be quiet! Dark Rock PRO TR4 review
MSI GeForce RTX 2060 Gaming Z review
Palit GeForce RTX 2060 GamingPro OC review
Promo: MS Office 2016 Pro Plus Key For $31.5
Corsair IronClaw RGB mouse review
Corsair Harpoon RGB Wireless mouse review
Corsair M65 RGB ELITE game mouse review
GeForce RTX 2060 review (founder)
WD Black NVME SSD (1TB) Review
Seasonic Focus SGX-650 Power Supply review
Nvidia Turing GeForce 2080 (Ti) architecture review
Posted by: on: 09/14/2018 02:00 PM [
130 comment(s) ]
It is almost time before we can present you our full review of the GeForce RTX series graphics cards. But first we'll be taking an architectural deep dive into the Turing graphics processors, and of course, we can share with you all specifications, you know the nitty gritty stuff.
Read article
Advertisement
Tagged as:
geforce rtx review,
nvidia
pages 1 2 3 4 5 6
Fox2232
Senior Member
Posts: 8133
Senior Member
Posts: 8133
Posted on: 09/14/2018 02:39 PM
DLSS =
- Take angle under which 2 edges intersect
- take color information at edge
- load resulting pixel values from database
Variable rate shading =
- 1/2 precision
- 1/4 precision
- 1/8th precision
- and lovely 1/16th precision
This for sure boosts performance. Gamers will have mandatory camera tracking their eyes...
TSS =
- Bake in results into texture on the fly
- use old information to skip actual work
- update baked in texture from time to time (or at rate you feel comfortable with)
- probably would not be as bad as it seems if we had higher than 16xAF
To sum it up: New features can be paraphrased as "Way to cheap out IQ for performance gain."
Maybe good for 8K, somewhat OK for 4K. But hit to per pixel quality on 1440p will be unpleasant. On 1080p unacceptable.
DLSS =
- Take angle under which 2 edges intersect
- take color information at edge
- load resulting pixel values from database
Variable rate shading =
- 1/2 precision
- 1/4 precision
- 1/8th precision
- and lovely 1/16th precision
This for sure boosts performance. Gamers will have mandatory camera tracking their eyes...
TSS =
- Bake in results into texture on the fly
- use old information to skip actual work
- update baked in texture from time to time (or at rate you feel comfortable with)
- probably would not be as bad as it seems if we had higher than 16xAF
To sum it up: New features can be paraphrased as "Way to cheap out IQ for performance gain."
Maybe good for 8K, somewhat OK for 4K. But hit to per pixel quality on 1440p will be unpleasant. On 1080p unacceptable.
SniperX
Member
Posts: 40
Member
Posts: 40
Posted on: 09/14/2018 02:40 PM
Same here, Duke all the way. Regarding the architecture, it's clear where the gaps are that will be filled by the 2080+ and the Titan X (looks like there won't be a 2070 Ti). Anticipation for the reviews at like at 11.
Same here, Duke all the way. Regarding the architecture, it's clear where the gaps are that will be filled by the 2080+ and the Titan X (looks like there won't be a 2070 Ti). Anticipation for the reviews at like at 11.
-Tj-
Senior Member
Posts: 15895
Senior Member
Posts: 15895
Posted on: 09/14/2018 02:45 PM
Wow this 2080rtx is crippled like shit with tensor cores compared to TI variant nvm 2070..
Wow this 2080rtx is crippled like shit with tensor cores compared to TI variant nvm 2070..
MK80
Member
Posts: 80
Member
Posts: 80
Posted on: 09/14/2018 02:58 PM
OC/2200 Ghz o_O nice
OC/2200 Ghz o_O nice
Robbo9999
Senior Member
Posts: 1159
Senior Member
Posts: 1159
Posted on: 09/14/2018 03:20 PM
Yeah, comparitively small gaps between 2070 up to 2080, and then 2080ti comes along & massively increases the gap in specs.
I like the look of the Duke in comparison with the other MSI card, like someone already mentioned, I also like the NVidia Founders aesthetic. Looking forward to the reviews, and will read up on the associated tech of the architecture a bit more once the card reviews are out, I understand/know some of it, but not all.
Wow this 2080rtx is crippled like crap with tensor cores compared to TI variant nvm 2070..
Yeah, comparitively small gaps between 2070 up to 2080, and then 2080ti comes along & massively increases the gap in specs.
I like the look of the Duke in comparison with the other MSI card, like someone already mentioned, I also like the NVidia Founders aesthetic. Looking forward to the reviews, and will read up on the associated tech of the architecture a bit more once the card reviews are out, I understand/know some of it, but not all.
Monchis
Senior Member
Posts: 1295
Senior Member
Posts: 1295
Posted on: 09/14/2018 03:27 PM
The dlss comparison shot is really tiny.
The dlss comparison shot is really tiny.
Kaarme
Senior Member
Posts: 1333
Senior Member
Posts: 1333
Posted on: 09/14/2018 03:30 PM
At least this way 2070 won't be made using second-rate chips that weren't good enough for 2080.
At least this way 2070 won't be made using second-rate chips that weren't good enough for 2080.
scoter man1
Senior Member
Posts: 4681
Senior Member
Posts: 4681
Posted on: 09/14/2018 03:36 PM
C'mon HH, show us the numbers =D
C'mon HH, show us the numbers =D
Agent-A01
Senior Member
Posts: 11318
Senior Member
Posts: 11318
Posted on: 09/14/2018 03:37 PM
@Hilbert Hagedoorn
Thanks for the article.
Can you post the full photo of DLSS comparison?
Also can you comment on this
"Another important change with NVLink SLI is that now each GPU can access the other's memory in a cache-coherent way, which lets them combine framebuffer sizes- something that was not possible with SLI before. The underlying reason is that the old SLI link was used to only transfer the final rendered frames to the master GPU, which would then combine them with its own frame data and then output the combined image on the monitor. In framebuffer-combined mode, each GPU will automatically route memory requests to the correct card no matter which GPU is asking for which chunk of memory."
You said that RTX cannot share memory.
Is this a software limitation where shared memory-access is limited to 'prosumer' cards?
@Hilbert Hagedoorn
Thanks for the article.
Can you post the full photo of DLSS comparison?
Also can you comment on this
"Another important change with NVLink SLI is that now each GPU can access the other's memory in a cache-coherent way, which lets them combine framebuffer sizes- something that was not possible with SLI before. The underlying reason is that the old SLI link was used to only transfer the final rendered frames to the master GPU, which would then combine them with its own frame data and then output the combined image on the monitor. In framebuffer-combined mode, each GPU will automatically route memory requests to the correct card no matter which GPU is asking for which chunk of memory."
You said that RTX cannot share memory.
Is this a software limitation where shared memory-access is limited to 'prosumer' cards?
Fox2232
Senior Member
Posts: 8133
Senior Member
Posts: 8133
Posted on: 09/14/2018 03:44 PM
@Hilbert Hagedoorn
Thanks for the article.
Can you post the full photo of DLSS comparison?
Also can you comment on this
"Another important change with NVLink SLI is that now each GPU can access the other's memory in a cache-coherent way, which lets them combine framebuffer sizes- something that was not possible with SLI before. The underlying reason is that the old SLI link was used to only transfer the final rendered frames to the master GPU, which would then combine them with its own frame data and then output the combined image on the monitor. In framebuffer-combined mode, each GPU will automatically route memory requests to the correct card no matter which GPU is asking for which chunk of memory."
You said that RTX cannot share memory.
Is this a software limitation where shared memory-access is limited to 'prosumer' cards?
It does not matter for gaming. Sharing even 1GB of VRAM via 100GB/s available for 2080Ti will add 10ms to rendering time (ignoring actual latency of random accesses themselves).
You would be better off paying extra 1GB of VRAM on each card and paying for additional 32bit on IMC for that 1GB.
@Hilbert Hagedoorn
Thanks for the article.
Can you post the full photo of DLSS comparison?
Also can you comment on this
"Another important change with NVLink SLI is that now each GPU can access the other's memory in a cache-coherent way, which lets them combine framebuffer sizes- something that was not possible with SLI before. The underlying reason is that the old SLI link was used to only transfer the final rendered frames to the master GPU, which would then combine them with its own frame data and then output the combined image on the monitor. In framebuffer-combined mode, each GPU will automatically route memory requests to the correct card no matter which GPU is asking for which chunk of memory."
You said that RTX cannot share memory.
Is this a software limitation where shared memory-access is limited to 'prosumer' cards?
It does not matter for gaming. Sharing even 1GB of VRAM via 100GB/s available for 2080Ti will add 10ms to rendering time (ignoring actual latency of random accesses themselves).
You would be better off paying extra 1GB of VRAM on each card and paying for additional 32bit on IMC for that 1GB.
H83
Senior Member
Posts: 2526
Senior Member
Posts: 2526
Posted on: 09/14/2018 04:02 PM
First of all, very nice article, i managed to understand most of it... Can´t wait for the review!
That Duke card is really a looker although my Duke card looks even better. Too bad MSI managed to ruin the looks of the gaming version... Nvidia´s FE look a lot like the cards from Pallit, at least to me.
Also 79$ for a SLI bridge??? Nvidia needs to stop copying Apple´s business practices...
First of all, very nice article, i managed to understand most of it... Can´t wait for the review!
That Duke card is really a looker although my Duke card looks even better. Too bad MSI managed to ruin the looks of the gaming version... Nvidia´s FE look a lot like the cards from Pallit, at least to me.
Also 79$ for a SLI bridge??? Nvidia needs to stop copying Apple´s business practices...
tunejunky
Senior Member
Posts: 771
Senior Member
Posts: 771
Posted on: 09/14/2018 04:10 PM
just wow at these prices.
given that there is no need for any of these products the enthusiast market will quickly be tapped. and this is exactly why the 10xx is slow-walking its E.O.L.
even a competitive gamer can buy a 1080ti (at the lowest ever price) for half as much and have better performance in some scenarios.
my gawd, you can even buy a 1080 and a g-sync monitor together for less money... i'm shocked i can even say that.
all of this just to beat out C.E.S. and AMD's announcement of the first mid-level card with high-end performance.
just wow at these prices.
given that there is no need for any of these products the enthusiast market will quickly be tapped. and this is exactly why the 10xx is slow-walking its E.O.L.
even a competitive gamer can buy a 1080ti (at the lowest ever price) for half as much and have better performance in some scenarios.
my gawd, you can even buy a 1080 and a g-sync monitor together for less money... i'm shocked i can even say that.
all of this just to beat out C.E.S. and AMD's announcement of the first mid-level card with high-end performance.
StewieTech
Chuck Norris
Posts: 2537
Chuck Norris
Posts: 2537
Posted on: 09/14/2018 04:20 PM
Yeah, what gives these prices? They´re stupid man...
For real, i can remember a time not too long ago when the top dawg went for 600 bucks.
Yeah, what gives these prices? They´re stupid man...
For real, i can remember a time not too long ago when the top dawg went for 600 bucks.
PrMinisterGR
Senior Member
Posts: 6970
Senior Member
Posts: 6970
Posted on: 09/14/2018 04:28 PM
Great job Hilbert!
I know that everybody is justified to talk about the prices, but the tech itself looks crazy, I am genuinely impressed, especially considering that Nvidia didn't have to innovate at all. Pascal with +40% Shader engines would be enough for everyone to be satisfied, and much cheaper to produce.
Applause for the leather jacket.
I expect the initial benchmarks to be underwhelming, as this is a new architecture, but if Nvidia has a good driver for this at a point, it should be night and day with Pascal. The potential of the Tensor cores alone is insane.
Great job Hilbert!
I know that everybody is justified to talk about the prices, but the tech itself looks crazy, I am genuinely impressed, especially considering that Nvidia didn't have to innovate at all. Pascal with +40% Shader engines would be enough for everyone to be satisfied, and much cheaper to produce.
Applause for the leather jacket.
I expect the initial benchmarks to be underwhelming, as this is a new architecture, but if Nvidia has a good driver for this at a point, it should be night and day with Pascal. The potential of the Tensor cores alone is insane.
wavetrex
Senior Member
Posts: 457
Senior Member
Posts: 457
Posted on: 09/14/2018 04:41 PM
I can do ~2100 with my 1080 if I want to, and running at 1930'ish out of the box in the games that I play.
2200 is unimpressive at all (but expected, since this is basically the same node, just slightly refined), and it will probably consume HUGE amounts of power at that frequency, given that Turing is a bigger chip with all those fancy semi-useless raytracing corez.
Really curious what the benchmarks will show compared to SIMILARLY PRICED Pascal:
So 2070 results vs 1080 Ti, and not vs 1070.
OC/2200 Ghz o_O nice
I can do ~2100 with my 1080 if I want to, and running at 1930'ish out of the box in the games that I play.
2200 is unimpressive at all (but expected, since this is basically the same node, just slightly refined), and it will probably consume HUGE amounts of power at that frequency, given that Turing is a bigger chip with all those fancy semi-useless raytracing corez.
Really curious what the benchmarks will show compared to SIMILARLY PRICED Pascal:
So 2070 results vs 1080 Ti, and not vs 1070.
JamesSneed
Senior Member
Posts: 401
Senior Member
Posts: 401
Posted on: 09/14/2018 04:44 PM
Can't wait to see testing results. I have this sneaking suspicion that the improvements are not that great unless games are making use of new features. Just going off the onslaught of pre-launch marketing that we have seen.
Can't wait to see testing results. I have this sneaking suspicion that the improvements are not that great unless games are making use of new features. Just going off the onslaught of pre-launch marketing that we have seen.
Denial
Senior Member
Posts: 11800
Senior Member
Posts: 11800
Posted on: 09/14/2018 04:45 PM
Great job Hilbert!
I know that everybody is justified to talk about the prices, but the tech itself looks crazy, I am genuinely impressed, especially considering that Nvidia didn't have to innovate at all. Pascal with +40% Shader engines would be enough for everyone to be satisfied, and much cheaper to produce.
Applause for the leather jacket.
I expect the initial benchmarks to be underwhelming, as this is a new architecture, but if Nvidia has a good driver for this at a point, it should be night and day with Pascal. The potential of the Tensor cores alone is insane.
Haven't seen you in a bit - how are things going?
@Hilbert Hagedoorn
Thanks for the article.
Can you post the full photo of DLSS comparison?
Also can you comment on this
"Another important change with NVLink SLI is that now each GPU can access the other's memory in a cache-coherent way, which lets them combine framebuffer sizes- something that was not possible with SLI before. The underlying reason is that the old SLI link was used to only transfer the final rendered frames to the master GPU, which would then combine them with its own frame data and then output the combined image on the monitor. In framebuffer-combined mode, each GPU will automatically route memory requests to the correct card no matter which GPU is asking for which chunk of memory."
You said that RTX cannot share memory.
Is this a software limitation where shared memory-access is limited to 'prosumer' cards?
I think that its more limited to prosumer workloads. The latency access across NVLink is too slow to complete rasterization tasks in the time required for realtime frame rendering. With certain workloads like data modeling and stuff, the card doesn't have to spit out a frame in X(ms).. it can take longer, in which case this can be useful. There may be upcoming graphics techniques where this can be applied but at the moment mGPU/SLI will continue to just share the frame buffer at the end.
Great job Hilbert!
I know that everybody is justified to talk about the prices, but the tech itself looks crazy, I am genuinely impressed, especially considering that Nvidia didn't have to innovate at all. Pascal with +40% Shader engines would be enough for everyone to be satisfied, and much cheaper to produce.
Applause for the leather jacket.
I expect the initial benchmarks to be underwhelming, as this is a new architecture, but if Nvidia has a good driver for this at a point, it should be night and day with Pascal. The potential of the Tensor cores alone is insane.
Haven't seen you in a bit - how are things going?
@Hilbert Hagedoorn
Thanks for the article.
Can you post the full photo of DLSS comparison?
Also can you comment on this
"Another important change with NVLink SLI is that now each GPU can access the other's memory in a cache-coherent way, which lets them combine framebuffer sizes- something that was not possible with SLI before. The underlying reason is that the old SLI link was used to only transfer the final rendered frames to the master GPU, which would then combine them with its own frame data and then output the combined image on the monitor. In framebuffer-combined mode, each GPU will automatically route memory requests to the correct card no matter which GPU is asking for which chunk of memory."
You said that RTX cannot share memory.
Is this a software limitation where shared memory-access is limited to 'prosumer' cards?
I think that its more limited to prosumer workloads. The latency access across NVLink is too slow to complete rasterization tasks in the time required for realtime frame rendering. With certain workloads like data modeling and stuff, the card doesn't have to spit out a frame in X(ms).. it can take longer, in which case this can be useful. There may be upcoming graphics techniques where this can be applied but at the moment mGPU/SLI will continue to just share the frame buffer at the end.
PrMinisterGR
Senior Member
Posts: 6970
Senior Member
Posts: 6970
Posted on: 09/14/2018 04:54 PM
Haven't seen you in a bit - how are things going?
I think that its more limited to prosumer workloads. The latency access across NVLink is too slow to complete rasterization tasks in the time required for realtime frame rendering. With certain workloads like data modeling and stuff, the card doesn't have to spit out a frame in X(ms).. it can take longer, in which case this can be useful. There may be upcoming graphics techniques where this can be applied but at the moment mGPU/SLI will continue to just share the frame buffer at the end.
Hey man, just fine, but a different job role, I have missed you guys
I actually believe that AMD will compete with all this very soon, just not in the ultra high end. 750+mm wafers are an insane risk.
Haven't seen you in a bit - how are things going?
I think that its more limited to prosumer workloads. The latency access across NVLink is too slow to complete rasterization tasks in the time required for realtime frame rendering. With certain workloads like data modeling and stuff, the card doesn't have to spit out a frame in X(ms).. it can take longer, in which case this can be useful. There may be upcoming graphics techniques where this can be applied but at the moment mGPU/SLI will continue to just share the frame buffer at the end.
Hey man, just fine, but a different job role, I have missed you guys
I actually believe that AMD will compete with all this very soon, just not in the ultra high end. 750+mm wafers are an insane risk.
Noisiv
Senior Member
Posts: 6499
Senior Member
Posts: 6499
Posted on: 09/14/2018 04:55 PM
This is just an appetizer! Let the crow eating begin:
https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/technologies/turing-architecture/NVIDIA-Turing-Architecture-Whitepaper.pdf
This is just an appetizer! Let the crow eating begin:
64x super-sampled AA is comparable to DLSS 2x
https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/technologies/turing-architecture/NVIDIA-Turing-Architecture-Whitepaper.pdf
PrMinisterGR
Senior Member
Posts: 6970
Senior Member
Posts: 6970
Posted on: 09/14/2018 05:07 PM
It all depends on the quality of the model training. Nvidia exaggerates always, so hold a smaller basket.
It is still impressive and the only logical way forward. Both the machine learning and the partial ray tracing. If AMD was more liquid, we would have had these a bit earlier.
People also forget that we will have an Intel GPU in a couple of years. Things are bound to be more interesting then.
It all depends on the quality of the model training. Nvidia exaggerates always, so hold a smaller basket.
It is still impressive and the only logical way forward. Both the machine learning and the partial ray tracing. If AMD was more liquid, we would have had these a bit earlier.
People also forget that we will have an Intel GPU in a couple of years. Things are bound to be more interesting then.
chispy
Senior Member
Posts: 8517
Senior Member
Posts: 8517
Posted on: 09/14/2018 05:56 PM
That's a lot of new tech pack into this new video cards , great review @Hilbert Hagedoorn , it looks good but that price it is what turns off many people. Nevertheless great to see new technology moving forward.
That's a lot of new tech pack into this new video cards , great review @Hilbert Hagedoorn , it looks good but that price it is what turns off many people. Nevertheless great to see new technology moving forward.
pharma
Senior Member
Posts: 907
Senior Member
Posts: 907
Posted on: 09/14/2018 05:58 PM
Any new RTX tech/process adopted by competitors will need to be different so as not to infringe on Nvidia's hardware and software IP. In that aspect it may take longer to develop as opposed to licensing.
Any new RTX tech/process adopted by competitors will need to be different so as not to infringe on Nvidia's hardware and software IP. In that aspect it may take longer to develop as opposed to licensing.
Reddoguk
Senior Member
Posts: 1605
Senior Member
Posts: 1605
Posted on: 09/14/2018 06:27 PM
I personally think these first RTX cards will be a flop. They will sell but people have already realized that these cards with RT On won't be pushing out the performance they expected.
Real-Time Ray Tracing might be a cool new feature but the hardware to run that type of tech is not here yet.
I don't care too much about Ray Tracing if you can only hit 30-60fps at 1080p.
I personally think these first RTX cards will be a flop. They will sell but people have already realized that these cards with RT On won't be pushing out the performance they expected.
Real-Time Ray Tracing might be a cool new feature but the hardware to run that type of tech is not here yet.
I don't care too much about Ray Tracing if you can only hit 30-60fps at 1080p.
Solfaur
Senior Member
Posts: 6938
Senior Member
Posts: 6938
Posted on: 09/14/2018 06:28 PM
Strictly aesthetically speaking, I really dislike the MSI Gaming this time around, the design is a mess. I wonder how much better the cooling is compared to the Duke, which I find to look a lot better. And of course I'm looking forward to all the Turing reviews like everybody else, despite the fact that I'll most likely skip this generation. Nice preview boss.
Strictly aesthetically speaking, I really dislike the MSI Gaming this time around, the design is a mess. I wonder how much better the cooling is compared to the Duke, which I find to look a lot better. And of course I'm looking forward to all the Turing reviews like everybody else, despite the fact that I'll most likely skip this generation. Nice preview boss.
pages 1 2 3 4 5 6
Click here to post a comment for this article on the message forum.
Member
Posts: 20
I like the look of the Duke better than the Gaming Trio.