chew*

Well first off not all of your runs are on equal terms.......so that tosses consistency out of the equasion. You changed stuff between boots. Dual channel, single channel or were so unstable you were dropping channels for starters. Not sure what point your trying to prove but your going about it wrong. All things equal between boots = 2 drastically diff scores = some form of bug or inconsistency.....however in my experience if you can boot chances are you will not bug..... The problem with bugged runs is people boot lower becasue they are not stable enough to boot higher, then they clock up in windows.......thats when the bugging shows up. Could be something as simple as the system clock going out of whack whilst running the bench due to instability. For me it was a simple case of hey the freaking asus board is limited to subs that can be used. Crucial timings are limited to 4-4-4 and I think 8 as lowest. I knew exactly what the max possible was, I used both boards, I knew which boards could do what. The giga could pull all those timings to 1-1-1-1 and for my runs they were certainly tighter than 4-4-4-8. Timing limitations that don't exist on the gigabyte. Combine that with another bench ( 32m pi from all parties involved ) also very sens to timings......and things just did not add up Of course I recently had that remedied for the M4E but prior to 1298 it just wasn't possible. Last but not least having not ever used that platform I could not make an accurate analysis. I only make analysis of scores after i have run the same platform at the same speed on the same board or equivalent ( many hours/days on end ). Just like you should be analyzing the same way before trying to "prove disprove" something.

I'm sure if I went back and labeled all my results other than corsair or gskill it would sway your charts I have never bothered using model in submissions. In all my testing PSC is always 2 secs behind...... That however doesn't change the fact that if the system in question was a daily rig I would without a doubt choose the flare. Effeciency can be measured in dollars to For instance my single 2500K purchase and a 5500 chip = effeciency. Mine and msimax's pooling money together for 18+ chips for a 5600 and 5700 2600K chip = sucks monkeyballs. Same can be said about $159 total set of ram versus a $400 set if we are talking 2x4gig. Edit, BTW where did all the flares go ?

For this card it's top clocks without pll, Cold bugs about -150, tested pll real quick pulls down full pit with. Prepping for a 4 way run so I need to know each of these ref cards in and out. 2 more to test. Linux, thats one run through vantage when i found safe speeds, not to worried about effec, second 2 cpu configs versus 1, compared to 1 cpu configs the score is spot on......with 10K less cpu score than a 2 cpu setup

The retail 990X that I tested ran 5700 through vantage cpu test........least when tuned for performance etc uncore ram etc etc......don't let the ES 990X results fool ya it's 980X ES all over again. Thats a ref card btw have to test 2 more ref then turn pll mod on see what that does, later I will try out the the CU II

Thats a ref card big guy. no tweaks, all results done on a big screen that was causing some performance issues

Very nice mark

I clipped this one thing becasue I feel you are misguided a little. So let me get this straight? You want to ban someone that does not do incredible effeciency at a live comp where they have limited time....... You do realize for instance lets use 32m pi for example the best results are obtained by tuning each timing one at a time and running bench, jotting down gains or losses and repeating. This could take days platform dependant and given your available time to test between a real life. Really the OS plays little role here, far less than you think. This is realistic in a live competition with time limits how? A perfect example of this is MOA, we were told the model of ram that was to be used, we bought said model to get a feel for it and dial in ballpark performance subs prior to event. When we got to comp model of ram was not what we purchased nor anything like it ( not even same ICS in fact we had single bank 3x2g and the model for event was dual bank 3x2g), nothing worked we had pretested for not to mention it seemed we had a bad stick, our results suffered greatly. On topic, splave check your post about expander on XS.

I think if a team is organized a bunch of guys will get together and say OK i have this this and this, what do you have? So for instance a bud of mine has a matrix 4870, I have only a reference.......well heck dude you tackle that class your matrix has way more potential. I save ln2, I can focus on another class where my cards will have more potential. When you bench for a hw specific class in rev 4 you truly are repping your team, thats the way I see it. Our team has already gotten together and said hey ok you have this and this and this, I have this and this and this. Lets figure out whats best then divvy it up My anaylsis for rev 4? Teams win, manufacturers lose. No need for a team to buy 20 samples of a 580 GTX + 990X to be competetive to spam the latest HW class for points anymore.

Absolutely it won't be sad, those 2 scores were in preperation of rev 4. We knew the highest would count the most so we tried to get the highest possible we could for our team with what we had at our disposal

So does intel cherry pick retail chips for benchers? This chip straight from intel marketing. L051B170 0586 53X, didn't waste time testing further other than to jump multis just in case..... Oh and to put a # to my 5664mhz L041C108 4323 Also we forgot to add all the junk from L041C108, forget the numbers but the count is 10 chips 53X to 54X

Nothing wrong with sharing cpu's mobos for 3d which is what I was referring to travis, which is also why I didn't quite understand the complaint. I'm only going by what you told me personally about testing a few chips with him for 3d. I think my above post was misinterpreted. So bottom line what I meant to say is if you share cpu's boards for 3d why complain if someone else does, thats all. I know other teams do, they even say so, nothing wrong with it. Hell I sent my own personal 980X and R III E ( from OC summit winnings ) around to XS to bench 3d on, the only difference really is the chip sucked, people only complain when the chip is good.

More or less ran this just for another compare of correct scaling.

Dumo's coming soon with a 5790 ish, then I whip out all the tricks.

Like it matters, if I didn't have his cpu and I had my 100 mhz less cpu with a better IMC + ln2 I would blast my own score which I intend to do anyway. Unless of course you don't think I can do it....... Also to "team bench" costs money, 3 hour+ ride, gas, rental or bus ticket. Not to mention we pooled money together for our cpu's till we found decent one's. Last I read to share a cpu in a 3d bench is completely legit so I don't even see why this was even brought up. Anyway thats the whole purpose of rev 4, so that teams can't dominate top 5 in a ranking, a xeon class comes to mind....... Don't you bench with Travis and share his errr I mean ASUS's cpu's boards etc etc

See tighter timings............

Becasue pro If i ran tight timings like you I would not only bug maxxmem but also my sytstem would be so unstable, I would get double posts like you. :

Becasue pro If i ran tight timings like you I would not only bug maxxmem but also my sytstem would be so unstable, I would get double posts like you. :

I should hum that trackball at you now for running the NB higher than 4500 :

Looks good pro, Right where i would expect it to be. I see you listened about not exceeding CPU clocks with NB clocks for this one, that or your cpu was a turd for NB :

To many, intel sux.

No challenge, phase is as far as I go, can't afford ln2 right now.

And your D9's

1x1 gig has it's advantages, NB clocks being one of them Cpu's not really that great, All my good 965's died under ln2.

On AMD? LOL its a miracle this can even run

OC windforce, I had a weird issue with one of my monitors in XP that would not run res at default of bench in 06, maybe that is whats wrong with your setup? Otherwise not sure.