Intel 4.5GHz & 5GHz LinX Stable Club

:explode2::explode2::explode2:

Core i5 2500k @ 5317.6 MHz | 1.592 v. | ASUS P8P67 PRO | Water (EK) | L051B314 |< Over 100 GFlops >


http://upic.me/t/lk/twk_ocz.jpg


http://upic.me/t/s5/p200411_21.53.jpg

Enjoy the OP update! :D

Quote:

---

Originally Posted by skuldarin

UPDATED

---

No, it's not, LinX was stopped before taking the screenshot. :(

Quote:

---

Originally Posted by ac001

ac001 | Core i7 990X [6C 6T] @ 4539MHz | 1.429 | ASUS Rampage III Black Edition | Water | -

---

Same thing, LinX has to be running when you take your screenshot. :(

Quote:

---

Originally Posted by newhit

OOPS! I hope my recently posted "challenging bit of fun" gets in Mr Zalbard :worship:.

---

Oh, yes, it surely does! It proudly kept #1 for a bit, actually! Ended up being 2MHz too slow, though! :mad:

Quote:

---

Originally Posted by hellper

Hi everybody!

A little bit old-fashioned, I know, but here is my application to the 4.5+ GHz club.

---

Hey! Nice first post, and welcome to the forums! :)
Nothing wrong with a 4.5GHz C2D, it's perfectly fast for pretty much everything, still.
Good job on the post, also, everything is well and in order. :up:

Quote:

---

Originally Posted by TWK_OCZ

:explode2::explode2::explode2:

---

:oscar:
:smoke:
:explode2:

Quote:

---

Originally Posted by zalbard

Enjoy the OP update! :D

Oh, yes, it surely does! It proudly kept #1 for a bit, actually! Ended up being 2MHz too slow, though! :mad:

---

+3MHz FTW? :up:

PS
Somehow I don't think so :D

Fire extinguisher at the ready............
L042B076
GA-P67A-UD7
Water
Win7-64bit AVX (SP1)

I have noticed that nearly all the submissions start "L0" for the batches. I know that in the big wide world you can see various batch numbers but it does seem a bit (OK, a tiny bit :)) interesting that on here "L0" starts the majority of sandybridge submissions.

ac001 | Core i7 990X [6C 6T] @ 4605MHz | 1.468 | ASUS Rampage III Black Edition | Water | -

http://picload.org/image/olrooi/linx4600.png

regards ac001

Quote:

---

Originally Posted by newhit

I have noticed that nearly all the submissions start "L0" for the batches. I know that in the big wide world you can see various batch numbers but it does seem a bit (OK, a tiny bit :)) interesting that on here "L0" starts the majority of sandybridge submissions.

---

Our batch is the best so far for low volts OC :up:

Dont kill him with so much volts :D , for high multi my does 55x ( max bench 5560 ) what can yours do ?

http://upic.me/i/mu/yociexp83.gif


Core i5 2500k [4C 4T] @ 5341.9 MHz | 1.608 v. | ASUS P8P67 PRO | Water (EK) | L051B314 |< Over 100 GFlops >


http://upic.me/t/4w/5341.9.jpg

Quote:

---

Originally Posted by st0ned

Our batch is the best so far for low volts OC :up:

Dont kill him with so much volts :D , for high multi my does 55x ( max bench 5560 ) what can yours do ?

---

55x is dreamland for me. Mine is now warped at the corners at 5332 :D

Can you get a LinX 20+ at 55x? That has got to be a killer chip or maybe 56x :up:

UPDATE: this is the absolute max i can do until i fix some thermal walls.

4.710ghz

same everything else. mem speed at 1884 and uncore at 2x (3768) CL9 T2
gflops at 101.2, down from 102.2 so there's a bit of a wall there too.


980x BATCH: 3013A540 | 6C/6T | 188x25 (4.710ghz) | 1.55625v | Asus Rampage III Extreme | Water (Koolance cpu360)

http://i3.photobucket.com/albums/y71...s/Jackpot2.jpg

Quote:

---

Originally Posted by newhit

55x is dreamland for me. Mine is now warped at the corners at 5332 :D

Can you get a LinX 20+ at 55x? That has got to be a killer chip or maybe 56x :up:

---

I wasnt talking about linx :rofl:

I wont put my chip under linX with 1.6v :shrug: But it can run at 5561 anything you throw him, 2D , 3D you name it :p: But Linx doesn't worth the LN2 :D

i7-2600K
1.360v (under load)
batch: 3101A112 (that's what the box says)

ASRock extreme6
Air (Silver Arrow)

Win7 64bit SP1

I also did it with HT on but the GFlops are lower and there's quite a bit of variance so I posted w/o HT

http://i52.tinypic.com/ofszrt.jpg

Quote:

---

Originally Posted by vafan13

i7-2600K
1.360v (under load)
batch: 3101A112 (that's what the box says)

ASRock extreme6
Air (Silver Arrow)

Win7 64bit SP1

I also did it with HT on but the GFlops are lower and there's quite a bit of variance so I posted w/o HT

---

Nice low voltage Costa Rica sample @ 50x. :up:

Quote:

---

Originally Posted by PolRoger

Nice low voltage Costa Rica sample @ 50x. :up:

---

Yeah I'm pretty happy with it. Apparently Costa Rica is good for more than just coffee. :D

Hi guys! It's been a while since I posted here, but I got a new toy and was thinking of joinin' it too.

donmarkoni | i7 990X [6C 6T] @ 4811MHz | 1.5V | Rampage III Extreme | Water | 3028A486 (not sure, forgot to take photo, will check)

http://img94.imageshack.us/img94/678...x48ghzlinx.jpg

Donmarkoni in the hiz-ouse! Your poor 990X :D Very nice run. Remember you cant RMA an ES!

EDIT - whats it need for 4.5GHz? Whats your loop consist of and what was the ambient temp for that run? Temps are nice given the vcore & speed.

^^
good example of how LinX shouldn't be used as a performance check. i scratched at 103gflops with 4.672ghz, 101 range with 4.710 and Donmarkoni there with 103 at 4.8 AND higher uncore / mem speed at the same time. i tell ya... somethin about those sandy's, they have their way with gflops.

Quote:

---

Originally Posted by st0ned

I wasnt talking about linx :rofl:

I wont put my chip under linX with 1.6v :shrug: But it can run at 5561 anything you throw him, 2D , 3D you name it :p: But Linx doesn't worth the LN2 :D

---

Sorry. I know I'm missing something. Are you saying that LinX is not worth the bother or that LinX is putting too much stress on the chip?

With your chip I'm sure you wouldn't need 1.6v........for one hell of a good submission.:up:

Also, I'm reading on various forums that Freezing doesn't do much that's why I've not connected my new SS yet...........but, what if I did?........ ;)

Thanx

PS
I have a backup chip (yes, only one) that runs Prime Blend and SST's for hours (yawn!) and handles all the usual benches without trouble at 5GHz which I keep for daily use. I use this chip for testing.

Quote:

---

Originally Posted by CryptiK

Donmarkoni in the hiz-ouse! Your poor 990X :D Very nice run. Remember you cant RMA an ES!

EDIT - whats it need for 4.5GHz? Whats your loop consist of and what was the ambient temp for that run? Temps are nice given the vcore & speed.

---

Thanks! I guess that X in 990X stands for extreme :D so I hope it can take the punishment. :hammer: Thanks for the reminder, of course.

I'll need to test at 4.5GHz, but for 4.6GHz HT ON it needs 1.425V and everything else pretty much the same as on the 4.8GHz screenshot.

WC loop goes like this: reservoir - 500lph pump - EK Supreme (restricted) - EK VGA Supreme (restricted) - 3x120 crossflow radiator - and back to res.
Outside was around 9-10°C according to forecast and ambient was around 15°C I guestimate. Standard spring night, t-shirt and shorts.

Quote:

---

Originally Posted by skuldarin

^^
good example of how LinX shouldn't be used as a performance check. i scratched at 103gflops with 4.672ghz, 101 range with 4.710 and Donmarkoni there with 103 at 4.8 AND higher uncore / mem speed at the same time. i tell ya... somethin about those sandy's, they have their way with gflops.

---

Well, yes and no. It all depends on system optimization and feeding CPU enough voltage. I got 100GFLOPS at 4560MHz, so I see your point, GFLOPS should be higher for 4.8GHz.
My theory is when CPU frequency stops scaling with voltage and GFLOPS are less then expected, then you have went over sweet spot.

I5 2500K-3048A446 @ 5.5 GHz
Asus P8P67 Pro
Custom build SS.

http://img855.imageshack.us/img855/2...3175928.th.jpg

Time for 5.5GHz LinX Stable Club.

Quote:

---

Originally Posted by lkozarov

I5 2500K-3048A446 @ 5.5 GHz
Asus P8P67 Pro
Custom build SS.

Time for 5.5GHz LinX Stable Club.

---

WHOA!!

So, there is an SS mark-up on SB's :up:

Now where did I put that rubber matting :)

I need a reality check please................

If my SB wont get into windows at 55x on water would it be useless to hook up my SS because my ceiling is still 54x whatever the cooling?

Was testing 1208 bios for my new 24/7 and decided might as well to submit :)

I7-990X 3045A643

Running
Attachment 113936

Finished
Attachment 113937

Tried to get 4.8, only passed 11 loops. Just 60MHz :shrug:

Core i7 980X [6C 6T] @ | 4.794.2MHz | ASUS Rampage III Extreme | Water | 3005F801

http://dl.dropbox.com/u/22074005/Xtr...tems/4.794.png

Update, I did it. Got my 4.8

Core i7 980X [6C 6T] @ | 4812.8MHz | ASUS Rampage III Extreme | Water | 3005F801

http://dl.dropbox.com/u/22074005/Xtr...ms/4.812.8.png

Updated! :cool:

Quote:

---

Originally Posted by newhit

Fire extinguisher at the ready............

---

:D
Good job! Tight competition, almost HWBot scale! :up: :ROTF:

Quote:

---

Originally Posted by ac001

ac001 | Core i7 990X [6C 6T] @ 4605MHz | 1.468 | ASUS Rampage III Black Edition | Water | -[/IMG]

---

How do you like your ASUS RIIIBE?

Quote:

---

Originally Posted by donmarkoni

Hi guys! It's been a while since I posted here, but I got a new toy and was thinking of joinin' it too.

---

Nice toy, and glad to see you back! :)

Quote:

---

Originally Posted by lkozarov

I5 2500K-3048A446 @ 5.5 GHz

---

Outstanding, congrats on such a huge milestone! :eek: :up:

Quote:

---

Originally Posted by lkozarov

Time for 5.5GHz LinX Stable Club.

---

You'd feel a bit lonely, wouldn't you? ;)

Quote:

---

Originally Posted by McDown

Was testing 1208 bios for my new 24/7 and decided might as well to submit :)

---

Nice job, but please remember to mention the way you're cooling your CPU. I had to look up some other posts of yours to find out.

Ok this is the best I have done with this chip so far :)

Core i7-920 [4C 8T] @ | 4504.5MHz | ASUS Rampage III Extreme | Water | 3949A372

http://img51.imageshack.us/img51/893/45ghzlinx.png

http://img864.imageshack.us/img864/8061/45ghzlinx2.png

YEA BABY!!! I guess I found the sweet spot. Wait a minute how is everyone getting twice the GFlops I am??

2500K
3101A127
Cooler Master Hyper 212

Quote:

---

Originally Posted by zalbard

How do you like your ASUS RIIIBE?

---

I like it pretty much. my old RIIIE died few weeks ago so I got this one. beside the black paint there are not so many differences. I dont have any free pci-e slot for that thunderbolt card, so I can only use the audio function but the quality is great. there is onboard usb3 connector on RIIIBE but I dont have any cable or adapter yet to use it.

there is a great bios feature called "memory recheck". with RIIE and RIIIE I always had the problem that it only detected 10 of 12 GB RAM first boot of the day and I had to reboot manually to detect all RAM. with activated "memory recheck" the RIIIBE will automatically boot twice.

cons are: - there is nothing newer than the initial bios and it has major issues like no temp readings for northbridge and southbridge. - no support for my asus oc station. - it has same problems with dip switches for pci-e slots like the RIIIE. if I want to run benchmarks with single gpu that dip switches are useless and I have to remove gpu 2 and 3 otherwise pci-e slot 1 will only use 8 lanes.

regards ac001

Quote:

---

Originally Posted by little_scrapper

Wait a minute how is everyone getting twice the GFlops I am??

---

You need Windows 7 SP1 and the most up-to-date LinX to benefit from AVX and get over 100 GFlops. :)

Quote:

---

Originally Posted by ac001

I like it pretty much. my old RIIIE died few weeks ago so I got this one. beside the black paint there are not so many differences. I dont have any free pci-e slot for that thunderbolt card, so I can only use the audio function but the quality is great. there is onboard usb3 connector on RIIIBE but I dont have any cable or adapter yet to use it.

there is a great bios feature called "memory recheck". with RIIE and RIIIE I always had the problem that it only detected 10 of 12 GB RAM first boot of the day and I had to reboot manually to detect all RAM. with activated "memory recheck" the RIIIBE will automatically boot twice.

cons are: - there is nothing newer than the initial bios and it has major issues like no temp readings for northbridge and southbridge. - no support for my asus oc station. - it has same problems with dip switches for pci-e slots like the RIIIE. if I want to run benchmarks with single gpu that dip switches are useless and I have to remove gpu 2 and 3 otherwise pci-e slot 1 will only use 8 lanes.

regards ac001

---

Interesting, thanks for the feedback! Memory recheck sounds like a must, it's the first time I hear about it. :up:

P.S. OP's been updated. :)

i'ma lap my 980 and shoot for 5ghz. i'm pretty sure i could do it... we'll see when it's done.

R3E has this option in BIOS 1301. Manual tRL adjustment has been removed since with mem recheck enabled the retraining is sufficient to properly set all channels. I still think rebooting twice is ridiculous and shouldn't be necessary, other boards dont need to, why does Asus? :rolleyes: Sounds like a bandaid fix to me.

Quote:

---

Originally Posted by ac001

I like it pretty much. my old RIIIE died few weeks ago so I got this one. beside the black paint there are not so many differences. I dont have any free pci-e slot for that thunderbolt card, so I can only use the audio function but the quality is great. there is onboard usb3 connector on RIIIBE but I dont have any cable or adapter yet to use it.

there is a great bios feature called "memory recheck". with RIIE and RIIIE I always had the problem that it only detected 10 of 12 GB RAM first boot of the day and I had to reboot manually to detect all RAM. with activated "memory recheck" the RIIIBE will automatically boot twice.

cons are: - there is nothing newer than the initial bios and it has major issues like no temp readings for northbridge and southbridge. - no support for my asus oc station. - it has same problems with dip switches for pci-e slots like the RIIIE. if I want to run benchmarks with single gpu that dip switches are useless and I have to remove gpu 2 and 3 otherwise pci-e slot 1 will only use 8 lanes.

regards ac001

---

Quote:

---

Originally Posted by zalbard

You need Windows 7 SP1 and the most up-to-date LinX to benefit from AVX and get over 100 GFlops. :)

P.S. OP's been updated. :)

---

I have SP1 installed and I just downloaded LinX (is 6.4 not latest?) for that run. Is there some setting i need to turn on because some others have slower clocks than I did and are still getting twice the Flops I am.

Download linX from the main page here. The libraries have been updated.

Quote:

---

Originally Posted by acebmxer

Download linX from the main page here. The libraries have been updated.

---

I downloaded it yesterday through the link provided on the first page. Is that not the right place?

Perhaps it's throttling? 98C is nuts. Other than that, I have no idea...
It looks like it does not use AVX for some reason.
Try an older version of LinX?
Could you have disabled something in UEFI or OS preventing the CPU from using AVX? Do you use an ancient or buggy UEFI, maybe? I am just guessing...

crossg | Core i7 990X [6C 6T] @ 4828.8MHz | 1.456 | ASUS Rampage III Formula | Air | 3045A752

Started LinX and went out and changed the front brakes on my car. Came back in and it was finished 23 loops. Thanks zalbard!

Quote:

---

Originally Posted by zalbard

Perhaps it's throttling? 98C is nuts. Other than that, I have no idea...
It looks like it does not use AVX for some reason.
Try an older version of LinX?
Could you have disabled something in UEFI or OS preventing the CPU from using AVX? Do you use an ancient or buggy UEFI, maybe? I am just guessing...

---

When i first turned it on I had Vcore at like 1.6V and it immediately spiked in temp. I then dropped vcore to the 1.4V that is shown to get temp down. That is also why i let it run for a couple of extra loops. Wanted to keep it an honest run. So no it wasnt throttling afaik.

I have the latest 1208 BIOS. And i have no clue what in the UEFI would cause it. If anyone knows I am all ears.

Creaking past 5400. I think this is it for me. Good fun tho :up:

OOPS! I've just read the first post again. I hope this 4C/4T is alright. :confused::)

Yep, it's fine. It's just forcing LinX to run on n threads (where n is the number of enabled physical cores of your CPU) with HT enabled is neither necessary nor legit since the most recent Linpack libraries came out (you have to run it on all the available threads, which would be 2*n).
Sorry about the confusion.

Just re-done my 5ghz run, at a higher QPI and higher Uncore speed.
Should bode well for a 100+ GFlop run when i turn HT off.

kitfit1
Core i7 990x
6C 12T
5009.7MHz
1.611 Vcore
EVGA Classified E760
Custom Single Stage
3044B119

http://oi55.tinypic.com/xpmrd.jpg

zalbard, can you move my listing to the "Over 100 GFlops" section. Thanks!

SimpleTECH | Core i7 2600k [4C 8T] @ 5000.0MHz | 1.456 | Asus P8P67 Pro | Water | L040B208

http://img163.imageshack.us/img163/1...01456v1.th.jpg

Quote:

---

Originally Posted by SimpleTECH

zalbard, can you move my listing to the "Over 100 GFlops" section. Thanks!

SimpleTECH | Core i7 2600k [4C 8T] @ 5000.0MHz | 1.456 | Asus P8P67 Pro | Water | L040B208[/URL]

---

Huge test ! I think you should try to work a bit on those mem timmings :up:

Quote:

---

Originally Posted by st0ned

Huge test ! I think you should try to work a bit on those mem timmings :up:

---

:wiggle:

Yeah I got my memory cheap, can't overclock a few MHz. Luckily I have some G.Skill PI 2200 stored somewhere that I can use for more GFlops.

Here is a new result from my 990X. Next milestone is 5GHz, but I think it might wait for some natural or unnatural cold. ;)

donmarkoni | i7 990X [6C 6T] @ 4903MHz | 1.554V | Rampage III Extreme | Water | 3026A486

http://img850.imageshack.us/img850/6...x49ghzlinx.jpg

I have a 2600k with 8GB@ 1600mhz with 9-9-9-24 and I'm getting roughly 65 Gflops. Does it seem kinda low? I assume getting faster ram will help?

At what CPU speed, HT on or off?

Quote:

---

Originally Posted by CryptiK

At what CPU speed, HT on or off?

---

My 2600k was running at 5.2ghz with ht on. To think about it I didn't install sp1 yet...

Quote:

---

Originally Posted by aznsniper911

My 2600k was running at 5.2ghz with ht on. To think about it I didn't install sp1 yet...

---

yes you need SP1 installed and make sure you are using the latest LinX with the updated library.

^This.

I'll be gone for 3 days, effective tomorrow. Will update everything once I'm back, so keep posting. :up:

Quote:

---

Originally Posted by newhit

Creaking past 5400. I think this is it for me. Good fun tho :up:

OOPS! I've just read the first post again. I hope this 4C/4T is alright. :confused::)

---

:eek: crazy LinX speeds but why so low Gflops at that speed?

I'm at ~133GFlops at 4800 (4C 4T)
http://tweakers.net/ext/f/jwcZonAaa3...snk/medium.jpg

Quote:

---

Originally Posted by Bullu$

:eek: crazy LinX speeds but why so low Gflops at that speed?

I'm at ~133GFlops at 4800 (4C 4T)
http://tweakers.net/ext/f/jwcZonAaa3...snk/medium.jpg

---

Memory speed makes a huge difference.

Yeah... memory increase HUGE Gflops in LinX!

Quote:

---

Originally Posted by SimpleTECH

Memory speed makes a huge difference.

---

If by 'huge' you mean 1-2 GFlops max, I agree 100%. I have run linx at 4.2GHz with memory speeds ranging from 1600 to 2000 and the difference is always very small. I have many screenshots to prove it, but I don't think this is the correct place for it. Maybe I'll make a new thread.

It certainly doesn't make 73GFlops worth of difference.

I think it's a myth that bad clockers use to excuse their poor performance.

73 - not.
but a few to a dozen.

Mine in 4/8 has 17gflops difference between both!

CPU: i7-2600k at 5.2ghz (1.53V and batch #3103B307)
Motherboard: Asus P8P67 Pro B3 (Firmware 1305)
RAM: Corsair Vengeance 2x4GB DDR3-1866
Cooling: Water
http://i595.photobucket.com/albums/t...202/52linx.jpg

Quote:

---

Originally Posted by Bullu$

:eek: crazy LinX speeds but why so low Gflops at that speed?

---

I was just trying for my max clock. I didn't pay too much attention to my memory. I have much better memory clox at lower MHz. I asked "Zalbard" above if it was all right and he agreed so I'm good to go.:up:

This is only an exercise into bios manipulation and cooling, not my everyday clock.

PS
Nearly forgot.........the rams in are Corsair Dominators rated at what the screen shows.

PPS
I'm not sure (yes, I'm really not sure....... as in don't know) why ppl are consentrating on the ram. I will say that the Corsair *seemed* to make the difference whether I got over 5.4 or not. Obvioiusly, the fact that I needed almost 1.7v is more the main difference although the cpu is still ok but maybe not in the long term.


Lower value..............

I will test my Baby with Linx the next days :up:

http://www.abload.de/thumb/5.4ghz1.448vcoreace5k6f.jpg

i7 970 es
Watercooled

Try again, this time using all of your memory channels... lol.
You should also be over 95GFlops with a config like that.

Quote:

---

Originally Posted by Sanousie

i7 970 es
Watercooled

---

Quote:

---

Originally Posted by zoson

Try again, this time using all of your memory channels... lol.
You should also be over 95GFlops with a config like that.

---

Very hard with such a low UnCore speed...

Quote:

---

Originally Posted by zoson

Try again, this time using all of your memory channels... lol.
You should also be over 95GFlops with a config like that.

---

I would if i was running triple channel ram, but i dont have any 3 matched pair, i get 94gflops at 4.4ghz with a good uncore speed. Only have 2x2gb patriot 2000mhz Hypers, which do 2000mhz cas7 and to be honest, triple channel over dual channel has no performance gains when it comes to gaming (which is what i use this pc for).

But for some reason this ES has a very weak IMC, nothing like my 950 which needs 1.4v for 2000mhz Ram/ 4000mhz uncore.

Quote:

---

Originally Posted by needforspeed

Update 3:
Can I be the first Gulftown to Join the: 5GHz Club <Over 100 GFlops> (Sandy Bridge and others)……;)

needforspeed| Core i7 990X [6C 6T] @ 5002.9MHz | 1.640 | ASUS Rampage III Extreme | Water | 3045B094

I have my eye on 5100Mhz.. but don’t know if I have the rocks.. to take the Vcore any higher.. although the temps are fantastic..

---

Good job!

Any idea if there's going to be a category for EVGA SR-2 owners and using a pair of 6 core CPU's(12 cores/24 threads) and 24 GB of ram onboard?


I've started linx using a 48194 problem size and using 17.75 GB or ram and it's been nearly 20 minutes and hasn't finished the first pass yet....:ROTF:


So i'd kinda hate to need to run it like this for over 400 minutes( about 7 hours) to complete 21 passes...:p:


Yay, first pass completed in just under 20 mins...:rofl:

Quote:

---

Originally Posted by Shadow0001

Any idea if there's going to be a category for EVGA SR-2 owners and using a pair of 6 core CPU's(12 cores/24 threads) and 24 GB of ram onboard?


I've started linx using a 48194 problem size and using 17.75 GB or ram and it's been nearly 20 minutes and hasn't finished the first pass yet....:ROTF:


So i'd kinda hate to need to run it like this for over 400 minutes( about 7 hours) to complete 21 passes...:p:


Yay, first pass completed in just under 20 mins...:rofl:

---

This is odd. It should be way faster.
Have you tried switching CPU configs (NUMA and such) to see if it affects the performance?
It takes around 2H for a good Gulftown run with 12GB of RAM (link). You have twice more memory, but also twice more processing power. So I'm not sure what's up.
Mind posting your current run?
12GB LinX run could be an option in your case, I guess... Need more info! :yepp:
You'd fit into "Over 100 GFlops" category, that's not too bad as is. I'll think about it, though. :)

Quote:

---

Originally Posted by zalbard

This is odd. It should be way faster.
Have you tried switching CPU configs (NUMA and such) to see if it affects the performance?
It takes around 2H for a good Gulftown run with 12GB of RAM (link). You have twice more memory, but also twice more processing power. So I'm not sure what's up.
Mind posting your current run?
12GB LinX run could be an option in your case, I guess... Need more info! :yepp:
You'd fit into "Over 100 GFlops" category, that's not too bad as is. I'll think about it, though. :)

---

Keep in mind though that with larger memory amounts also comes options to calculate larger problem sizes too, wich aren't even available to other LinX users if they don't have 24GB of ram onboard in the first place, so that leaves pretty much everyone running a dual channel memory setup out....


I'll have to see if NUMA is posing a problem performance wise, but the only way i get well above 100GF/sec is by calculating small problem sizes and using much lower memory amounts too...Here the picture using a 48194 problem size at 17.7 GB of ram:

http://i765.photobucket.com/albums/x...s/DSCN0062.jpg


CPU cores floored to 100% at nearly 4.2 Ghz, with each CPU using only 1.35 volts and the temps have a hard time exceeding 60*C with the fans set at 60%, so the system is actually pretty silent, but that's to be expected with water cooling....

Yeah, as you see, GFlops is way too low. Not sure what's causing this...
By the way, just use PrintScreen the next time, much easier, IMO. :)

Quote:

---

Originally Posted by zalbard

Yeah, as you see, GFlops is way too low. Not sure what's causing this...
By the way, just use PrintScreen the next time, much easier, IMO. :)

---

I made 2 more runs, with numa disabled using the largest problem size and maxed memory amount, and a second run using a more standard problem size and about 5 GB of memory and the results were surprising to say the least:


http://i765.photobucket.com/albums/x...roblemsize.png


This is the large problem size with maxed memory usage and the overall performance fell another 50% with numa disabled compared to my last run with it enabled, so i thought that numa is a requirement for best performance and should not be disabled in the bios at all....Until i did the smaller run that is:


http://i765.photobucket.com/albums/x...eandmemory.png


The last time i ran this at those exact settings, i had about aa 90 GF/sec result, and with only numa being disabled, it shot up to 106 GF/sec, so it seems that at least for dual socket systems and when using large memory amounts,Numa is mandatory, but when using LinX with smaller problem sizes and much less memory, disabling numa actually improves performance overall...Weird, but there you have it.


The thing that i noticed when running the large problem/maxed memory size amount though, was that the CPU's were running at lower temperatures for most of the 30 minute run(think 15~20*C lower here), and only occasionally having peaks every few minutes, while running the smaller problem size and less memory, the temperatures were consistently higher for the duration of that entire pass as there's obviously less memory accessing going on...


It gives the impression like the CPU's are spending most of the time in that large problem/max memory scenario accessing memory, rather than spending it on the actual problem calculations(where the CPU temperatures also peak), hence the much lower GF values overall....Just something curious i noticed.

Made another run with numa enabled using the same settings as the second picture in my previous post:



http://i765.photobucket.com/albums/x...umaenabled.png


There it is 90 Gflop with numa enabled and 106 GFlop with it disabled for these specific settings....The more the application uses memory along with an O/S that recognizes Numa, the more usefull Numa becomes performance wise, at least in dual socket systems.


Wich brings up a good question, is windows 7 64 bit professional a Numa aware O/S?...:confused:

I get inconsistent performance in Linx (read as lower GF and less core temps) when I have a problem with low vtt voltage, bad memory timings, a bad RTL setting for one of the memory channels, too low of PCIe freq, incorrect IOH voltage, or even wrong CPU PLL among many other things...

you my friend, are dealing with the sole reason I haven't submitted my 4.5Ghz+ run.... Cuz I wasn't happy with the gigaflop performance. The highest I've seen with my W3520 is 68GF and I'm fairly certain I can get higher and hit it more consistently. The problem is, just when, I'm almost there, Asus will release a new BIOS for my R3G and all my settings won't carry over or I'll learn something new about tweaking a setting in the BIOS like memory RTL or CPU differential amplitude and I'll have to start the benching process all over again. Its fun to me and I'm in no rush to post a result in this thread for that reason.

Quote:

---

Originally Posted by R37ribution

I get inconsistent performance in Linx (read as lower GF and less core temps) when I have a problem with low vtt voltage, bad memory timings, a bad RTL setting for one of the memory channels, too low of PCIe freq, incorrect IOH voltage, or even wrong CPU PLL among many other things...

you my friend, are dealing with the sole reason I haven't submitted my 4.5Ghz+ run.... Cuz I wasn't happy with the gigaflop performance. The highest I've seen with my W3520 is 68GF and I'm fairly certain I can get higher and hit it more consistently. The problem is, just when, I'm almost there, Asus will release a new BIOS for my R3G and all my settings won't carry over or I'll learn something new about tweaking a setting in the BIOS like memory RTL or CPU differential amplitude and I'll have to start the benching process all over again. Its fun to me and I'm in no rush to post a result in this thread for that reason.

---

I suppose it's possible considering that in my case,pretty much everything gets doubled up, from the extra CPU to 6 more memory slots for that CPU and all the Bios settings relevant to that CPU and memory, so there's more chances for strange stuff to happen....


On the bright side, i've had Gflops results using settings seen in this thread, wich the majority use smaller problem sizes and memory amounts, where it reaches over 150 Gflops, so slow it isn't....:ROTF:

Today I switched out my memory from 8GB to 4GB but they had tighter timmings but I got a lower Gflop, any reason for this?

Just a couple of reminders...

Quote:

---

Originally Posted by SimpleTECH

zalbard, can you move my listing to the "Over 100 GFlops" section. Thanks!

---

Just noticed your LinX test is finished on the screenshot, had to pull it. :(
Please sort this out, mate! :up:

Quote:

---

Originally Posted by Sanousie

i7 970 es
Watercooled

---

Nice job! :)
Need your motherboard info, you forgot to list that. Then I can add you right away.
By the way, amazing job on that Thuban in your sig! Clearly a golden chip! :up:

Quote:

---

Originally Posted by aznsniper911

Today I switched out my memory from 8GB to 4GB but they had tighter timmings but I got a lower Gflop, any reason for this?

---

Larger problem size usually means more gflops if everything else is equal.
For example, my 12gb kit at 9-9-9-25 gets ~86GFlops at 2GHz with cpu at 4.2, while my 6gb kit at 7-9-7-20 gets ~83GFlops at 2GHz with cpu at 4.2. The difference is the problem size with 12gb is ~9GB, and the problem size with 6gb is only ~5GB.

hi all ..
this is my a new z68x ud7 board and 2600k

http://img190.imageshack.us/img190/8558/lnxs.jpg

Are there any comments about the voltage ?
thanx .

LinX has to be running (calculating) when you're taking your screenshot per the rules in the OP.
You obviously want to let LinX use all the memory for maximal stress.
Do you really need to overvolt the PLL? vCore seems quite high, but this might be required, depends on the chip...

sorry .. I will test again After adjusting the voltage .

Quote:

---

Originally Posted by zoson

Larger problem size usually means more gflops if everything else is equal.
For example, my 12gb kit at 9-9-9-25 gets ~86GFlops at 2GHz with cpu at 4.2, while my 6gb kit at 7-9-7-20 gets ~83GFlops at 2GHz with cpu at 4.2. The difference is the problem size with 12gb is ~9GB, and the problem size with 6gb is only ~5GB.

---

Thanks! Guess I'm going back to 8GB of slower memory.

How can I be sure that AVX is enabled? I have installed SP1 and am using the newest version of LinX. My GFlops are between 55 and 60, but I expected them to be in the 80 to 100 range. Cpu is 2500k at 4GHz, mem at 1067.

Quote:

---

Originally Posted by dr_sharp

How can I be sure that AVX is enabled? I have installed SP1 and am using the newest version of LinX. My GFlops are between 55 and 60, but I expected them to be in the 80 to 100 range. Cpu is 2500k at 4GHz, mem at 1067.

---

Did you update the LINPACK binaries in the LinX folder? The original ones don't have AVX support, you need more recent ones from Intel.

Quote:

---

Originally Posted by danielkza

Did you update the LINPACK binaries in the LinX folder? The original ones don't have AVX support, you need more recent ones from Intel.

---

Excellent! Thank you for the link. Now 106GFlops.

Should these files be updated anywhere else in Windows?

Updated LinX with the new Linpack libraries (10.3.3.006 -> 10.3.4.007).
Replaced the link in the OP.

Quote:

---

Originally Posted by dr_sharp

Should these files be updated anywhere else in Windows?

---

No.

Quote:

---

Originally Posted by zalbard

Just noticed your LinX test is finished on the screenshot, had to pull it. :(
Please sort this out, mate! :up:

---

:doh: My bad mate. I remember reading the OP and seeing that but after getting higher GFlops than before, it must have went past me. lol

I'll resubmit a new run with my G.Skill PI 2200. Want to see if I can squeeze out some more GFlops than before. :up:

How does LinX calculate GFlops? I am curious because I am able to achieve 112 GFlops @ 4GHz while others cannot achieve 112 GFlops unless they are at 5GHz... :confused:

with 4threads u get higher GFlops than 8threads.

And you have AVX.

Quote:

---

Originally Posted by kimandsally

Here's my best shot took a while to get here

http://img442.imageshack.us/img442/8...ram20pa.th.jpg

---

My score seems to have been removed from the database is there something wrong?

Quote:

---

Originally Posted by kimandsally

My score seems to have been removed from the database is there something wrong?

---

Yes

Quote:

---

Originally Posted by zalbard

Please make sure that LinX is running / calculating (21st or higher number loop) when you're taking your screenshot (on at least one of your screenshots).

---

Quote:

---

Originally Posted by Bullu$

with 4threads u get higher GFlops than 8threads.

---

So are you saying HT is BS?

Seems a bit rough at the time I think it met the rules so to change them later is not really fair, in my case I do not have the rest of the rig to do it again.
It's like changing the result of a race after a new set of rules are made a year later.
Maybe next year it will be changed to 25 then all will be cancelled who knows:mad:

13/12/2009 was the date mine was submitted the rules changed, (Update: 14 Dec, 2010) a year and a day later and said it would be for all new submissions, looking through the ones done at the time mine was there are many more that will need removeing or mine should be put back.

Please make sure that LinX is running / calculating (21st or higher number loop) when you're taking your screenshot (on at least one of your screenshots). You obviously want a CPU-Z window with 4.5GHz+ (5GHz+ for Sandy Bridge) frequency on the same screenshot, as well.
This rule will be enforced for all the new submissions.

Quote:

---

Originally Posted by kimandsally

My score seems to have been removed from the database is there something wrong?

---

Odd, probably got lost in an OP formatting update... :confused:
Sorry! :(
I'll add it back in a moment.

Just as a note, all the submissions get to stay as long as they follow the rules at the time of posting.

P.S. What's your Mobo? I put it down as E760. And do you know your batch number?

Quote:

---

Originally Posted by zalbard

Just as a note, all the submissions get to stay as long as they follow the rules at the time of posting.

---

Oh, OK. That seems fair :)

Quote:

---

Originally Posted by dr_sharp

So are you saying HT is BS?

---

The problem is LinX keeps the math units fully occupied: there are no time slices left for the extra 4 threads to work with, and therefore, HT may actually cause performance to drop. That doesn't mean it doesn't work for many other situations with different load patterns.

hello

^^ did you not read thread title? LinX stable not OCCT stable.

MIVE, water, the batch # is: 3103B373

http://img29.imageshack.us/img29/6863/59335176.jpg

Good job, and welcome to the club! :)
Don't suppose you can look up the batch number? I'm guessing it's probably too late, hehe.

Edited the submission post to include the batch number. I have not sold it, yet :D Max multi was 52x :(

http://img171.imageshack.us/img171/2815/snap5a.jpg

AsRock Z68 Extreme4
Titan Amanda Tec ( peltye off )

Quote:

---

Originally Posted by machKim

http://img171.imageshack.us/img171/2815/snap5a.jpg

AsRock Z68 Extreme4
Titan Amanda Tec ( peltye off )

---

You've got to be kidding me...