thuban mem controller testing

[[XC] Oj101]

Any reason why the HT link has to be above 2G for better performance?

In regard to the efficiency: I suppose that Thuban still has the same issues as Deneb has when it comes to memory data transfer. AMD people confirmed that an elevated memory frequency is no use if you don't increase the IMC frequency accordingly. An appropriate ratio would be IMCRAM as 3:1, which would mean a 6GHz IMC if you want to make propper use of 2G memory.

Why 3:1? I was always under the impression that it was 2:1.

Hmm...

Can I ask what bios? I just booted into windows @ 1775 CL7 on my Gigabyte 890GX board and I'm currently at loop 6 in 32M

It's Micron GTS @ 2.0v

Why are you running at stock voltage? I run my GTS at 1.9x to 2.0x volts 24/7 for many months now

[massman]

This will be corrected in the next BIOS but the main reason for maxing HT is to sync with NB speed (which you obviously want as high as possible on this platform) to keep the current memory table at optimum settings. Right now the rules are set to ensure stability/compatibility instead of maximum performance. Now that the AGESA code is stable for Thuban, we will set different rules when the BIOS "sees" performance settings being utilized.

So, if I understand correctly, the sole action of raising the HT link isn't gaining any performance (which got me confused in the first place), but it makes sure the memory tables remain at optimum settings. Something in lines of making sure the system sees the NB/HT link combination as 'close to default' so it enables the high performance tables.

Bios is tuned for optimal performance at default settings, and tuned for overclocking headroom at non-default settings. Hey, that makes sense even for me!

Thanks for the heads up!

Why 3:1? I was always under the impression that it was 2:1.

Might be the difference between 'okay to run' and 'saturation'. 2:1 might be delivering enough bandwidth to have some sort of efficient subsystem and 3:1 might be the saturation level where the bandwidth provided by the memory is used to the fullest.

I should be getting a C-IV this afternoon, I'll run the memory tests to check this out .

[BeepBeep2]

Why 3:1? I was always under the impression that it was 2:1.



Why are you running at stock voltage? I run my GTS at 1.9x to 2.0x volts 24/7 for many months now

I don't know... I just use what I need

[bingo13]

So, if I understand correctly, the sole action of raising the HT link isn't gaining any performance (which got me confused in the first place), but it makes sure the memory tables remain at optimum settings. Something in lines of making sure the system sees the NB/HT link combination as 'close to default' so it enables the high performance tables.

Bios is tuned for optimal performance at default settings, and tuned for overclocking headroom at non-default settings. Hey, that makes sense even for me!

Thanks for the heads up!



.

You got it, in fact in some of the numbers around here you will see NB at 2100 and HT at 1900, for some odd reason not having those two in sync will throw off the table that is used and memory performance will drop. We figured it out and it should be updated in the next BIOS. This only occurs in auto mode and clocking HTT (bclk) up, but I wanted to mention it so users can experience the full potential of the board.

We will tighten up the tables in the next release. Depending on NB speed and memory utilized the 0505 BIOS might actually be a little faster on the memory side but then some of the disk/3D/four-dimm enhancements are lost.

[Lightman]

You got it, in fact in some of the numbers around here you will see NB at 2100 and HT at 1900, for some odd reason not having those two in sync will throw off the table that is used and memory performance will drop. We figured it out and it should be updated in the next BIOS. This only occurs in auto mode and clocking HTT (bclk) up, but I wanted to mention it so users can experience the full potential of the board.

We will tighten up the tables in the next release. Depending on NB speed and memory utilized the 0505 BIOS might actually be a little faster on the memory side but then some of the disk/3D/four-dimm enhancements are lost.

With very little time I had yesterday playing around CH-IV + 1090T I've noticed setting NB and HTT in sync gives best memory latencies and write speeds.

One thing puzzled me. BIOS is reporting CPU/NB voltage @1.252V, but AOD in Vista says it's 1.15V... A bug in 0707?
Also why enabling Line Calibration to cure Vdrop can cause instabilities? I just enabled it for my last boot very late last night and it worked correctly eliminating Vdrop. I ran out of time to check if stability is where it should be...


Can't wait to get back home tomorrow and dig under the hood of my new toys

[massman]

Been testing this board since yesterday-night. Some issues with unganged mode at the moment.

2100 CL7-7-7-21 (but ganged)

[radaja]

WOW,that is insane massman

[Riska]

In my test of the 790FX chipset the ganged mode has a little better performence than unganged havent tested it on the 890FX chipset yet

[Lightman]

WOW!

I'm starting from DDR3 1800MHz and will move upward once I get hang of DDR3 timings. 2100 CL7 is crazy on AMD
What chips Massman?

[massman]

In my test of the 790FX chipset the ganged mode has a little better performence than unganged havent tested it on the 890FX chipset yet

Depends on the benchmark, really. I think 2100 ganged should be better than 1920 unganged, though .

WOW!

I'm starting from DDR3 1800MHz and will move upward once I get hang of DDR3 timings. 2100 CL7 is crazy on AMD
What chips Massman?

The super-dooper ultra-expensive Corsair Dominator GTX2

[Sanousie]

Sorry to hijack the thread, im was just wondering what would be the max safe voltage for the nb on a 1055T under a phase change?

Cheers.

Hmm...

Can I ask what bios? I just booted into windows @ 1775 CL7 on my Gigabyte 890GX board and I'm currently at loop 6 in 32M

It's Micron GTS @ 2.0v

What bios version what that f7b? may i ask the trfc settings you were using what about the rest of the timings, If you can provide this info much appreciated, trying to tweak these g.skill trident 2000mhz.

[richierich]

I'd like to run the Dominator-GTs but the heatsinks are too high

[BeepBeep2]

Sorry to hijack the thread, im was just wondering what would be the max safe voltage for the nb on a 1055T under a phase change?

Cheers.



What bios version what that f7b? may i ask the trfc settings you were using what about the rest of the timings, If you can provide this info much appreciated, trying to tweak these g.skill trident 2000mhz.

F7B. The board locks up at 940/1880, however I've gotten 932 stable. I'll post a BIOS shot in a bit.

[Riska]

just unscrew the little alu sink at the top of the GT mem then you will be fine with the mem

[Smartidiot89]

Been testing this board since yesterday-night. Some issues with unganged mode at the moment.

2100 CL7-7-7-21 (but ganged)

Bull

Would love to see read speed with higher CPU/NB clocks under LN2!

[massman]

You mean like this? Think I'm being limited by the Ganged mode, though.

[[XC] Oj101]

Oh my sweet mother of AMD, you're nanoseconds away from my best with BH-5!!!

[i found nemo]

nice nb there!! lol

[BeepBeep2]

I think that latency reading just made me crap my pants

[Lightman]

That's the bad CPU you've killed but it did Jesus on you later on Massman?

Nice!

[massman]

Yes, Lightman :-).

Next target is 2000CL7 unganged.

[[XC] Oj101]

No. 2000CL6 please

[massman]

Either need a different memory kit, or Leeghoofd to agree on 1.8V+ on the GTX2

[Sanousie]

Those speeds are insane, Very nice massman

[Zeus]

Either need a different memory kit, or Leeghoofd to agree on 1.8V+ on the GTX2

I'm sure he's willing to sacrifice his ram for the sake of some WR's.

Nice bandwidth and latency there mate!