Unigine Valley 1.0 Benchmark Thread

[RussK1]

It's just to get a higher score. It just forces settings that aren't controlled very effectively by the application. I guess it would give you better performance technically, but you lose some quality.

Technically it's cheating. Lowering settings via control panel to inflate the scores while everyone else is running quality modes. Tantamount to disabling tesselation via CCC and getting huge scores in VAlley and Heaven yet the screenshots show Extreme HD...

 

[nikoli707]

most the the benchmark threads online encourage users to optimize 3dmark/unigine/etc. via their control panel. its the same as forcing AA or AF in catalyst/control panel for a certain game because many times amd/nvidia algorithms are more efficient or effective than the ones built in games. radeon pro or nvidia inspector have settings the are considered cheating though. also lod and mipmap tweaks or wireframe or speed hacks are cheating.

 

[smeezekitty]

Well if you do tweak the settings make sure that you state that

 

[RussK1]

most the the benchmark threads online encourage users to optimize 3dmark/unigine/etc.

Really? Which one's? Because any bench thread I've participated in encourages honesty and default settings in both CCC and nVidia control panels. Manipulating the settings to bypass a benchmarks settings is deceptive and counter productive. Why even run the bench to begin with other than to say "hey my score is higher than yours"? And then posting a screen shot as if you ran it's settings?

Try disabling tesselation in CCC and running 3Dmark. It won't even submit the score because it knows tesselation has been manipulated at the driver level. It's cheating...

 

[Mirakledba]

I did state so. Its ok, Ill remove those settings right now.

 

[king smp]

I have to say for a fair benchmarking comparison the driver settings should be the defaults set by Nvidia or AMD.
I didnt realize on one benchmarking thread that I had really tweaked the settings on my 660Ti and I was getting some killer scores thinking I had a golden 660Ti handed down by the Gods of Olympus LOL
then I set my driver to default and was cast back into the Mortal Realm

 

[jimthenagual]

An example of a thread that encourages tweaks and how they manage it
http://www.overclock.net/t/1360884/official-top-30-unigine-valley-benchmark-1-0

Not saying we need to do this, just that it can be done, and done well.

 

[nikoli707]

i agree with jimthenagual. ive used that thread for years. but to make anyone happy, i will run my valley extreme hd on stock nvidia settings as a comparison to 3349. so you have to trust me that im using 1.212v core and unigine should confirm im using 1350/1650 clocks. by legal standards tho, any setting used in "normal" amd ccc/nvida cp are legal, specifying one monitor, forcing no vsync, prefering performance over quality.... those are "legal" world record settings, so i doubt my clocks will change more than 1% or so.

 

[RobCrezz]

Anyone else noticed how valley is dependent so much on vRam speed/bandwidth? Or is it more how crippled my card is with the 192bit bus? Lucky my vram overclocks so much (thanks hynix!)

I have had to dial back my vram speed to about 1820 (7280mhz from 5400mhz stock 😀), but am now managing gpu speeds of 1250mhz but still barely improving my score.

 

[magicjam]

I hope I am not reviving anything.
magicjam - FX8320 @ 3.8Ghz - 7870 Myst Tahiti LE @ 1230/1500 - 38.4 - 1605

No driver tweaks, no optimizations! Everything is default. Only gpu core clock was moved from 975 stock to 1230mhz, stock voltage.
GPU temp @ 46C , VRM temp 64C during benchmark.

 

[Renzkuken1]

My gtx680 is scoring 2124 on the valley extreme hd test. Is that good for a single card or should I be getting more? I will post the screen shot later.

 

[nikoli707]

without tweaks 2124 is roughly where a 680 should be depending on the core/mem clocks

 

[Renzkuken1]

without tweaks 2124 is roughly where a 680 should be depending on the core/mem clocks

Really? :/ that's an msi lightning gtx680. I have it overvolted 100mv core is over clocked +80mhz and memory overclocked +500 I would love to see a 680 beat that score. Just out of pure curiosity to your statement.

 

[nikoli707]

without tweaks 2124 is roughly where a 680 should be depending on the core/mem clocks

Really? :/ that's an msi lightning gtx680. I have it overvolted 100mv core is over clocked +80mhz and memory overclocked +500 I would love to see a 680 beat that score. Just out of pure curiosity to your statement.

well you can definitely get a bit higher on your core clock, i would imagine you can get at least to 1250mhz core with the max stock voltage. but many of the 770/680s start getting into the 2300+ range and i imagine its with a custom bios and a few driver tweaks.

 

[oOSlushyOo]

I believe I have resolved my issues. After looking everywhere but the obvious...drivers....dur. I will get a score posted soon!

 

[oOSlushyOo]

Both of those screenshots are from the same system, the only difference between these two is two months time. No hardware changes. Trying to figure out why my performance has dropped so dramatically. I have confirmed SLI is enabled. Both runs are using the preset Extreme HD setting. My gpu temps top at 80c, cpu at 60c. For both runs the graphics cards are at stock speeds.

Anyone have any guesses as to what's going on with my system? I'm at a loss.

I truly believe it may be driver related. That is what plagued my 780 SLI from the get go. Safe Mode it and remove the Nvidia drivers via the Device Manager and do a disk cleanup, then restart (all in safe mode...at least it worked for me). Re install the drivers under normal windows and restart once more. You can either manually DL the latest drivers or simply let Ge Force Experience re install the necessary drivers. By doing this in my current SLI setup, I gained an immediate 2k points Xscore on 3DMark.

 

[oOSlushyOo]

Here you go:
i7 4770K @ 4.3, EVGA 780 Reference SLI @ +250 on Core and + 375 on RAM. Max Temp of 71C w/ fans upped to support the OC and Power Point set at 103% at 87C.

 

[nikoli707]

there is no way you were running 1437mhz core, that would require a very high vcore and much enhanced cooling. but still a good score none the less. use gpuz and/or msi afterburner to log and confirm your actual core speeds. without a custom bios, valley and some others will report wrong clocks.

kingpin 780ti's under water can break 4000 at around 1.5+ghz fyi

 

[oOSlushyOo]

there is no way you were running 1437mhz core, that would require a very high vcore and much enhanced cooling. but still a good score none the less. use gpuz and/or msi afterburner to log and confirm your actual core speeds. without a custom bios, valley and others will report wrong clocks.

kingpin 780ti's under water can break 4000 at around 1.5+ghz.

"based on the Valley Bench". I was simply reading what the benchmark was showing. I did not record the values via another program like GPU-Z, sorry bout that. I take it will a grain of salt. I will edit that part out so that it is not misleading.

 

[RussK1]

No driver tweaks. Just 500MHz more on OC of CPU and different board -----> Asus Maximus V Extreme... much better minimum FPS.



Put another machine together today and it's rather impressive for what it is.

i7-3770K 4GHz, ASRock Z77 Extreme 9, 8GB Patriot Sector 5, XFX DD 7870's CF 1150/1475MHz.

 

[stickg1]

NiceRuss! Those are impressive!

 

[king smp]

Darn Russ you make me want to rob a bank so I can do upgrades LOL

 

[RussK1]

What's weird is there is nearly a 15fps difference between the Extreme 9 and V Extreme. Could it have been the CPU overclock or just the boards? The minimum also went up...

 

[Mirakledba]

king smp - Im with you man.

 

[Mouldread]