Hello everyone,
Since people are obtaining the highly wanted GTX 980M, a new fresh thread for the performance doesn't seem so bad. So I made one(instead of spamming the official release page).
Lets obtain here all kind of benchmark results/ frames on new games.
I will get started with the request for the OC GTX 980M(MSI 8GB/GT72 dominator pro).
+135Mhz on core
+50 Mhz on memory(I didnt feel like messing with the memory yet).
NVIDIA GeForce GTX 980M video card benchmark result - Intel Core i7-4710HQ,Micro-Star International Co., Ltd. MS-1781
10340 Graphic score!!!
GPU temperatures maxed out at 73. Most of the time it barely hit the 69.
I can test FPS with OC later today on request, also other benchmarks. Writing this from the train atm.
-
Amazing score for a single mobile GPU!
[drooling icon] -
Looking good
Can you also do a 3dmark 11 benchmark.
Thanks :thumbsup: -
Sure, I can this evening
, working some things out during my 2,5 houre train trip hahaha.
edit: also gonna check for memory OC but dont think it can go high, and will barely increase frames.TBoneSan likes this. -
Found this 3dmark with oc 1173 on core and 1553 on memory NVIDIA GeForce GTX 980M video card benchmark result - Intel Core i7-4710HQ,Micro-Star International Co., Ltd. MS-1781 , im actually curious to see how high the memory can clock since its the biggest downclock compared to desktop gpus.
Cloudfire, Cakefish and massixline like this. -
1553MHz on the VRAM is extremely good. It means we can easily get memory bandwidth you find on desktop GPUs.
I propose you rename the thread to "GTX 980M Benchmark thread" BTW. I thought it was just another Maxwell discussion threadCakefish likes this. -
Robbo99999 Notebook Prophet
-
Alsof managers yo extract vbioskeep you guys up to date. Stil traveling so respondimg from phone, apologize for this poorly writtrn reply.
Will post all these scenarios output tonight.Robbo99999 likes this. -
Meaker@Sager Company Representative
Coming soon: Fire Strike Ultra
Looks like 3dmark have a new benchmark setting to make everyone cry at the screen.deadsmiley, Tonrac, jaug1337 and 1 other person like this. -
-
-
Short update, memory overclocks are unstable, I got no clue how that other guy got the memory this high and got his score improved(firestrike).
NVIDIA GeForce GTX 980M video card benchmark result - Intel Core i7-4710HQ,Micro-Star International Co., Ltd. MS-1781 check it out. Memory still unstable, but got score up to 10525 which is something xD. Will be working on it later, midterms are comming up.
edit: how do I change title to GTX 980M Benchmark thread? or can any mod do this pls -
Robbo99999 Notebook Prophet
-
WOW this card is just insane!
Would love to see the performance between the 4GB VS 8GB GTX 980M.
Is there any advantage to having 8GB's of VRAM at 1080p, I know more is useful when you increase screen res or use more than 1 monitor but I would love to see some hard numbers. -
edit: I foudn my score on higher clocks http://www.3dmark.com/3dm/4349303
It is lower then my 1414mhz overclock -
considerign these badly optimized games and engines.
-
Robbo99999 Notebook Prophet
-
Robbo99999 Notebook Prophet
-
And as HTWingNut said earlier, you will benefit a little more from that extra VRAM with high frequency RAM. So, having a ton of VRAM is pointless if your RAM is just going to bottleneck it. That's partially why these new X97/X99 boards support up to 2800MHz memory because GPU's are starting to have more and more VRAM. Having anymore than 4GB's at this point is unnecessary, but I would get something with at least 4GB's (or more) VRAM if you plan to keep it longer.
P.S. I'm pretty sure HT said something along these lines. Might have interpreted it wrong, but yeah... Don't quote me on this. -
Robbo99999 Notebook Prophet
(The only context of system RAM speed having an effect on gaming performance where GPU VRAM amount is concerned is if you don't have enough VRAM, at which point textures and information has to be swapped from system RAM to VRAM - this is where fast system RAM might help in this relationship, but probably wouldn't help a great deal. Just important to have enough VRAM really.)octiceps likes this. -
Yep, GDDR5 is up to 8 GHz now. Typical DDR3 and DDR4 speeds are, what, a quarter of that? LOL
-
So, again, don't quote me on any of that. -
-
Fast GPUs at fast trains
-
on-topic. this GPU is sick!!! my god -
Tee Hee...
NVIDIA GeForce GTX 980M video card benchmark result - Intel Core i7-4810MQ Processor,Notebook P377SM-A
3DMark11 23621 P graphics score: http://www.3dmark.com/3dm11/8869856
loldeadsmiley, maxheap and massixline like this. -
When I saw 20k GPU score on Fire Strike, I started coughing, then I realized SLI, I kept on coughing
-
It's kind of odd watching the combined test run fluidly.
Cakefish likes this. -
Is this a sager?? Do you have any reviews upcoming??
EDIT: Just saw hardware info, when will you be reviewing it?? -
-
NVIDIA GeForce GTX 970 video card benchmark result - Intel Core i7-4930K,ASUSTeK COMPUTER INC. X79-DELUXE
And here's one with +120/140 core and +400 vram; actual boost was 1506MHz and mem was 7800MHz: http://www.3dmark.com/fs/3056349
I used an older driver (344.11) while HT used the latest one (344.48). Also HT is running Win8 while I'm running Win7, so slight advantage to HT. Interestingly HT's graphics score is actually higher than mine LOLdeadsmiley and massixline like this. -
Hmmm scaling seems sub-optimal. Driver issues or CPU bottleneck? Your Gigabyte 970's at stock (1329/7000 amirite?) should do quite a bit better than 980M SLI, overclocked or otherwise.
-
You have an amazing desktop replacement and an amazing desktop?!
-
-
I'm 99.9% certain a 4930K @ 4.5GHz is not going to bottleneck 970 SLI lol.
Ningyo likes this. -
Also keep in mind that Fire Strike favors Windows 8 greatly.
P.S. is 4930K a bad overclocker or you're just unlucky? Both you and vulcan have 4.5 GHz @ ~1.4V which is very middling. That's like Haswell-E numbers LOL. -
That's what I've been trying to find out lately, just how much does Firestrike favor Win8, and whether it only manifests in graphics score, or physics score, or both. Maybe we should ask Mr. Fox, but IIRC it was not more than 10% overall score difference.
As for the cards themselves, these were actually replacements I got after I RMA'd the old ones because of ridiculous coil whine. (no that's my rig nor is that my video, but my cards sounded exactly like that, now imagine having 2 of those in your rig:cry
These ones still whine (UGH), but they're much better, and I actually can't hear them while gaming unless it's a dead silent quiet scene. And at least I found somewhere between 80-90 frames they stop whining, and below 80 FPS they're completely silent. So a workaround for now is to cap the FPS at 80, or push settings so high average frames drop below 80.
Mr. Fox likes this. -
Yeah, 10% sounds about right, about the same difference between W7 and W8 in BF4 IIRC.
Mr. Fox likes this. -
Dude your better than average desktop beats 99% of the scores submitted to FireStrike
Still kudos for amazing systems! You should change your name to n=4 (number of gpus!)
-
@octiceps: My 4930K is a pile of bovine excrement, probably in the bottom 10%. Needs about 1.37-1.4V fore 4.5GHz (using offset voltage for overclocking), and IMC appears to have degraded to the point it now flat out refuses to to run 2133 ram. Well it posts and I can boot into Windows and do light tasks just fine, but try gaming and the system locks up and crashes in 10 minutes. Thought about using my overclocking warranty (PTPP) and sending to Intel for replacement, but decided I'm going to let it run as it is now until end of year and see how it holds up until then. Bottom line is I'm going to keep adding volts to keep a 4.5GHz OC, even if it means going up to 1.5V. If it dies it dies period.
On a brighter note, these dom plats can manage 1866 @ 8-8-9-20 with tREF maxed out (32.767 us), so at least the ram is solid.
@maxheap: Yeah but only because I'm using the latest technology. Just wait until the 980/970 shortage dies down and you'll see it slowly crawl back. Also I refuse to call my desktop "amazing" because of the above reasons. -
Ok so my Firestrike score is definitely not what it should be due to borked drivers. I present you this: Result
One result belongs to cards that had comparable ASICs (66% vs 64%), care to guess which is which?
So it turns out this has to do with the driver not being to compensate for overly large differences in ASIC quality leading to different VDDCs. One of my cards has an ASIC of 74.1%, and only needs 1.156V to run a stable boost of 1366MHz under load. The other card has an ASIC of 66.1%, and needs 1.206V to do 1366MHz boost. The problem is the driver either can't, or doesn't know how to compensate for this VDDC difference, so scaling gets completely messed up. In fact, to get the cards to match boost and volts, I actually have to make it so the lower VDDC card runs 20MHz behind the higher VDDC card when OCing. More details in this thread so I won't go into it, but I will add my own results from testing.
With SLI disabled and +120 core and +300 mem on the lower VDDC card, boost increased to 1524MHz, while voltage also jumped to 1.225V when running Firestrike. Now keep in mind this lower VDDC card is 20MHz behind the higher VID card. At +140 core/+300 mem, boost further increased to 1545MHz and voltage stayed put at 1.225V.
So what this means is that if everything were to have worked properly, SLI should give a boost of 1524MHz in Firestirke. But it doesn't, and instead with SLI on boost only goes up to 1506MHz, and the reason for that is because the lower VDDC card is downvolting to 1.200V from 1.225V, so it simply does not have enough juice to eek out another boost bin or two. Even with the exact same OC, the cards that had comparable ASICs and voltages worked much better. Clearly the "mismatched set" is losing about 40MHz of core clock in Firestrike (1506 vs 1545) even though on paper everything looks to be the same, and this would certainly explain the discrepancy in score.
On the bright side, +120/+140 core and +300 mem (=1506 boost 7600 mem) is completely game stable without even touching volts, and cards stay at 1.200V/1.206V. So basically stock 980 level performance without needing to add any juice for this new set of Gigabyte 970s. I guess I really shouldn't complain.
Sadly 344.48 does NOT seem to fix this, so still need to wait for more mature drivers. nVidia, get your **** together and unbork these drivers plz kthx.moviemarketing and Ningyo like this. -
-
I am still coughing
20k FireStrike on a laptop... tech is evolving fast indeed
-
-
Anyhow, for some reason Fire Strike all but ignores the physics results when calculating the overall score. You can have a really horrible CPU and low physics score and still get what looks like an impressive result with Fire Strike if you have great GPU(s). I guess Fire Strike was only intended to be a GPU benchmark and the physics result is just there to look at all by itself. 3DMark 11 better reflects overall system performance because a low physics score will put the hurt on the overall score in a big way. For DX10, 3DMark Vantage is the best test of overall system beast-factor.
I would like to see more 3DMark 11 scores from some of these new Ultrabooks that have 980M paired up with crippled low-TDP HQ i7 CPUs. So far I can only find one example of a good 3DMark 11 score for 980M SLI, and while the Graphics score is nice, the Physics score is not very good. That's with 4800MQ, so the HQ results will probably be even worse. It will be interesting to see what my 3DMark 11 score looks like if or when I upgrade my M18xR2 to 980M SLI.
Comparison of 3DMark 11 Results -
Had a go with DSR and it is pretty awesome. With a 1506 core 7600 mem OC on my desktop Gigabyte 970s, at a DSR of 2.25x (effectively 2880x1640 downsampled to 1080p), I could consistently get 65-85 FPS in Watch Dogs on Ultra with SMAA. (this is vanilla Watch Dogs with no mods or optimizations outside of Ubisoft's own patches)
Gameplay felt smooth with minor stuttering only during extremely action packed scenes (think 3 cars exploding at the same time with lots of people running and multiple enemies shooting at me). vRAM usage peaked at 3.55GB as well, so these desktop 970 cards appear to be powerful enough to push 60 frames @ 3K with a moderate OC.
Going by HT's results, 980M SLI OC could probably match desktop Gigabyte 970 G1 at stock. As expected these cards are overkill for 60 FPS at 1080p, and requires you to really crank up the AA (or play a poorly optimized title) if you don't want a CPU bottleneck.
I'd say 980M SLI OC is likely enough to pump a steady 60 FPS at 1440p for most titles, and for less demanding DX10 games you could probably even run them at 3K.
EDIT: Played Far Cry 3 for 40 minutes at 2.25x DSR on max settings with 2x MSAA. 75-110 FPS with an average of about 90 FPS. Experimenting with AA showed that each 2x increase in AA resulted in a loss of 15-20 average FPS. 4x MSAA was still smooth, but 8x MSAA definitely induced stuttering. Still pretty impressive considering this is basically 3K downsampled to 1080p. -
Old news but still... never again W8, never again. -
Copied from the main thread:
Shadow of Mordor
imgur: the simple image sharer
Max Temp of 75 degrees which is a good 10 degrees less than my 880m. Decent. Avergae of 55FPS in the benchmark with everything maxed. Ignore the minimum FPS reading - that's where the FPS starts off at the initial load of the level, the actual bench dropped to 42 during a bunch of explosions.
Max Payne 3
imgur: the simple image sharer
Never went above 61 degrees, remained at a constant 60 at those settings.
Ryse
Only the forest level has weird stuttering. Every other level sticks between 50-60FPS. This is at 1080 with all settings maxed. I could barely maintain a solid 30 with the 880m. That's a huge leap.
Crysis 3
imgur: the simple image sharer
See above settings for a rock solid 60FPS at 1080. If I turn shading/post processing or object to very high, things drop to the low 50s/high 40s. AA is at SMAA x2.
Crysis
http://i.imgur.com/of3LgvQ.jpg?1
Everything at Very High and 8x AA at 1920x1080. Average of 60FPS. Drops down to 50 once you're in the more taxing areas and stays in the low 70s when things are less busy. That 8x AA is just ludicrous though and super taxing, not practical and no difference over 4x so not worth it outside of benchmarking! And yes, my afterburner is now showing stock clocks as 540Mhz which is wrong. -
Does it means every piece have a different performance? Nearly 9 percent is a bit too much imo.
I am asking because i had to send my lappy for RMA and they gave my money back so i need to buy a new one.
GTX 980M Benchmarks thread
Discussion in 'Gaming (Software and Graphics Cards)' started by naldor, Oct 12, 2014.