bar. (and again, I don't like the way the time is aligned to the far left of the blue bar)

 

 

Also RIP AMD who decided to call in sick on RTX3080 Benchmark Day.

[Rohith_Kumar_Sp]

I'm disappointed that linus's reviews doesn't cover in depth as it uses to do before. 

 

I got more out of Jayz 2 cents video than linus'a video as the benchmarks linus use is not sufficient and a bit confusing 

 

 

 

[GoldenLag]

11 minutes ago, Rohith_Kumar_Sp said:

I'm disappointed that linua's reviews doesn't cover in depth as it uses to do before. 

they havent covered GPUs or CPUs in dept for a long time. as far as im concerned, they are an entertainment channel, and not for those that want deep dives. 

[porina]

39 minutes ago, HenrySalayne said:

Noise? Nvidia claimed it will be "10x" quieter than the previous generation (so probably 10 dB, SPL is not measured linear). 

Level is just a level, and dB is one way of expressing those levels on a non-linear scale. Without going back to check exactly what was claimed, a 10x change in sound pressure level is a 20 dB change.

 

A 10 dB change is roughly 3x level change.

[Mateyyy]

Why would you use a 3950X for the test bench? Or rather, why only a 3950X, and not also a 10700K/10900K to see whether Nvidia's claim that "a faster CPU still is more important than PCIe 4.0" holds true (which it does)?

 

It was already known that Zen 2 holds back the 2080Ti even at 1440p, if not lower-end SKUs as well, so imo the testing methodology for gaming benchmarks made no sense. Just like including CS:GO in the benchmark suite, which is known to be a mostly CPU bound title regardless of resolution, making the CPU choice even more questionable.

[Rohith_Kumar_Sp]

7 minutes ago, Mateyyy said:

Why would you use a 3950X for the test bench? Or rather, why only a 3950X, and not also a 10700K/10900K to see whether Nvidia's claim that "a faster CPU still is more important than PCIe 4.0" holds true?

 

It was already known that Zen 2 holds back the 2080Ti even at 1440p, if not lower-end SKUs as well, so imo the testing methodology for gaming benchmarks made no sense. Just like including CS:GO in the benchmark suite, which is known to be a mostly CPU bound title regardless of resolution, making the CPU choice even more questionable.

This is true. Pci 3rd gen 10900k is giving more performance than a ryzen 4th gen in digital foundry's review. Hence my confusing and disappointment with the review. 

[SolarNova]

My comments based not only on LTT vid, but other benchmark vids form other sources aswell

 

Went in ready to be disappointed, and i wasnt disappointed in being disappointed.

 

Im coming from a 4k 120hz standpoint. It just cant do it unfortunately.

 

Great gains vs the 2080 for sure, but the 2080 was such a disappointing card due to its price and competition vs the 1080ti, so it really didnt stand a chance. A roughly 60% on average gain vs it is nice, but expected, though somewhat disappointing as 60% is not 100% (doubling)

 

The biggest disappointment for me is its 4k results, which it can now safely be said that the 3080 is THE 4k gaming card, but its not 120hz capable in the most demanding AAA titles commonly tested which is a shame.

 

The 3090 being so highly overpriced wont be any help here either even if it can handle 4k 120hz, however the 3080ti which we should expect can hopefully fill that gap so long as Nvidia doesn't try to increase the price of the X80ti tier like they did during the 20 series, it should slot in at the current launch price of the 3080, and the 3080 at that time brought down in price. Same as they did in previous generations.

 

Still, the 3080 is a great option for those looking to finally upgrade form a 1080ti, just not really as good as what those running 4k were hoping for. if it had truly been a doubling of the 2080 performance, it probably could have been good for 4k120hz, alas, and not unsurprisingly, Nvidia Lied.

[Shreyas1]

You all are missing the greatest part of the RTX 3080 - the design means that we will finally be able to see one of the GPU fans from a normal case layout

[spartaman64]

37 minutes ago, Mateyyy said:

Why would you use a 3950X for the test bench? Or rather, why only a 3950X, and not also a 10700K/10900K to see whether Nvidia's claim that "a faster CPU still is more important than PCIe 4.0" holds true (which it does)?

 

It was already known that Zen 2 holds back the 2080Ti even at 1440p, if not lower-end SKUs as well, so imo the testing methodology for gaming benchmarks made no sense. Just like including CS:GO in the benchmark suite, which is known to be a mostly CPU bound title regardless of resolution, making the CPU choice even more questionable.

so many other reviewers use intel its nice to see an amd test

[SolarNova]

This review below is very interesting.

It shows the effect of CPUs on the 3080, the most important part IMO is the 4770k performance and its performance of the 3080 at 4k.

https://www.tomshardware.com/features/nvidia-geforce-rtx-3080-ampere-cpu-scaling-benchmarks

 

TLDR: if ur running an older CPU and want to get a RTX 3080, 4k should be ur target resolution as it negates, almost completely, the increase in CPU performance over the generations.

[Mateyyy]

1 minute ago, spartaman64 said:

so many other reviewers use intel its nice to see an amd test

It's a GPU review. Logically you'd expect the respective GPUs to be paired with hardware that'll allow them to run at their full potential, so the viewer can get an idea of how fast GPU X is in comparison with GPU Y.

Might as well run the benchmarks on Bulldozer at 900p and say "look, the 3080 performs the same as a 1660!!".

[spartaman64]

4 minutes ago, Mateyyy said:

It's a GPU review. Logically you'd expect the respective GPUs to be paired with hardware that'll allow them to run at their full potential, so the viewer can get an idea of how fast GPU X is in comparison with GPU Y.

Might as well run the benchmarks on Bulldozer at 900p and say "look, the 3080 performs the same as a 1660!!".

except the difference isnt that extreme and theres a significant amount of people running amd instead of intel

[Results45]

On a related note, here's first RTX 2080 to be sold for under $350:

 

Spoiler

https://merc.li/jMND8QHsb

 

EDIT: it just sold FOR $305 

[Mateyyy]

16 minutes ago, spartaman64 said:

except the difference isnt that extreme

It's still a difference regardless.

16 minutes ago, spartaman64 said:

and theres a significant amount of people running amd instead of intel

That irrelevant in this context.

[Helpful Tech Witch]

Lol at 9:55 he said  "That Microsoft is implementing into their Xbox Series X and Xbox Series X"

[Stromholt]

Hey hey, just crusin' by lo leave an interesting 3080 FE vs 2080 FE benchmark done in Black Desert Online.

Pretty demanding title, mostly CPU-Bound than GPU-Bound, but apparently 2080 FE was a huge bottlecap (just see CPU utilization in both cases).

The difference is around 33% raw-FPS increase, but it's pretty interesting too see nonetheless (fun fact: In BDO one of the playerbase sayings is more FPS = more damage, mostly due to how poorly optimized the game is, so after the OCs are released, I'll probably go out for one, since this is my most played title so far). 
 

 

 

[TetraSky]

"Double the performance! in very specific scenarios and workload. "

 

At $1k CAD, it's way out of my budget, but it's nice to see that they've at least made it cheaper than their previous top of the line cards. The used market in 5 years from now looking very attractive.

[Rohith_Kumar_Sp]

26 minutes ago, SolarNova said:

This review below is very interesting.

It shows the effect of CPUs on the 3080, the most important part IMO is the 4770k performance and its performance of the 3080 at 4k.

https://www.tomshardware.com/features/nvidia-geforce-rtx-3080-ampere-cpu-scaling-benchmarks

 

TLDR: if ur running an older CPU and want to get a RTX 3080, 4k should be ur target resolution as it negates, almost completely, the increase in CPU performance over the generations.

welp, i have 4790k planning to get 3080 but for 1080/1440p at most, do i need an cpu upgrade? that'd cost a lot since my rig is like 6 years old 


edit  dafuq? 

holy batman 

 

[SolarNova]

31 minutes ago, Rohith_Kumar_Sp said:

welp, i have 4790k planning to get 3080 but for 1080/1440p at most, do i need an cpu upgrade? that'd cost a lot since my rig is like 6 years old 

As mentioned in the review i linked, the difference in FPS on a GPU as powerful as a 3080 at 1080p or 1440p when comparing a 4790k or equivalent CPU vs a more modern CPU like a Ryzen 3600 or better is rather significant, as much as multiple 'average' GPU generational performance jumps..

 

What i'd suggest is that if u cant afford to do a full system upgrade, and u do get a 3080, that u make sure u run games at an internal rendered resolution of 4k, at least that way ur gaining some amount of visual improvement whilst pushing all the processing to the GPU. You wont get the full effect of 4k, but it certainly look better than native 1080p. (you'll most likely want to disable any Anti Aliasing (AA) when running 4k DSR , u likely wont need it, and AA nowadays is usually limited to the crap versions (FXAA , TAA etc),

 

This is in fact what i already do with my 1080ti, in games that can get over 120fps I make sure to use DSR to render them at 4k fully utilizing my 1080ti. Im still rocking a 1080p Plasma until i buy a 48" OLED later this year / early next, which will coincide with my purchase of a 3080/3080ti/BigNavi so i can use HDMI 2.1 with the OLED for 4k 120hz.

On a 42" 1080p Plasma, there is a significant noticeable difference between rendering at native 1080p vs 4k, so its certainly worth it for me.