Nvidia Pascal Reviews [1080XP, 1080ti, 1080, 1070ti, 1070, 1060, 1050, and 1030]

CSI PC said:
Well just like Maxwell 2, seems the 1080 is the poor value card.
According to interview with pcgameshardware, the 1070 will have the same number of ROPs, L2 cache, Delta-C compression.
The only difference according to the NVIDIA representative is the use of GDDR5 memory instead of GDDR5X.
Ok so bandwidth limited at higher resolutions, but might still pull off good results at 1440p.
http://www.pcgameshardware.de/Nvidi...cials/Geforce-GTX-1080-GTX-1070-KFKA-1195567/
Click to expand...
It's actually not an Interview, but an FAQ we set up for the readers. But correctness is confirmed with Nvidia. Just for clarity's sake.
 
Ext3h said:
I'm only complaining that they are apparently not putting the hardware to FULL use yet. Now that they fixed Pascal, it's about time that they move the DX12 compute queues to the GMU as well. Till now, the hardware queues in that are still reserved for CUDA only. In hindsight it makes sense why they didn't do that for Maxwell yet, it just wouldn't have worked properly at all. But with Pascal, that limitation is gone, and there are still *actual* gains to be achieved there.
Click to expand...
There must be use cases where the GCN way is able to exploit idle cycles where Pascal can not. Sebbbi suggested some of those cases a bit ago.
 
Razor1 said:
Just for fun what is the cost of this new chip and how is AMD going to price it lower to current products and maintain margins....

I can't see them keeping margins close if they place it at $300 for their top end Polaris if they cut prices less that $300, I'm sure they would be cutting into current margins, if they have the performance so they can bump up their price, I think they would do it wouldn't you?
Click to expand...

Volume...
 
pjbliverpool said:
It also has 80% for the 1080's memory bandwidth and almost all of it's fill rate and geometry throughput. And even if it didn't, since when did performance scale perfectly linearly with CU's and clock speed?

Also, I think the claim is that it will be faster than Titan-X, not a Titan-X killer. However, in sales terms, it should be a Titan-X killer (and a 980Ti killer) since it will be about as fast, or faster, at a vastly lower price.
Click to expand...
And, a better scheduler then Maxwell (hurray!).
I may start to thing that it could be better to completely disable the so called "async compute" on Maxwell via driver update, I bet NV could do a better job with a little of heuristic on a software serialization..
 
Alessio1989 said:
And, a better scheduler then Maxwell (hurray!).
I may start to thing that it could be better to completely disable the so called "async compute" on Maxwell via driver update, I bet NV could do a better job with a little of heuristic on a software serialization..
Click to expand...
According to NVIDIA, async is already disabled on Maxwell.
 
Alessio1989 said:
And, a better scheduler then Maxwell (hurray!).
I may start to thing that it could be better to completely disable the so called "async compute" on Maxwell via driver update, I bet NV could do a better job with a little of heuristic on a software serialization..
Click to expand...

If, it have been enabled ( i was sure it have never been so far ), why will you they disable it ? it show a small perf gain with it for pascal, so.. And tomorrow, anyway everyone will have forget about Maxwell and Async compute declaration by nvidia.
 
ToTTenTranz said:
All sources so far point to Polaris 10 having Hawaii performance at lower price and power consumption. AMD has claimed 2.5x better power efficiency for Polaris.
Hawaii's TDP is 275W. 275/2.5 = 110W.
Click to expand...
I wouldn't say all sources. One rumour put it near 980 Ti performance. If the SiSoft 1.3Ghz entry is correct and a partially disabled card, it could get quite close to the GTX 1070. In any case, we shouldn't have long to wait to find out. I hope AMD haven't hobbled it with 32 ROPs.
 
Ryan Smith said:
According to NVIDIA, async is already disabled on Maxwell.
Click to expand...
So, why does Oxide performs a vendor id check for the render path to avoid atrocious performance penalties. A software serialization should not impact so much performance after all. At least it does not on intel iGPUs... I do not trust hardware vendors any-more...
 
LordEC911 said:
Volume...
Click to expand...

Volume is one way to get over all better bottom line, but margins matter too, they can't be cutting down margins for the sake of gaining marketshare (there is an equilibrium to meet, where cutting down margins vs. extra marketshare gained equal out or is cost effective by increasing net profits) it hasn't worked for them in the past, and this is why they shied away from it for the Fury launch an the r3xx launch.
 
Alessio1989 said:
So, why does Oxide performs a vendor id check for the render path to avoid atrocious performance penalties. A software serialization should not impact so much performance after all. At least it does not on intel iGPUs... I do not trust hardware vendors any-more...
Click to expand...
I think it is a rather complex situation, and one I asked if Ryan in PM could clarify with Oxide.
Ext3H mentioned in response to my thoughts on this that Oxide removed the NVIDIA VendorID that disabled async compute in the render path, just that unfortunately the only public information we have goes back to Feb '16 from Kollock where it was mentioned to still be disabled in the code for the public game.
And I assume Ext3H's information relates to that NVIDIA is meant to handle this more recently in the driver - I assume the 364.51 one and that Oxide is meant to have also responded by re-enabling async compute universally.

But clarification from Oxide to the public would be great.
That is a nudge to you Ryan :)
Cheers
 
Well thankfully we have our response from Oxide.
Thanks Ryan.
Ryan Smith said:
Just checked with Dan Baker. Async is still functionally disabled on Ashes when it detects an NVIDIA card, including the GTX 1080 (since they don't have one to test against yet).
Click to expand...
From: https://forum.beyond3d.com/threads/...nd-analysis-thread.57188/page-64#post-1915300
So as I thought this really complicates any testing done and conclusions reached with Pascal by reviews and forums, and also gaming on NVIDIA hardware in future when supporting pre-Pascal from from both a driver and game engine perspective.

Ext3H, was it a source that suggested the opposite?
https://forum.beyond3d.com/threads/nvidia-pascal-announcement.57763/page-36#post-1912853
Thanks
 
Last edited:
Alessio1989 said:
So, why does Oxide performs a vendor id check for the render path to avoid atrocious performance penalties. A software serialization should not impact so much performance after all. At least it does not on intel iGPUs... I do not trust hardware vendors any-more...
Click to expand...
Because even if NVIDIA doesn't have full async mode enabled, it's still possible to screw up performance with a card that is already humming along at full utilization by giving it work it doesn't have the time to handle. Async is not going to be a set-it-and-forget-it kind of thing.
CSI PC said:
That with the 364.51 driver?
Cheers
Click to expand...
364, 368, etc. All drivers. Async has never been enabled for Maxwell, from what NVIDIA has told me. It is however enabled for Pascal.
 
Ryan Smith said:
Because even if NVIDIA doesn't have full async mode enabled, it's still possible to screw up performance with a card that is already humming along at full utilization by giving it work it doesn't have the time to handle. Async is not going to be a set-it-and-forget-it kind of thing.
364, 368, etc. All drivers. Async has never been enabled for Maxwell, from what NVIDIA has told me. It is however enabled for Pascal.
Click to expand...

It was what i think, and indeed, i really think it will never happend for Maxwell gpu's ...
 
Last edited:
Ryan Smith said:
Because even if NVIDIA doesn't have full async mode enabled, it's still possible to screw up performance with a card that is already humming along at full utilization by giving it work it doesn't have the time to handle. Async is not going to be a set-it-and-forget-it kind of thing.
364, 368, etc. All drivers. Async has never been enabled for Maxwell, from what NVIDIA has told me. It is however enabled for Pascal.
Click to expand...
Thanks Ryan,
there is a lot of wrong info/conclusions being used out there on various mainstream forums and critically reviews, this definitely helps to clarify it.
Cheers
 
Last edited:
silent_guy said:
I don't think the power increase of GDDR5 will be anywhere close to the power decrease of disabling 25% of the shaders.
Click to expand...

It is still likely that the default bios won't allow the same power limit on 1070 as it does on 1080. Third party models will likely have more headroom, but also on the 1080.
 
You must log in or register to reply here.

Similar threads

Rys
GPUDB integration FAQ
Replies
8
Views
3K
kaiser77_1982
K
BRiT
PSA: EVGA Cards using Nvidia 1080 and 1070 catching on fire
Replies
4
Views
3K
DuckThor Evil
DuckThor Evil
S
Pascal vs. Maxwell - a Generational Performance Comparison & Analysis
Replies
19
Views
13K
Kaotik
Kaotik
A
GP102/GP100 FP16 support vs. Performance in old DirectX 8 and DirectX 9
Replies
12
Views
5K
Razor1
R
R
Building a PC with XB1X specs *spawn
Replies
10
Views
2K
Blazkowicz
B
Back
Top