Overclock.net banner

3D Fanboy Competition 2016: nVidia vs AMD

101K views 2K replies 195 participants last post by  neoroy 
#1 ·
3D Fanboy Competition: nVidia vs AMD!!



Summary: As done previously in 2010 and 2014 we call to arms again this epic battle. In 2010 ATI dominated the competition winning by 939,215 points. In 2014 nVidia made a comeback winning by the narrow margin of 207,108 points. Who will take the title for 2016?

2011 Competition link:

2014 Competition link:

DATES:

Starts: 2/1/2016 EST

Ends: 3/31/2016 EST

Benchmarks: FUTUREMARK Link

(Click for HWbot rules)

3DMark Vantage - Performance

3DMark11 - Performance

3DMark - Fire Strike

Team Captains:

RED: Rasparthe

GREEN: strong island 1

Prizing:

1st $40 in each class

Random 1: $40

Random 2: $25

Random 3: $25

Rules:

:clock:
Pick one team Red or Green. Once you submit for a team, you may NOT switch teams.

:clock:
Use SystemInfo Version 4.40 or newer

:clock:
All hardware is allowed including engineering samples.

:clock:
All submissions must be formatted and comply with hwbot standards. For example - you need to have 2 tabs of CPU-z open on the cpu, and memory. The benchmark score needs to be visible along with one screen of gpu-z open.

:clock:
4 classes per team, x1, x2, x3 and x4 gpu categories. Enter all four if you choose.

:clock:
FutureMark validation link required

:clock:
Use competition wallpaper

:clock:
Must use CPUz verion 1.74 or newer. LINK

:clock:
Multiple setups can be used, but only one setup per GPU class

:clock:
Please refrain from editing submissions post and make a new post instead.

:clock:
Prizing will only be awarded to the winning team

:clock:
Used Competition wallpaper

AVNRev2.png 1,976k .png file


Scoring:

:clock:
3DMark scores tally together to create team total. Each individual score is added together.

:clock:
Only Graphics score counts

:clock:
Vantage score is divided by 2

:clock:
Firestrike score is multiplied by 2

Example screenshot:



SCOREBOARDS

Team Red

https://docs.google.com/spreadsheets/d/1qCz20xLGSA5_nFeWEkgCi4hBCqHpxHzhWQKvtKQE6_M/pubhtml?gid=0&output=html&widget=true

Team Green

https://docs.google.com/spreadsheets/d/1VZuyDty9IiLxw1zhRbbr6hAij8q0tpum4Yvz0dNkSBw/pubhtml?gid=0&output=html&widget=true
 

Attachments

See less See more
19
#5 ·
Quote:
Originally Posted by rt123 View Post

Interesting competition, but very poor choice in Benchmarks.

1) All of the benchs favour 8Core CPUs. Maybe try FireStrike Ultra instead of regular FireStrike.

2) All benchs favour Nvidia GPUs. Maybe have atleast 1 AMD favouring Bench like Unigine Heaven was last time.
GPUPi is a good option, which tends to favour AMD.
There was much consideration in choosing benchmarks and these points were discussed.

1. We wanted to limit resolution size so more members can participate. We also choose benchmarks that can also be ran on W8 and W10. This competition is of a different flavor focusing just as much on participation as shear gpu power. High performance systems will yield high scores for the overall tally, but a team with larger number of entires can shifts the ranks in their favor. Hense the 'fanboy' theme.
wink.gif


2. GPUPi would have been a great choice, but faster GPUs yield lower score values, which do not tally up well for the present scoring theme. Instead of conjuring up an algorithm to score this benchmark, we choose to leave thing simple and use point scoring benchmarks.
 
#6 ·
Quote:
Originally Posted by PizzaMan View Post

There was much consideration in choosing benchmarks and these points were discussed.

1. We wanted to limit resolution size so more members can participate. We also choose benchmarks that can also be ran on W8 and W10. This competition is of a different flavor focusing just as much on participation as shear gpu power. High performance systems will yield high scores for the overall tally, but a team with larger number of entires can shifts the ranks in their favor. Hense the 'fanboy' theme.
wink.gif


2. GPUPi would have been a great choice, but faster GPUs yield lower score values, which do not tally up well for the present scoring theme. Instead of conjuring up an algorithm to score this benchmark, we choose to leave thing simple and use point scoring benchmarks.
What do you think about just using GPU scores rather than overall? Also, Firestrike Ultra renders in 4K but only displays at that resolution if it can. It can be run on a 1680x1050 screen if you would like (that's how I've done it in the past).
 
#7 ·
Quote:
Originally Posted by PizzaMan View Post

There was much consideration in choosing benchmarks and these points were discussed.

1. We wanted to limit resolution size so more members can participate. We also choose benchmarks that can also be ran on W8 and W10. This competition is of a different flavor focusing just as much on participation as shear gpu power. High performance systems will yield high scores for the overall tally, but a team with larger number of entires can shifts the ranks in their favor. Hense the 'fanboy' theme.
wink.gif


2. GPUPi would have been a great choice, but faster GPUs yield lower score values, which do not tally up well for the present scoring theme. Instead of conjuring up an algorithm to score this benchmark, we choose to leave thing simple and use point scoring benchmarks.
1) Bass addressed the resolution point above. And the competition can't be focusing on "shear GPU" power when its so heavily CPU reliant. If you want to focus on GPU power, you choose benchmarks where GPU is the most important.
Agreed on the number of entries part.

2) GPUPi bench is dependent on integer performance of the GPU & the cards with the best integer performance score the most. There is no "faster GPU yield lower score values" going on. The benchmark is testing a different part of the GPU architecture unlike traditional 3D benches, so the faster cards list just looks a bit different, but its still represents a reality.
 
#8 ·
GPUPi is time based. Faster scores equate to a smaller number.

Using gpu scores from the benches is a possibility though....
 
#9 ·
Quote:
Originally Posted by rt123 View Post

2) GPUPi bench is dependent on integer performance of the GPU & the cards with the best integer performance score the most. There is no "faster GPU yield lower score values" going on. The benchmark is testing a different part of the GPU architecture unlike traditional 3D benches, so the faster cards list just looks a bit different, but its still represents a reality.
What he is saying is that a better score in GPUPI is lower than a worse one. How do you add up scores from a benchmark that gets lower the better it gets? It would require an algorithm or complex scoring mechanism which was what was trying to be avoided. It was of course, possible, but when you maybe dealing with 100s of submission for many different cards and setups it could become problematic. Trying to keep it simple for the captains and for the guys updating the scoreboard.
 
#10 ·
Quote:
Originally Posted by PizzaMan View Post

GPUPi is time based. Faster scores equate to a smaller number.

Using gpu scores from the benches is a possibility though....
Give love a chance! This will allow Haswell systems to score on par with 5960X systems!
 
#11 ·
Probably the best suggestion Bass has made all year. ;p
 
#12 ·
band.gif


All downhill from here
biggrin.gif
 
#13 ·
Interesting to say the least. Once again saying for the record "Blech, Vantage"
tongue.gif
 
#14 ·
Unfortunately the choice of benchmarks as is, is going to skew the results in favor of the CPU, and Vantage is basically a CPU benchmark. Benchmarks which level the field a bit across CPUs IMO:
Heaven 4.0
Firestrike Ultra (virtually rendered and downsampled, you do not need a 4K monitor to play)
ffxiv-heavensward benchmark... this may need a bit more data to verify CPU leveling.

Or just use the graphics sub-score from Mark11 or Fire Strike as Bass recommended.
my
2cents.gif


edit: oh.. for GpuPi... just do "1/time x 100" or change 100 to any number for aggregate score weighting.
thumb.gif
 
#15 ·
Many, many thanks for the suggestions guys. It looks like some changes are going to be made before the start of the competition. Glad to see the overlords here are open and willing to hear logical arguments and improvement.

This is going to be one epic of a competition! Luckily Red Team will easily crush this one.
thumb.gif
 
#16 ·
Quote:
Originally Posted by Rasparthe View Post

Many, many thanks for the suggestions guys. It looks like some changes are going to be made before the start of the competition. Glad to see the overlords here are open and willing to hear logical arguments and improvement.

This is going to be one epic of a competition! Luckily Red Team will easily crush this one.
thumb.gif
GG Green Team. Red Team 4 Lyfe!
 
#17 ·
#18 ·
So W8.1 is all fine and dandy for this right? If so I'll probably chip in some points with my 2x 780 Ti's.
 
#20 ·
Quote:
Originally Posted by Rasparthe View Post

Yep, just make sure your Futuremark Systeminfo is newer 4.40 and you are fine for Win 8 and Win 10
Alright, perfect. Now I have to go flash 1.212V bios' on the things.
thumb.gif
 
#21 ·
Feb 1-8 is the FFW.... See ya after that.
thumb.gif
 
#23 ·
Interestingly enough, it's actually tough to choose a side. Oh wait, probably not, since my best red team card is an HD 6850, but I do have a pair of them.
 
#24 ·
#26 ·
What about Intel
sozo.gif


I wanna run my i5 4300U
sozo.gif


lachen.gif
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top