James, did you experience ANY boost from Giga to the MVF?
Printable View
dude both boards are basically identical, its all about what bandwidth you can extract from each board and depending on your cpu/board that will also be very similar
This thread will now be allowed to get back on topic.
I've tried both boards (mvg and ud3h) the ud3h I feel keeps you less dependent on the ic specific options in the bios that the mvg has. memtweak works on both boards. if you already have one or the other just stay with it.
I had to send my mvg back and I got the ud3h in the meantime. both are rad. i cant say one is better than the other.
sorry to do the cane analogy but its the best I can describe them..
Do I need to take out the devil?
http://youtu.be/nPWvInwk6vw
or do I have to take a crap really badly?
http://youtu.be/HxTjM8iEbZg
the pimp cane can take out the devil in 6.17.15s
the hurry cane got you to the toilet in 6.17.14s
both are good. ;)
just over a month to go... hope to see your results before Aug1
tuning mem on my daily settings i found a nice results
andressergio - 6:36:860 - 50*100 - 1300mhz - 10-12-12-25-137 1T - M5G - Team X LV 2400C9 - 1.655v Air
http://i.imgur.com/A9PeO.jpg
ever since Trans_AM told me that mem tweakit works on giga boards... ive been playing around and discovered something very interesting. I was rather impressed by pro's read score and wanted to know what the f*** i was doing differently. after about 10 min of tweaking with the asus tool, i discovered something very fun and strange at the same time :)
tRRSR setting in mem tweakit is a very strange one. it defaults to 5 on UD5 + C7 flares @ 1277 (106.4bclk *47 multi)
the strange: when setting a value of 3, it then sets itself to 8 and bandwidth goes down, when set to 2, it goes to 6. i dont remember if these were the EXACT values BUT you get the idea... something very strange is going on
the fun: when changing the value from the default 5 to 4.... IT STAYS AT 4... then read goes up 1000mb/s :up:
5
http://i48.photobucket.com/albums/f2...are-tRRSR5.jpg
4
http://i48.photobucket.com/albums/f2...are-tRRSR4.jpg
Finally got my mem clocking issues resolved at least for the time being, unfortunately my efficiency is still terrible :p:
Kinda at a loss, I'm at least 1.5 sec slow, I'm sure some is waza but even without waza I'm still at least a second off. Doing what I used to for GT/SB doesn't doesn't seem to cut it anymore, probably doing something obvious wrong :shrug:
http://i.minus.com/jbcRMC8523Y8N0.png
i think the blue ripjaws was defaulting a 4 value. to make matters more interesting, setting it in bios results in bsod at windows boot. setting in windows = 32m stable. :shrug:
im on psc flares right now giving them a spin and was trying to figure out some issues im getting when trying to reach 2600 c8... using 2600 strap = no post, using 2400 strap + bclk in windows gets me to 2600 8-11-8-27 enough for aida to run but not 32m stable. fail after loop 4 @ 1.775v. there is alot of work that i have to do to try to catch any old runs so im just posting any stuff that i find odd for those that might be running the same board and similar ram configuration.
sometimes 32m isnt the tighest settings but the fastest :D
^ Does the Gigabyte board still have the memory profiles? Curious as to what the Extreme profile sets the sub-timings to.
Cant find the blue gskill 2133c8 kit, so Ive been trying other kits from ebay. 2000 Tridents, Ripjaws, HyperX, PI's, anything I could find but it looks like nothing is going to get me past 1225 8-10-7. Have an old kit of PI's, PC3-12800 c7 that will do 1300 8-12-7 fairly easy, but the 32m time is slower than 1225 8-10-7.
@youngpro
#7 result was yours...I remember you submitted that. do you want me to remove #7 and update your result with your recent submission?
whatever you want mate, i dont mind ;)
Lol @ Dumo.... yup i been bidding on all the old bbse & psc kits & they been going for crazy money.
Deux - 6:19:593 - 50x100 - 1300 MHz - 6-11-7-28-80-1 - ASUS Maximus V Formula- G.Skill Pi 2200mhz 2x2gb/PSC - 1.92V - LN2 mem, SS on CPU
http://i.minus.com/jm1OV6yGQ9fcN.png
Had a few liters of ln2 to waste, now to actually fix my awful efficiency :p:
Attachment 128013
I tested another set today, the MSI GD65, it is with poor performance because I need throtllestop on the mobo. and my worst cpu / memcontrol .... :(
asus MVG looks like this because it took 2 of the good 3770k so it Benches no more the mobo, nor the parts that sat on it. :shakes
in the second asus mvg has one channel RAM passed, so are wating for maximus :yepp:
Nice run deux
wow Splave..nice run..:yepp:
pro..aircooling :eek: awesome!
angoholic- 6:17:985 - 50x100 - 1300 MHz - 7-10-8-25-80-1 - ASUS Maximus V Gene- G.Skill RipjawsX 2x2gb/PSC - 1.885V - LN2 (-80°C), CPU air
http://www.abload.de/thumb/32m_6m_17_984_5ghzvrzq7.jpg
Thats quite some update angoholic :up:
This is the fruit of 3 hours work a .046 gain :rolleyes:
These G.Skill mems are crazy good they are 18 months old & have been used in my daily runner all that time running tight & fast. I've probably spent weeks over 1.9v on air & when I'm done messing about just back them down again to 2133 1.65v & they still spot on!!!!
C-N - 6:19:312 - 47x106.4 - 1277MHz - 8-11-7-22-71-1 - Gigabyte UD5H - GSkill C8 2133mhz 2x2GB/BBSE - 1.91v - Stock Air
http://i106.photobucket.com/albums/m...2M/6m19312.png
http://i106.photobucket.com/albums/m...9312AIDA64.png
Just thought I would add a couple tips after benching with fredyama's 32m mastery and raja's extreme technical knowhow you can put a lot of bs Internet tips to rest.
-slim os not necessary
-theme doesn't matter
-exact max mem is not important if you are in the ballpark
-page file not hugely important
-cdt cdt iv etc not needed simple copy and paste once is sufficient
-huge wazza file is excessive under 2gb is more than enough
-maxing out available mem is not the key, system balance is most important
-running 16k ever is dumb lol
-worrying about not using your mouse once you start wazza doesn't matter
Flame suit on but I'm just trying to share for those of you that don't know. I experienced a whole week of this with the man that is safe to say the 32m king.
-slim os not necessary - true, did't see difference between slim & slimmer )
-theme doesn't matter - true
-exact max mem is not important if you are in the ballpark - what is ballpark?
-page file not hugely important - true
-huge wazza file is excessive under 2gb is more than enough - true
-worrying about not using your mouse once you start wazza doesn't matter - true
ballpark meaning close to the correct amount 500-700 should be fine
yes ramdisk will help
Display drivers installed, ati / nvidia doesnt matter
very nice run angoholic!
just to chime in on what splave was saying
16k is to "wipe" the left over junk from your last run, if you do that manually you dont need 16k, otherwise 16k is advised
maxmem NEVER really matters, it just has an impact on your waza size, so once you get your waza size/wait time right, stick with your maxmem
theme/slim OS also impact your waza, so adjusting your waza based on these will give you additional gains/loses, but as splave said, maximizing available has no baring on if you have your waza right or not
for me waza is the hardest part of 32m these days and getting it right is really hard
Awesome run C-N:up:
Same sticks here but green and black pcbs
I'm still waiting for nice pot. Probly easier to just score a dime bag and get over it. Lol
Dumo - 6:19:422 - 48x104.2 - 1250.5MHz - 8-10-7-21-72-1T - MVF - Gskill 2133 C8 - 1.875v - air
http://img853.imageshack.us/img853/8942/another422.png
Thanks Dumo.
Nice run there with tRCD 10..... Mine are the black PCB's but as far as I can tell looking at others results green/black are as good as each other.
I have been itching to chill mine too I see more lol.... they can run 32M faster 2650 & tighter C7 but not quicker. 2650 starts fast but slows down & C7 is 7 or so seconds slower than C8 although that might be something to do with having to move to tWL 8 to get C7 to run.
Brian y would send one of his goons in upstate NY after you if he read this.Quote:
-slim os not necessary
Bit of RamDisk action.... tbh I didnt find it helped a new waza gave me most of this but interesting experiment for me anyway.
C-N - 6:19:203 - 47x106.4 - 1277MHz - 8-11-7-22-71-1 - Gigabyte UD5H - GSkill C8 2133mhz 2x2GB/BBSE - 1.91v - Stock Air
http://i106.photobucket.com/albums/m...2M/6m19203.png
http://i106.photobucket.com/albums/m...9203AIDA64.png
^ very clean run and nice ram
C-N
good runs. what ramdisk you used?
Thanks guys :up:
Gunstar I used the AR Soft RAMDisk here http://www.arsoft-online.de/index.ph...=fileinfo&id=3
Thanks angoholic :)
:D Thanks mate..... I did kind of home in on you as the next one on the list lol but I'm sure you aint done yet ;)
It looks the same difference to me near enough 12.689 v 12.875s last loop
RAMDisk left v Normal right
http://i106.photobucket.com/albums/m...iskvNormal.png
The last two runs I posted are not exactly comparable as I used a different waza which did knock another .1** off my .312 score & I was also running a couple of tweaks differently hence I think the difference is very small if not just margin of error.
thanks, but there is even more in it.. did not think that i could go this far with cl7..last time i made only two quick runs for pretesting psc cold..was suprised i got dino..today i spent some more ln2 on that and did a couple of runs, last was this one:
angoholic- 6:17:625 - 46x108.7 - 1304.5MHz - 7-10-8-25-72-1T - MVG - GSkill 2200 C7 PSC - 1.885v - all LN2 cooled
http://www.abload.de/thumb/32m_6m_17_625_5ghz0ydyi.jpg
Attachment 128114
give him better memory this is nearly full optimized :D
n1 one
sweet run Angoholic:up:
dont be surprised mate, cas7 1300 is the :banana::banana::banana::banana: :)
well done!
i do have some issues with wazza there as i caught you at 12th loop and then dropped off badly but part of it is cas7 1300+ & other part is wazza :) :)
how does one get a 4.250 first iteration, crazyness ;)
first iteration doesnt mean anything, well it could mean someone started at higher clock but i doubt anyone would do that here, we're benching for fun
ive seen it start at that also a few times but i prefer it starting slower hehe
Cheapseats final loop is waza and mem timings
i read discussions about tweaks, OSes etc. Hehehehe if it was only that simple. Wait till you start testing a :banana::banana::banana::banana:load of OSes and other parameters and then come back with your head itching
-updated
*angholic, nice to see germans grabbing #2 :D we appreciate your frozen pics :D
*c-n and dumo .... airsanity!
*team.pure in the house, Deux welcome aboard
about OS's... its alot of chasing the dragon. doesnt matter what, someone will always be faster in pi with same clocks and similar timings. in the end one would think the only variable is the OS. my advice here is this, dont act like a tweaker when you are tweaking... you cant ever catch that dragon because once you do it will slip through your fingers once again. just use what you know works well and tune out your subs and waza. once that limit is reached, then go back to the OS for the finishing touches.
super sauce TW S2K3? meh, tried it... not much faster than a well tuned XP
ramdisk? slightly faster until waza is fine tuned. then gain is very minimal, mabye im doing it wrong? (not sarcastic, i actually question this)
maxmem? a specific value always worked best for me. over or under results in *slightly* slower scores... talking like .02 or so
stripped os? can make regular XP just as fast by disabling services... but if im disabling them anyways... why not just cut them out and make install go faster?
themes? .02 gain olive vs silver consistantly
keyboard vs mouse? i think it gains... though i havent havent tested in quite some time though. i get through setting up 32m faster using keyboard anyways. less thinking more doing.
in the end... the most gain is in bios, the least is in OS (excluding the basics like LSC, maxmem, etc). like 90/10. there is a gain, but is it worth the weeks of building testing rebuilding testing borking os starting over testing retesting....
if you know what you got works well, stick with it. there are alot of different ways to the finish line. you can blast through or finnesse your way in :)
Well I've lost myself with waza & consistency!!!! I see it makes a massive difference just cant work out what I'm doing wrong consistency wise.
These runs are with the same bios settings only thing different was the waza.....
In this run everything was the same as my .203 entry using RAMDisk like I mentioned to cheapseats.
http://i106.photobucket.com/albums/m...2M/6m19234.png
In this run the only difference was waza..... looks like a crazy fast start then slows down!
http://i106.photobucket.com/albums/m...9203Normal.png
thanks :)
thank you :)
thx, yup, i know this wazza trouble :D ..with my runs, most of them from 1-5 is fast, then dropping, but there are also better ones..
i suppose that the initial loop does not mean anything for the result. at air settings i had .265 to .313. with cold mem the probability seems to be higher to have a faster initial, but that is not usual, i also had 17-18s times with slow initial loop. some runs 1-5 are ultra fast then dropping very bad, seems that it also depends on wazza.
as far as i observed initial loop depends on:
cpu speed but fixed here..
memory timings (with bbse vs psc on air i got different typical values)
memory speed
wazza effect
whether results from previous runs got deleted or not
yap, it is all about the fun. if someone would change speed during a run, he needs the RogConnect Server running->impact on PP->need really more clock speeds to get gain->change speed during run-> distribution of loops would be different greater than a bad wazza effect.
in case of doubt to obviate that someone could think the run was at other speed i recorded a video of the first run from yesterday which was cookies time (initial was .281 by the way ;) )
http://www.youtube.com/watch?v=p5PqZDtQBCg
just in case anyone would still have doubts, i could make another one re-running showing that ROG Connect was NOT involved. :yepp:
want more? :D
I wasnt doubting mate, I was impressed :) flushed out my current best backup till I get my stuff back from taiwan
And some improovement from me with Formula
Aristidis - 6.18.735 - 106.4X47 - 1276.5Mhz - 8-10-7-22-72-1 - ASUS MAXIMUS V FORMULA - GSKILL RIPJAWS PC17000CL8 2X2GB/BBSE - 1.84V - AIRCOOLING
Attachment 128126
No lol he wont let those out of his sight ;) I meant my chips and benching suitcase
Splave - 6:17:640 - 48x104.2 - 1354.7MHz - 6-11-7-25-96-1 - ASUS Maximus V Extreme - G.Skill Trident 2400mhz 2x2gb/PSC - 1.98v - LN2 All
http://i.imgur.com/llNen.jpg
limit of IMC, will try another chip
pretesting for 6.8 32m? :D
-updated
*Ari is king of Airsanity!
*splave and angholic battle for 2nd and 3rd crazeee
Crazy result Splave. Getting over my 6m 20s is getting impossible for me. Wazaa driving me nuts. :D
My first "good" score on this bench:
ToyTen - 6:34:260 - 48x104.2 - 1146.3MHz - 9-10-8-21-79-1 - Asus Gene V - Gskill Perfect Storms 17066CL9 2x2GB - 1.73v Air Cooling
http://img706.imageshack.us/img706/4654/634260.jpg
Lots of improvements to do yet, as i cant have XP on this computer :confused:
Manage to lower on my 24/7 settings
andressergio - 6.34.260 - 100x50 - 1300Mhz - 10-12-12-25-1T - M5G - PATRIOT VIPER II XTREME 2400CL11 2X4GB - 1.665V - AIRCOOLING
http://i.imgur.com/XUJcQ.jpg
yes, i just couldnt figure that there was an incompatibility with my Raptor 74Go & the rest of my hadware ... just couldn't beleive it :)
it is fixed now. I have XP but i have terrible results... The one i posted was made with a fast SSD but on my h24 OS, now im back with a simple old HDD. that chould explain why i guess.
If i cant get decent score i might need another fast SSD.
Thx
Hm, shouldn't my g.skill 2133 c8 be able to handle TRRSR = 4 at 1200MHz? I notice a huge improvement in bandwidth compared to 5 but I can't seem to make it stable...
^yeah it should no problem unless some of your other second/third timings are off
I just saw this competition, but luckily there's still plenty of time to get in on it. I'll see what I can come up with.
Lets see some OCN flame :D
hey guys I bought these mem sinks and just so happens they line up perfect with the dominator pot holes! Just need to run to the hardware store to get some screws, the threading on the sinks is slightly smaller then dominator screws
http://www.performance-pcs.com/catal...0bafd0261abf99
Will test later today!
http://www.performance-pcs.com/catal...ALIV-BK_01.jpg
why not use the included screws!Quote:
Included:
Thermal PAD.
Screws x 1Set.
they look legitt
Just about to try these sinks. It seems to have good mass...
http://img210.imageshack.us/img210/3...nshot004oi.jpg
Whilst we're on the subject of cooling I zip tied my Vapo to my BBSE the other day to try a bit more voltage/speed.... it looked fast but didn’t work fail test 18 was my best at 2600 but here’s a pic anyway.
http://i106.photobucket.com/albums/m...32M/BBSESS.jpg
a pic from the other night sadly nothing productive to post atm, but will be retrying tomorrow.
http://i.imgur.com/qKEp1.jpg
I do have to say the memory pot has some mass to it:up:
Hey Spalve, what sinks are those?
Man, i'd really love to participate but unfortunately my CPU won't do 5GHz, no matter what voltage i throw at it. I can run 32M at 4.9GHz, got 6 minute 30.094 so i've got some work to do but i think that translates to around 6:21 at 5G. This is with PSC at 2400 9-10-9-28 1T Latency Boundary 5. (crappy ram to match crappy cpu)
Anyone got a good 3770K left? :)
^Yes found one with stronger imc
http://i.imgur.com/4UcMT.jpg
ive found IMC's for memory to be quite interesting on this platform,
I have a very strong IMC at high cas levels, but ive got a better chip for tight timings, and the tighter timings chip doesnt clock so well for high frequency, got me confused
so I think I broke superpi:p:
http://i.imgur.com/0Jos3.png
not really sure what happened, but that is the first time I have ever seen that:shrug:
Had to look ten times to see what was wrong, lol. I have seen that before though, an extra loop.
lmao WR for 25 loops Im sure of it
wtf...
i found someone of i4 had the same in good old 775 days but it seems that person did not recognize that..
http://i447.photobucket.com/albums/q...M_25Loop_1.gif