BenchZowner

We are live!

Here's the channel's link - http://www.livestream.com/benchzowner p.s. go go go tin

Crio and I plan to bench 'tomorrow' with my i7 950 and various graphics cards ( mainly 5870, GTX 280, 3850, 9600 and possibly some 8800GT 512MBs ) on LN2 & dry ice. Chris tossed an idea to broadcast the session on the net via livestream, and I agreed I'm currently setting up the account & channel, and will post again later on with a link to the channel. Live benching & nice dance/house/progr music altogether, join us

I picked up some dry ice today ( 15.5 lbs / 7kg ) to do some VGA pretesting as a preparation for the next LN2 session. I decided to use an i7 920 just to grab a few more HWpoints for the sake of it : I started with low CPU clocks and did some VGA OC digging, hitting 1355MHz GPU @ AquaMark 3 at... hold on... 1.672V ( probably stable with less voltage, I was just rushing and wanted to be ok in terms of ext. param. that could affect the stability ), 1300MHz GPU @ 3D Mark 03/05/Vantage at -60C to -65C. Then I pushed the CPU a lil' further, and tried some 3Dm03/05/Vantage just for fun. 40069 3D Mark05 - http://service.futuremark.com/compare?3dm05=5135996 104126 3D Mark03 - http://service.futuremark.com/compare?2k3=6262310 P23075 3D Mark Vantage ( Performance Preset ) - http://service.futuremark.com/compare?3dmv=1858927 I think I'm ready for my next LN2 session now

The same. All FutureMark benchmarks feature a save score and submit later function.

In wPrime 32m dual channel is as fast ( if not faster ) than triple channel.

Had a short session today ( didn't went really well, stupid kneaded eraser insulation on the VGA didn't work well : ), the scores aren't tweaked at all, so don't give me a lecture about the efficiency. I had to tweak the CPU settings from the beginning since I can't remember the settings I used the other time : I'm BCLK limited ( and NB temp limited probably, 69C @ boot lol ) and got lots of stuff to check ( ideally on a -130C cascade : ) to see if I can get past that stupid BCLK wall, check the settings & stuff affecting the CBB & CB ( Cold Boot Bug & Cold Bug ) and tweak the RAM sub-timings & cool the RAM with LN2 as well. The CPU can do 5300+ 3D Mark06, 5500+ AquaMark 3 & Pi 32m, so unless someone can hook me up with a 5.7GHz+ 32m / AquaMark 3 - 5.5GHz 3D Mark06 chip, I'm fine ( somebody ?! ) Will come back soon with proper scores with CPU + VGA/s + RAMs @ LN2 ( maybe LN2 on teh chipset too )

Just warming up : Monstrukill ... CPU temp got under -55C and lost the... +60MHz validation & Pi1M. Don't worry though, I'll be back soon

Nice start man If you post a review of them make sure you switch the sticker on the DIMMs with the one found inside the blister package I just finished cleaning up my Classified and put some ice cubes in the freezer ( my good CPU needs to be close to 0C to hit any BCLK over 212MHz in dual/triple channel ) and then I'll join the club

Thanks Petri I will try to get more LN2 next week and push things the proper way this time. Here's the CPU info in case you guys want to hunt down the same batch of 950s:

I can't believe this is true, but fortunately ( for me : ) it is A few days ago... ok, ok! 2 days ago, I tested a Core i7 950 that did 4.8GHz SuperPi 32M on ice ( the regular one, not dry ice ) [ CoreTemp 37C/35C/36C/33C ] and the results were very encouraging ( 5.4GHz-something SuperPi 1M, 5.3GHz AquaMark 3, 5.16GHz Pi32M @ -58Cpot_temp, wPrime at 5.05GHz ). Today I had a few liters of LN2 ( 12 to be precise ). Yesterday I started preparing some Windows images with Norton Ghost ( something that made me waste a lot LN2 today :shakes: ) and then recovered the HDD to an all-around Windows XP Pro SP3 installation. I wanted to test the CPU, but I got tempted to do some AquaMark 3 with a LN2 cooled 5870, so I prepped both the mobo & the VGA for LN2. I started with CPU on LN2 and an empty Tek-9 3.0 "Fatbody"-cooled 5870 @ stock to check the overclocking capabilities & the "coldbug/coldboot" of the CPU. I didn't try hard, neither really paid any attention ( I had an old friend here and we had some good laughs during this benching session ) so there might be more or even more than a lot of OC headroom left ( we shall find out soon ) After fooling around in Windows XP I decided to load the Windows 7 x64 image to get some higher AquaMark 3 & 3D Mark03 numbers, but that wasn't the best of ideas I must admit : Norton Ghost 14 had problems restoring my image, and after 2 tries and lots of waiting, I decided to give Windows Backup a go. Finally, after 3 minutes or so the image was finally restored and we were ready to go. Well... that's when I realized that I lost a few scores & screenshots because I forgot to transfer them from the HDD to my USB stick... I was bored and didn't mod the 5870, so it was SmartDoctor & Afterburner. Well, the result was 1287MHz / GDDR5-5224 AquaMark 3 @... -179C ( -180C = coldbug : ) @ 1.45Vsomething. I didn't push the UnCore, nor the RAM ( slacky 8-8-8-28 80 tRFC and 2/8 divider, B2B 12(!) ) just the CPU clock I lost a nice 393k AQ3 with the card at 900/1200 and 5.54G CPU while saving the screenshot ( damn GPU-z lol and lack of LN2 ). I also benched SuperPi 1M @ 5.5GHz / DDR3-2400 8-8-8-28 Triple Channel just for fun I really need a cascade to spend some time testing the CPU and finding the best clocks/timings & settings I'll leave you with a few screenshots ( well... some of the saved ones ) One of the cb lost screenies A cr*ppy AQ3 ( full screenie, yay ) A... not so tweaked ( just LSC ) SuperPi 32M on dice 5.57GHz 4cores/8threads CPU-z validation ( pending, x86-secret is down ) I shall be back soon

LOL Isaie, it's DANCING IN THE RAIN, not BENCHING in the rain You should leave the system exposed to the rain, and hybrid cool the system ( water...cooling & air-cooling in the mix ). Now all we need is a... RickRoll

massmansucks

MIP runs are accepted on ORB ( doesn't detect it ) AFAIK. So yes, it is possible.

Every CPU will have performance issues if it's unstable or throttling due to high voltage or temperatures.

Shouldn't as long as you don't motivate it to be a jerk if you keep it cool and fed with the appropriate voltage and BIOS settings it should be fine and play along.

That sounds like an unstable overclock. nearly 1GHz higher clock and less performance indicates that.

We need more information. Which preset did you run ? The performance preset ? How much did you OC your CPU ?

haha, never thought of that. Disabling post-processing is so lame.

D-PP ? Say what ?

AMD ? Who ? ( just kidding ) A drop-down menu to select AMD or Intel would be nice, and useful to quickly tie the batch numbers with the manufacturer instead of linking them with the processor name.

I'd say a single alphanumeric value ( e.g. L741A485 ) And don't limit the input characters, because there are longer batches out there, like the ones ending with a T ( e.g. L741A485T )

No Andre he's not talking about OVP ( obviously you misread that ), he wants the OCP ( OverCurrent Protection ) mod. OCP should be kicking in at around 1.15GHz GPU regardless of the voltage applied to the GPU ( I could be wrong, I haven't tried any 5870 on LN2 yet )

You didn't get my point man. Let me make this clear ( although I wouldn't like to mention this at all, but... ): I'm a smart cheater ( that's an example, not speaking about myself lol ), so I visit HWbot and check the wPrime 1024m rankings. I found a Dual Xeon system clocked at 4050MHz with X amount of RAM clocked at Y MHz, scoring 100s. I find a screenshot of wPrime 1024m with two CPUz windows open, I open it up in paint, change the results in the wPrime window, change the details in the CPUz windows to nearby/similar frequencies and then post the score. Yes, here's my Dual Xeon system at 4010MHz with X amount of RAM clocked at Y-z MHz, scoring 101s

You can see the test details on speedhacked 3D Marks and spot a cheat. With a plain screenshot, how will you be able to detect if for example I edit the wPrime v1.55 result window and copy paste edited CPU-z windows with altered info ?