ASUS Maximus Formula SE

[Tony]

I will flash the bios chip and plug it in the board...see what happens. AMIbcp won't open the file to edit the string which means they moved to new source code on core 8

[S1mon-]

I will flash the bios chip and plug it in the board...see what happens. AMIbcp won't open the file to edit the string which means they moved to new source code on core 8

Nice

[eSp!s0]

Running 4 x 1 Gb Transcend AXeram 1200+'s at 600mhz with this board. Had to up NB v to 1.49 from 1.45 and FSB term V from 1.4 to 1.51.

5,5,5,12,42 no problem at 2.2vdram TRD 7 4:3 divider FSB 450, strap 400 Q6600 at 3.6ghz. Tried CAS 4 but I think that is impossible unless someone can suggest settigs :-) Couldn't get 700mhz either at CAS6 :-(

My Gskill 8000's 2x2Gb gave up at just over 500mhz with a TRFC of 52 programmed into the SPD. So not sure extra TRFC will make much of a difference.

I have 2x1GB of the Transcend 1200 and I don`t come higher with clock than 1200Mhz but I am not good in ram ocing and I only tried till 2,20V bios which is 2,24-2,26V windows.
Using 450fsb and same divider=1200mhz(400strap), too. Also giving ram 2,16V bios, this is 2,20-2,22V windows and having TRD 7 with TB enabled 0.
5-5-5-12 is also best I can get in main timings. But I am glad to hear, that 2 pir of them are no problem with this board and I will think about a second pair Transcend sometime.

[Aldy402]

@Tony

Im eagerly awaiting your results..

[Tony]

will be a few days m8...needed to buy a parallel extension cable for the bios flasher...then I need to build a PC with a parallel connection..LOL

I will have a play though and see whats what

[Tony]

TRD changes as FSB ramps up, there is a change it seems at 500fsb and probably more lower down..so yes it is related to straps but I doubt its off the BSEL like I used to think it was.

Asus 965 changes at 401, TRD losens and fsb is allowed to rise, issue is for 60fsb you lost memory performance. Now what we need is the ability to see preprogrammed Auto TRD changes Vs MCH clock so we can decide to overclock MCH or not. I will look into this more over the next few weeks.
While i think AT did an awesome job of explaining TRD I fel there is still something missing and some simple mapping out should have those who are still confused a little grasping what is happening.

I will say those looking for 500fsb+ are totally wasting their time, TRD kills performance as FSB ramps up, there is a sweet spot for sure and its a lot less than 500fsb
buy cpu's with as high a multi as you can, buying cheap and pushing fsb may get you CPU MHZ but it kills MCH performance and so can end up a lot slower than what you though.

[cadaveca]

WEll Tony, I must say...you're the man!


I have been begging for years(literally) for TRD and TRFC adjustments in bios. When 975x came out, I knew these two settings were key...I mean really, playing with memset, et al really highlighted this...but due to bios control, it was only good for benching!

So, again, thanks...many aspiring clockers are going to thank you!

[Tony]

TRD was not my doing really, i have been going on about straps and performance tweaking with Intel hinting but never telling , performance level (my friend Roger ) was the key to getting the ball rolling for TRD to finally start showing its face...FCG has pushed the envelope much further though even coming out with little calculations showing where things "should" work and where things just won't. I still think we need to map out TRD Vs FSB adding Vmch maybe into the mix so that we can see whats damn fast and what is slow... this has to be done per board though as its obvious from Asus 965 changing tRD at 401fsb and GBT changing at 435FSB different boards by Auto are going to give quite different petrformance.

It also shows what a mess memory reviews were, there is no way you could compare dimms tested on different boards with different bios files altering tRd at different points...easy way of making X brand dimms look good on Y brand board me thinks if engineers and PR work together.
You can also see this could be a way for X memory company to work with bios engineers to add code to spd that tells the board to increase tRD at a given FSB to allow "their' dimms to do the super high clocks and V brand does not due to the rules of TRD and FSB.
Not saying this has happened but it makes you start thinking as it would be dead easy.

tRFC, yes this is my doing, I pushed some reviewer friends to help me get Asus to sort the issue out as reviewers seem to have more influence then engineers (specially me, i can't even get boards to test with unless i beg for days, sometimes weeks)

I latched onto a product OCZ developed a few months ago (4GB memory kit) and started pushing it around 1000MHZ+, what I noted though was tRFC had a massive impact on how the IC's clocked up,lucky for me I was using DFI and Abit P35 boards with fully open tRFC in steps of 1, so I did not have any issues pushing OCZ to produce what you see as Reaper X EB and 8000 Reaper X, the issue was when I finally started testing on Asus boards (P35 and x38)
and found the boards severely tRFC limited, so much so 1000MHZ rated dimms were hitting 950 and stopping and in some cases not even allowing a boot.
Lucky for us Asus are fixing it, and the bios I have even though some parts are broken does allow tRFC to scale and I was testing at 1066 with dimms stuck at 940 with the previous bios build.

Overall "us engineers" in the background are really trying to get you guys what you need and trying to explain how things should work, we may not pour LN2 and have massive clocks to brag about but we at least contribute to how boards work and give the tools for those who play super cooling something to play with.

[cadaveca]

Overall "us engineers" in the background are really trying to get you guys what you need and trying to explain how things should work, we may not pour LN2 and have massive clocks to brag about but we at least contribute to how boards work and give the tools for those who play super cooling something to play with.

Creating a monster, just to have it in chains, is never good. It's always more enjoyable when it smashes crap.

[JohnMike]

TRD was not my doing really, i have been going on about straps and performance tweaking with Intel hinting but never telling , performance level (my friend Roger ) was the key to getting the ball rolling for TRD to finally start showing its face...FCG has pushed the envelope much further though even coming out with little calculations showing where things "should" work and where things just won't. I still think we need to map out TRD Vs FSB adding Vmch maybe into the mix so that we can see whats damn fast and what is slow... this has to be done per board though as its obvious from Asus 965 changing tRD at 401fsb and GBT changing at 435FSB different boards by Auto are going to give quite different petrformance.

It also shows what a mess memory reviews were, there is no way you could compare dimms tested on different boards with different bios files altering tRd at different points...easy way of making X brand dimms look good on Y brand board me thinks if engineers and PR work together.
You can also see this could be a way for X memory company to work with bios engineers to add code to spd that tells the board to increase tRD at a given FSB to allow "their' dimms to do the super high clocks and V brand does not due to the rules of TRD and FSB.
Not saying this has happened but it makes you start thinking as it would be dead easy.

tRFC, yes this is my doing, I pushed some reviewer friends to help me get Asus to sort the issue out as reviewers seem to have more influence then engineers (specially me, i can't even get boards to test with unless i beg for days, sometimes weeks)

I latched onto a product OCZ developed a few months ago (4GB memory kit) and started pushing it around 1000MHZ+, what I noted though was tRFC had a massive impact on how the IC's clocked up,lucky for me I was using DFI and Abit P35 boards with fully open tRFC in steps of 1, so I did not have any issues pushing OCZ to produce what you see as Reaper X EB and 8000 Reaper X, the issue was when I finally started testing on Asus boards (P35 and x38)
and found the boards severely tRFC limited, so much so 1000MHZ rated dimms were hitting 950 and stopping and in some cases not even allowing a boot.
Lucky for us Asus are fixing it, and the bios I have even though some parts are broken does allow tRFC to scale and I was testing at 1066 with dimms stuck at 940 with the previous bios build.

Thanks for every bit of info and knowledge that you share with us. I have been for long time reading your posts and learn a lot.

[Overall "us engineers" in the background are really trying to get you guys what you need and trying to explain how things should work, we may not pour LN2 and have massive clocks to brag about but we at least contribute to how boards work and give the tools for those who play super cooling something to play with.

[cstkl1]

yeah ty tony..
a lot of ure explanation is making sense of somethings which i always thought mobo limit without any reasoning to it..

another thing y e8400 on 500fsb cant be run on this board without having the mem's go nuts...
499 ok.. mem stable.. proc stable..

that 1 freaking fsb and thats it..

[Nuckin_Futs]

update: Still no luck on my bad BIOS flash. Can anybody share their results of any type of BIOS flash problems on this mobo?

I can only bot to XP x64 SP2 if I press F2 to load default and continue, Yet, if I press F1 to enter CMOS, I cannot reboot after "save and exit". I have to clear and use F2 to load OS.

Anybody eperience something close to this?

[Aurhinius]

TRD was not my doing really, i have been going on about straps and performance tweaking with Intel hinting but never telling , performance level (my friend Roger ) was the key to getting the ball rolling for TRD to finally start showing its face...FCG has pushed the envelope much further though even coming out with little calculations showing where things "should" work and where things just won't. I still think we need to map out TRD Vs FSB adding Vmch maybe into the mix so that we can see whats damn fast and what is slow... this has to be done per board though as its obvious from Asus 965 changing tRD at 401fsb and GBT changing at 435FSB different boards by Auto are going to give quite different petrformance.

It also shows what a mess memory reviews were, there is no way you could compare dimms tested on different boards with different bios files altering tRd at different points...easy way of making X brand dimms look good on Y brand board me thinks if engineers and PR work together.
You can also see this could be a way for X memory company to work with bios engineers to add code to spd that tells the board to increase tRD at a given FSB to allow "their' dimms to do the super high clocks and V brand does not due to the rules of TRD and FSB.
Not saying this has happened but it makes you start thinking as it would be dead easy.

tRFC, yes this is my doing, I pushed some reviewer friends to help me get Asus to sort the issue out as reviewers seem to have more influence then engineers (specially me, i can't even get boards to test with unless i beg for days, sometimes weeks)

I latched onto a product OCZ developed a few months ago (4GB memory kit) and started pushing it around 1000MHZ+, what I noted though was tRFC had a massive impact on how the IC's clocked up,lucky for me I was using DFI and Abit P35 boards with fully open tRFC in steps of 1, so I did not have any issues pushing OCZ to produce what you see as Reaper X EB and 8000 Reaper X, the issue was when I finally started testing on Asus boards (P35 and x38)
and found the boards severely tRFC limited, so much so 1000MHZ rated dimms were hitting 950 and stopping and in some cases not even allowing a boot.
Lucky for us Asus are fixing it, and the bios I have even though some parts are broken does allow tRFC to scale and I was testing at 1066 with dimms stuck at 940 with the previous bios build.

Overall "us engineers" in the background are really trying to get you guys what you need and trying to explain how things should work, we may not pour LN2 and have massive clocks to brag about but we at least contribute to how boards work and give the tools for those who play super cooling something to play with.

I am very interested to see what you can come up with Tony. The AT review shed a lot of light on TRD however I have found that plugging in the numbers they provide as a guide I still cannot seem to get the results they do. Usually I am finding my TRD is always one higher. I guess this may just be a result of the older MCH on the X38 compared to the X48.

I've tested 4 x 1GB at 900mhz CAS4 TRD 6. It provides a lot less bandwidth than CAS 5 TRD 7 1200mhz. Sisoft approx 7400 to 8400 approx.

I for one would love to see a graph of TRD Vs FSB with vmch and CAS. So we can determine whether 900mhz CAS 4 TRD 6 is faster than 1000mhz CAS 5 TRD 7 for example.

We need to find that sweet spot as you suggested.

Get working on it

[cstkl1]

can anybody do tRD 7 on 500 fsb 1:1..
i cant?? seems lowest is tRD 8.

the rest of AT tables seems right..
ps foxconn p35 mars seems to be able to do tRD 7..( some call this the ram magic board.. so was this a magic??)

[Spyrus]

P35 can do easily tRD=7 on 530+
X38 (asus) can do that at around 490.

[cadaveca]

the rest of AT tables seems right..

That's because of the X48 "binning".

I mean really, I was so shocked to see so many people buy into the story about INtel binning X38 to X48 for FSB!!!!

It became quite clear that there was some confusion out there as to what Intel chipset really are, and why, in comparison, AMD does not make thier own chipsets(although they do now, via ATI...)

It's all about memory control.


AMD, having the memory controller on cpu, had control. Intel, with no IMC, had to make thier own chipsets to ensure control of memory interface.

Bring into the picture...790i...I wonder what's the difference between Ultra and Ultra SLi? Better memory control while running two cards, which makes for a hotter chipset, maybe?


So don't take AT's article entirely to heart. Intel specifically binned those chipsets with faster memory performance in mind. DDR3-2000 is far more work than DDR2-1200...

X48 used to DDR2 will probably the fastest boards ever. Why? <---Won't see me asking THAT question.

[cstkl1]

Finally got my rams working ... thanks guys...
had to sacrifice some vcore.. 0.008v.. but what the heck...
the solution was weird
i was running 0907 with E6850 at 4ghz..
and just popped a new proc.. and it was causing all the issue

reflashed it to 0902.. got it stable.. ( temps were less accurate with the elite and the condensation i was having) and reflashed it back to 0907.. all this on the E8400.
guess what stable.. and best part temps also dropped..... reminds me of xfiles theme song....



btw so whats the reasoning on Asus X38 not having the ability on doing tRD 7 on 500fsb 1:1

i suppose AT tested the rampage on x48 on that .. so it must be working..

reasons anyone??

[Aldy402]

@Tony

-my F2-8000CL5D-4GBPQ set from Gskill is severly limited due to the tRFC, issue. It wont let me go over 5mhz of stock speed no matter what voltage. It refuses to boot on any divider besides "AUTO" even when within/under stock speeds. Should I try flashing the SPD to a higher tRFC?

-would you know if the next bios rls addresses the tRFC issue ? or is that too far ahead?

[eSp!s0]

TRD was not my doing really, i have been going on about straps and performance tweaking with Intel hinting but never telling , performance level (my friend Roger ) was the key to getting the ball rolling for TRD to finally start showing its face...FCG has pushed the envelope much further though even coming out with little calculations showing where things "should" work and where things just won't. I still think we need to map out TRD Vs FSB adding Vmch maybe into the mix so that we can see whats damn fast and what is slow... this has to be done per board though as its obvious from Asus 965 changing tRD at 401fsb and GBT changing at 435FSB different boards by Auto are going to give quite different petrformance.

It also shows what a mess memory reviews were, there is no way you could compare dimms tested on different boards with different bios files altering tRd at different points...easy way of making X brand dimms look good on Y brand board me thinks if engineers and PR work together.
You can also see this could be a way for X memory company to work with bios engineers to add code to spd that tells the board to increase tRD at a given FSB to allow "their' dimms to do the super high clocks and V brand does not due to the rules of TRD and FSB.
Not saying this has happened but it makes you start thinking as it would be dead easy.

tRFC, yes this is my doing, I pushed some reviewer friends to help me get Asus to sort the issue out as reviewers seem to have more influence then engineers (specially me, i can't even get boards to test with unless i beg for days, sometimes weeks)

I latched onto a product OCZ developed a few months ago (4GB memory kit) and started pushing it around 1000MHZ+, what I noted though was tRFC had a massive impact on how the IC's clocked up,lucky for me I was using DFI and Abit P35 boards with fully open tRFC in steps of 1, so I did not have any issues pushing OCZ to produce what you see as Reaper X EB and 8000 Reaper X, the issue was when I finally started testing on Asus boards (P35 and x38)
and found the boards severely tRFC limited, so much so 1000MHZ rated dimms were hitting 950 and stopping and in some cases not even allowing a boot.
Lucky for us Asus are fixing it, and the bios I have even though some parts are broken does allow tRFC to scale and I was testing at 1066 with dimms stuck at 940 with the previous bios build.

Overall "us engineers" in the background are really trying to get you guys what you need and trying to explain how things should work, we may not pour LN2 and have massive clocks to brag about but we at least contribute to how boards work and give the tools for those who play super cooling something to play with.

Yes thank you for your explanations!
We would know nothing without guys like you.
Ok, I don`t understand everything, but that`s only because I am not that good in english, the technical things are very good to understand

[cstkl1]

Yes thank you for your explanations!
We would know nothing without guys like you.
Ok, I don`t understand everything, but that`s only because I am not that good in english, the technical things are very good to understand

ty to quite a few of u guys
since i got the 500fsb stable and ram
been upping the multis

[eSp!s0]

ty to quite a few of u guys???
Do you mean "Thank you for quoting a few guys here"?

[cstkl1]

ty to quite a few of u guys???
Do you mean "Thank you for quoting a few guys here"?

yup some for helping me on the forum, some for PM some heck for posting..

thats how i solved the issue..
no harm right..

hmm is the e8500 out yet in the us?? on retail??

[Tony]

@Tony

-my F2-8000CL5D-4GBPQ set from Gskill is severly limited due to the tRFC, issue. It wont let me go over 5mhz of stock speed no matter what voltage. It refuses to boot on any divider besides "AUTO" even when within/under stock speeds. Should I try flashing the SPD to a higher tRFC?

-would you know if the next bios rls addresses the tRFC issue ? or is that too far ahead?

you need to talk to gskill m8, i can't comment on their product

[NBK]

yup some for helping me on the forum, some for PM some heck for posting..

thats how i solved the issue..
no harm right..

hmm is the e8500 out yet in the us?? on retail??

Hi

The retail E8500 was out in the US and then it vanished. I have heard that it was supposed to reappear on 11 Feb 2008 or in April 2008. The truth is no one knows...

HTH
NBK

[Aurhinius]

My findings

FSB 400 266 STRAP Mem 1000mhz 4,4,4,12 TRD 6 (Board just won't do TRD 5) divider 5:4 - Everest bandwidth 8800 latency 53ns

FSB 450 400 STRAP Mem 1200mhz 5,5,5,12 TRD 7 divider 4:3 - Everest Bandwidth 9996 latency 51ns

So far it seems the extra mhz on the FSB is still the winner and I think there is something going on with the X38 that is different from the X48.

Always thought lower strap was better... guess not or the board has been optimised better for higher straps.

can you shed any light on this Tony?

Okay did a bit more testing and managed to get a TRD of 5 in certain circumstances by dropping AI clock twister to moderate instead of strong.

Tested FSB 400, 266 STAP, 5,5,5,12 , 3:2 divider 1200mhz TRD 5 everest 9500 52ns
will not boot 400, 266, 5:4, 4,4,4,12 TRD 5