Nvidia RTX Series

[Zoraptor]

They're kind of claiming the reverse I think, with RTX I/O. Don't need so much vram anyway because they'll be pulling stuff off the ssd super quick? Dunno, I kind of glaze over after too much marketing speak.

I simply don't see how a 30$ more expensive 3070Ti can work as a product though- there's little to no extra profit margin built in and in makes the standard 3070 look like a bad buy. My assumption is that a 3070Ti would be the top binned chips as well as have the extra memory to justify a ~600$ price tag.

Edited September 3, 2020 by Zoraptor

[Azdeus]

Oh yeah, they'll be top binned no doubt about it, but it's going to put it way to close to the 3080 in price, there'd be no real reason to buy it.

Edit; I think it more likely that they lower the price of the base 3070 to kill any possible contender AMD might put there.

Edited September 4, 2020 by Azdeus

[Azdeus]

On monday when Gainward did their booboo, I downloaded the datasheet for the 3090.

Compare it to the one that is there now; https://www.gainward.com/main/product/vga/pro/p01090/p01090_datasheet_14305f4ef3d9da1d1.pdf?s=726

There are a few differences...

3090 datasheet.pdf

[Malcador]

The cooler airflow is interesting, wonder if CPU coolers are going to get adversely affected a lot in practice.

[Sarex]

9 minutes ago, Malcador said:

The cooler airflow is interesting, wonder if CPU coolers are going to get adversely affected a lot in practice.

I don't think there will be much difference, the non-reference coolers scatter the exhaust inside the case, this one is just a little more direct on the second fan, besides the popular choice for coolers nowadays is AIO water coolers which in most setup scenarios pull the air from outside of the case.

[AwesomeOcelot]

I'm going to try to get a founders edition just to see what this new cooler is about. In terms of blowing air into the CPU cooler. It's been my experience, and the opinion of smarter people than me, that the delta is large enough that it shouldn't matter. I hope to find out though.

[Sarex]

2 hours ago, AwesomeOcelot said:

I'm going to try to get a founders edition just to see what this new cooler is about. In terms of blowing air into the CPU cooler. It's been my experience, and the opinion of smarter people than me, that the delta is large enough that it shouldn't matter. I hope to find out though.

Let us know how it works out. Really interested in that cooler design.

[Azdeus]

I want to buy all my stuff at the same store, it's close by and I have 0% faith in shipping these things, so the FE version is out of my hands.

[Malcador]

21 hours ago, Sarex said:

I don't think there will be much difference, the non-reference coolers scatter the exhaust inside the case, this one is just a little more direct on the second fan, besides the popular choice for coolers nowadays is AIO water coolers which in most setup scenarios pull the air from outside of the case.

Fair enough, I haven't come across any test with it (granted have not looked much) but was something that stuck out to me as being a bit odd.    Sure sucks in Canada, NCIX used to be good for me but they died (and lost all my personal info...) so have to go with Newegg pretty much or CC which has dubious customer service.

[Sarex]

48 minutes ago, Malcador said:

Fair enough, I haven't come across any test with it (granted have not looked much) but was something that stuck out to me as being a bit odd.    Sure sucks in Canada, NCIX used to be good for me but they died (and lost all my personal info...) so have to go with Newegg pretty much or CC which has dubious customer service.

At least you have Newegg. I could only wish the prices were anywhere close to that here.

[LadyCrimson]

On 9/2/2020 at 1:51 PM, Malcador said:

I have vastly inferior hardware to everyone here, I see.

Only a year and a half ago I still had an ancient i7-920 CPU (that still worked overall fine for 1080, really).

4-5 years from now 'net trolls will probably be telling me "lulz a 9900k, what a potato, gtfo."

Then maybe a few years after that, for a year or so I'll be top-dog again. Then back to a potato.

And round and round she goes ....

[ComradeYellow]

https://nvidianews.nvidia.com/news/nvidia-to-acquire-arm-for-40-billion-creating-worlds-premier-computing-company-for-the-age-of-ai?ncid=so-twit-58914#cid=gnl_so-twit_en-us

Lotta money just for tech sheeet.

[Bokishi]

Some reviews are up on the 3080. Seems like a 4K champ

https://www.tomshardware.com/news/nvidia-geforce-rtx-3080-review

[LadyCrimson]

^  Oh? BL3 is their "second-most demanding game in our suite?"   I mean I know it's harsh on lower end gpu's but...

I guess EVGA's built-in OC on my FTW 2080ti version is why my fps at Ultra on that game is so much higher than their graph? I get more like 70 even with all-Ultra (nothing disabled), not 47. And I use the same CPU as they're testing with.

At any rate, certainly looks like a nice 4k card, especially for the lowered price, although reading that the minimum can still drop a fair bit (at least on all-Ultra) is a bit pfft.  Even if the average is 100 doesn't that mean if you tried to cap it at 100, you'd probably still see drops below that? One day they'll have a card (and cpu) that can maintain 120+ without ever dropping below 100, like how I can play at 60 and never have it drop. But by then game dev's will have sucked up the extra leeway and everyone will be talking about 8k/60 and not 4k/100 and ... never mind. Another never ending cycle. 

 

[LadyCrimson]

^  I durped and forgot 70-ish fps was with V. Fog off, because I hate the way that looks. If I have that on Ultra it drops down a bit over 60 fps. Still more than 47 tho.

...also, for me, DX11 vs. DX12 makes almost zero difference performance-wise. Only thing DX12 did in BL3 was make me wait 15 seconds longer on startup....

Edit: but I sure wouldn't have minded the lower price of the 30xx series. 

Edited September 16, 2020 by LadyCrimson

[AwesomeOcelot]

Benchmarks are consistent and average frame rates are over several runs. You can launch a game and see 70 ish fps, but that's one area with a random viewport position, it's not multiple runs of averaged frame rates, that get averaged. You're probably not keeping an eye on the frame rate in action. Some games vary in frame rate over each benchmark run. Another reviewer, using a different run of the same game, will get a different frame rate, just by looking at different things in game. The better reviewers will try to control for all these factors, the average gamer will not.

[LadyCrimson]

No, I'm aware of that. You don't look at your FPS while standing around town picking your nose with your gun/sword. I go out and combat, run, jump, try a few different areas. Is there a very specific area or bossfight in BL3 that might be the worst fps spot in the game and it might drop to 50 for me? Sure. That wouldn't be "average" tho, either.

That's the base 3080 early benches, ofc.  I'm sure some of the aftermarket OC versions will be higher/better (for more price tho). I'll be curious what EVGA's FTW versions (or similar) of such are.

 

Edit: I'd also be interested what you'd get with DLSS 2.  DLSS1 was dogpoop but DLSS 2 is much better.  Using that I can get 80fps ultra in Death Stranding. So that one I might get steady 100fps with a 3080. 

 

Edited September 16, 2020 by LadyCrimson

[AwesomeOcelot]

Digital Foundry benchmark of BL3 4K on a 3080 is over 60, 53 on 1% low. 2080 Ti is 44 mean, 39 1% low.

Edited September 16, 2020 by AwesomeOcelot

[LadyCrimson]

Btw, I'm not ragging on the 3080.  It's a great sounding card.  If one hasn't upgraded a gpu for a bit and wants to move to 4k it's a good buy seems like.

I'm just coming from the perspective of "have a good gpu for my needs already/skip generation" so I may sound more critical or nitpicky than I intend to, lol.

If there's a 3080ti that'd (probably) be the one I'd want if I was buying ... unless the 3090 is the new ti? I'm not clear about that.

[Bokishi]

I think the 3090 replaces the Titan X in this series, has the extra price premium and the VRAM

[Zoraptor]

24 minutes ago, LadyCrimson said:

If there's a 3080ti that'd (probably) be the one I'd want if I was buying ... unless the 3090 is the new ti? I'm not clear about that.

3080S/Ti at some point seems pretty likely since there's a huge price gap to the 3090 and one of the more persistent rumours is of a 3080 with more memory being produced.

I wouldn't expect too much from factory overclocked models though. Per Gamers Nexus the last 10% stock performance increases power draw by 100W, and you'd have to presume reviewers have had binned cards sent to them. An overclock is going to be very hungry.

[LadyCrimson]

I have an 850wPS, I'd probably be good. Unless the 40xx series needs a 1000w. 

So likely there will still be a ti or something similar in the future then. Good to know.

Also ... man that recent MS Flight Sim must need a future 20900k CPU or something (3080 won't even do it 4k/60). 

 

[AwesomeOcelot]

4 hours ago, LadyCrimson said:

Btw, I'm not ragging on the 3080.  It's a great sounding card.  If one hasn't upgraded a gpu for a bit and wants to move to 4k it's a good buy seems like.

I'm just coming from the perspective of "have a good gpu for my needs already/skip generation" so I may sound more critical or nitpicky than I intend to, lol.

If there's a 3080ti that'd (probably) be the one I'd want if I was buying ... unless the 3090 is the new ti? I'm not clear about that.

I don't care about that, it seems you're talking about something else to what the benchmarks are. "I get 70 fps" while playing isn't what a benchmark is, if your PC ran this benchmark, you'd get around 44 fps, 4K, "bad ass" settings (I think is ultra in BL). Maybe a bit more, as the 2080 Ti does over clock, but under 50. You seem to be saying, the game is a very playable 60 fps at most times. Digital Foundry do those kind of videos too, they have settings guides and target minimum frames.

I wouldn't understand upgrading from a 2080 Ti right now. 3090 is for 8K and AI compute as far as I can tell, it's a terrible gaming card apart from that, even worse value for money than the 2080 Ti was. The 2080 Ti will probably drive what you want for a long time. A 3090 will only burn a hole in your wallet unless you upgrade to 8K.

I'm a value orientated builder and patient, I couldn't care less about bragging rights. 3080 is 30% faster than a 2080 Ti for a little over half the price. I also want it because my OLED 4K needs HDMI 2.1 to drive 120fps HDR. I also ordered a Valve Index for HL Alex. I would have never gotten a 10900K because the 10600K would runs games virtually the same. Actually the 3950X is only 1% different at 4K according to TechPowerUps many benchmarks. Most games at 4K are GPU limited.

 

[LadyCrimson]

8 hours ago, AwesomeOcelot said:

I don't care about that, it seems you're talking about something else to what the benchmarks are.

 

*scratches head*  
No, I know that's how benchmarks look/are run. BL3 does let you do that from an ingame setting actually, forgot about that (most of my games do not have such a feature, I forget more games do that now). But "flying" at rapid speed in/through the game is not playing, no one plays a game that way outside of fly-cheat codes.

I guess my confusion then stems that I thought benchmarks gave a decent representation of what kind of actual real average gameplay FPS one might see/expect with given settings vs. only an identical stress test to compare % increases. Sounds like that isn't necessarily always going to be the case, at least not imo? I suppose that would explain why I look at benchmarks and wonder why my rigs (over the years) often seem to be (real playing) higher fps etc. than indicated. And here I thought I just kept such a clutter free rig or something.

Running BL3's benchmark looks like 52ish, 69ish if I turn V. Fog all the way Off (the harshest game setting). I only ran them once each, dunno how much variance there is/might be if repeated a lot.

Edited September 17, 2020 by LadyCrimson

[AwesomeOcelot]

46 minutes ago, LadyCrimson said:

*scratches head*  
No, I know that's how benchmarks look/are run. BL3 does let you do that from an ingame setting actually, forgot about that (most of my games do not have such a feature, I forget more games do that now). But "flying" at rapid speed in/through the game is not playing, no one plays a game that way outside of fly-cheat codes.

I guess my confusion then stems that I thought benchmarks gave a decent representation of what kind of actual real average gameplay FPS one might see/expect with given settings vs. only an identical stress test to compare % increases. Sounds like that isn't necessarily always going to be the case, at least not imo? I suppose that would explain why I look at benchmarks and wonder why my rigs (over the years) often seem to be (real playing) higher fps etc. than indicated. And here I thought I just kept such a clutter free rig or something.

Running BL3's benchmark looks like 52ish, 69ish if I turn V. Fog all the way Off (the harshest game setting). I only ran them once each, dunno how much variance there is/might be if repeated a lot.

In my experience most benchmarks aren't like what you describe. A specific subset of benchmarks are like that, Digital Foundry do them from time to time. Most benchmarks are even worse than that in being representative of actual gameplay. They're designed purely for showing difference and consistency. If hardware reviewers only used real world testing they'd go out of business, because in real world testing, CPU's are not very important for games (there are a few exceptions but not many), and GPUs from model to model e.g. 2080 vs 2080 super, wouldn't produce noticeable differences in a real world setting.

Reviewers actually use a fresh install for benchmarking, they are as clutter free as could possibly be. They almost always use an open air bench, which might be better in terms of thermals than the average user.

The 10900K reviews use 1080p medium settings with a 2080 Ti. Not saying that people don't do that, but that's probably 1% of owners. I swear a lot of benchmarking sites are about benchmarking for benchmarkers, people who build PCs to get 3d Mark scores. Which I am not against, it's a hobby, like drag racing and car tweaking.

Edited September 17, 2020 by AwesomeOcelot