Support NeoGAF

[Cute Dragon Here]

OK, got my results with my i5-4590 and GTX 970.

At stock, I get 3340 points.

Overclocked to the edge of stability (the point just before errors show up) and locking the CPU to 3.7 GHz via multi-core enhancement, I get 3745 points. The GPU ran as high as 1493 MHz in observed clocks; the GPU boost clock was set to 1418 MHz, and the memory clock was set to 7950 MHz (offsets: +165 and +940)

Now I kind of want to run a 1080. Or two.

 

[Dictator93]

Core i7 930 @ 4.2 Ghz
Titan X @ 1429 Mhz and 4001 mhz on the VRAM

Not bad I must say... but I really need to upgrade this CPU at some point...

 

[lordfuzzybutt]

Core i7 930 @ 4.2 Ghz
Titan X @ 1429 Mhz and 4001 mhz on the VRAM

Not bad I must say... but I really need to upgrade this CPU at some point...

That CPU score is around mine with a 4670k at 4.2. Granted mine is an i5 but it's 3 gen later.

Man that CPU is a legend.

 

[Arulan]

Core i7 930 @ 4.2 Ghz
Titan X @ 1429 Mhz and 4001 mhz on the VRAM

Not bad I must say... but I really need to upgrade this CPU at some point...

Keep the 930 alive!

I just upgraded from the 920. What an amazing CPU that was. I was going to benchmark that system to see where it fell, but it's not on Windows 10.

 

[UltimateIke]

i7 3770k @ 4.5 GHz
GTX 970 @ 1517 MHz / 4001 MHz

http://www.3dmark.com/3dm/13234608?

Upgrading to a GTX 1070 next week!

 

[OBias]

GTX 1070
FX 8300
8GB RAM 1600MHz
no overclock anywhere

 

[Parsnip]

Seems correct, no overclocks. Once upon a time I had some issues with CPU scores due to a bios bug (I think) but I think that looks right.

3DMark Score
5282
Graphics Score
5893
CPU Score
3329
Graphics Test 1
38.62 fps
Graphics Test 2
33.63 fps
CPU Test
11.19 fps

http://www.3dmark.com/spy/36070

Such a great idea for a bench setting by the way.

 

[Raticus79]

60% off is decent. Finally picked this up.

Core i7 930 @ 4.2 Ghz
Titan X @ 1429 Mhz and 4001 mhz on the VRAM

Not bad I must say... but I really need to upgrade this CPU at some point...

Same here. I'm pretty much just taking the 2 x Titan X out of my current PC and making a new build around a 6950x within a month or so. I'm probably getting bottlenecked pretty badly on PCI-E with this 2600k and old motherboard. It's been a long wait...

"Before" - 6220 - http://i.imgur.com/THIgo2j.png
Edit: hang on, 1429MHz? Oops, still had these at the stock 1000, lol. Let me try that again.

A bit better.

Moral of the story is this 2600k isn't holding up as well as I thought it was. Dang.

 

[ZOONAMI]

1070 FE ocd

6344 graphics score.

http://www.3dmark.com/3dm/13274963

 

[tenchir]

Fury Nano at stock.

 

[Slick Butter]

Seems correct, no overclocks. Once upon a time I had some issues with CPU scores due to a bios bug (I think) but I think that looks right.

CPU Score
3329
CPU Test
11.19 fps

Huh. Your 4670k performed a decent bit better than my 4690k (also not OC'd). I wonder whats up with my CPU, though at least, unlike my GPU, it's frequency can go pretty high.

 

[mephixto]

Hmm I'm okay with this:

 

[LelouchZero]

Huh. Your 4670k performed a decent bit better than my 4690k (also not OC'd). I wonder whats up with my CPU, though at least, unlike my GPU, it's frequency can go pretty high.

Could be your ram speed, what clock-speed is your ram running at?

 

[DonMigs85]

OK, got my results with my i5-4590 and GTX 970.

At stock, I get 3340 points.

Overclocked to the edge of stability (the point just before errors show up) and locking the CPU to 3.7 GHz via multi-core enhancement, I get 3745 points. The GPU ran as high as 1493 MHz in observed clocks; the GPU boost clock was set to 1418 MHz, and the memory clock was set to 7950 MHz (offsets: +165 and +940)

Now I kind of want to run a 1080. Or two.

Try to hold on, remember that Volta and HBM2 will bring significant gainz.

 

[oxidax]

5508 / 6033 Graphics Score
980ti
i5-4690K

Im ok with that.

Shout out to 3D Mark for charging $5 to unlock "extra features" that should be given free specially to those of us who purchased the full edition in the first places.

 

[SmartWaffles]

Scored 6438

i7 4790K @ 4.4GHz
Zotac GTX 1080 @ 2080 - 2100MHz

http://www.3dmark.com/spy/52735

The CPU test is really dragging my score down.

 

[EatChildren]

i7 6700K @ 4.6GHz
16GB DDR4
MSI 980 Ti @ 1316MH(Core) 3506MHz(Mem)

Total Score: 5,489

Graphics score: 5,425
Graphics test 1: 35.75 FPS
Graphics test 2: 30.80 FPS

CPU test: 5,888
CPU test: 19.78 FPS

http://www.3dmark.com/3dm/13285071?

Your score was invalid for the following reasons
- Time measurement inconsistencies detected during benchmark run. This result is not eligible for Hall of Fame or overclocking contests.

?

 

[Cute Dragon Here]

?

Sometimes happens with Windows 8 or later systems. If you see that, run the benchmark again until it doesn't appear.

There also is a bug with SystemInfo in Windows 10 where sometimes it fails and you end up running the benchmark only to end up with no hardware data (and rendering the benchmark invalid).

 

[legend101]

i7 4700K @ 4.4GHz
32GB DDR3 @ 1600Mhz
Asus Gtx Strix 1080 @ 2060MHz(Core) 10400MHz(Mem)

Total Score: 7011

Graphics score: 7717
CPU test: 4619

Still some room for improvement.

 

[longdi]

Not approved drivers? /shrugs

 

[lordfuzzybutt]

Scored 6438

i7 4790K @ 4.4GHz
Zotac GTX 1080 @ 2080 - 2100MHz

http://www.3dmark.com/spy/52735

The CPU test is really dragging my score down.

Your CPU score seems suspiciously low.

 

[Slick Butter]

Could be your ram speed, what clock-speed is your ram running at?

1600 MHz, I see theirs is faster. Would it affect it that much? My 4690k got a bot 200pts less on both of my benches than their 4670k.

 

[SmartWaffles]

Your CPU score seems suspiciously low.

Yep, might even be the 1600 RAM? My CPU is actually overclocked to 4.4GHz, dunno why the score is lower than other similar CPUs.

 

[jabuseika]

Dang my system feels kind of old now.

 

[RickyRozay]

MSI 980 TI @ 1470MHz on the core clock and 200MHz on the memory. Definitely could overclock it more but I'm leaving the voltage at stock for now.

2600k oced to 4.4.

 

[Cute Dragon Here]

In the meantime, you guys can all laugh at my laptop's GTX 760M's non-performance in the benchmark. 720 points, and I even managed to get the sub-5 FPS achievement.

Tip for people with hybrid graphics in any form: force the dGPU to be used in your driver's control panel, or you'll always crash or run poorly since there are no profiles yet for GPU switching in Time Spy.

 

[lordfuzzybutt]

Yep, might even be the 1600 RAM? My CPU is actually overclocked to 4.4GHz, dunno why the score is lower than other similar CPUs.

My RAM is 1600 too. CPU score is 3300. I suspect something else is off.

http://www.3dmark.com/spy/42981

4.4 Ghz 4790k should score at least over 4500 as far as I have seen.

 

[Woffls]

3DMark Score - 1563

5820K @ 4.3GHz auto voltage - CPU Score: 6786

GTX 750 Ti SC - Graphics Score: 1377

Don't worry; GTX 1070 arrives Monday.

 

[Parsnip]

Huh. Your 4670k performed a decent bit better than my 4690k (also not OC'd). I wonder whats up with my CPU, though at least, unlike my GPU, it's frequency can go pretty high.

The latest (and last I guess) bios for my gigabyte mobo has this issue that if I have everything in "auto", my uncore/NB frequency (as reported by cpuz in Windows) would be significantly lower than it should be. I don't remember the exact number it was down to, nor do I know what uncore affects exactly, but it would significantly affect my cpu performance.

If I manually set either ram or cpu to the exact same value as the automatic value was, everything's fine and my NB frequency is as it should be and cpu performance is inline with others.

No idea if you have the same issue of course, but it's something to check.

 

[Phreakuency]

Seems to be right for the specs I think.

 

[MuchoMalo]

I really wish I could push this card and CPU just a little bit farther.

That means your overclock isn't stable.

No it doesn't. It just means that there was some kind of slight glitch. Unstable overclocks cause crashes or artifacting.

 

[plagiarize]

Though, I have to say, as great as those raw numbers looked, frametimes were *all* over the place and obvious microstuttering was happening through every graphics test making the perceived smoothness way below the framerates I was reportedly getting.

I'm sure a graphics driver update will fix that, but these raw numbers are mondo misleading.

Still, I'm very happy with this PC I overbuilt in late 2014 and just got to leverage allllll of this power a couple of days back in the Nvidia VR FunHouse demo. The plan was to build something VR ready back in 2014. I'm happy to say I hit that target (but not by as much as you might think).

Oh, and this is at stock clock on my CPU I think. I like to keep it at stock during the summer. I may re run this later over clocked to see if I can push this over 10,000.

 

[SmartWaffles]

My RAM is 1600 too. CPU score is 3300. I suspect something else is off.

http://www.3dmark.com/spy/42981

4.4 Ghz 4790k should score at least over 4500 as far as I have seen.

Redid the test with several background apps closed and it's a lot more like a normal score:
http://www.3dmark.com/spy/57177

 

[cripterion]

i7 4790K manually OC @4.4
Palit GTX 1080 GR Premium stock

Dunno about what inconsistencies they talking about, I just had upgraded the drivers so maybe that's why? Will try another time with different OC settings.

Just redid the test with slight OC to the card (+96Mhz GPU offset +38hz Memory) , no background apps besides Steam.

 

[Exentryk]

Seems fine. Don't really see the need to upgrade as I'm not into VR (yet), and no big PC RPG on the horizon.

 

[SuperHans]

So glad I chose an R9 380 over the 960.

 

[demented waffle]

Total score: 4133
Gpu: 4033

R9290x

http://www.3dmark.com/3dm/13307006?

 

[Gattsu25]

No overclocks

 

[irishcow]

Total Score: 5463, Graphics Score: 6315, CPU Score: 3097

i5 2500k @ 4.5 Ghz
Gigabyte G1 Gaming 1070 Core @ 2000 MHZ, Memory @ 9216 MHZ

http://www.3dmark.com/3dm/13308800?

 

[Traxtech]

http://www.3dmark.com/3dm/13309703?

The CPU test eats my CPU haha.

Beat it slightly second time around

http://www.3dmark.com/3dm/13310505?

 

[Justinh]

Man, I love these demos. Reminds me of the cool stuff they use to make to show off new graphics cards. Remember those? With like... the fairies and stuff?

Whatever. Those were cool...

I was gonna buy a new CPU (6800K) next month, and get the mobo and RAM in September, but I really want a new(er) Blackberry so I think I'm gonna put it off a little further. I miss having one with a keyboard.
Oh yeah: 2500K is at 4.6Ghz.

 

[ZOONAMI]

Man, seems like there a huge variance in 1080 scores, some people reporting super minor ocs and getting nearly an extra 1500+ points? What's up with that? Some stock 1080s at 6500 or so. Seems like a ton of variance even with oc when people are getting nearly 8000 with a 1080.

My 1070 was like 5600 gpu score stock iirc and 6,334 graphics score ocd, which is super close to some 1080 scores. So a 900 point gain with a heavy oc (+220 core and +1200mhz memory). Seems like way less swing in the scores. Guess there are a few people with 6500 scores for 1070s but not seeing many. Seeing a lot of 7800+ 1080 scores.

 

[ZOONAMI]

Hmm I'm okay with this:

That seems crazy, like 100% scaling basically right?

 

[Dictator93]

Yeah, the two 970 SLI scores posted so far had like 99% scaling evidenced I think. Right?

 

[Durante]

Man, seems like there a huge variance in 1080 scores, some people reporting super minor ocs and getting nearly an extra 1500+ points? What's up with that? Some stock 1080s at 6500 or so. Seems like a ton of variance even with oc when people are getting nearly 8000 with a 1080.

Are you looking at the graphics scores? Because CPU has a pretty huge impact on the overall score.

As for graphics scores, overclocking is really only a part of what a 1080 will end up clocking at in a benchmark like this. It's a function of power budget, cooling (and therefore also case temperatures), and of course base/OC clocks. Between all of those you can probably get around 400 MHz from the top-end of the OC deck to a default FE in a hot case, and that makes for a difference of ~20%.

 

[Stevey]

i7 3770k @4.1ghz
GTX 780

Geting a 1080 when EVGA release the classified version.

 

[ZOONAMI]

Are you looking at the graphics scores? Because CPU has a pretty huge impact on the overall score.

As for graphics scores, overclocking is really only a part of what a 1080 will end up clocking at in a benchmark like this. It's a function of power budget, cooling (and therefore also case temperatures), and of course base/OC clocks. Between all of those you can probably get around 400 MHz from the top-end of the OC deck to a default FE in a hot case, and that makes for a difference of ~20%.

Yes, the gpu scores seem to be ranging from 6500 - 8000 roughly. One example in particular was like a +70 core clock and a +38 memory OC and the gpu score was like 7,700. Stock 1080s some are around 6500. Seems crazy to me.

 

[Durante]

Yes, the gpu scores seem to be ranging from 6500 - 8000 roughly. One example in particular was like a +70 core clock and a +38 memory OC and the gpu score was like 7,700. Stock 1080s some are around 6500. Seems crazy to me.

+70 core from what though? There are 3rd party designs of the 1080 which boost up to ~2GHz reliably without any additional OC -- they are ~18% faster out of the box than a FE. Add a small additional OC, get 20% total, 6500*1.2=7800, mystery solved?

 

[ZOONAMI]

This one:

i7 4790K manually OC @4.4
Palit GTX 1080 GR Premium stock

Dunno about what inconsistencies they talking about, I just had upgraded the drivers so maybe that's why? Will try another time with different OC settings.

Just redid the test with slight OC to the card (+96Mhz GPU offset +38hz Memory) , no background apps besides Steam.

 

[ZOONAMI]

+70 core from what though? There are 3rd party designs of the 1080 which boost up to ~2GHz reliably without any additional OC -- they are ~18% faster out of the box than a FE. Add a small additional OC, get 20% total, 6500*1.2=7800, mystery solved?

Yeah makes sense. Cool, thx for explanation. I guess some of the 1080s really just aren't boosting very high with no OC. I guess about a 1000 point variance for 1070s is about 18%, so not much different.