• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

NVIDIA Pascal Thread

Page 179 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
I initially thought it was my PSU going bad as I've never experienced GPU coil whine before. So after getting a new one it still persisted 🙁

Sent from my Nexus 6P using Tapatalk
 
Same name instead of rebrand, probably slightly higher core clock, new/faster memory chips, keeps only 1 x 6-pin power connector (updated GTX 1060). 🙂

MSI also has new models:

MSI-GTX-1080-1060-GAMING-X-PLUS.jpg


https://webcache.googleusercontent....01060-gaming-x-plus+&cd=2&hl=nl&ct=clnk&gl=nl

GTX 1080 looks much better at $499 with faster memory, I actually prefer it over GTX 1070 at current pricing now.
 
Interesting tidbit for any EVGA owners thinking about stepping up to the 1080 Ti. As of now, EVGA is only planning on including the 1080 Ti FE in the Step-Up program (which is how I got mine). While EVGA just announced custom models, none run at reference clocks, and EVGA only includes reference-clocked models (with either reference or non-reference coolers) in the Step-Up program.
 
Just noticed that techpower up add some new games and the gap between GTX1070 and 1080 is 26% avg now.It was 21% at launch.They still keep some very old games like GTA5 so the gap would be even greater with only new games close to 30% probably.
Also gap vs furyx was 14% at launch and its only 7% now
GTX1070 launch
perfrel_2560_1440.png

Today
perfrel_2560_1440.png
 
interesting 11,4 Gbps Ram, so 7% more shaderpower and 19% more Bandwidth. Should yield in a bit more than 10% faster speed as the 1080Ti, probably a bit faster than the best Custom 1080Tis. But of course still so much overprized, that i don't know who is buying this.
 
Last edited:
Samwell said:
interesting 11,4 Gbps Ram, so 7% more shaderpower and 19% more Bandwidth. Should yield in a bit more than 10% faster speed as the 1080Ti, probably a bit faster than the best Custom 1080Tis. But of course still so much overprized, that i don't know who is buying this.
Click to expand...

People with money that like top performance. Pretty simple actually.
 
Rumor: GV104 scores ~18.300 points @ FSE

2-3dmark_firestrike_extreme.png

Comes from a source who previously leaked Pascal info on chinese forums. I assume it's the graphics score. Also says it sports 256-bit GDDR6.
 
Sweepr said:
Rumor: GV104 scores ~18.300 points @ FSE

2-3dmark_firestrike_extreme.png

Comes from a source who previously leaked Pascal info on chinese forums. I assume it's the graphics score. Also says it sports 256-bit GDDR6.
Click to expand...

That's 1.78X GTX 1080, or about the jump we saw from GTX 980 to GTX 1080.
 
Similar to the ratio between GP104 and GM200.

256-bit GDDR6 14 GT/s means 448 GB/s: memory bandwidth is reduced from GeForce GTX 1080 Ti.
 
Hmm, don't see anything on the included 3DMark chart that scores 18K in FSE. Glitch? Anywho - I avoid 'ru' domains like the plague, or I'd look it up my self.
 
Pretty sure it's fake. It's way to early for Volta scores. Scores normally leak at last, when you know nearly everything else. From Volta we don't know anything.
Another point is that we don't have Gv104 in Nvidias drivers yet. With Maxwell/Pascal the gpu names leaked 6-12 Months in advance in drivers (In 11/2015 you could see the whole range with GP102, which no one heard before). So far only GV100 has been in drivers in february this year, so other Voltas will be later and in no way there are real benches existing.
 
Fake and gv104 wont be 35% faster than full GP102.It will be similar to kepler-maxwell around 10% for GTX1180.
But if they want catch up with GTX1170 atleast GTX1080TI they will need cutdown it very little otherwise it will be slower.
1080TI is around 60-70% faster than GTX1070 now so they need bigger jump from GTX1070 to GTX1170 than it was from GTX970 to GTX1070(only 55% avg).They need atleast 60% to match 1080TI and it will be hard for same node.
 
That's a pretty huge jump from GT 730 GK208 to GP108 on the desktop side and GM108 to GP108 on the mobile side. It brings a lot of new features that have been missing from the entry level GPUs, DP1.4, HDMI 2.0b, fixed function HEVC Main10/Main12 hardware decoding & VP9 hardware decoding, HEVC Main10 hardware encoding, Feature Level 12_1 support, CUDA Compute Capability 6.1 instead of CC3.5 on GK208 & CC5.2 on GM108.
 
Samwell said:
Pretty sure it's fake. It's way to early for Volta scores. Scores normally leak at last, when you know nearly everything else. From Volta we don't know anything.
Another point is that we don't have Gv104 in Nvidias drivers yet. With Maxwell/Pascal the gpu names leaked 6-12 Months in advance in drivers (In 11/2015 you could see the whole range with GP102, which no one heard before). So far only GV100 has been in drivers in february this year, so other Voltas will be later and in no way there are real benches existing.
Click to expand...

Although it's certainly within plausible performance parameters.

Remember the 1080 was 30% faster than the 980 Ti, but part of that was due to the fact that Nvidia simply clocked the card higher and left less overclocking headroom (two cards are closer to 10-15% apart when both are max overclocked).

A GV104 may in fact pull off a 18,300 score in the end, it just might be the max-overclock score (even if the 18.3k score here is just randomly made up).
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top