GTX 1080 Benchmarked

D.P. · 8 May 2016 at 21:52

fs123 said:
according to the leak article in the OP http://videocardz.com/59558/nvidia-geforce-gtx-1080-3dmark-benchmarks:

The 1080 can run higher clocks due to the 16nm process but clock for clock it is slower than the 980ti so the architecture is not really faster with less CUDA cores as someone mentioned. So the fact is that a 1080 will need to run at over 2GHz to match a 1.8GHz 980ti. The stock 1080 just about matches a 1.5GHz 980ti.

The base clock of the 1080 is actually 1600MHz.

Phixsator · 8 May 2016 at 22:15

D.P. said:
The base clock of the 1080 is actually 1600MHz.

But do we know at what boost clocks the 1080 benchmarks have been done at? considering just running the card is classified "Stock" even though it may as well boost high thanks to optimal conditiones.

Safetytrousers · 8 May 2016 at 22:19

fs123 said:
. So the fact is that a 1080 will need to run at over 2GHz to match a 1.8GHz 980ti. The stock 1080 just about matches a 1.5GHz 980ti.

People keep on talking about a highly clocked 980ti being near the 1080. My 980ti will not get anywhere near 1.5, and 1.8 is the realm of very exotic builds. The 1080 is at those high speeds as standard. If a much higher clock was the only positive difference a 1080 could give that alone would be of interest to me.

Tommy_Here · 8 May 2016 at 22:22

NirK said:
So on the 27th apparently the only card will be the founders edition? The other cards will be later and you will not be able to buy the normal reference edition or custom cards on the 27th?

I've read custom cards are to be available on the 27th and yes if they have enough stock after all the preorders you'll be able to pick one up I know I'll definitely be preordering

D.P. · 8 May 2016 at 22:34

Safetytrousers said:
People keep on talking about a highly clocked 980ti being near the 1080. My 980ti will not get anywhere near 1.5, and 1.8 is the realm of very exotic builds. The 1080 is at those high speeds as standard. If a much higher clock was the only positive difference a 1080 could give that alone would be of interest to me.

People are saying that because desperate to justify their Nvidia bashing and endless "Pascal is 6 months later than Polaris" talk.. "This card that is way faster than the previous gen is not actually faster if you underclock it to values it was never design to run at and you can get Liquid Nitrogen to run the old cards at a 50% over clock...."

queamin · 8 May 2016 at 22:39

No the point people are saying without the large over clock 1080 is only just ahead of the 980ti and people should wait until reviews to see how much better a 1080 is before saying a 980ti going to be £200.

Brazo · 8 May 2016 at 22:46

I think it looks a nice upgrade point and can't wait for pricing!

Andromeda1983 · 8 May 2016 at 22:47

Moving from a 970 to a 1080 at 2560x1440 seems like a pretty solid upgrade if you ask me

D1craig · 9 May 2016 at 00:07

waiting for proper benchmarks is the way to go.

why cant we be under the assumption that nvidia has done this because they are scared of what they have seen from AMD? from every peice of evidence nvidia was a good month or so behind, this seems like it is going to be like the fury release and very slowly coming into stock.

i would wait for reports and reviews from actual gamers and not ones that wear their nvidia caps and tshirts while making reviews. even these will not want to tell you the gpu is rubbish because they want you to think they have not wasted their money.

D.P. · 9 May 2016 at 00:56

D1craig said:
waiting for proper benchmarks is the way to go.

why cant we be under the assumption that nvidia has done this because they are scared of what they have seen from AMD? from every peice of evidence nvidia was a good month or so behind, this seems like it is going to be like the fury release and very slowly coming into stock.

i would wait for reports and reviews from actual gamers and not ones that wear their nvidia caps and tshirts while making reviews. even these will not want to tell you the gpu is rubbish because they want you to think they have not wasted their money.

Thete were absolutely no reliable reports that Nvidia would be behind AMD, only red-eyed AMD enthusiasts regurgitating baseless rumours from the least accurate rumour site.

The same people.were saying that since nvidia didn't demo anything at CES and AMD did that AMD must be behind. While every rational person was syai Gb that a lack of CES demo is meaningless.

They were saying that since TSMC were a few months later to 16nm that NVida would be a frw.months behind. Despite the fact that the process.was ready late 2015 and sending samples back to Nvidia.

People were say ok bg GDDR5X would be here until September when Micron claimed shipping in summer, and then said ahead of schedule, and the said they will have a significant uptick in Q2 financials due to Shipping said memory.

There was never any reliable information about a delay in the slightest. April was the often touted release window, 1 week in to May and we get the release.

We have to see what stocks like but insider ing Nvidia can already offer a hand pi ked Founders edition alongside the regular from day 1 I expect their yields are plenty good.enough.

At the end of the day you have to consider the possibility that Nvidia had a plan that they have executed without delay and now is the time to release. If AMD tours are reliable them AMD simply don't have a product to compete in the same market, however great Polaris is. They have products for different market segments. Moreover, even if Polaris was some killer GPU that they can sell dirt cheap Nvidia doesn't gain anything by paper launching a product with no stock for months. They don't magically get sales of non existent products by being first. If nvidia were really threatened they would delay a launch and try a response with higher clocks and as many small tweaks as they can muster, get yields really high so they can drop prices so their offering is competitive. Conversely, if you beleive you have the best product amd it's ready you don't wait for the competition!

Nvidia biggest concern now is likely getting GP100 out ASAP to stop AMD getting the low-end slaes, laptop deals etc.

jakspyder · 9 May 2016 at 01:02

Fairly promising early results, for sure though, its the full fat card i'll be waiting on, mainly for the 4k clout im after from a single GPU.

shamus21 · 9 May 2016 at 06:08

jakspyder said:
Fairly promising early results, for sure though, its the full fat card i'll be waiting on, mainly for the 4k clout im after from a single GPU.

Yes that’s what I am looking forward to. No more multi card set up `s Sil has become poorly supported over the last 18 months. Looking at numbers for the 1080 if valid, 1 full fat pascal should do 4K.

JediFragger · 9 May 2016 at 06:34

17th for reviews apparantly, hopefuilly so!

$h@d0w · 9 May 2016 at 07:22

Ashes 4k (and others in the dropdown):

http://www.ashesofthesingularity.co...-details/ef668b20-4060-4caf-bb44-7f3ff6fc1bc5

Edit - not sure how legit they are as they seem to have been done before the announce.

AlamoX · 9 May 2016 at 07:31

$h@d0w said:
Ashes 4k (and others in the dropdown):

http://www.ashesofthesingularity.co...-details/ef668b20-4060-4caf-bb44-7f3ff6fc1bc5

it scored lower than 200 series/fury/980ti single gpu ?
another user at 1080p here

fs123 · 9 May 2016 at 07:58

AlamoX said:
it scored lower than 200 series/fury/980ti single gpu ?
another user at 1080p here

I assume the benchmark is legit unless you can fake your gpu id somehow.
Not quite 'twice' the performance of a TitanX/980 whatever. On the leaderboard the 1080 is not even in the top 10. I think the actual reviews will drag many into reality very soon. It may be faster in DX11 but this DX12 result indicates that Pascal in essentially just Maxwell at a higher clockspeed. With less CUDA cores than a TitanX the AoTS results are to be expected, especially if Async Compute is still software scheduled.

There's one 1080 at number 23 in the 1080P Extreme test :

http://www.ashesofthesingularity.com/metaverse#/ladders/benchmark/overall/High_1080p?viewType=myself

andybird123 · 9 May 2016 at 08:04

They could be either faked, or early driver tests at low clocks. Running slower than a FuryX i would be concerned. Running lower than a 980 and i call shenanigans.

NirK · 9 May 2016 at 08:34

Does not look right as a 980ti is massively faster than the 1080 in those results.

fs123 · 9 May 2016 at 08:59

NirK said:
Does not look right as a 980ti is massively faster than the 1080 in those results.

Those are legit results.

http://www.ashesofthesingularity.co...-details/473c2154-f520-4c70-9de0-2a5ded72442b

The username search for PellyNV on google leads to this twitter account. Sean Pelletier - Senior Product Manager for GeForce Game Ready Drivers @ NVIDIA

https://twitter.com/pellynv?lang=en-gb

It's pretty safe to say that the benchmark result is real and uploaded by a senior Nvidia employee. The lower than 980ti result is explained by the lower number of CUDA cores. The 980ti/TitanX higher up in the leaderboard will be overclocked of course and the 1080 may be lower clocked too.

Gregster · 9 May 2016 at 09:13

There is a Pelly_NV on Steam and AVS forums that don't strike me as an NVidia employee.