2013 Autumn Folly

[HiVizMan]

2013 Autumn Folly

Caps Off Bonus



Two day burn - there is no cap limit at all for this special 2 day part of the contest. Just push your hardware as hard as it can go and show us what you could have done if there had not been such a mean cap. Set by the wicked witch of the South. That is me if you are not sure.

Cooling restrictions still apply of course.

Summer is well and truly behind us and time to start using the cooler weather and get benching.

This time round it is going to be slightly different.

To compete in this contest you need to form a team. One Intel CPU and one AMD CPU. The choice is there to bench solo if you own or have access to both. But the idea is to have a benching buddy and bench as a team.



Both CPU's have to run each of the benchmarks. The two scores are submitted and the points or times are combined.

Benchmarks

Wprime 1024 mil

Super Pi 4mil

ROG Realbench

Futuremark


As always there are limitations. This time it is no different.

Rules

• Air or Water cooling only - no sub ambient.
  
• Same CPU for all benchmarks
  
• Screenshot to include CPUz main tab, memory tab and motherboard tab
  
• Offical background to be used
  
• Intel CPU cap
  Sandybridge 5000
  Ivybridge 4700
  Haswell 4400
  SB-E 4900
  IVY-E 4600
  
• AMD CPU cap
  990FX 5000
  FM2 No limit
  
• Weekly 5 points bonus for submissions (and a bonus bench week 2 and 4)
  
  

Challenge Wallpaper courtesy of Chris as always.



http://www.myminifile.com/images/af1080.jpg

Contest starts 1st October and end 31st October.


Due to forum being down for a few hours last night the contest will close 9am 1 November 2013 London time.


Important notice: i5 Processors get an additional 100MHz extra to the cap.

[nikosa43]

Menthol, I say jump in. Good op to use your Pi memory sticks. Also, until Det come back, maybe Vizman or anyone else find an AMD guy and you both continue as a team. So, your team will have the bonus points too. It's pitty not to have you in this and it's still in the beginning.

[meankeys]

NP here 🙂

Glad to have ya in the comp

[solaufein_gr]

More people and teams, more fun and more sportsmanship. Noone will complain.

Got a question. Changing an element like "NB/CPU" freq. is reason for running all 3 bench again for results to be valid? Or i can keep e.g. the wPrime from 2400 NBfreq. and superPi from 2600 NBfreq.?
Everything else but NBfreq. is the same.

[nikosa43]

solaufein_gr wrote:
More people and teams, more fun and more sportsmanship. Noone will complain.Got a question. Changing an element like "NB/CPU" freq. is reason for running all 3 bench again for results to be valid? Or i can keep e.g. the wPrime from 2400 NBfreq. and superPi from 2600 NBfreq.?Everything else but NBfreq. is the same.

I think that if you don't change your CPU clock you can do anything you want. That's the reference, all other settings can be altered to suit the specific bench.

[solaufein_gr]

Only the clock? So i also, can change FSB and multi (and change RAM divider for instance?) between superPi and realbench?

Ohh, and what is the "tolerance" in CPU freq? I saw some runs with 4615 (i know you set it correct in bios, have to consider about +20Mhz when setting the BIOS), or so, Mhz, but that may be significant for sensitive bench like superPI.

[nikosa43]

Yeap, that's why there is a cap (limit) for the clock. That's the only one. Anything else can be altered to suit your needs or the benchmarks's needs to be more efficient. It's in your knowledge, or experiments you make to find the best settings for your system and for every benchmark. With the same CPU clock and the same FSB or RAM timmings one benchmark may have good results but another not. So, make sure you reach the same clock for all and try to find which settings make each benchmark more efficient with higher scores 🙂 As for tolerance, I try to be as close as can be. A few Mhz will not be a problem, let's say 5- 10. Vizman is more suitable to give an answer for that. 30s and 50s, I believe is too much IMO.

[Zka17]

Guys, the rule is to use the same CPU for each benchmark - the settings of the CPU may be particular for benchmarks... at least that's how it was in other Challenges...

[HiVizMan]

Menthol wrote:
Vizman,
Do you have a problem with me filling in for Det until he returns? or anyone else for that matter

Hell no you are very welecome and thank you for being a knight in shining armour. And just for the record your daughter is not known to me regardless of her being born in 1975 😛

solaufein_gr wrote:
More people and teams, more fun and more sportsmanship. Noone will complain.

Got a question. Changing an element like "NB/CPU" freq. is reason for running all 3 bench again for results to be valid? Or i can keep e.g. the wPrime from 2400 NBfreq. and superPi from 2600 NBfreq.?
Everything else but NBfreq. is the same.

I was going to comment in your submissions about the NB and HT you will get way better performance with changes made to both. And that can be benchmark dependant.

solaufein_gr wrote:
Only the clock? So i also, can change FSB and multi (and change RAM divider for instance?) between superPi and realbench?

Ohh, and what is the "tolerance" in CPU freq? I saw some runs with 4615 (i know you set it correct in bios, have to consider about +20Mhz when setting the BIOS), or so, Mhz, but that may be significant for sensitive bench like superPI.

You can change every thing you like as long as you are below the CPU frequency limit.

The CPU limit needs to be followed guys, I can appreciate the pushing for good times but the point is to control your systems and stay below that limit. Not on the limit but below the limit. First week I do not stress to much as we all know the scores will get heaps better as we all learn more about our benchmarks and our systems. But below the limit is your watchword.

Zka17 wrote:
Guys, the rule is to use the same CPU for each benchmark - the settings of the CPU may be particular for benchmarks... at least that's how it was in other Challenges...

Quite correct - we want you guys to be trying to find every ounce of performance from your settings.

Tip of the day for Super PI look in your bios for settings that need not be active to run the bench. Some surprising settings will impact on the efficiency of the benchmark.

[Menthol]

HiVizMan wrote:
Hell no you are very welecome and thank you for being a knight in shining armour

More like an old nut in stinky boxers

WTH is that bonus bench Viz, you're crazy dude, lol

[solaufein_gr]

edit:
Just in time, i've posted too early 🙂

Let me put clear examples:
I got 1min28,500sec on sPI and 229.000sec on wPrime with 23.5x195 = 4598Mhz
i got 1min28.300sec on sPi and 229.300sec on wPrime with 20.0x229 = 4595Mhz
we can keep the sPI from second run and wPrime from first as valid?
Or we need teh whole set of 3 benchmarks for given cpu speed?
I'm sorry if i'm annoying, i think i've misanderstood the rules and i'm loosing my time doing runs that i shouldn't :rolleyes: