A short story behind Corsair Dominator @ DDR3-2462 CL9

[saaya]

Wonder what the results are on the released 0402 BIOS instead of a custom BIOS spin? I bet they might just be a little different.

huh? theese results are with a custom bios?

hmmm good point kasparz...
and even if you find the settings that let one cpu run 533fsb, i doubt it will work for many cpus... even if they can run 533, they might need different vtt and vcore and vpll hmmmm the fsb is becoming more and more a problem for intel, its really time for nehalem! imagine nehalem would still use fsb
im glad intel finally went IMC

[Kasparz]

First of all, there were no SPD extensions for DDR. Second, there were DDR625 too. DDR600 could be achievable on good K8 IMC, but 625 was really hard to achieve on most K8 IMC.
I know what our customers needs, and i know who need fast memory, and who does not. I'm not going to recommend product to anyone if its clearly not designed properly and could cause stability problems.
You might know that 95% of all people who purchased memory over jedec standard that needs to set timings/voltages manually, never even heard what BIOS is and how to get there. Even majority of people who purchases very expensive memory like DDR2 1200MHz+ or DDR3 1800MHz+ is same as described above. 70% of customer built computers is returned to RMA due to this problem.
And you might know that EPP you must enable manually in BIOS to work, XMP is enabled automatically without customers/system builders assist.
There were lots of people whom i should tell long story about why they need to set settings manually, and lots of people who returned memory for money due to this.
Thats just our community is experienced people with knowledge, lots of potential this memory buyers aren't.

[saaya]

Well with our 1866 so far nobody needed help getting them to work, and with our 1800 kits maybe 5 or 10 people worldwide, and they dont have an EPP or XMP profile.
But yes, i see what you mean kasparz. i dont think its a good idea to go that close to the limit cause you will get swamped with customer calls and emails... but then again, if this is just about a small quantitiy of kits for the top ocers... why not?

[Kasparz]

Thats not close to limits, thats way over limit.
Lets say, quads on stock with that FSB would not work...
95% of Wolfdales/Conroe at 533FSB and stock Vpll would not work.
99% of X38/X48 chipsets at 533FSB and stock voltages would not work, or will work very unstable.

Sascha, people asks most of their questions in shop, not to manufacturer directly

[George_o/c]

im glad intel finally went IMC

I have a question for you guys as far as the Intel CPUs and IMC are concerned ... Won't these CPUs with IMC, experience cold-bug problems at high temperatures (for example -20*C ...) ...

[saaya]

Thats not close to limits, thats way over limit.
Lets say, quads on stock with that FSB would not work...
95% of Wolfdales/Conroe at 533FSB and stock Vpll would not work.
99% of X38/X48 chipsets at 533FSB and stock voltages would not work, or will work very unstable.

does pll voltage really help that much? i thought for 45nm it doesnt help at all or barely helps? From what ive heard a lot of 45nm chips can hit 533fsb, but if they can run it stable... without tweaking i guess your right, only very few can. then again, on 790 you dont need a huge fsb, so i think thats what corsair is actually aiming theese sets at.

Sascha, people asks most of their questions in shop, not to manufacturer directly

with that i cant argue
i know its true... which is why usually shop owners tend to know a lot more about the products and the problems people have with it than the manufacturers

I have a question for you guys as far as the Intel CPUs and IMC are concerned ... Won't these CPUs with IMC, experience cold-bug problems at high temperatures (for example -20*C ...) ...

i dont think so...
only some batches of a64 chips had an imc related cold bug i think.
there are newer K8 K9 and K10 chips that have cold bugs which are unrelated to the imc, and there are some that can run down to -50C or lower.

and dont forget, intel chips have cold bugs as well, and they dont have an imc some intel cpus have cold bugs at -60 already!

i think it actually doesnt have to do with temps directly but with over-current protections of the circuit. or its just a bi-product of the mfg process and silicon doping... intel and amd and tscm etc are doing really weird stuff
and they are looking for improvements in a certain temp and voltage and current range, so maybe some new technique works there, but causes probs at low temps or higher volts dont scale etc...

back on topic!
was this really achieve with a special bios?
is this bios available for download somewhere?
whats the difference between this special bios and the public bios?

[Tony]

Bios 0001 is a debug bios, possibly all options enabled, all options manually adjustable.
The bios was never made public, OCZ do not have it and I doubt ever will. Like end users OCZ use what is in the public domain, and always have done, Cellshock look to do the same.

[Tony]

Thats not close to limits, thats way over limit.
Lets say, quads on stock with that FSB would not work...
95% of Wolfdales/Conroe at 533FSB and stock Vpll would not work.
99% of X38/X48 chipsets at 533FSB and stock voltages would not work, or will work very unstable.

Sascha, people asks most of their questions in shop, not to manufacturer directly

You know man you are bang on the money, FSB is beyond stretched now and the options in bios for Dram ratio mean you have to run 500+fsb to get these ram speeds.

XS users don't see the big picture, ram manufacturers don't make 10 kits for the extreme users who are able to run 2200fsb, they make 5000 kits in the hope that many people will be able to enjoy the added speed, issue is now we are able to provide technology that out paces 90% of the FSB capability of most processors and boards...but end users don't understand that and just RMA.

FSB is dead, its time for IMC if we want to run these uber high ram speeds.

[saaya]

Bios 0001 is a debug bios, possibly all options enabled, all options manually adjustable.
The bios was never made public, OCZ do not have it and I doubt ever will. Like end users OCZ use what is in the public domain, and always have done, Cellshock look to do the same.

have you ever used that bios or seen it?
do you think asus minds if michal shares that bios with us?
I guess so... but why exactly would they care?
a big selling point of x48 was extra bios options, so ok, you dont wanna spill all the beans at once... but they dont intend to release yet another x38/x48 board, do they? so why would they care if this bios goes public?
hell, its prolly a mess and most people would only look at it and then delete it

its time for IMC if we want to run these uber high ram speeds.

what do you mean with IF

790 would be the only real platform for 2133 memory i think.
but then again, several people seem to have problems reaching even 2000 stable on 790. if anybody cares about my personal opinion, i think even 2000 is too much for an official spec... there are so many, especially older, p35 and x38 boards out there that wont even run 2000 stable and even the latest p35/p45/x38/x48 and 790 boards dont all run ddr3 2000 stable.

its really a shame that no mainboard mfg tests and guarantees ddr3 2000.
the highest ive seen is gigabyte and they only test and guarantee ddr3 1900...

thats why the fastest mem im making here at cellshock is 1866... everybody knows it can run way more than that...
but i thought its not a good idea to rate it higher than that.

anyways, if those results are really with a special bios, whats the max with the retail bios?
looking forward to some more results!

[bachus_anonym]

Wonder what the results are on the released 0402 BIOS instead of a custom BIOS spin? I bet they might just be a little different.

What would you like to bet, Gary? But seriously, the BIOS that was used on this system is in essence, as I mentioned earlier, 0301. There's nothing super-magical about it, and is as good as 0301 can be. It contains an update or, a fix if you will, that I suggested some time ago and it's up to Asus to decide if they include it in a final BIOS at some point. I think they will, or should. But from my testing here, it didn't make any difference for the numbers you see here.

Bios 0001 is a debug bios, possibly all options enabled, all options manually adjustable.
The bios was never made public, OCZ do not have it and I doubt ever will. Like end users OCZ use what is in the public domain, and always have done, Cellshock look to do the same.

What I said above... I wish there was a BIOS like it, Tony. That's would be nice, wouldn't it?

1066MHz XMP profile? Why do i smell FAIL when using XMP?
What cpu can do 6x533 with stock cooling/voltages(fsb wise), and what board can do 533FSB with everything stock?
I would like to know how Corsair did that, seriously...
Everything over 933MHz should be with EPP2.0 extension for nVidia, due to having unlinked mode therefore not touching cpu and FSB to sky high frequencies.
I know that enthusiasts over there would ever use XMP, but average Joe getting fast memory would put computer together on Intel chipset and it just won't work.
That speedbin for memory is only useful for Nehalem.

Kasparz, dude.. Don't pay attention to SPD, you too hang on it. SPD on those modules is an experiment, OK? And this is not a final product. Don't get too hot in the water, please?

[saaya]

the BIOS that was used on this system is in essence, as I mentioned earlier, 0301. There's nothing super-magical about it, and is as good as 0301 can be. It contains an update or, a fix if you will, that I suggested some time ago and it's up to Asus to decide if they include it in a final BIOS at some point. I think they will, or should. But from my testing here, it didn't make any difference for the numbers you see here.

but if the bios didnt give you a boost... then why did you use it?
and why should asus make it public if it doesnt give a boost?
so with a retail bios you get the same ocs?

[Tony]

Its still a developer bios, its still something Cellshock and OCZ etc have not had access to so its still unfair to post any results using it. Period

Are you guys working on helping Intel getting quads to run 566FSB m8 as you are going to need it?

[bachus_anonym]

but if the bios didnt give you a boost... then why did you use it?
and why should asus make it public if it doesnt give a boost?
so with a retail bios you get the same ocs?

... because it was on this system, and I did not flash it with anything never as there was no need for it. Sascha. It's that simple, believe me, that it doesn't make a difference.

[Tony]

Competitors never believe each other Bachus, I told you that when you started working for Corsair...

All Corsair have done here is damage the situation with DDR3 at 2000 or higher, not helped it. Blind posting like this gets end users hopes up and they clearly will be utterly disappointed by this when they struggle like crazy to get their systems running the same.

Well done guys!

[bachus_anonym]

Its still a developer bios, its still something Cellshock and OCZ etc have not had access to so its still unfair to post any results using it. Period
Are you guys working on helping Intel getting quads to run 566FSB m8 as you are going to need it?

Eh come on, Tony. It's only a presentation of what can be done on this platform. Exploring the limits of X48 chipset...

[saaya]

michal, i trust you...
i still think it wasnt a good idea to use that bios, even if it doesnt give you a boost.
you should at least have mentioned that its a special bios, even if it doesnt boot.

Im a bit concerned how many memory reps are posting in this thread now
i hope this doesnt go wrong and i hope we can all get along without bashing each other

and im still hoping for some more results to maybe push this thread back on topic
(hopefully with a retail bios to convince the doubters and end that discussion)

[bachus_anonym]

and im still hoping for some more results to maybe push this thread back on topic (hopefully with a retail bios to convince the doubters and end that discussion)

I will flash 0301 and 0402 BIOS, and rerun 2200+ HyperPI 32M, and 2133C9 ORTHOS (if I have time, I still have work to do, you know ) as soon as I get to work this morning, OK? Just to prove that the BIOS is as good as any newer rev

[Tony]

DDR3 action is 7-7-7-22 at 1800, we all know it, tRD dictates it and the limit on quad FSB forces it....

X48 has FSB limitations but enhanced tRD, faster performance at lower FSB, this is exactly why Intel releases Extreme edition CPU's, high FSB no longer needed

Im done with this now, I think you were bold to post 2400MHZ publicly, we have done the same at our labs but what is the point of getting end user hopes up???

ta ta

[DMH]

DDR3 action is 7-7-7-22 at 1800, we all know it, tRD dictates it and the limit on quad FSB forces it....

X48 has FSB limitations but enhanced tRD, faster performance at lower FSB, this is exactly why Intel releases Extreme edition CPU's, high FSB no longer needed

Im done with this now, I think you were bold to post 2400MHZ publicly, we have done the same at our labs but what is the point of getting end user hopes up???

ta ta

Then why not show it?For me this is just a normal o'cer who got a magical number and wanted to share it nothing,else if he says the BIOS didn't gave him a boost i believe him

[IFMU]

Interesting post.
Think I'm gonna follow this one a bit.

[eva2000]

woah Bachus man crazy clocks.. and crazy CPU PLL volts there

Just when I thought finding my E3110 that does 545FSB 32M stable at stock CPU PLL 1.5v on Blitz Formula was enough for DDR3 testing, now I need >550-615FSB LOL

Anyone ask if Intel will give us P35/X38/X48 users a higher mem divider than 1:2 i.e. 2:5 or 3:7 for 500FSB 3:7 DDR3-2333 or quad core 450FSB 3:7 DDR3-2100 heh

[saaya]

Anyone ask if Intel will give us P35/X38/X48 users a higher mem divider than 1:2 i.e. 2:5 or 3:7 for 500FSB 3:7 DDR3-2333 or quad core 450FSB 3:7 DDR3-2100 heh

nooooooooooooooooooo they should just drop the whole divider stuff already and go unlinked IMC
its not easy to get an unlinked imc working well, but if anybody can do it, its intel

[Tony]

nooooooooooooooooooo they should just drop the whole divider stuff already and go unlinked IMC
its not easy to get an unlinked imc working well, but if anybody can do it, its intel

Its not that easy man, what gives you the idea it is?
Unlinked means clock crossing and all the issues it brings with it, welcome back tRD etc!!

Sync bus 1:1 is the ONLY way forward, plus Intel have gone for width not speed with IMC.

[Kasparz]

You know man you are bang on the money, FSB is beyond stretched now and the options in bios for Dram ratio mean you have to run 500+fsb to get these ram speeds.
XS users don't see the big picture, ram manufacturers don't make 10 kits for the extreme users who are able to run 2200fsb, they make 5000 kits in the hope that many people will be able to enjoy the added speed, issue is now we are able to provide technology that out paces 90% of the FSB capability of most processors and boards...but end users don't understand that and just RMA.
FSB is dead, its time for IMC if we want to run these uber high ram speeds.

Yeah, i totally agree, but...
...as long as there is no Nehalem in market, EPP2.0 for high speed rated memory is the only way to go. And you know that.

[QuadDamage]

DDR3 action is 7-7-7-22 at 1800, we all know it, tRD dictates it and the limit on quad FSB forces it....

X48 has FSB limitations but enhanced tRD, faster performance at lower FSB, this is exactly why Intel releases Extreme edition CPU's, high FSB no longer needed

Im done with this now, I think you were bold to post 2400MHZ publicly, we have done the same at our labs but what is the point of getting end user hopes up???

ta ta

And you got screenshots to back this up?