Printable View
here is my rig
core i5 ivy bridge 3570k @4.5ghz
corsair vengeance 1600mhz memory9.9.9.24
gigabyte z77ud5 mobo
gigabyte windforce gtx 670
corsair force 3 gt ssd 120gb
bios settings all set to stock but
cpu @1.220v bios
cpu pll@ 1.7v bios anything lower and my gflops go down to about 100gflops
here is my proof (hopefully i have done it right) i know its not 5ghz but i really dont neet that much :)
plese let me know if i qualify for this thread
Attachment 129045Attachment 129044
My first IB:
http://img.techpowerup.org/120803/3570_50_140.jpg
Watercooled, 25c ambient.
3219B510
^^ That chip needs an IHS removal ;)
It is removed ;-)Quote:
Originally Posted by tool_462
Temps are fine if I run 1.33v @4.9, or dont use avx.
Then either you have a weak watercooling or you have some problem in terms of TIM/IHS remont your temps shouldn't be nowhere near that high at those volts.
The mount is good, every cpu is different - and so is the internal core temp circuits. Temps went down 10-12 by removing ihs.
LinX with avx is hot, this is prime at 4.9:
http://img.techpowerup.org/120804/prime 49x.png
That chip is starving for bandwidth, lol. What watercooling and what tim if i may ask sniper?
Quote:
Originally Posted by Sniper
Well with 1.38v I only hit 62ºC on the hottest core so yes... something must be wrong with your setup.
The tim is ic diamond, and simple 360 rad with silent fans in 25-27 ambient. Havent started on ram clocks yet.Quote:
Originally Posted by Zeus
And what do you mean by "starving for bandwidth", is the GFlops too low ?
No i Gflops is pure calculation power by the cpu. It's starving for bandwidth as you're only running 800 9-9-9. Running 1100+ provides much more bandwidth for the cpu to utilize. Run Aida's bandwidth benchmark to see what i mean.
There's no point in getting the cpu that high without having the ram running fast(er), it bottlenecks the system so to speak. ;)
Ok, running 1100 7-11-7-28-1T now - does not affect gflops. Only 2x2gb so far, see what I can get with 4 sticks.
Stuen4y|i5 3570k@4.7GHz with 1.39V (Ultra High LLC)|L208B070|TRUE Spirit 140 + 2xTY140@~900RPM|ASUS P8Z77-V|OCCT 4.3 Linpack +AVX|Stock TIM
Generally a bad overclocker, needs almost 1.45V for any benches @4.8GHz and anything above that is a no go only for validations and SPi like stuff... I think the most prime/LinX/OCCT I got @ 4.8GHz is 15 minutes with insane temps. 24/7 settings are 4.6GHz with 1.3V but I think I'll lower it to 4.5 to get more sensible voltages.
http://img822.imageshack.us/img822/8...3aug051204.jpg
I'm not here to make it into this club (I already made it on my x58 anyways) but I'm currently satisfied with this (components in my signature): http://i.imgur.com/0slux.png
Dr_Dx | Core i7 3770K [4C/8T] @ 5000.1MHz | 1.288v | ASUS Maximus V Formula| Water | L221B009 | LinX 0.6.4 AVX w/ Linpack 10.3.11.019 | Liquid Pro
Another one - going for low voltage 5Ghz 100+Gflops Lynx stable
Attachment 129317
med1kl 4.85ghz i7 3930k SandyBridge-E 6C+6T 1.4 Vcore 16GB Ram Asus Rampage 4 Formula Bios 2105 Koolance Exos 2.5 External Water Cooling with CPU 370 Waterblock
http://i.imgur.com/7mtgv.png
Dr_Dx | Core i7 3770K [4C/8T] @ 5100.0MHz | 1.328v | ASUS Maximus V Formula| Water | L221B009 | LinX 0.6.4 AVX w/ Linpack 10.3.11.019 | Liquid Pro
Attachment 129387
Why is realtemp opened when you don't show the real load temps?
I think it is showing real load temps but dr_dx is running a water chiller. 5.1 at such low voltage is absolutely insane, what a chip!
May be you are right, I just saw it's opened 1 minute ago, but checked his other picture and saw it was opened during the whole test.
Dr_Dx | Core i7 3770K [4C/8T] @ 5300.3MHz | 1.488v | ASUS Maximus V Formula| Water | L221B009 | LinX 0.6.4 AVX w/ Linpack 10.3.11.019 | Liquid Pro
Attachment 129397
I currently run my rig with an OC that includes full-featured ops, all the way through four levels of Turbo, that peaks at 5 GHz single core. It's stable and reliable enough to lend confidence to an all-out attempt at full 5 GHz rigging for a qualifying run submission, but I have a question on the requirement for using "All" of the RAM within LinX's setup in a system sporting much more of a RAM load than is the norm for most gaming rigs.
Right now I have 32 GB installed, and after the OS and an 8 GB RAMDisk for Temp files there's usually a bit over 22 GB remaining. With that info, here's a snap of the first iteration of such a LinX setup with my system at 4.625 GHz which may immediately highlight what I'm about to ask:
Attachment 129662
Since there won't be too much more time trimmed off each iteration at full OC, can there be a reasonable ceiling to the amount of RAM used for LinX runs so as not to tie down the system completely for 12+ hours just to generate a qualifying screenshot?...
The last few screenshots in this thread show that with more than 8 Gb memory people still use around 7Gb in LinX equaling just about 1 hour of LinX run.
Thank you for pointing that out.
I shall return...
the community decided on 9216 awhile ago. zalbard hasn't been around. also 99% of the recent submissions won't be accepted.
Do you know what post that is? I didn't see it on Zalbard's original post. If in fact that was decided, it would be good to indicate that in post #1 so people will know what the "current" requirements are (like using AVX, etc.). Running "MAX" RAM is not a good solution. I know I have 16g of ram but have been testing at 7168 simply because I didn't want to sit around for 3 + hours (much less 12 hours like the above poster) to complete 20 passes and that is a very common amount for folks running 8G RAM to test with. Historically, most of the "accepted" runs are between 60~90 minutes. IMHO, since RAM amounts vary greatly, a better solution would be to set a minimum time that LINX has to run without an error. That way guys that run a lot of RAM aren't penalized by having to run for 12 hours and conversely you don't have these 13 minute runs of less than 2G RAM either. Somewhere between 60~120 minutes should be sufficient to verify system stability.Quote:
Originally Posted by zoson