ASUS Striker II Extreme

[Qwik]

I have been able to get my MEM to 2000MHz but I have to run unlinked and CPU at 1800MHz, can't get 2000FSB no matter what settings I try either.

[Roger_D25]

I finally got core 1 sorted, then core 4 failed.
GTL settings seem to be the answer here. Currently Priming with VTT on Auto, and -41 on GTL, been OK for an hour! Must admit, haven't played around with the different bios files. Currently running 9 x 450MHz. Also I'm not sure this board likes OCZ memory, I can't get anything higher than 1800MHz with my 2000MHz rated memory, even if I just use two sticks. I'm disappointed that I can't even get this board to post properly at 2000FSB, strange that in the reviews they seem to be able to do this no problem - prolly a hand-picked board!! Gonna have a play around today and see what happens.

Tyke - Sorry to hear about that, you spend all this time trying to get one core stable and as soon as you do a different core fails, lol! I think your very right about GTL's being the answer, especially at the higher FSB speeds your running at the moment! The more I read about how GTL settings on this board work the more I think they are important. The only detailed information I've found regarding the GTL's on this motherboard in particular is Kris's article over at Anandtech (great article by the way)!

This article explains very clearly how to first calculate the relationship between VTT and GTL's, then how to convert that setting into mV which the Striker II Extreme uses. Using this information can get you very close to where you want to be, then its up to experimentation to fine tune them and get things stable.

Does anyone remember what page or what post # had the included GTL voltage table in it (I believe it was this thread but I could be wrong)?

Qwik - I assume you were using your QX9770 when you were testing the settings above? With that said, although you mentioned you haven't been able to run at FSB2000Mhz how high can you go?

[Tyke]

Tyke - Sorry to hear about that, you spend all this time trying to get one core stable and as soon as you do a different core fails, lol! I think your very right about GTL's being the answer, especially at the higher FSB speeds your running at the moment! The more I read about how GTL settings on this board work the more I think they are important. The only detailed information I've found regarding the GTL's on this motherboard in particular is Kris's article over at Anandtech (great article by the way)!

This article explains very clearly how to first calculate the relationship between VTT and GTL's, then how to convert that setting into mV which the Striker II Extreme uses. Using this information can get you very close to where you want to be, then its up to experimentation to fine tune them and get things stable.

Does anyone remember what page or what post # had the included GTL voltage table in it (I believe it was this thread but I could be wrong)?

Qwik - I assume you were using your QX9770 when you were testing the settings above? With that said, although you mentioned you haven't been able to run at FSB2000Mhz how high can you go?

Only problem with this is knowing what percentage of standard voltage to set. If you read the Anandtech article it seems to suggest setting a positive GTL for Quad Core, other sites have said to set a negative GTL if you have 45nm Quad Core. I'm confused!!

[Roger_D25]

Yeah Anandtech says the following, "Historical data has shown that dual-core CPUs (in particular 45nm dual-cores) often clock to higher FSB levels when the GTL reference values are closer to 63%-64% of VTT while quad-core CPUs usually need the full strenth value or even alittle more voltage (67%-70% of VTT)". So for example if you were running a VTT value of 1.20v (we'll assume this is the measured VTT value) the GTL value for 67% would be,

1.20v x 67%(0.67) = 0.804v(804mV) To get this GTL value just set BIOS from "Auto" to "Default". (I could be wrong here, although I fully understand the information below its the "default" setting that confuses me)?

Now if you wanted to increase your GTL value to 70% (one of the value that Anandtech mentions for 45nm quad-core CPUs) you would do the following,

70%(0.70) - 67%(0.67) = 0.03, we then multiply this value by 1000 to convert the value into mV (0.30 x 1000 = 30mv). Since we are increasing the GTL value from 67% to 70% we would set the BIOS GTL value to +30mv.

Another example (just to show how the math would be done) would be lowering the GTL value from the default of 67% to 63% (one of the values Anandtech mentions for 45nm dual-core CPUs) you would do the following,

67%(0.67) - 63%(0.63) = 0.04, we then multiple this value by 1000 to convert the value into mV (0.40 x 1000 = 40mV). Since we are lowering the GTL value from the default value of 67% down to 63% we would set the BIOS GTL value to -40mV.

Below is a quick breakdown of each GTL settings in BIOS and what it controls,

Each die receives two GTL reference votlages (quad-core = 2 dual-cores)
CPU GTL_REF0 Ratio: For Core 0/1 Data Bus (Runs at full FSB transfer rate)
CPU GTL_REF1 Ratio: For Core 2-3 Data Bus (Runs at full FSB transfer rate)
CPU GTL_REF2 Ratio: For Core 0/1 Address Bus (Half of FSB transfer rate)
CPU GTL_REF3 Ratio: For Core 2/3 Address Bus (Half of FSB transfer rate)

*Note: Both CPU GTL_REF2 Ratio and CPU GTL_REF3 have no effect of dual-core processors.
*Note: As a rule, adjustments are normally made to FSB Data Bus values first. These lines are constantly heavily loaded and as such are more susceptible to the effecs of reduced signaling margins
*Note: Because there are only a handful of specific GTL values in BIOS you may need to slightly adjust your VTT up or down to get closer to one of these specific offset values.

Almost all of the information above came from Anandtech Striker II Extreme motherboard review HERE.

Please correct me ASAP if any of the information above is incorrect or misleading. Most of it was copied from Anandtech's article but not all of it was word or word.

[Tyke]

Yeah Anandtech says the following, "Historical data has shown that dual-core CPUs (in particular 45nm dual-cores) often clock to higher FSB levels when the GTL reference values are closer to 63%-64% of VTT while quad-core CPUs usually need the full strenth value or even alittle more voltage (67%-70% of VTT)". So for example if you were running a VTT value of 1.20v (we'll assume this is the measured VTT value) the GTL value for 67% would be,

1.20v x 67%(0.67) = 0.804v(804mV) To get this GTL value just set BIOS from "Auto" to "Default". (I could be wrong here, although I fully understand the information below its the "default" setting that confuses me)?

Now if you wanted to increase your GTL value to 70% (one of the value that Anandtech mentions for 45nm quad-core CPUs) you would do the following,

70%(0.70) - 67%(0.67) = 0.03, we then multiply this value by 1000 to convert the value into mV (0.30 x 1000 = 30mv). Since we are increasing the GTL value from 67% to 70% we would set the BIOS GTL value to +30mv.

Another example (just to show how the math would be done) would be lowering the GTL value from the default of 67% to 63% (one of the values Anandtech mentions for 45nm dual-core CPUs) you would do the following,

67%(0.67) - 63%(0.63) = 0.04, we then multiple this value by 1000 to convert the value into mV (0.40 x 1000 = 40mV). Since we are lowering the GTL value from the default value of 67% down to 63% we would set the BIOS GTL value to -40mV.

Below is a quick breakdown of each GTL settings in BIOS and what it controls,

Each die receives two GTL reference votlages (quad-core = 2 dual-cores)
CPU GTL_REF0 Ratio: For Core 0/1 Data Bus (Runs at full FSB transfer rate)
CPU GTL_REF1 Ratio: For Core 2-3 Data Bus (Runs at full FSB transfer rate)
CPU GTL_REF2 Ratio: For Core 0/1 Address Bus (Half of FSB transfer rate)
CPU GTL_REF3 Ratio: For Core 2/3 Address Bus (Half of FSB transfer rate)

*Note: Both CPU GTL_REF2 Ratio and CPU GTL_REF3 have no effect of dual-core processors.
*Note: As a rule, adjustments are normally made to FSB Data Bus values first. These lines are constantly heavily loaded and as such are more susceptible to the effecs of reduced signaling margins
*Note: Because there are only a handful of specific GTL values in BIOS you may need to slightly adjust your VTT up or down to get closer to one of these specific offset values.

Almost all of the information above came from Anandtech Striker II Extreme motherboard review HERE.

Please correct me ASAP if any of the information above is incorrect or misleading. Most of it was copied from Anandtech's article but not all of it was word or word.

Yes I read that article. My point is that in it he seems to be saying to set GTL at around 70% of VTT for Quad Cores. However in other Anandtech reviews such as the for the MSI X48 it is suggested that a 45nm Quad Core should be 63% of VTT. That is my confusion!

[X2beach]

0701 Is Out

ftp://ftp.asus.com/pub/asus/mb/socke...treme/0701.zip

Going to flash to it in just a few

[Roger_D25]

I could be wrong here but I think the differences between the two suggested quad-core GTL settings must have to do with the chipset and how it handles these things (not to mention how the different BIOS's interact with the chipsets). So even though both settings are different they are both relevant to each motherboard. That is just my personal opinion on why there might be different suggested GTL settings. You would think however that these values were the same no matter what motherboard you were using since a percentage of the VTT voltage is always going to be the same value,

67% of 1.10v = 0.737v(737mV)
70% of 1.20v = 0.840v(840mV)
63% of 1.26v = 0.794v(794mV)
etc, etc............................

Then again people with the same hardware have had different results using the same GTL settings in BIOS, on the other hand people have had great clocking results on the same hardware using different GTL settings? I only posted all that information in my last post to hopefully help out some people who don't know anything about GTL settings and what they are or how they work.

0701 Is Out

ftp://ftp.asus.com/pub/asus/mb/socke...treme/0701.zip

Going to flash to it in just a few

Thank you so very much my friend! I clicked the link you provided and it was a direct download link, was this BIOS from the FTP server or from the actual website? I'm curious as to why there are a handful of available BIOS files that people have but Asus's website only has two available? Basically if your just a normal joe who doens't have access to their FTP server your stuck with old crappy BIOS files! Very dissapointing I think, however I'm glad that some of you guys have acess to them and share

X2beach how are you going to flash your board, what method?

[X2beach]

I just flashed to the new 0701 bios

I tried EZFLASH got the same ROM ID Mismatch

so I flashed it with my USB Drive in Dos mode with awdflash

[BenchZowner]

Faster mirror for BIOS v0701.
Fixes & Bugs list to be included...soon

[Roger_D25]

I incorrectly posted earlier about Asus's website only having 3 BIOS files avaiable, after just checking I realized that they recenlty added BIOS File 0512. Looking forward BenchZowner to fixes and bugs for this BIOS file!

[slim142]

Ok guys here it is the changelog for 0512, nothing interesting. Basically a 0511.

Striker II Extreme BIOS version 0512
- CPU multiper will no longer be set to default value in the BIOS setting after CPR.

Note: Please use either AWDFLASH or AsusUpdate version 7.14.01 to upgrade BIOS when your currentlyl installed BIOS is prior to version 0402. You could use EZflash to upgrade BIOS from version 0508.

Minor upgrade... but at least we now know what it does

[tekjunkie]

YEEHAW!!! 0701 here I come

[slim142]

Hmmm, I wouldnt be THAT excited, yes since is a new series (070x) then it means there is a good change/fix/update in this BIOS. However, we can expect a lot from ASUS like worthless updates...

[BenchZowner]

Been using 0701 for 3 hours or so.
Looks better than the any other BIOS version I've tried before.
I've been benching on the limits ( real limits, like over the "known" stable FSB, memory @ ~1900 7-6-5-21-1T-1-1-5-1-1-48-1, and 45nm Vcore over 1.7V on watercooling ), and had no problems, and no corruption ( yet )

[Lu(ky]

Been using 0701 for 3 hours or so.
Looks better than the any other BIOS version I've tried before.
I've been benching on the limits ( real limits, like over the "known" stable FSB, memory @ ~1900 7-6-5-21-1T-1-1-5-1-1-48-1, and 45nm Vcore over 1.7V on watercooling ), and had no problems, and no corruption ( yet )

Great news I will be trying this new bios asap... Will get back to you guys tomorrow with results..

[Roger_D25]

BenchZowner - First of all thanks for posting some initial results from the new BIOS, much apprecaited! With that said I'm curious if you have had much luck lowering your "tWR" at all? Looking at the timings you just posted it seems obvious that you have (looks like your tWR is 5 from your post but not sure) but I'm curious how because no matter what I set it to it stays at 12? I've even tried changing all other settings in relation to it thinking that it might help but no go? I have a feeling it may be a bug with BIOS version 0511 but I'm not sure. Any help would be greatly apprecaited because I've had great luck with very tight timings except of course tWR!

[BenchZowner]

I will check 0511 tomorrow and see if the tWR setting is bugged ( which is more than likely from what you're saying )

p.s. Yeah, my tWR is at 5 ( works fine with both 0601 and 0701 )

[Roger_D25]

That seems like an awful lot of work for you to do, heck I'll just try a new BIOS (should probably update anyways)? With that said thanks very much for the suggested help! I'll definately post back tomorrow with results. By the way I'll be very surprised if you keep benching at those settings and don't get major corruption, lol!

[BenchZowner]

Imagine that I was benching at the point where AquaMark 3 was running once after one or two fails, alt + tab, open paint, paste & save, return to AquaMark 3, press Exit... after the exit sequence...BSOD. Reboot, altered BIOS settings ( yes, even higher, frequency and voltage ) windows booted fine.
Bench 3D Mark03 + AquaMark 3, exit AquaMark 3... fine this time, no BSOD.

Now I gotta check if this BIOS fixed the 6-5-5-x issue ( my sticks can run SuperPi 32M @ DDR3-1800+ 6-5-5-18 on the P5K3 Deluxe, but on the S2E they fail at any setting higher than DDR3-1650... )

[Roger_D25]

Jes*s what sticks are you running there BenchZowner and what kind of VDIMM did they take on the P5K3-D to run at 1800+ 6-5-5- (at least 2.2v-2.25v)? I'm very surprised they can't run at anything over 1650Mhz (unless of course your talking about keeping those same settings, 6-5-5-)? If you loosened things up a bit they will clock over 1650Mhz right?

[BenchZowner]

CellShock CS3222580 ( DDR3-1800 CL8 ) 2GB Kit.
Vddr = 2.3V

They do clock nicely with 6-5-5, 6-6-5, 7-6-5, 7-7-6 but those timings don't really work on the Striker II Extreme ( I mean it cannot reach the same frequencies, for example with the P5K3 Deluxe and 7-7-6-21 they're stable at DDR3-2024, on the Striker II Extreme with the same timings the max stable I can reach is around DDR3-1850 ).
8-7-6-21 isn't as good as those new kits with high binned D9JNL ( reaching DDR3-2100+ 8-7-6-21 stable ).

I'd really like to make things work fine and bench AquaMark 3 & 3D Mark2001SE with high CPU clocks and the DIMMs at DDR3-1800 6-5-5-16 1T... but unless a new BIOS fixes the issue ( I doubt it ) I can't see that happening

[Roger_D25]

Despite the lower clocks on the S2E though aren't you finding that the overall bandwidth is considerably higher on this board over the P5K3 (despite having higher + tighter timings)? I find that even at DDR3-1600 6-5-5- I get higher bandwidth than I did on my old Blitz Ex with 7-6-6-1900+? Granted I'm certainly no expert with this stuff (just enjoy it very much) what tends to be more important when benching, higher RAM speeds with tighter timings or maybe lower clocks with higher bandwidth? Here is another instance where I could just do the testing and find out for myself but I have a feeling you know the answer?

[BenchZowner]

That depends on the benchmark in question, and your max FSB.
For example if you can reach high FSBs, like over 550MHz then 7-6-5-21 or 8-7-6-21 depending on the max stable frequency for each timings-set it'd be better than lower clocks and tighter timings.

If you cannot reach high FSBs, let's say because of a Quad-Core CPU, then the tighter the timings the better ( linked and synch at DDR3-1800 and 445MHz FSB would be awesome for SuperPi, 3D Mark2001SE and AquaMark 3 and Hexus PiFast ).

So in general, it always depends on your max FSB and your memory kit's sweet spot ( frequency + timings )

[BenchZowner]

Kinda woot
The newest BIOS ( v0701 ) "kinda" solved the 6-5-5 issues.
I can bench SuperPi 1M & AquaMark 3 & 3D Mark2001SE with my RAM at DDR3-1760 6-5-5-16 1T.
Still unstable ( memtest errors, and not fully stable in windows ) but it's definitely better than nothing.
2.1k points gain over 7-6-5-21 ( at higher frequency ) in AquaMark 3.

Awesome ( considering the FSB ) rates as well.
"Pics" ( not tweaked windows installation... actually a gaming performance review installation )

[tekjunkie]

Good to see that people are getting good results with 0701. Unfortunately after a bit of testing with 0701, I went back to 0603. 0603 works for me very well. Maybe I will come back after more testing with 0603. This does not mean anything. I just felt that there is 1 more variable in my testing that I need to worry about, hence decide to keep the BIOS version constant and as I said 0603 works for me very well.