Nvidia Turing Product Reviews and Previews: (Super, TI, 2080, 2070, 2060, 1660, etc)

I know that we're in a graphics centric forum, but honestly why is the discussion about benefits of RT reflections so limited to IQ? The biggest advantage that I saw or thought about RT reflections, the moment I saw them in the BF V demos back then were related to gameplay, being able to see people behind you and such...
 
And if AMD punished tech sites for unfavorable review scores then that's terrible and it doesn't excuse nvidia for doing the same.
I do not know the facts, but if Hardware Unboxed signed Nvidia's NDA and did not break a recent agreement then they should have received a card unless there was a review shortage (which does happen). Other than that they might have to get in queue similar to sites like phoronix.com, which tend to post reviews much later than mainstream sites.
 
BTW I decided to check this conspiracy thing about reviewers not doing 4K for myself and out of the 5 that I have bookmarked and typically use for reviews: Anandtech, TechPowerUp, Guru3D, Tom's Hardware and PcPerspective, only Tom's didn't do 4K. I fail to see the conspiracy, sorry. I'm pretty sure those are among the biggest tech sites, top 10 surely... If Nvidia didn't want us to know 4K performance, they did a really poor job, IMO.
 
It's a guide, not a set of orders. You don't have to follow them, unless I'm mistaken and there's a contract stating what reviews are and are not allowed to report on. What was stopping videocardz.com benchmarking at different resolutions to the guide recommendations.
If this is anything like review guidelines for games, and I believe it is, they might not have an actual contract, but effectively they pretty much have an unwritten one.
Gaming publications receive review copies for free and way before launch, but under guidelines of when they can post content (the famous review embargo) what content they can show or what should not be spoiled etc. Aledgedly, these guidelines have been getting more and more specific through the ages as well. There is nothing keeping a reviewer from breaking those guidelines and do whatever the fuck he wants, they don't sign a contract (or as far as I know, they didn't do it the last time I saw journalists talk about that) but they get sort of black-listed with that publisher and will have less media access to future news from them, like: less opportunities for interviews, preview events, exclusive videos, and in extreme cases they might not even get review copies for the next few games from that publisher, meaning they would have to wait for the consumer launch of a title to even start playing it to wright a review, if they even wanna bother at that point. Some journalists that have consistently broken these types of guidelines even started being black listed by publishers they didn't even break any guidelines with directly, but they had become so infamous other companies were just being more cautious about those guys.
So yeah, if the relationship of hardware manufacturers and journalists is any similar, it's not far fetched to think that, even if there isn't a explicit contract legally keeping reviewers from breaking the guidelines, they have strong incentives to follow them. The consequence of going against Nvidia could mean them not getting their review copy as early as his competitors next time.
 
It's not that far fetched, but that's not really a useful discussion point for this thread as it'll apply to every hardware review thread. That discussion should be held in its own thread and reviews just discussed as we find them.
 
(I had posted this on this thread but it was teleported elsewhere, so here goes again: )


Looking at the mobile versions on the just announced MSI laptops, it seems the RTX 2060 model uses the same 180W adapter as it was needed for the GTX 1070 Max-Q model.

It'll be interesting to see how the GTX 1070 Max-Q compares to the laptop RTX 2060, but so far it points out to Turing being substantially more power efficient than Pascal, at lower TDPs.
 
Not just behind you, I can spot people on the reflection of floors and puddles even though they are not directly in my line of sight, in the mansion of Rotterdam I can cover a certain door while spotting enemies peaking from the windows to the sides of it on the floor.

Are you saying you couldn't see those things before, because they were not clear enough..? Or are you saying RTX adds players into the reflections, instead of offering more fidelity..?

I have always seen such things in my reflection and shadows. I just never cared how detailed they were in a FPS game because you don't have time to actually look at such thing, only react.
 
Palit GeForce RTX 2060 GamingPro OC review
If Palit can keep the price of the GamingPro OC card under control, they might have a winner here. the standardized factory tweak offers 4% more performance over reference and once manually tweaked you can boost 105 extra perf out of it, and that is RTX 2070 level performance folks. The GeForce RTX 2060 series positions will sit best in the FullHD and (W)QHD resolution monitor ranges.
...
Performance wise you are looking at the GTX 1070 (Ti) / 1080 on raw shader perf, added benefits are of course the RT and Tensor cores. If you stick to the aforementioned resolutions your games will run at proper framerates with the very best image quality. The 6 GB graphics memory is sufficient here.

We've been able to push another 10% of extra perf out of the card compared to reference. Both traditional overclocking, as well as the OC Scanner, functions bring us close to that value. The combination of memory, power and voltage settings will bring you a proper overclock. Once you've applied it, you get a few percent more performance. Nice to see is that we have been able to pass 16 Gbps on the memory, and that does help.
https://www.guru3d.com/articles_pages/palit_geforce_rtx_2060_gamingpro_oc_review,1.html
 
Are you saying you couldn't see those things before, because they were not clear enough..? Or are you saying RTX adds players into the reflections, instead of offering more fidelity..?

I have always seen such things in my reflection and shadows. I just never cared how detailed they were in a FPS game because you don't have time to actually look at such thing, only react.
Without RT players are only reflected if they apear directly on screen. With RTX, it's possible to have scenarios where you see reflctions of players that are not directly on screen, affording a possible advantage on those scenarios.
 
Rumours everywhere today stating that the GTX1660Ti will be only 20% faster than a GTX1060, while costing more or less the same. Lame. AMD please come back, we've missed you. Preferably with a GTX1660Ti competitor that mops the floor with it. Plenty of space between it and the GTX2060 to cover.
 
Rumours everywhere today stating that the GTX1660Ti will be only 20% faster than a GTX1060, while costing more or less the same. Lame. AMD please come back, we've missed you. Preferably with a GTX1660Ti competitor that mops the floor with it. Plenty of space between it and the GTX2060 to cover.
It was ATOS bench with mobile version (GTX 1060 in comparison is the mobile variant too, of course)
 
Looks like GTX Turing is real:

dckB4DM.jpg


GTX 16xx family incoming.
Remains to be seen if these are repurposed TU106 chips or TU116 is in fact real.

If there are no RT units in there, I wonder what differentiates it from Volta.
GDDR6? Turing also has an incremental update to the tensor units IIRC.
 
I wouldnt buy a 1660 over a 2060, its essentially a low end gpu akin to 1050/ti at a higher price. Without RT its less future proof.

Nvidia has nice gpus, but their pricing is really annoying.
 
Looks like GTX Turing is real:
If there are no RT units in there, I wonder what differentiates it from Volta.
GDDR6? Turing also has an incremental update to the tensor units IIRC.
Standard-Volta would have half-rate FP64 as well as more L1. And yes, it's Tensors would not be able to accelerate Inferencing Workloads with INT8/INT4 precision as much. If the small chip is using standard Volta ingredients, that is.
 
Looks like GTX Turing is real:
GTX 16xx family incoming.
Remains to be seen if these are repurposed TU106 chips or TU116 is in fact real.

If there are no RT units in there, I wonder what differentiates it from Volta.
GDDR6? Turing also has an incremental update to the tensor units IIRC.
Supposedly Tensors are cut off too
 
Last I heard it was still in the air whether tensors were included, which is likely since they would be needed for DLSS.
 
Last I heard it was still in the air whether tensors were included, which is likely since they would be needed for DLSS.
They could just aswell want to keep DLSS as buying incentive for the more expensive RTX cards.
Assuming TU116 chip is indeed separate chip I don't think it can really be 'in the air', they either included the cores on the chip or they didn't
 
It has to be a smaller chip and they will probably use it for more SKUs. No chance they will go close or under $200 with a 445mm2 chip. Maybe around 250mm2 for this one.
 
Back
Top