• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Nvidia announces the GTX 980 Ti | $650 £550 €605

Tnecniv

Neo Member
Beat me by 67 points, and I had my 2600K @ 4.7 on air :p
http://www.3dmark.com/compare/fs/5235096/fs/5560039
Nice, running neck and neck. My processor is overclocked on air too with a Corsair push-pull (forget the exact model number). But even though I have been running it stable over 4 years and counting, I lost the chip lotto with the 2600K. I need high voltage, around 1.4, to keep it stable. On the bright side, I've had great luck with my gpu purchases. My last four, up to current 980Ti, have all been great chips.

I haven't seen the need to upgrade from 2600K yet, the only game on my end I've seen induce a slight cpu bottleneck is GTAV (and while I haven't reinstalled to test, I imagine Crysis 3 as well perhaps).
 

Tnecniv

Neo Member
I had to wait until today due to a minor paycheck hiccup (switching banks, direct deposit change is still pending so I had to wait for the manual check). Pleasantly surprised to see the Batman: Arkham Knight offer replaced with MGSV:TPP. I went with the ZOTAC ZT-90503-10P since it was only $10 above the $649 base price and has better cooling than the reference boards. I was a little leery because I haven't heard of ZOTAC before, but it's got solid reviews, so I'ma risk it.

Even splurged on 3-day shipping, so I'm hoping to have it by Friday. :)
I was in the same situation when I purchased a Zotac 980 (amp omega w/far cry 4) last October.

It was a great first impression. It overclocked well, core boost at 1475mhz and memory at 1953mhz stable in all games I played and tested. The gpuz asic rating for it was 78.7%

The resale value was decent as well, I sold mine to a co-worker for around $350.
 

Tovarisc

Member
I got my new system set up with this card. I have left all speeds at stock so far. I'm playing at 1440p and getting nearly 60 fps with everything but hairworks turned on Witcher 3. However, my card is running at 83C on this game at clock speeds. That seems pretty high. Should I be worried about this?

What 980Ti model you have and does it have reference cooler? If it's using reference cooler then hitting temps around 83C under load is normal, but if running e.g. ASUS STRIX then hitting so high temps isn't normal.

So is anybody planning on a Skylake build?

o/

Getting i5 6600K, 16GB DDR4 and Gigabyte GTX980Ti G1 with 500GB SSD to run it on. Some parts already in hands (even box for GPU is sexy), some in order and last will go into order within next 2 days. I hope I get to build around early next week.
 
I had to wait until today due to a minor paycheck hiccup (switching banks, direct deposit change is still pending so I had to wait for the manual check). Pleasantly surprised to see the Batman: Arkham Knight offer replaced with MGSV:TPP. I went with the ZOTAC ZT-90503-10P since it was only $10 above the $649 base price and has better cooling than the reference boards. I was a little leery because I haven't heard of ZOTAC before, but it's got solid reviews, so I'ma risk it.

Even splurged on 3-day shipping, so I'm hoping to have it by Friday. :)

Zotac makes great cards from my experience. A little bit barebones in the packaging department, but high quality cards and good customer support.
 

paskowitz

Member
So is anybody planning on a Skylake build?

Nah, I'm waiting for Cannonlake. My 4790K is comfortable at 4.8ghz and I can't imagine Skylake will perform any better on average. Heat and power consumption are not big concerns. Maybe once devs make DX12 games that can start utilizing the iGPU, but that is a long way off. I would have to replace my CPU and motherboard and if I wanted to do things properly, my ram as well. All that money could be put towards a second 980 ti Classified, which would be significantly more beneficial.

On that note, I sold my first Classy and got a second. 75% ASIC (vs 70% for my original). Pretty big difference in performance. My first card pretty much peaked at 1450/7500Mhz @1787mV and would not take any additional voltage. In addition, the voltage would fluctuate all over the place, causing the clocks to fluctuate and consequently my framerates. It would also get pretty hot (77c at that clock).

New card does 1535/8000Mhz @1200mV in Firestrike and 1500/8000Mhz @1200mV in games. At that clock heat output is the same but if I limit my framerate to 61fps heat drops to the lower 70's. The best improvement is the stable voltage. I now get 1.2 all the way up to 80c. Now it is just a waiting game for somebody to make an approved waterblock. Considering the cost savings I am probably just going to hook it up to a Swifttech Rad/res/pump combo unit instead of a custom loop.
 
I'm gonna call again because Newegg is giving the code to all 980tis. I will tell them that if I don't get the code I'll return it and buy it from newweggt
Edit :Newegg os not giving the code to gigabyte 980 ti...

EDIT: Read your edit wrong. That sucks, figured they would be applying it to all cards.
 
On that note, I sold my first Classy and got a second. 75% ASIC (vs 70% for my original). Pretty big difference in performance. My first card pretty much peaked at 1450/7500Mhz @1787mV and would not take any additional voltage. In addition, the voltage would fluctuate all over the place, causing the clocks to fluctuate and consequently my framerates. It would also get pretty hot (77c at that clock).

I don't know why that would be related to ASIC. It's lower ASIC cards that respond better to voltage changes. Unless you were not talking about the things as being related.

In any case, awesome that you got a better performing card.
 

paskowitz

Member
I don't know why that would be related to ASIC. It's lower ASIC cards that respond better to voltage changes. Unless you were not talking about the things as being related.

In any case, awesome that you got a better performing card.

I wasn't stating either. From what I have read, I tend to think there is a moderate correlation but it is not like 5% better ASIC = X% better overclocking. Certainly there will be a consistent difference between a 55 ASIC card and an 85 ASIC card but outside of huge gaps like that, I don't think it really matters (in regards to clocks or voltage). I just put it there for the record.

Just hit 17164 of Firestrike and 5138 on Ultra on air. Probably an extra couple hundred to squeeze out (17400/5300-ish). IIRC the top ranked score with a 4790K and 1x 980 ti is 18057. That was definitely on water if not something colder. The Classified cooler is pretty good for benchmarking (open air environment). While I am only at 1.200mV (according to software), temps never got above 65c with the fans at max.

However, I would hasten anyone looking to purchase a Classified if they have a case with low airflow. This thing dumps hot air like a plane taking off, especially out the side of the card. Under full OC/load there is a 10c difference with the side panel on vs off. And that is in a Define R4 (window) with a bottom intake and top back exhaust Noctua fans. The air around the heat pipes at the top of card gets super hot as well. My first card was somewhere between opening the door of a car with black leather in the sun and putting your hand over boiling water.
 

Two Words

Member
What 980Ti model you have and does it have reference cooler? If it's using reference cooler then hitting temps around 83C under load is normal, but if running e.g. ASUS STRIX then hitting so high temps isn't normal.



o/

Getting i5 6600K, 16GB DDR4 and Gigabyte GTX980Ti G1 with 500GB SSD to run it on. Some parts already in hands (even box for GPU is sexy), some in order and last will go into order within next 2 days. I hope I get to build around early next week.

Yeah, it's a reference cooler. I set up a more aggressive fan curve and now I'm hitting around 73C under load.
 
New high on Firestrike graphics score just now and I think it's about at the ceiling.

21473 - http://www.3dmark.com/fs/5616493

You can see the FS demo and combined test as the tallest peaks here. 1534mhz boost and 4212mhz memory locked like a bouss. On air! Sweet!

image1gdkj2.jpg
 

PFD

Member
Got this yesterday


68.5% ASIC sadly, but there's zero coilwhine (even at 1000+ FPS) and I managed to hit 1450Mhz core with no problems (running Witcher 3.)

Overall I'm pretty happy about it. It's roughly 2.5X the performance of my old GTX 770 2GB
 
New high on Firestrike graphics score just now and I think it's about at the ceiling.

21473 - http://www.3dmark.com/fs/5616493

You can see the FS demo and combined test as the tallest peaks here. 1534mhz boost and 4212mhz memory locked like a bouss. On air! Sweet!

image1gdkj2.jpg

Wow, very nice. One of the highest graphics scores on a single Ti that I've seen.

That CPU is holding you back though! I think you could hit 18k total score with a 4790K or 5820K OC'ed.
 
Anyone know an estimated restock date for the evga 980 ti hybrid card? How is it still out of stock?

Restock doesn't even seem to matter with this card. From the time you get the notification that it's in stock, you have like 30 seconds to buy it before it's sold out.

I don't even know why people are still bothering. Buy a reference card and install an AIO cooling solution.
 

Sevenfold

Member
Anyone know an estimated restock date for the evga 980 ti hybrid card? How is it still out of stock?

If you are talking about the UK, the last time Hybrids had a due date at Scan they got about 20% of their back orders on the day, similar story with the Classy, the Gigabyte G1 is in stock now though but I've got a 780 ti Classy so want same/hybrid and don't mind waiting. Been the same with all custom cooler 980ti's since release, but EVGA especially. It's going to be September I think before supply meets preorders. (I choose Scan because it's round the corner from home so I can take stuff back, there are cheaper places but in this case I'm pretty sure it represents the picture across the UK)
 
Weird. I tried running just a single card in Windows 10 and my score went down from Windows 7. I can't break 21,700 graphics score now.

It's a small difference, but I expected the score to go up, not down.
 
Weird. I tried running just a single card in Windows 10 and my score went down from Windows 7. I can't break 21,700 graphics score now.

It's a small difference, but I expected the score to go up, not down.

Same problem I have that I posted some pages back.

I did everything I can think of. Finally I formatted the drive and the score finally went back up.
 

AmyS

Member
Nvidia Announces Financial Results for Second Quarter Fiscal 2016:

http://www.wsj.com/articles/nvidia-revenue-tops-company-estimate-1438894780

Nvidia Corp. said revenue for the July quarter was above its projection as the graphics chip maker saw strong demand for its GeForce GTX gaming processors, reflecting new games and growth in e-sports.

The company also experienced strong growth in the customer base for Nvidia Grid graphics virtualization. Nvidia shares rose 9.4% to $22.37 in after-hours trading.

http://www.reuters.com/article/2015/08/06/us-nvidia-results-idUSKCN0QB2DH20150806
Nvidia Corp reported a surprise rise in second-quarter revenue and gave a better-than-expected revenue forecast for the current quarter, helped by strong demand for its graphic chips used in gaming and cars.

The company's shares rose nearly 10 percent in extended trading on Thursday. Nvidia's revenue increased 4.5 percent in the quarter ended July 26, while analysts on average were expecting revenue to decrease about 8 percent. The company forecast sales would fall in the third quarter, but the decline was again less than analysts' expectations.

Nvidia gets a majority of its revenue from its graphic chips made for personal computers, and there were fears that the fall in PC sales would hurt Nvidia just like it has Intel Corp and Advanced Micro Devices Inc.

But, Nvidia said gaming revenue rose 59 percent, helped by strong sales of its popular GeForce series of gaming chips.

"The near term story continues to be PC gaming," said Wedbush Securities analyst Betsy Van Hees. "That is the key driver for them and they continue to dominate in that area."

Nvidia has also been increasing its focus on making chips that allows people to play graphics-heavy games over the internet and chips used in a car's dashboard display and in self-driving cars.

Nvidia conference call webcast: http://edge.media-server.com/m/p/ogsezhku

@ 29:50 there is a comment + question for Jen-Hsun about 980 Ti having very good quarter and very competitive with HBM products from their main competition, Nvidia having good software overall. and asked what Nvidia's roadmap looks like for the next 12-18 months. Jen-Hsun said "you can't just be a good transistor-slinger, if you will". Usual talk about increasing performance by 2 or so every year, even on the same node. No mention of Pascal or 16nm FinFET but didn't really expect him to say anything at this point.

Jen-Hsun also mentioned President Obama's executive order on exascale supercomputing.
 
Wow, very nice. One of the highest graphics scores on a single Ti that I've seen.
.

Not even 2nd on this page! jim2point0 and paskowitz both got a few hundred on me.

Did get a little higher today though. Clean install of new drivers - and it might be coincidental/anecdotal - but I got the highest stable OC (in Firestrike) that I've ever had and bumped up about 70 points.

+142 core
+799 memory

This card pushed (and sustained) 8.6ghz on memory (!!!) and 1536mhz boost with max temp at 74.


8.6ghz!

This calls for a dance

bitdance.gif
 

paskowitz

Member
Not even 2nd on this page! jim2point0 and paskowitz both got a few hundred on me.

Did get a little higher today though. Clean install of new drivers - and it might be coincidental/anecdotal - but I got the highest stable OC (in Firestrike) that I've ever had and bumped up about 70 points.

+142 core
+799 memory

This card pushed (and sustained) 8.6ghz on memory (!!!) and 1536mhz boost with max temp at 74.


8.6ghz!

This calls for a dance

bitdance.gif

Great score!

Your results are actually pretty interesting. Your memory clock is much higher than mine and the core clock is only 1Mhz less in benchmark. (My boost is 1408 in GPUz). Jim2point0's is even higher (at least according to 3DMark). It looks like those clocks were consistent as well. I am running windows 10 and a different CPU, so that may be making the difference on the graphics side. You may want to try reducing your memory clock and seeing if you can crank out a little more on the core. Also, disable shadow play and any intensive background services.

GlkDOpP.png

Quick snapshot in valley for time's sake:
VfEw7T8.png
 
Great score!

Your results are actually pretty interesting. Your memory clock is much higher than mine and the core clock is only 1Mhz less in benchmark. (My boost is 1408 in GPUz). Jim2point0's is even higher (at least according to 3DMark). It looks like those clocks were consistent as well. I am running windows 10 and a different CPU, so that may be making the difference on the graphics side. You may want to try reducing your memory clock and seeing if you can crank out a little more on the core. Also, disable shadow play and any intensive background services.

I do need to play with some more variations of core vs memory but from the short tests I've done this card is just not happy when I start going above +142. I think I did like +158 core and +299 memory and I got artifacts in FS. I could be confabulating; I need to go back. I just noticed that I could juice up memory consistently so I decided to leave core at 142 and play with that instead.

The max boost between yours and mine is very interesting though. You're sitting at 40mhz over my boost but max frequency is ending up the same. Could it be because of the differences between Heaven and FS?

And actually I feel like my score is low given the OC. I still haven't been able to find a straight answer for this but does ANY component other than GPU affect graphics score? Or just total score? I'm still rocking a 2600k.
 

ZOONAMI

Junior Member
Not even 2nd on this page! jim2point0 and paskowitz both got a few hundred on me.

Did get a little higher today though. Clean install of new drivers - and it might be coincidental/anecdotal - but I got the highest stable OC (in Firestrike) that I've ever had and bumped up about 70 points.

+142 core
+799 memory

This card pushed (and sustained) 8.6ghz on memory (!!!) and 1536mhz boost with max temp at 74.



8.6ghz!

This calls for a dance

bitdance.gif

I feel like with those clocks you should be easily breaking 22000. My G1 will hit in the 2100s at 1506/8012. You might have some background processes or something going on.

That said, gpus are individuals and anything above 21,000 is pretty awesome.
 
I feel like with those clocks you should be easily breaking 22000. My G1 will hit in the 2100s at 1506/8012. You might have some background processes or something going on.

That said, gpus are individuals and anything above 21,000 is pretty awesome.

That's what I was thinking when I first started adding to the clocks. 22k seemed like it was well within reach at first but that doesn't seem to be the case now.

As you said, GPUs are individuals. This could just be the nature of my particular card, but I'd also be interested to pop it into a newer system just to see if there's any difference.
 

Fess

Member
The ASUS card seems to win over the Gigabyte card in pretty much all test reports, why is people over here and on other boards still recommending the Gigabyte card? Is it because the G1 is easier to overclock or does the ASUS card have some issues I haven't heard about yet?
 

Zaph

Member
The ASUS card seems to win over the Gigabyte card in pretty much all test reports, why is people over here and on other boards still recommending the Gigabyte card? Is it because the G1 is easier to overclock or does the ASUS card have some issues I haven't heard about yet?

Probably because the G1 was one of the first (if not the first) non-reference card to come out and easily obtained.
 

Tovarisc

Member
Probably because the G1 was one of the first (if not the first) non-reference card to come out and easily obtained.

It was first non-reference design to surface and has kept its spot at top-ish of benches where it makes appearance. In comparison ASUS non-reference design is basically hot off the presses, not many have experience with it and availability is even dodgier than with G1.
 
The ASUS card seems to win over the Gigabyte card in pretty much all test reports, why is people over here and on other boards still recommending the Gigabyte card? Is it because the G1 is easier to overclock or does the ASUS card have some issues I haven't heard about yet?

On overclockers.net there are some rumblings about some kind of issue with Strix cooler. I've seen it going through the 980Ti owners thread; not sure exactly what the problem is though.

As for card recommendations: unless you've used several versions of the 980Ti, assuming your card isn't obviously flawed like horrible coil whine, you'll probably recommend it. I recommend the MSI because I got a great one, but I saw a review on newegg where 1 guy said his had awful coil whine and when he RMA'd it the second one was even worse.

Just depends what you end up with.
 
The ASUS card seems to win over the Gigabyte card in pretty much all test reports, why is people over here and on other boards still recommending the Gigabyte card? Is it because the G1 is easier to overclock or does the ASUS card have some issues I haven't heard about yet?

Most of these cards perform very similarly. Especially when you take into account overclocking. I doubt there are many Asus cards that could perform better than either of my Gigabytes (on air cooling). At least not from the benchmarks I've seen people posting.

Right now, if I were to buy a pair of 980TIs, I'd still probably go for the Gigabytes. But if I had a build with very specific aesthetics going on, I'd buy the cards that matched the style the closest because.... like I said, they all perform similarly.
 

Fess

Member
Most of these cards perform very similarly. Especially when you take into account overclocking. I doubt there are many Asus cards that could perform better than either of my Gigabytes (on air cooling). At least not from the benchmarks I've seen people posting.

Right now, if I were to buy a pair of 980TIs, I'd still probably go for the Gigabytes. But if I had a build with very specific aesthetics going on, I'd buy the cards that matched the style the closest because.... like I said, they all perform similarly.
Yup, from the tests I've seen where overclocking has been done G1 seems to be the best. And I have a Gigabyte 780Ti right now which has had zero issues, so I've been leaning toward another Gigabyte this time too. But non-overclocked tests seems to go to the ASUS just about everywhere now, in both performance, noise and power-consumption. So I'm not sure what to do anymore. I get that many recommendations might come from when the G1 was the top card, but what would you say is the reason that you would still go for that one? And would you choose the same card if you would get a single card instead of a pair as you said?
 
Yup, from the tests I've seen where overclocking has been done G1 seems to be the best. And I have a Gigabyte 780Ti right now which has had zero issues, so I've been leaning toward another Gigabyte this time too. But non-overclocked tests seems to go to the ASUS just about everywhere now, in both performance, noise and power-consumption. So I'm not sure what to do anymore. I get that many recommendations might come from when the G1 was the top card, but what would you say is the reason that you would still go for that one? And would you choose the same card if you would get a single card instead of a pair as you said?

For just a single card, I'd probably get a Gigabyte still.

First, look at the differences in performance when overclocked.


That's a fraction of a single FPS in 2 different games.

And then look at the temperatures:


That's why. They run the coolest out of all the non-reference cards.

I think the reason the Asus performs better than the Gigabyte in non overclocked tests is because it has a higher factory overclock. But that really means nothing because any of these cards should be able to go higher than the factory overclock.

I think if I were going SLI now, I'd probably get reference cards and slap AIO coolers on them. Even the Gigabytes heat up like a mofo in SLI.
 

Fess

Member
For just a single card, I'd probably get a Gigabyte still.

First, look at the differences in performance when overclocked.



That's a fraction of a single FPS in 2 different games.

And then look at the temperatures:



That's why. They run the coolest out of all the non-reference cards.

I think the reason the Asus performs better than the Gigabyte in non overclocked tests is because it has a higher factory overclock. But that really means nothing because any of these cards should be able to go higher than the factory overclock.

I think if I were going SLI now, I'd probably get reference cards and slap AIO coolers on them. Even the Gigabytes heat up like a mofo in SLI.
Thanks, I actually hadn't noticed the temp differences, that made the G1 recommendations make way more sense.
No problem with SLI heating, the price is too high for two ones for me, I'm still on the fence for one single 980Ti :/
 
I can vouch that the g1 runs hot as hell in SLI, the top card could hit 80 degrees when oc'd

All the non-reference coolers will hit similar temps in SLI depending on airflow in your case. Reference coolers will hit the throttling temp of 83-84C overclocked when in a single card configuration, so as long as you have a case with good airflow, there is never a reason to go reference.
 
The DSR+GSync+SLI limitation has become very frustrating and it is even worse now because at one point I was able to enable all 3 and as long as I was playing at the native resolution, Gsync would be enabled but if I wanted to downsample, I could change it to a DSR resolution and all it would do is disable Gsync. Now they've blocked DSR completely if you have Gsync and SLI enabled so I don't even have the chance to pick and choose what I want. This will be less of an issue when I replace my 1080p monitor with a 1440p monitor but for now, I'm playing Alien Isolation with a machine that is getting 250-450 fps when I completely uncap the framerate at 1080p. What a waste.
 
Top Bottom