NVIDIA Pascal Thread

[Sweepr]

NVIDIA GeForce GTX 1060 official performance leaked

http://videocardz.com/62122/nvidia-geforce-gtx-1060-official-performance-leaked

 

[Sweepr]

Geforce GTX 1060 6GB listed in US stores ($259)

6GB model is already available for pre-order in the US, for $259 and $263:

https://www.sabrepc.com/catalogsearch/result/?q=gtx+1060&order=relevance&dir=desc
http://www.shopblt.com/search/order_id=%21ORDERID%21&s_max=25&t_all=1&s_all=gtx+1060

Not bad at all for pre-launch prices, closer to MSRP than other Pascal cards.


Geforce GTX 1060 6GB OC - 3DMark Fire Strike Extreme & Ultra Results

Fire Strike Extreme Graphics Score: 6517
Fire Strike Ultra Graphics Score: 3116

http://www.3dmark.com/compare/fs/9258975/fs/9249139/fs/9247799#
http://www.3dmark.com/compare/fs/9259050/fs/9248958/fs/9247684#

 

[Sweepr]

Geforce GTX 1060 3GB Specs

https://benchlife.info/nvidia-geforce-gtx-1060-3gb-spec-07122016

GP106, 1152 SPs, 3GB GDDR5. Launch 2-3 weeks after the 6GB model.


NVIDIA's Geforce GTX 1060 Reviewer's Guide

http://videocardz.com/62138/nvidia-geforce-gtx-1060-reviewers-guide-leaked

GP106 = 200mm²

 

[LTC8K6]

http://i1024.photobucket.com/albums/y303/martmail55/gp106-300_zpssbkucixh.jpg

Memory and core looks a tad slow. I think that will hold the 3gb cards back a bit.

 

[96Firebird]

Memory compression man!

 

[Sweepr]

Mobile Geforce GTX 1060 Specs

1280 SPs
80 TMUs
48 ROPs
6 GB 192-bit (192 GB/s)
Core Clock: 1405 MHz (base), 1671 MHz (boost)
65W TDP

www.purepc.pl/notebooki/geforce_gtx_1060_mobile_pelny_uklad_gp106_trafi_do_laptopow

Almost identical to the desktop version, at (nearly) half the TDP.

 

[LTC8K6]

Mobile Geforce GTX 1060 Specs

1280 SPs
80 TMUs
48 ROPs
6 GB 192-bit (192 GB/s)
Core Clock: 1405 MHz (base), 1671 MHz (boost)
65W TDP

www.purepc.pl/notebooki/geforce_gtx_1060_mobile_pelny_uklad_gp106_trafi_do_laptopow

Almost identical to the desktop version, at (nearly) half the TDP.

GPU-Z says 106 TMUs.

 

[Sweepr]

First rumour about Volta.

Nvidia Volta gets 16nm FinFET treatment

A little birdie told us that Nvidia is giving its Volta the 16nm FinFET treatment. This product uses stacked DRAM too so it looks like the whole thing will be pretty bleeding edge.

Our same deep throat told us that the performance per watt is expected to increase tremendously. Although this might be vague, little is known about Volta other than it is arriving after Pascal so any information we get is news. The earliest we expect Volta is 2017.

www.fudzilla.com/news/graphics/41122-nvidia-volta-is-16nm-finfet

Fuad says it's a 16nm part with stacked DRAM, 'tremendous perf/watt increase'. Considering all Pascal GPUs were taped-out as of mid-2016, the first Volta GPU(s) could arrive in 2017, while the competition might still be catching up with GP104/GP102.

GPU-Z says 106 TMUs.

And it's wrong. GP106 is 80 TMUs.

 

[Sweepr]

Geforce GTX 1060 3DMark Time Spy DirectX 12 Results

http://www.3dmark.com/compare/spy/20652/spy/11620/spy/19654/spy/21277/spy/25212/spy/11101#

Different scores from different users.
3 on the right are at stock, 3 on the left OC.
Graphics score of 3923 / 3994 / 3999 pts at stock, matches the RX 480:

Geforce GTX 1060 3GB launching next month?

- Geforce GTX 1060 6GB launching July 19th
- Geforce GTX 1060 3GB launching August 10th

https://world.tmall.com/item/535668...176416&cm_id=140105335569ed55e27b&abbucket=15

 

[FORTHEWIND]

Does anyone know how Pascal handle Async Compute execution? Some say no, some say with preemption and Ryan Smith said in Time Spy comments section that Pascal really does support it. (Confused)

 

[Deders]

From what I gather it's down to granularity. AMD can do it per shader.

Nvidia on the other hand used to be able to split different kinds of calculations between different gpc's, effectively lopping off a large portion of rendering power in order to allocate it to compute calculations.

With Pascal different types of calculations can be allocated to each SMM cluster within each GPC group, and the amount of shaders within each cluster has been halved compared to Maxwell allowing for finer granularity.

 

[renderstate]

If you understand what preemption is then you know it makes no sense to say it's used to run concurrent tasks.

Let me tell you how concurrent tasks run on Pascal. It's really easy. You see, some SMs are idle and the HW scheduler decides to schedule some independent compute (or graphics) work on them.

Done. It wasn't that hard If you cast away all the unfounded statements we heard from AMD less technically savvy supporters you see there is really no mystery.

Now we just need to repeat this for three or four months and with a bit of luck the message will percolate. No wait, who am I kidding

 

[FORTHEWIND]

If you understand what preemption is then you know it makes no sense to say it's used to run concurrent tasks.

Let me tell you how concurrent tasks run on Pascal. It's really easy. You see, some SMs are idle and the HW scheduler decides to schedule some independent compute (or graphics) work on them.

Done. It wasn't that hard If you cast away all the unfounded statements we heard from AMD less technically savvy supporters you see there is really no mystery.

Now we just need to repeat this for three or four months and with a bit of luck the message will percolate. No wait, who am I kidding

1. I didn't said anything about used for concurrent workload.

2. Jee, what was I posting again? Right, a question about How Pascal does Async Compute . Does that doesn't qualified as a question? If so, how could you accused me of saying the bold.

3. Pascal used a software-based scheduler. Where did it get the HW scheduler from? Fermi?

4. A more technical answer would be more helpful and educating would you agree?

5. Don't be a douche. I asked nicely for an explanation and did not try to offend your Pascal Overlord k.

From what I gather it's down to granularity. AMD can do it per shader.

Nvidia on the other hand used to be able to split different kinds of calculations between different gpc's, effectively lopping off a large portion of rendering power in order to allocate it to compute calculations.

With Pascal different types of calculations can be allocated to each SMM cluster within each GPC group, and the amount of shaders within each cluster has been halved compared to Maxwell allowing for finer granularity.

Thanks @Deders for a genuine explanation . But, isn't Pascal can do it per shader now too (Maxwell was GPC only, Pascal per SM inside GPC) What's confused me is some said it's a clever hack (Pre-emtion, NV driver) so... yeah. Confused.

 

[rgallant]

If you understand what preemption is then you know it makes no sense to say it's used to run concurrent tasks.

Let me tell you how concurrent tasks run on Pascal. It's really easy. You see, some SMs are idle and the HW scheduler decides to schedule some independent compute (or graphics) work on them.

Done. It wasn't that hard If you cast away all the unfounded statements we heard from AMD less technically savvy supporters you see there is really no mystery.

Now we just need to repeat this for three or four months and with a bit of luck the message will percolate. No wait, who am I kidding

you forgot to flush [the data that is , on nv cards what I heard]

 

[Deders]

Thanks @Deders for a genuine explanation . But, isn't Pascal can do it per shader now too (Maxwell was GPC only, Pascal per SM inside GPC) What's confused me is some said it's a clever hack (Pre-emtion, NV driver) so... yeah. Confused.

I think basically the pre-emption decides which SM clusters are needed to calculate which kinds of tasks in order to maintain the highest framerate.

 

[Sweepr]

Mobile Geforce GTX 1080/1070/1060 will be released in a month.

Nvidia brings Pascal graphics chips for Mobile in August

Mobile Geforce GTX 1060
1280 CUDA cores
6GB GDDR5 192-bit
65W TDP

Mobile Geforce GTX 1070
1920 CUDA cores
8GB GDDR5 256-bit
85W TDP

Mobile Geforce GTX 1080
2560 CUDA cores
8GB GDDR5 256-bit
125W TDP

www.golem.de/news/notebooks-nvidia-bringt-pascal-grafikchips-fuer-mobile-im-august-1607-122153.html

Same core count as desktops, at lower clockspeeds.

 

[railven]

Mobile Geforce GTX 1080/1070/1060 will be released in a month.

Nvidia brings Pascal graphics chips for Mobile in August

Mobile Geforce GTX 1060
1280 CUDA cores
6GB GDDR5 192-bit
65W TDP

Mobile Geforce GTX 1070
1920 CUDA cores
8GB GDDR5 256-bit
85W TDP

Mobile Geforce GTX 1080
2560 CUDA cores
8GB GDDR5 256-bit
125W TDP

www.golem.de/news/notebooks-nvidia-bringt-pascal-grafikchips-fuer-mobile-im-august-1607-122153.html

Same core count as desktops, at lower clockspeeds.

Is Nvidia launching first in mobile? I don't follow that sector as much (only buy a laptop during my December vacations which I then promptly give to my nephews since I have little use for them.)

If true, WTH were people saying about NV being so far behind?

 

[godihatework]

Is Nvidia launching first in mobile? I don't follow that sector as much (only buy a laptop during my December vacations which I then promptly give to my nephews since I have little use for them.)

If true, WTH were people saying about NV being so far behind?

same FUD as it ever was. people were rumormongering based on unsourced forum posts and treating it as gospel...

pretty much the AT VC forum modus operandi in a nutshell.

 

[sontin]

And they had just believed AMD's fud campaign. Look at this:

https://www.computerbase.de/2015-12/amd-radeon-support-fuer-hdr-und-4k-displays-mit-120-hz-ab-2016/

And the Raja guy:

We believe we’re several months ahead of this transition, especially for the notebook and the mainstream market. The competition is talking about chips for cars and stuff, but not the mainstream market.

http://venturebeat.com/2016/01/15/a...o-full-graphics-immersion-with-16k-screens/2/

AMD press slides and discussion has no business in the Nvidia subforum.

-Rvenger

 

[Rvenger]

First rumour about Volta.

Nvidia Volta gets 16nm FinFET treatment



www.fudzilla.com/news/graphics/41122-nvidia-volta-is-16nm-finfet

Fuad says it's a 16nm part with stacked DRAM, 'tremendous perf/watt increase'. Considering all Pascal GPUs were taped-out as of mid-2016, the first Volta GPU(s) could arrive in 2017, while the competition might still be catching up with GP104/GP102.




And it's wrong. GP106 is 80 TMUs.

Volta discussion has no business in this thread. If you continue to attempt to derail your own thread I will lock it and remove you from VC&G. I have warned you TWICE regarding this. Make a new thread if you want to talk about Volta.


-Rvenger

 

[kithylin]

So where are all the GTX-1060 "full reviews" based on review samples sent to all these different hardware sites to do reviews for us?

The NDA's been lifted for 11 days (12 if you count today) and the retail versions are going to the stores here in a couple days and I can't find solid, objective reviews on the things yet. :|

 

[96Firebird]

So where are all the GTX-1060 "full reviews" based on review samples sent to all these different hardware sites to do reviews for us?

The NDA's been lifted for 11 days (12 if you count today) and the retail versions are going to the stores here in a couple days and I can't find solid, objective reviews on the things yet. :|

The NDA lifts at 9AM EST today...

 

[Termie]

The NDA lifts at 9AM EST today...

First reviews are up:

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_1060/

http://www.techspot.com/review/1209-nvidia-geforce-gtx-1060/

 

[kithylin]

The NDA lifts at 9AM EST today...

I heard a lot of people (Somewhere even in this thread) said it was on the 7'th.

EDIT: Right here. So NDA's been off for nearly 12 days and hardly no info.

Some new info/rumour about Geforce GTX 1060 by Bench-life. Better translation would be appreciated!

- NDA lifts July 7 (next thursday)
- On sale around July 14
- 3GB and 6GB versions of GP106-300, and two versions of GP106-400 (3GB and 6GB?)
- GP106 with 1250 CUDA cores (GP106-300?) doesn't support SLI
- Geforce GTX 1060 3GB for $249, Geforce GTX 1060 6GB for $299



https://benchlife.info/nvidia-geforce-gtx-1060-launch-mid-of-july-06292016



DigiTimes says they might reduce prices to make them more competitive.

 

[guskline]

A number of GTX 1060 reviews have been posted at various websites.