Massman

Board will be in full retail in June, if I remember correctly. That's probably why it's not available yet

Not yet (blame the CPU), but it will certainly be there. If you need the mod for OCP ... let me know.

Because I get so so many requests, I'll just throw the MSI Afterburner Extreme software in here. This is the tool that allows higher Vgpu for the extreme overclockers under us. Beware however ... with great power comes great responsibility: do not use this application if you have no clue what you are doing. I'm only going to say this once: if you break you card because you've been using excessive voltage on air cooling, that is YOUR OWN FAULT so don't even bother to complain about the software. There's a lighter variant of the software available as well here: LINK MSI Afterburner 1.51 Extreme: LINK (R5870 compatible) MSI Afterburner 1.6b6 Extreme: LINK (GTX400 series compatible)

Around 1.80V ... 1.85V or 1.9V just adds more instability in the equation :-/

The voltage isn't visible in CPU-Z for some hardware/software related issue. MSI's looking into the issue right now. Whaaaa? People are now bashing me because my keyboard looks. This is high-school all over again . Quite frankly ... it's the one keyboard I never have issues with.

Three hoorays for FM's update system? Nice one Rick ... I'm happy you're actually posting this in a public forum and not holding the info for yourself!

All feedback was sent and today I received a new bios. The biggest issue of the previous bios was that the default B2B was set to 13(!) which allowed rediculous memory frequencies, but coming at no performance whatsoever. To give you an example: 2400CL8 was giving me 16k bandwidth. The new bios comes with a default B2B at disabled (or 0), which makes it clock a little less high, but comes at a higher performance level. Oh, and I managed to -finally- flash my OC Dashboard. It's now perfectly working with this mainboard I've focussed on memory and BCLK this night - only air cooling. First some picture of my dirty workspace. Now some results. 230MHz BCLK CL6-7-6 (220MHz air // 240MHz phase with X58A-UD7) 2100CL7-7-7 air Findings and Remarks: Again, I've send my feedback to MSI. Just some things I wanted to share with you. - VTT over 1.430V set in bios (~1.48V real) doesn't post. You can increase VTT with the OC Dashboard in Windows no problem - 230MHz BCLK is straight from boot to windows and 32M stable, however to get it there you need to follow this routine: --- boot into bios --- set all correct multipliers and memory timings I want to use with 230MHz BCLK --- set BCLK to 195MHz --- save bios --- boot into bios --- set BCLK to 230MHz --- soft reboot = I can boot at 230MHz into windows - 210MHz 2:10 is perfectly stable, but not straight from boot to windows. Boot @ 210MHz = no loop stable; boot @ 195MHz and increase to 210 in windows = fully stable - 230Mhz BCLK was done with 105Mhz PCie, 1.48V VTT and 1.3V IOH. Nothing tuned, nothing tweaked. - At the moment, MEM/UNC ratios lower than 1/2 doesn't post. This should be fixed in the next bioses. - The V-switch works differently (imho better) than on previous boards. Now it doesn't increase the voltage anymore, but removes the voltage securities in the bios. Eg: vswitch off = max Vcore of 1.7V, vswitch on = max vcore of 2.3V Tomorrow I will switch back to the Single Stage to see if I can get more from BCLK and memory. For 3D comparison I will have to wait untill a new high-end VGA lands on my desk. Let me know if I need to test things in particular or make some pictures or so. //edit: before someone kicks in the open doors - no, the efficiency of the 32M is not superb, I didn't care. No, the results are not extraordinary, they are where they should be.

Seems like I forgot to post the LN2 testing The quick tests to check for stability. Didn't have a propper 3D card at my disposal, so only two Wprimes from that session.

We will cut the prize in X equal parts, with X = the amount of people who hit the target. J/k: 2) If multiple people hit the target, a lucky draw between the people who hit the target will determine who wins the prize"

Ouh, nice one!

Who knows ... maybe at 6.4G 3D

Are you using 980X or any other very recent VGA card? This might be due to the FM Systemscanner-thing not working properly

Yes.

Yeah. Gautam pointed out that the Tcl will lower the RTL timing, so that might hold a performance increase

Yes

The main stages are indeed AMD limited, the precision challenges are not platform limited.

I have just the one

Nice! Seems like I'm missing out on a lot of performance because of the ganged mode. Roughly 1.5K MB/s read bandwidth :-/

The final runs for the upcoming NB/MEM scaling article here at the bot are done. Expect something to appear on the frontpage this week. Until then, here's what I was working on last night.

You're the Andre of ancient hardware

The question is ... how efficient is it. Any gains to be seen from CL6 to CL5?

I'll update the Xpower LN2 stuff today.

That's just madness ... 2000CL5!!

Andre didn't pull his score.

Rig picture. Next week you can expect a small article on the memory and memory controller frequency scaling ...