Ok here's the deal: We'll have a "tweak it" versus "OC knob" battle in 1m but you are NOT allowed to use that cpu. :D
Printable View
Ok here's the deal: We'll have a "tweak it" versus "OC knob" battle in 1m but you are NOT allowed to use that cpu. :D
I'm afraid the better cpu I was going to get is no longer available.
That being said, efficiency on the gigabyte board cannot compete with that of the MSI or ASUS in SuperPi 1m.
I wish I could still compete.
Just to let the battle commence i will sacrifice one of my back ups:
http://img.photobucket.com/albums/v6...actory/844.jpg
Ok guys, shoot me.:D
my AIR:
http://valid.canardpc.com/cache/screenshot/600533.png
superpi crash with black screen, i tried too 4140MHz (2898NB) and crash at last round 1M superpi-holy sh*t :( I´ve seen my record....
http://valid.canardpc.com/cache/banner/600533.png
i must thank you guys for all your very usefull information, you are helping me get to my goal of sub 16
i am struggling a bit but im closer
http://i31.photobucket.com/albums/c3...1204/16250.jpg
the cpuz only shows 38 something but i am using core 2 at 4185mhz for spi.
More contenders, nice!
Slappa, thanks. Can't wait to see you going flat out on your new cpu. ;)
Flanker, what time did you see before it crashed or did it not finish?
Equinox1, nice time but can't you just raise all 4 cores?
I see you have the Crosshair III in your sig, just use the Tweak it function man, it works like a charm.
Man... You guys kick Azz... :up:
I've been watching this thread, and I would like to think I'm a decent Overclocker, with pretty good gear....
I actually thought I could compete with these numbers, but after a few hours and a few beers I realized I'm just not worthy... :rolleyes:
Here's my best so far, sub 16's seem out of reach for now...
http://i30.photobucket.com/albums/c3...09_SuperPi.png
I dont understand how you guys with 955 chips are getting so 'slow' PI scores. Is it due to DDR3 is slower then DDR2?
My OCed (not fully stable yet) Q6600 at 3.52ghz with ram at 1057mhz is just under 15 secs. Even at 3.2ghz with ram at 400 mhz I were seeing low 15 sec scrores of super PI.
Why am I seeing these scores when you guys are in the high 15s/low 16s?.. :confused:
Because Super Pi has always been an Intel bench :fact:
Cache!
It's about how SPi was written(a lot of legacy fp code that runs surprisingly well on Merom derivatives).Legacy fp code is becoming extinct in modern applications.Add in a usefulness level of SPI(extremely low) and the fact that there are many other programs that can calculate Pi in a fraction of a time it takes SPI to do it and you get a very very funny benchmark :) .Fun part is that is scaling well with clocks and can tell you quickly if more MHz scales good on one MPU;"not fun part" is that the "benchmark" is slow in what it is doing and wildly favors certain types of uarchs(the ones with low latency fp instructions which are becoming ever so less used).
Edit; I Was Wrong!
Ok swapped cpu's, this guy requires more voltage, little slow on the gun but lets get this started, now let me see your first backup.
Boot 40xx k10 stat up screenshot is of K10 speed not OC knob speed. 3 notches up by middle of run.
http://chew.ln2cooling.com/MSI%20OC%...I%2015.843.JPG
What the F did you do to get such efficiency?
http://img.photobucket.com/albums/v6...actory/813.jpg
It's back to the drawing board i guess. :confused:
Next backup.
The efficiency comes from not cracking 3k yet once you crack 3K NB effecieny dies until about 4500 mhz cpu.
Bench went higher not taking screenies at OC knob speed.
4 clicks up by middle of run.
http://chew.ln2cooling.com/MSI%20OC%...I%2015.812.JPG
how does that knob work is it just bumping the bus? were those 2 runs identical besides the fact that the 2nd you went to the '4th click' rather than the 3rd?
yah preety much.....trying to figure out how much i'm losing to delay time...I think if I hit it by 9th iteration it will apply at 10th.
So every click = Higher ram, higher cpu and higher NB higher HT.
So 4 notches up means CPU-Z screen speed+4x20.5MHz= 4264.4+82= 4346.4MHz, right?
http://img.photobucket.com/albums/v6...actory/735.jpg
Something like that but not for entire bench.
Wow you smoked by one of my backups.
5 clicks up a little sooner than halfway point.......
http://chew.ln2cooling.com/MSI%20OC%...I%2015.719.JPG
Dang, i'm out of back ups. :mad:
Well done chew* :clap:
I will shoot for more but i have a busy schedule for the couple of next days.
Why do you turn the speed up halfway the run instead of at the start?
So you start out at ~4260MHz and then somewhere halfway the run you just up the frequency about 100MHz?
How come the entire run is still fast and not just the second part?
And why does CPU-Z not update?
With tweak it CPU-Z reads out correct, or do you turn the speed down again right after the run finished?
Well done guys. I'd love to play with tweak it or an OC knob.
I expect to see you rebuttal to chew*'s backup Zeus, you can do it.
Because I Tweaked my OS and hardware for PI 1m only now ;) 74MB system ram usage and alot of junk removed from OS, subtimings tight now etc etc. Planning on flattening it so I didn't care. Still a few more things I could do. Fresh install, SP1, wazza, max mem etc etc.
Turning it down to avoid a possible crash screen shotting at rather high speeds with all cores.
You cheater! :D I want to see a screen with the real speed! :D
Still, i think my efficiency is better, non of your runs had a <68K Performance Product but i'm not sure since the real speed is not in the screen. :shrug:
It might take some time i will make you break out that damn back up score, hahaha. :D
I'm gonna need need some cold ambient though, that last screen was pretty much the max at that moment, ambient then was ~15ºC.
The forecast show some cold nights (~10ºC) so i hope i can find some time to fight back. :)
DICE cooling
5016MHz 1 core, all other cores @ 4807MHz
NB 2926MHz
Ram 1672MHz 7-7-5-15-22-1T
http://i74.photobucket.com/albums/i2...11/13812-1.png
Wheres the rest of the screenshot? Nice times!
Nice one Slappa!
Ok, Ive been working on this all day! 25*c ambient. Still not as good as chew* or Slappa, but its getting there! I gotta get some better ram.
http://img159.imageshack.us/img159/4976/pi5z.jpg
My Best was this:
http://img193.imageshack.us/img193/2555/pi7.jpg
Did manage to get a validation @ 4.3Ghz though :)
http://img193.imageshack.us/img193/3700/603094.png
http://valid.canardpc.com/show_oc.php?id=603094
My best with this ram!
http://img17.imageshack.us/img17/6080/pi1y.jpg
Been putting the screws to a new set of 2x2g ram........No 1m becasue that won't tell me much stability wise....
A little tweaking at the same speed goes a looooooooonnngg way.
http://chew.ln2cooling.com/elpida/elpida%209.JPG
http://chew.ln2cooling.com/elpida/elpida%2010.JPG
http://chew.ln2cooling.com/elpida/elpida%2011.JPG
http://chew.ln2cooling.com/elpida/elpida%2013.JPG
Wow man, what did you do between the 2nd and 3rd screen? Quite a big step you took there!
Nice going and lovely cpu, assuming it's still on water. :up:
my new one the best for me :)
AIR cooling, 4120MHz / 2884MHz NB 1M=16.705s
http://img148.imageshack.us/img148/1...z2880nbrek.jpg
nice works guys :yepp:
im using a k9a2 plat v1, do you guys know how i can adjust the NB multi, in my bios (which i think is v1.5 or 1.6 im not at home and i dont remember) i dont currently have any way to tweek that, is there an app that will possibly let me adjust it, or a bios that opens that option up?
my new one record with AIR :)
-finally, under 16.6s 16.593s
http://img9.imageshack.us/img9/7573/...hz16upload.jpg
How much is RAM speed playing a factor in superpi times here? does having DDR3 help?
what will be better in supepri 1M? CPU 4100MHz, 2870MHz NB, RAM 1093MHz or CPU 4141MHz, NB 2828MHz, RAM 1077MHz? I want to broke my record post in my last post....But think, i cant get to 4160MHz...
CPU speed will count most, NB at 2870 or 2828 won't make that much of a difference, just like ram speed. Just try to have the timings as tight as possible.
FlanK3r, what kind of temps were you dealing with at boot? I'm noticing if my BIOS HW monitor says 37c or up, I have problems booting into windows. :\
I dont know :). But in idle in windows, i have normally 36 C with defaults and with high OC (1.488V and 4GHz or more) about 43-45
Well I was able to reach 16.688s @ 4.1GHZ
http://sites.google.com/site/apokali...i1M@4.1GHz.PNG
I had my RAM at 1:2.66, 5-5-5-15, 2.2V.
you think it would be better at 1:2, 4-4-4-12?
i don't think that will improve your score much but you'll only know when you try.
Do you really need 1.6Vcore for that? Most Phenom II cpu's max out somewhere at 1.5-1.55V or even lower.
I tested yesterday some superpi 1M in x86 and in x64 and i found overall in x64 OS better score! Diferent is about 0.06s.
16.594 @ 4160MHZ (208x20), using 1:2 RAM divider @ 4-4-4-12
http://sites.google.com/site/apokali...pi_4160MHz.PNG
OK, time to revive the "DEAD THREAD" :D
OK, I worked on this chip for a couple days and I think I hit the wall @ 4.5Ghz with 1.58v @ -67c. 25 minutes prime stable then I stopped it and was able to pull a 15 second superpi run and get a validation with cpu-z! BTW, this is a Phenom II 940 plugged into an ASUS M3A79-T..
I know it isnt any where near the best or even close but what the hell! :up:
CPU-Z validation
Some pics:
http://i120.photobucket.com/albums/o...605_163613.jpg
http://i120.photobucket.com/albums/o...605_163531.jpg
http://i120.photobucket.com/albums/o...1b/15secpi.jpg
Nice! Now gun harder!
Get that ram and NB clocked.
Here is my DICE result from some time ago
http://i74.photobucket.com/albums/i2...erPi1mDICE.png
Heres my 940
http://i271.photobucket.com/albums/j...1/pi4261-1.jpg
that was during the winter in January
and heres 32m
http://i271.photobucket.com/albums/j...771/pi32m1.jpg
DDR2
hello guys i can To take part with my 550Be:p:
http://img23.imageshack.us/img23/8065/16218.jpg
here 2M with my best NB, i Forget 1M at 3122mhz NB frecuency:
http://img188.imageshack.us/img188/9701/38797.jpg
thanks :)
now sub 16 with 4250mhz
http://img196.imageshack.us/img196/6899/15969.jpg
:D
very nice! whats your board and cooling?
FINALLY. It took some goddamn low ambient temperatures. And that is why I thank Canadian winter.
http://i74.photobucket.com/albums/i253/Wop11/15657s.png
Whauuuu, congratulations, with AIR cool! What temps? :) U have higher 1M clock than my max validation :D
Gratz Slappa! About time mate. :) I can't compete with ya now, sold the AMD stuff.
my 14,789s (dice)
http://img39.imageshack.us/img39/3577/14s2.th.jpg
HT was up to 2.7 and NB to 3.5, but no tested
http://img39.imageshack.us/img39/7023/semttulonw.th.png
Here is mine it was done with a quick run non tweaked win7 and lots of things running
http://hwbot.org/result.do?resultId=906793
but with AIR, Slappas run is maybe the best (or Chews?:) )
great 550 overclock with air :up:
Same here only 550 with 4 cores enabled on water cooling
http://valid.canardpc.com/show_oc.php?id=766225
http://i34.tinypic.com/14vmft2.jpg
How about a sub 16's 32m on air ;) Beat up vista, not my PI setup, no tweaks.
http://chew.ln2cooling.com/qdig-file...0III/lrg_6.jpg
Hi chew, I have a question for you if you are willing to answer because I see that you like to help people alot... :) And thanks in advance! :up: First, sorry for my english, it's not my native language but I hope you will understand my post... :)
In first very big post (middle pics are cut because of very big quote) you are showing how memory tweaks are helping to achieve better super pi results with Phenom II 955, in last post you are showing much better score with Phenom II 965 with very low differences in used frequencies, cpu freq +58mhz, nb freq +147mhz, ram +31mhz with the same timings, only big difference that counts is much bigger htt with 965 ( I personally like bigger htt ) and Asus instead of Gigabyte... So the question is what is the reason for so much better score (~33s) with 965, htt or different motherboard with different options in bios setup or something else like tweaks or something?
For first post with 955 results I must say that these results are not so impressive if you look them like 24/7 kind of guy that looks for best price/performance ratio with some pleasure of overclocking inside that ratio :) Hardware used is very expensive, the best mbo from Gigabyte, 955, Corsair Dominator gt (?) ram, second setup with the best mbo from Asus, the best cpu from AMD, the same expensive memory (?)... Take a look at this hardware and result, looking at prices in my country 3 times cheaper cpu Athlon II 250 than Phenom II 955 or 965, 3 times cheaper Gigabyte 770T-UD3P than 790FXT-UD5P or Crosshair III, 2 times cheaper Kingston ram than Corsair ram with the same amount of them... No L3, slower ram frequency and timings, in 24/7 these frequencies no one use, so for the pleasure of overclocking and benchmarking with casual air or water, nothing extreme, this setup is much more better, almost the same performance for overall 3 times less money invested... Setups like yours needs to be shown at extreme conditions to see them in all their glory with every tweak out there :)
http://i38.tinypic.com/fdqzpl.jpg
The motherboard is just that good in 32m...however it's awsome 32m results has a price. It seems slow in cinebench and a few other benches......there is no perfect board, only perfect boards per application.
I used an AC tossed on a bucket for a Competition yes to cool video cards and was 100% open about it......
If you honestly think I would go that far for a basic untweaked vista 32m run you would sadly be mistaken.........Why bother when I have an arsenal of cooling at my disposal that is far easier and far less hassle..........
My results all scale linear......my air results are slower than my water which is quite a bit slower than my -20 phase.......
Maybe if you tried finding your good core you would be able to run 32m fast on 72F ambients.
Assumption is the mother of all screwups.
FWIW sanyo denki and megahalems lapped + 72F ambients go a long way on a chip that was binned out of 6 chips.
whoa getting a little testy there........all that because I asked if you used the ac? sorry if you dont like being questioned lol.
Turning on an ac is going that far? you throw the box on top press a button and shazam......air cooled, what an ego......to good for ac lol.
I didnt say anything about your results being linear.......but its good you keep track when showing your results;):D
So if your ambiant is 72F what did the cpu temp go to.........had to go into the high the 100's with 1.52v on just air. Did you use lots of fans like those big ones you had over the graphics cards?
Who said I didnt find my good core?:cool:
Hmm best out of 6 cpu's thats a nice one.......is it the best of the bunch?
I sold all the other chips......that should answer your question. AOD is not installed nor do I use it......its a nice application that is user friendly but I do all my handiwork in bios.....temps I don't care......32m does not have any real impact on temps anyway.
Its just as easy to slap the head of my mach 1 phase change on as well as my waterblock on benchstation 2.........and the results are far better.....
Like I said why bother ;)
Air results don't interest me.........they interest you guys which is why i do them........There is no such thing as an air WR.......there are Just WR's period.
Speaking of WR's time to bench more ln2.....
http://chew.ln2cooling.com/The_Laboratory/IMG_1695.JPG
well thanks for letting us be in your presence
break it guys! :D here's mine, not efficient yes but i don't mind
http://img36.imageshack.us/img36/4712/superpi1mbest.png
Damn fine win 7 run ridney.
Already done on the last page :P
However, I am just fooling as that would not be comparing apples to apples. I give you credit for achieving such a great time using a tri-core as they are slightly slower in 1M. Not to mention you are on win 7 as well.
Excellent Job :up:
What type of cooling may I ask? Just normal air? On my run I had to use cold ass winter air. My chip just can't pull the same numbers on standard ambients.
nah, i was trying to break chew and titan from their conversation here hehe not that i was asking someone to break my time :p: but thanks anyways
we don't have winter in singapore so it is definitely not cold air. i used a pack of ice cubes and a cooler to cool the chip using an apogee drive pump/block, pic on my avatar ;) it was a fun two hour bench session
Actually you a little nasty with responses, you going to get it right back.
We dont do this out of necessity we do it for fun.We not trying for WR so to downplay people running air and saying you dont care about how they do things is really being a little nasty towards all running air and water.
Theres nothing hard about running on air or switching on an ac or using chilled water.:shrug:
You seriously need to buy a small quiet single stage phase change and use it. Only then will you see where I am coming from.
If your trying to maxximize your clock you should grab a diff board as well versus the current one your on that needs a huge amount of voltage just to compensate for its droop/poor voltage regulation.....
You will find then that you don't and won't run near as hot nor will you need as much voltage.
I don't downplay anyones results nor have I ever.
Try grabbing one of these fans, lapping your xigamatek with 800 grit wetsand ( higher grit makes it perform worse :shrug: ) and tape the sides.....It will be alot more effecient. Its a very quiet fan for its CFM. Quite tolerable.
In light load apps it will beat the TRUE 120 lapped with same fan. Haven't tested it versus megahalems in light load yet but megahalems with same fan and lapped seems to be winning in load apps.
http://www.petrastechshop.com/12x38sadesan.html
Im back:
and..i have new record 1M for me 16.297s, think with lower ambients at winter i got run about 4240 MHz and about 16.1s
http://img48.imageshack.us/img48/686...kord16297s.jpg
17m 28s ... :-(((
http://img371.imageshack.us/img371/7...mhz2856mhz.jpg
http://img371.imageshack.us/img371/4...superpi32m.jpg
temps:
http://img383.imageshack.us/img383/1676/temps.jpg
And 32M (so far from DDR3 runs :-/ )
Nice runs.
I wouldn't worry to much about 32m.
Giga is just not that effecient at it.
That asus crosshair seems to be very good at it plus it seems to OC my chip about 50-60 mhz more with the same voltage although NB clocking is a tad strained, a newer bios might fix that.
On the other hand its not gonna win any cinebench battles, Giga really shines there.
I'm back to my M4A78T-E for benching purposes, sold the MSI 790FX-GD70.
I have soldered some SMD caps to the back of the socket, I hope it improves stability under cold and buys me another 50-100MHz.
http://www.shrani.si/t/3J/IT/2xNMjfRP/dsc0192.jpg http://www.shrani.si/t/3s/CN/32qTOjgO/dsc0193.jpg
<--- Before | After --->
And yes, I know it looks horrible :)