:O is your hearing damaged from those fans?!
Printable View
:O is your hearing damaged from those fans?!
Haha the Nidec sounds like a jet engine...my wife was on the phone the other day and the caller asked her if she was vacuuming LOL! My hearing is OK because I made a controller for the fans, so ramp them up to full speed only when benching. The loudest I have are my 6000rpm San Aces, which are just crazy - they are so powerful that the frame is made of solid metal, and they weigh a freaking TON :eek:
http://img713.imageshack.us/img713/1196/1jpgfz.jpg
My last thread-derailing tangent: I managed to get a 5214MHz verification after plucking up the courage to boot at 1.5625v, and having achieved all I set out to with this chip, it's being sold! :up:
I'm afraid it's already got a buyer dude - transaction in progress for 385 GBP which is about 600 USD.
The batch 93036C128) is not very well known but very good. I had about four CPUs lined up on ebay and every seller got back to me with batch info, and when I squared this batch with google it came back with sparse but VERY nice results, so I knew I had a good chance of getting a golden chip. It's not as gold as others in this thread (eg CryptiK's 4.5GHz with HT, 86 GFlops @ 1.343v :shocked: ) but it's not half bad either! :up:
I was going to get the golden 990X but asking price is too high so I'll go back to my crumby 920 for now and see if I can squeeze 4.3 or 4.4 out of it. It will happily bench 4.74GHz @ 1.6v, 226 bclk, but it's a very QPI-limited chip and needs 1.4 vcore for 4.2GHz LinX/Prime stable :down:
Got my 920 into the club but it took serious persuasion: 1.55v CPU :shocked: 1.5v QPI :shocked: Copper TRUE with 5C air :up:
LennyRhys | Core i7 920 DO [4C 4T] @ 4510.1 | 1.55 | Asus P6X58D-E | Air | Worst batch ever
http://img600.imageshack.us/img600/1...510mhzlinx.png
l0ud_sil3nc3 / 3960X @ 5.1 / Asus Rampage IV Extreme / Single Stage
http://i.imgur.com/VYN1U.jpg
New quick try, HT ON. For somehow my another OS doesn't run LinX properly, it doesn't stress the CPU at all, finally found the motivation to boot my another computer's SSD on this system. And on this I still cannot see Vcore through CPU-Z. Well it didn't seem to make 4500 under 1.4 V, seems to be fluctuating between 1.39 - 1.4, but almost there. I think I promised zoson some results a few months ago, hopefully this was on the right track :)
http://img36.imageshack.us/img36/7678/4509a.jpg
Happy holidays
I agree also but, at 600.00 a pop just for a 3930k it isnt worth it to run linx at 5ghz. Especially for the length of time is takes to run 20 passes with 16-32gigs, max mem selected.
He didnt use max mem for his run, 3080mb of 8192 available. Thats a lot of juice for 4500.
It's not just easy, it's the rules!
At 200+GFlops it should cut the time down significantly.
Never stopped us before. I just did that 1.6v Core 980x run. 980x's were expensive CPU's last time I checked, also. ;)
This is the 4.5GHz and 5GHz club. 4.5GHz for chips it's hard to hit 4.5GHz on, 5GHz for chips it's hard to hit 5GHz on. SB-E is easycakes to hit 5GHz, so if you want to get in, you gotta man up like l0ud_sil3nc3.
^LoudSilence didnt use max mem either. Its easy to "man up" on SS with 38second passes and max mem not selected.
No offense to loudsilence, he has a great chip, Ive seen results with it in other threads.
It takes 44 minutes to run 8 passes (failed on this) with max mem selected, water cooled. It would take over 2hrs to run 20 loops with 16gb max memory selected. Hardly worth it to re run it again to pass with a little more vcore so someone will think we are men :rofl:
http://img31.imageshack.us/img31/9828/failfw.jpg
Uploaded with ImageShack.us
Here is 5ghz max mem, 1024mb sticks in quad channel. 3930k, Rampage IV Extreme, water cooled.
http://img560.imageshack.us/img560/3...ealmanrofl.jpg
Uploaded with ImageShack.us
http://img580.imageshack.us/img580/1687/25passes.jpg
Uploaded with ImageShack.us
http://img850.imageshack.us/img850/2360/dualrads.jpg
Uploaded with ImageShack.us
Now THAT's a run OC Nub!!! Very, very impressive!
As to the maxmem thing... I think it should be max within reason for the platform... When going for super high OC's, it's not necessary to have lots of memory! It's why all my submissions are with my 6GB kit, and not with my 12GB kit or both kits together.
I think the question is if 1GB sticks can actually stress your cpu fully? I think the evidence is that you have higher GFlops with your bigger sticks... What if you put your bigger sticks in and test 7168MB memory size(say you had 4x 2GB for 8192 total mem, so 7168 would be ~ max mem)?
It takes about an hour to run 5120Mb memory size on a gulftown(basically all mem on a 3x2gb kit).
^ Thanks!
My kits with 2gb sticks are either mismatch (3x2gb kits from x58) or arent good for 2133 mhz. The 8gb's Dominators I have are good for 2000 tops on this platform, hard to get hypers stable, system shuts off during the run, very weird. I get just under 200gflops at 1866mhz with 2gb sticks so gflops look bad, and 1gb sticks seem to limit gflops as well, tried 6x1gb but still limit gflops. I know the 16gb kits are good for 2133+ but the time it takes to run 20 passes is way too long.
I havent tried linx since s1366 for this thread, have good 2600k's but saved them for benching.
I like this platform, its very stable but unless you are using cold you wont do much in the way of benching, so I thought Id give this a shot.
nice goin OC NUB
is that a/c on ?
Jesus 5GHz on water DDR3-2000MHz. Killing it.
Here is my entry for the 5Ghz under 100 GFlops club:
http://img209.imageshack.us/img209/5798/5000mhz.jpg
CPU: i7 2600K Batch# 3135B460 @ 5000Mhz on 1.416v
MB: ASUS Maximus IV Extreme-Z
RAM: 4GB DDR3-1886 [3.06GB used]
Cooling: Corsair H70; fan on chipset
Thank you.
Thanks!
Yes the ac is on, chip gets pretty warm at 1.5+vcore. I need to insulate the sensor on the ac again to stop it from shutting off. If you can keep the thermostat warm the compressor will just keep running, temps stay 40-45c that way during linx.
Updated to Service Pack 1 and the GFlops didn't go up by much, what up with that? lol
They went from 63.0051 to 63.2597; oh well, guess the title of my entry still correct at least.
Running 25 loops instead of 20 since I don't really look at the run, and most of the times I miss
the end of the 20th run; hopefully I'll catch it now since I got 4 more loops to take the shot.
Here it is with SP1 installed, and Linx still running while the shot was taken:
http://img215.imageshack.us/img215/3296/redone5ghz.jpg
CPU: i7 2600K Batch# 3135B460 @ 5000Mhz on 1.415v [BIOS]
MB: ASUS Maximus IV Extreme-Z
RAM: 4GB DDR3-1886 [3.03GB used]
Cooling: Corsair H70; fan on chipset
I'll try for something higher later on.
Telo,
There are two other reasons your GFlops could be low.
1. You have not updated your LinX Binaries from the Intel Math Kernel Library page. The LinX listed on the front page of this thread is outdated, and the one on the regular LinX thread is SUPER outdated.
2. You're not booted into diagnostic mode: start -> run -> msconfig -> click diagnostic mode -> ok -> restart
Either way, neither of your two runs are valid for this club.
It appears I was using the wrong LinX shortcut...
http://img820.imageshack.us/img820/7770/redone5ghz2.jpg
CPU: i7 2600K Batch# 3135B460 @ 5000Mhz on 1.44v [BIOS]
MB: ASUS Maximus IV Extreme-Z
RAM: 4GB DDR3-1886 [3GB used]
Cooling: Corsair H70; fan on chipset
Had to jack up the voltage from 1.415 to 1.44v :shakes:
I doubt anything else will stress this CPU like this so gonna bring it back to 1.415v
[XC]Atomicpineapple / 2500k @ 5.0 3145B254 / EVGA Z68 FTW/ Single Stage
http://i202.photobucket.com/albums/a...le/th_5GHz.jpg
One more:
i7 980x Batch 3031A243
Gigabyte X58A-OC Rev 1.0
6GB Mushkin PC3 12800
EK HF + Black ICE GTX Xtreme 480
http://i1139.photobucket.com/albums/...58A-OCLynx.jpg
RRunner that is a very nice chip, maybe one of the best gt for low volts, great job:up:
Very nice RRunner! What was the load vcore? Also very nice temps especially considering 26.5*C ambient. That a single 4 x 120mm rad setup? What fans/rpm?
RRunner that's pretty impressive if it's LinX stable below 1.3v! My 990X can boot into windows at 4.5GHz with 1.288v in CPU-Z but it's not stable. These 3031A2xx batches really are something else!
Thanks :)
In full load is 1.305v ... Yes, it´s a single Black Ice ... Fans Delta 3200 RPM´s/ 130 CFM :)
I did not tested with lower voltage than 1.28v (LLC Level 2 + Dynamic Vcore) ... It´s hot in here ... 29.5ºC at the moment! Holy crap :(
Wow, 1.305v load at 4500MHz - that's the most impressive Gulftown I've ever seen!
With a good bit of tweaking I think my 990X will match CryptiK's in vcore, doing 4.5GHz with 1.33-1.34v... but it definitely doesn't have as good an IMC and needs more QPI volts. I did a quick LinX run this morning with 4096MB to speed things up a bit, and this is what I got:
BIOS values:
CPU 1.3625v
QPI 1.2875v
IOH 1.2v
180x25
1800MHz 8-9-8-24-110-2T
3060MHz uncore
Copper TRUE, 15C ambient (no outside air or turbine cooling this time - just a very cool chip!)
http://img849.imageshack.us/img849/5...0xlinx4500.png
Next step for me is to test the IMC to see how the uncore scales with voltage. Looks like 4000MHz uncore will need about 1.4v but I'll see what I can do. 200x23 with 2000MHz memory would be a very nice daily OC :D :up:
Not sure if its the same for gulftown but qpi=vcore used to aid stability.
Also how come 1.2v IOH? I never needed over 1.16v even @ 215+ bclk. Not that it matters really, but its less heat in the chipset if you go low on volts.
Tried lowering CPU PLL? Easiest way to test for added stability is to get linx to a point where it will reliably fail within 5 or 6 runs, then drop your pll as low as it can go and start raising it. You should find a voltage range where you suddenly become more stable if not totally stable. Takes a while but you've got the patience :). 1.62v to 1.65v for me depending on qpi/vcore.
Cheers Pilsy - that run was focused mainly on lowering the CPU volts as much as possible, and 1.3625 is the best I can do. Unfortunately my board isn't quite "premier league" and won't allow me to drop CPU PLL, which is a real PITA frankly lol. I know this CPU will do 4.5GHz under 1.35v but not in my board! If Asus were to release a BIOS that allowed lowering of PLL voltage, that would be wonderful... as it stands I'm really tempted to get a better board, but it would be ridiculous for the sake of one voltage option! Argh!
As for IOH, I tried it at 1.16v and GFlops dropped to 82, so the higher voltage definitely seems to be doing something for the memory - could just be one of my board's many quirks. I guess I could repeat the same run as before and see what happens... all in due course. :)
My next task is underway: tighter memory and higher uncore for better memory performance - I've run it a few times and am taking voltages down each time. Took a quick screen grab to show increase in GFlops. It looks like 1.3125v will be needed for QPI, but that's not half bad compared to my 980X which needed 1.35+
Furthermore it's a very cool chip considering this is air cooled and in a 20C room - an earlier run maxed out at 81C, and that was with higher voltages!
http://img215.imageshack.us/img215/8...3600uncore.png
4.7GHz, 12 threads, air cooled. :up: Concentrating on CPU first as this looks like areally nice daily OC with the CPU at 1.4375v in BIOS, 1.424v idle and 1.432v load in CPU-Z, so memory and uncore are very low at 1720MHz (8-9-8-24-110-2T) and 3060MHz respectively. QPI is 1.325v and IOH 1.2v. PLL at 1.8v as I can't lower it grrrr... :shakes:
http://img72.imageshack.us/img72/2663/4719linx.png
Anyways after getting this run more easily than I anticipated (first try lol), I then started to tweak the memory and uncore. Without changing any voltages I tightened mem to 7-8-7-24-88-2T and raised uncore to 3440, and GFlops went to 89 consistently. I got 19 runs from LinX and just before it started to process the 20th run the system froze... ARGH!
I then tried memory at 2000Mhz (8-9-8-24-110-2T) and uncore at 3575MHz, and that gave me 90 GFlops but no stability, so I may try to get it stable with a lower uncore speed.
Since my first successful run (pictured) I've not been able to get anything stable - I've had the x101 BSOD several times, and twice I've had the "linpack_xeon64 has stopped working" popup about 10 loops into a LinX run that looked perfectly stable with very little fluctuation between GFlops. Weird.
Yet another submission; this time with a 2500K
http://img703.imageshack.us/img703/5456/2500k5ghz.jpg
CPU: i7 2500K / L143B578 / 1.43v
MoBo: ASUS Maximus IV Extreme-Z
Cooling: Corsair H70 / dedicated fan to the chipset
RAM: 4GB Mushkin 998750 @ 1866Mhz 8-8-7-24-2T
Did it :cool:
http://www.sieben.at/linX/linx5GHzcropped.jpg
Full Pic here
i5-2500k @1,475V >5GHz Prime stable, batch #3145B294
a hot one but aircooled by Thermalright HR02 Macho
on Asus P8Z68V-Pro/Gen3
2x 4GB GeiL EVO Corsa 1866MHz
Saphire Radeon 6850
Seasonic S12-430W
Have a good TimE!
http://farm8.staticflickr.com/7158/6...bb9e786d_b.jpg
Long time lurker, first time poster. Finally got it stable! 4.5ghz HT on @ 1.36vcore
i7 920 D0 @ 1.36 vcore 4.5ghz - Cooled by Corsair H100 (is that considered water cooled?)
Asus P6T SE
12GB DDR3 Corsair Ram
1x Radeon 5790
1x Radeon 5870
Corsair AX1200
Yes - "closed loop" water cooling. ;) 4.5GHz at 1.368v is very nice, but it doesn't look like you're stable - GFlops vary too much from 52-56. If the system is stable you should be getting less than 0.5 GFlop fluctuation, which means you may still need more vcore or QPI... unless you were using your system whilst LinX was running. Were you? :D
On another note, my Rampage III Extreme arrived today so now it's time to really let my 990X loose. 4.8GHz with HT enabled on air? It's possible. :up:
I sent you messages regarding those parts on the hardforum and got no answer. what's the batch of the cpu and what's the kit model of your mushkin?
Will the list on page 1 be updated or is the "club" already closed?
It's been a while but we still haven't got any 3930K or 3960X to fulfil the 5GHz stress?
Yeah, I'm running at 4.9 on my 3930K, beacase the damn VRM overheats on the Sabortooth.
I popped an R4E trying and settled for 4.75GHz for the top drawer systems.
I just bought The Nemesis' 990x. He hit 4.74 with 1.462v cpuz. I wonder what I'll be able to get this chip up to with 1.6v? :D
First quick test with new I7-2600 5004MHz Batch L145A899 Motherboard P8Z68-V Pro. Cooling with water
http://valid.canardpc.com/show_oc.php?id=2203464
Screenshot because i cannot upload image or URL.
http://aijaa.com/007209380664
I saw his a while back and thought it looked like a nice chip, and fortunately I now have one that is very similar.
The big question with Gulftowns is... how good is the IMC? I suppose even if the IMC isn't golden (mine isn't), the cores still are, like mine. I'd say any Gulftown that can do 4.5GHz at or below 1.35v (real) is golden. You should get 5GHz with 6 threads, definitely... if not with HT. :yepp: Very interested to see how you get on with it. :up:
guys, is this version of linx ok to enter on the list?
http://imup.se/i/Ws61NAiEtz.jpg
Moreover rules state 5 GHz for Sandy Bridge. 4,5-4,7 GHz is a joke using SB. I laugh at those people having 4,5 GHz SB and actually being excited about that, while so many people achieved much harder 4,5 GHz using Nehalem :)
BOOM:
zoson | Core i7 990x [6C 12T] @ 4811.8MHz | 1.508 | ASUS Rampage III Extreme | Water | 3101A239
http://i144.photobucket.com/albums/r...-9336GFlop.jpg
zoson | Core i7 990x [6C 6T] @ 5011.0MHz | 1.581 | ASUS Rampage III Extreme | Water | 3101A239
http://i144.photobucket.com/albums/r...10853GFlop.jpg
I honestly have no idea. I booted it at 4.725 right out the gate using the voltage on thenemesis screenshot of 4.74ghz. It does 3.8ghz UCLK with 1.35v QPI, which is great!!! It can also run my 18GB at 1866 instead of 1833.
I'm betting I can tighten the timings down a little more on the 6gb redline kit with this chip. I'm so happy I finally have a golden cpu! :D Had to rely on my skills alone before, but now I can FLYYYYYYYY.
Yes that is the correct version of Linx, however you need to update your intel math kernel library binaries.
You can get them here:
http://software.intel.com/en-us/arti...pack-download/
Download the windows package.
Open the zip.
Open your LinX folder.
Overwrite all files in your LinX folder with the files that come in the linpack update.
I think I'm going to spend some time tweaking 4.5GHz because higher clocks are just too dang hot for this chip with air cooling, and tbh I'd be happier running it below 1.4v if possible, so 4.6GHz may be on the cards. I'm at a 2-3C disadvantage since going from the TRUE Cu to a regular TRUE, but it does take a fair bit of strain off the CPU socket.
I'm guessing your 990X will do 4.5GHz at about 1.35-1.375 like mine, but ofc it depends how hard you want to push it in LinX in terms of performance rather than stability.
What about those on SB-E with retarded amounts of RAM. Running a 64k problem size takes a very long time. Can we get in this running 16GB (seems to be the standard) for the tests?
It's unbelieveable how much cooler this CPU runs at 4.5GHz compared to 4.7GHz, and all at the expense of only 3 GFlops (90.6 down to 87.6). This time it's in our living room at 20C ambient and instead of being cooled by a monster server fan it's simply two Gentle Typhoon AP-15s in push/pull on my vanilla TRUE. Max temp was 83C, so only 63C above ambient compared to 85C... Astounding! :D
136.5 x 33 for 4504MHz, uncore @ 3685MHz, mem @ 1910 8-9-8-24-110-2T, C-states enabled
CPU 1.35v (BIOS), 1.337v load (real)
QPI 1.35v (BIOS), 1.334v load (real)
CPU PLL 1.6v
Gutted that this run crapped out at such a late stage and with such consistent GFlops... oh well, nearly there. :yepp:
http://img864.imageshack.us/img864/5...504mhzlinx.png
True story. Last night I set up a 4512MHz run 25x180. Voltage was set at 1.35v BIOS and real was about 1.330v. It made it through 20 loops and i wasn't watching, crashed on the 21st. I decided not to save it when i noticed that a few times during the run my GFlops dropped ~1GFlop randomly. So yeah, looks like you're right, between 1.35v and 1.375v for stable 4.5ghz.
This is with CPU 1.349v (real) and QPI 1.334v (real); PLL still at 1.6v. Dropped uncore to 3550MHz and got 86.6 GFlops pretty consistently, but I'd really rather run my memory at its rated 2000MHz so I'll try other BCLK frequencies and see what works best. I could push the uncore a bit but I'm quite happy with an 85C temp ceiling... not sure if I can get 3680MHz uncore with 1.334v QPI...
I think with more tweaking my CPU might do 4.5GHz at 1.343v real, but definitely not lower!
http://img51.imageshack.us/img51/6274/4500t2.png
ASUS Rampage IV Extreme
EK on the CPU and Board.
Batch #3133B429
1.425v in the BIOS, High LLC, CPU CC 180%.
http://i39.tinypic.com/2hs3n2c.png
MMaggo if you have SP1 you it will need more volts to pass linx/prime and score nice GFlops, if you dont you can do it on a bit less volts scoring less GFlops...for what i see your CPU takes 1,433 LOAD and not 1,475 as u told me, aint that bad at all
cheers
Sergio
what tweaks ?
and why you say isnt performing good ? for linx ? linx its just a torture program and can fail, as they guys here said update it and should run ok, but i would care more on using it like in gamming like me :) thank linx
i wont linx mine when i get it just put mighty 5GHz and game game game lol, hope i can make it :)
Ok, I just looked and all the other entries had HT off.
Try decreasing the memory size to 9216 and retry. Either that or turn off HT. I think there's a bug in linpack for memory sizes > 9216...
Also andre... LinX is just a GUI interface that issues CLI commands to the linpack libraries that are maintained and produced by Intel themselves. Aside from the memory size bug... It's usually correct. Intel uses it themselves for testing chips.
Ok, I just looked and all the other entries had HT off.
Try decreasing the memory size to 9216 and retry. Either that or turn off HT. I think there's a bug in linpack for memory sizes > 9216...
Also andre... LinX is just a GUI interface that issues CLI commands to the linpack libraries that are maintained and produced by Intel themselves. Aside from the memory size bug... It's usually correct. Intel uses it themselves for testing chips.
Just got home and did some tests HT disabled, basically the same GFlops, 108.9. :shrug:
where did you get your linpack update? directly from intel i hope?
just google intel math kernel library download
Very nice chip zoson, so you got the 990X after all :)
I took my computer to the balcony, but I guess it's not worth it with the setup I have right now, the Corsair H100 is just not enough to pull high clocks and of course I'm afraid to kill/degrade my ES chip. :P
Well at least the IDLE temps were promising, but damn this winter is too hot! :D
http://img513.imageshack.us/img513/6646/lammotidle.png
Max core temps were around 50c with LinX and @ 4.5 GHz HT, but nothing special...I'll wait the -20 to -30 temps C like last year... ;)