• 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
What's happening at BTR *16*
#51
Well you have till the 19th so get writing!
#52
(09-14-2018, 06:11 AM)happy medium Wrote: Well you have till the 19th so get writing!

You mean get benching!

I was supposed to post about the Turing architecture this morning ...
Blush

But I am not sure if I will have time ... I have some pics of the 2080/Ti ... shall I post them along with a quickie write up today?
- or shall I just save everything for Wednesday's reviews?
Undecided

I was tired yesterday and went to bed early .. I am glad that I did ...
Bomb

Here is the Turing Whitepaper ... maybe I should just post the review on Wednesday and skip the features/architecture
https://www.nvidia.com/content/dam/en-zz...epaper.pdf
#53
(09-14-2018, 01:33 PM)apoppin Wrote: I was tired yesterday and went to bed early .. I am glad that I did ...
Bomb

Here is the Turing Whitepaper ... maybe I should just post the review on Wednesday and skip the features/architecture
https://www.nvidia.com/content/dam/en-zz...epaper.pdf

Saturday ... Sunday ... Monday ... Tuesday ....
- Wednesday, 6 AM

There isn't a lot of time left for me to evaluate 2 cards so I will not post anything until Wednesday - and I will incorporate what I didn't post yesterday into the review. But I was up until 2 AM last night and I "slept in" to nearly 7 AM. It's all benchmarking for the next 3+ days - then writing
Black_eye

Fortunately I have all of the other video cards benched using 36 (pancake, no VR) games (Titan XP/GTX 1080 Ti/GTX 1080/Vega 64 + GTX 1070/FuryX) and started to bench the lesser target card before I went to bed. Some surprises; wait for the review. Of course, I have to also overclock. So timing is tight to get everything done ... and I have to also write a full traditional review (features/editor's day video/architecture/unboxing/images/charts w/percentages/analysis/conclusion).

Beee

After Wednesday, I have to post the overclocking review(s) followed by the Vive Pro, Part 2 with RTX - before the end of September
Gamer4



Does anyone have any last minute requests? I will probably be following up the launch, overclocking, and VR evaluations with more because I want to test AA and image quality in depth .. and there are probably further things to explore with the new architecture. I'll be busy if my health holds.
Paladin
#54
If you could please analyze the IQ difference between DLSS AA and the other various forms of AA that we have traditionally had, that would be fantastic. Take some 4k screenshots if you can. I think a lot of readers would really enjoy seeing that.

I would love to see how DLSS AA compares to DSR.

These are going to be the first cards that can run FFXV at 4k 60fps so that is the game I would use for the IQ comparison.
#55
(09-15-2018, 05:36 PM)SickBeast Wrote: If you could please analyze the IQ difference between DLSS AA and the other various forms of AA that we have traditionally had, that would be fantastic.  Take some 4k screenshots if you can.  I think a lot of readers would really enjoy seeing that.

I would love to see how DLSS AA compares to DSR.

These are going to be the first cards that can run FFXV at 4k 60fps so that is the game I would use for the IQ comparison.

I will do this in a follow up. There is simply no time to explore IQ in any launch article - not with 36 game benchmarks. Maybe if I had another 3 days.

First launch, then OC, then VR, then IQ .. it's the natural order for a very small review site with a very large benching suite. The launch doesn't answer the question of "value" - it just gives a preliminary report with a strong emphasis on performance. Features are also taken into account, but these need to be tested over time.

It's the same cycle every couple of years. And usually when we look back we see generally decent value with new features - this one is a pretty radical departure into using Tensor cores for hybrid ray tracing - and it is going to take awhile to declare if it is a "success" or not. And usually, there is a RTG launch either before or after Nvidia's to bring their own features and performance, but that is sadly missing.
Beee
#56
(09-15-2018, 06:05 PM)apoppin Wrote:
(09-15-2018, 05:36 PM)SickBeast Wrote: If you could please analyze the IQ difference between DLSS AA and the other various forms of AA that we have traditionally had, that would be fantastic.  Take some 4k screenshots if you can.  I think a lot of readers would really enjoy seeing that.

I would love to see how DLSS AA compares to DSR.

These are going to be the first cards that can run FFXV at 4k 60fps so that is the game I would use for the IQ comparison.

I will do this in a follow up.  There is simply no time to explore IQ in any launch article - not with 36 game benchmarks.  Maybe if I had another 3 days.  

First launch, then OC, then VR, then IQ .. it's the natural order for a very small review site with a very large benching suite.  The launch doesn't answer the question of "value" - it just gives a preliminary report with a strong emphasis on performance.  Features are also taken into account, but these need to be tested over time.  

It's the same cycle every couple of years.  And usually when we look back we see generally decent value with new features - this one is a pretty radical departure into using Tensor cores for hybrid ray tracing - and it is going to take awhile to declare if it is a "success" or not. And usually, there is a RTG launch either before or after Nvidia's to bring their own features and performance, but that is sadly missing.
Beee

This is just making me realize that Greg would be such an asset to the site if he would analyze IQ. He would be so good at it. I wonder if the two of you can work something out. I just don't know how comfortable he would be in signing the NDA. And I would miss our heated discussions if he did that.
#57
(09-15-2018, 06:18 PM)SickBeast Wrote:  I just don't know how comfortable he would be in signing the NDA.  

I don't know how comfortable Nvidia would be offering it.
(actually, I do ... it's not going to happen)

There is a matter of expensive hard to get cards. I can't get a second card for myself.

And then there is the matter of not being able to see a difference using a deprecated OS.
#58
(09-16-2018, 03:37 AM)gstanford Wrote: I'm not a reviewer and nvidia certainly wouldn't appreciate an assessment from me on Turing.

The tensor processors need ripping out and replacing with standard SM's for gaming GPU's.  The space the RT core takes up within an SM could be extra L1 cach and  make the die smaller by its complete removing the rest as I've mocked up in the image below.

[Image: LwkVpNA.jpg]

Perhaps then we might have a product for gamers that will cater to their existing library of games with performance worth actually talking about.

Just one final note: if I did decide I wanted to do an IQ comparison like suggested then I simply would and there is no force or nvidia NDA clause on the face of this earth that would stop me from doing so.

Apparently the Titan V did just that, and if you look, it didn't have great gaming performance. Turing looks much faster.
#59
I guess to satisfy you, Nvidia should have just refreshed Pascal and forgot about ray tracing. The same "basic" architecture has proved to be rather successful and its continued evolution makes sense until there is a reason to completely change it.
Beee

I am glad Nvidia is not paying any attention to you. I like progress.
#60
Maybe Greg should demand satisfaction:



I'm not sure if the glove would work on nVidia, though. JC-LOL


Forum Jump:


Users browsing this thread: 6 Guest(s)