UnOfficial Asus Rampage II Extreme Thread

[maGSImum]

Currently stressing my i7 920 D0 FPO#3845B071:

LinX max mem, 5x, passed.
Stressing for the night with P95 small ftt.

Settings:

BCLK: 214
Multi: 21 (4494Mhz)
Memory: 1284Mhz 8-8-8-20-1T (will oc the memory tomorrow)
Uncore: 2782
QPI: 7704

Vcore: 1.3625
CPU PLL: 1.88
QPI/DRAM: 1.375
IOH: 1.365
IOH PCI: 1.696
ICH: 1.3
ICH PCI: 1.6

Turbo: off
Hypertreading: on

Stressed temps are:
Core's: 75~80 C
NB: 61 C
SB: 58 C

Let's hope it's still running in the morning...

[GAR]

I'm gonna say this again, overclocking PCI-E bus is completely pointless. If you want to get better graphics performance OC your graphics card, not the PCI-E bus.

And who are you, the god of PCI-E? overclocking the pci-e bus is known and from my own experience to give you a bit of extra performance, this is XS forums where every single point or fps matters, so enless you have facts, i suggest you dont act so confident on your opinion.

[BLHealthy4life]

I'm gonna say this again, overclocking PCI-E bus is completely pointless. If you want to get better graphics performance OC your graphics card, not the PCI-E bus.

If I remember right, overclocking the PCIe bus slightly can actually increase overall stability and enhance system overclocking in some cases.

Just to strengthen this point, The EVGA Classified review over at Anandtech showed that increasing the PCIe bus on that board, they were able to increase bclk...

[dejanh]

And who are you, the god of PCI-E? overclocking the pci-e bus is known and from my own experience to give you a bit of extra performance, this is XS forums where every single point or fps matters, so enless you have facts, i suggest you dont act so confident on your opinion.

No, I'm not the god of PCI-E overclocking and no you will not gain any performance from overclocking the PCI-E that you cannot gain by simply overclocking your video card.

Do you even know what the point of overclocking PCI-E is actually? In fact, do you even know what you are "overclocking" by OCing the PCI-E?

Now, on the Anandtech note I am aware that they mention PCI-E as a way of pushing the BCLK past a certain point but this is very much related to EVGA mobos and E-LEET software. You will not be able to boot with a BCLK past 222MHz in 99.99% of cases. If anything, you will simply end up making the system very unstable in addition to quite likely corrupting your drives, if not immediately over time for sure. It most certainly does not increase system stability nor does that review even imply that.

Unfortunately, unless you have an extremely BCLK friendly chip you are pretty much out of luck doing this on the R2E, and OCing PCI-E will be a useless effort that will give you more headaches than anything.

[dinos22]

we do see gains in 3d benching dejan by OCing the PCIe bus
at least that was the case with all previous generations of chipsets

i would like to hear your take on PCIe bus OC and why it doesnt affect anything though

[cstkl1]

dejan overclocking pcie bus does allow u to go higher bclk

but this also depends on ioh/ich voltage

and certain cards like the 9000 series for nvidia their clock gen is based on pcie clock. clear example would be 9800gx2.

also if u want to see the relationship of the board clockgen lookup in set fsb.
when u hit a certain bclk hang.
when u reboot try then again before increasing to that bclk that hang. increase the pcie clock.

for 3dmark benching
example 9800gx2 756/1654/2200 on pcie 100
is the same as 702/1584/2200 on pcie 110. requires less voltage also.

try it.

PS GAR u crack me up kekekek

" God of pcie "

feel like putting that on a siggy

[tistou77]

hello

what is the best bios for the current performance of the ram, 1204 or 1306?

and is what it's best to disable the LLC to avoid the load vdroop in vCore or not?

with LLC enabled:
VCore in bios 1.24v in idle: 1.24v and load: ~ 1.28v

thank you very much

[dreamaxx]

Mid May... then late May... maybe June... maybe... as always, who always wait, doesn't buy anything

Rofl... Thats me :P

[GAR]

No, I'm not the god of PCI-E overclocking and no you will not gain any performance from overclocking the PCI-E that you cannot gain by simply overclocking your video card.

Do you even know what the point of overclocking PCI-E is actually? In fact, do you even know what you are "overclocking" by OCing the PCI-E?

Now, on the Anandtech note I am aware that they mention PCI-E as a way of pushing the BCLK past a certain point but this is very much related to EVGA mobos and E-LEET software. You will not be able to boot with a BCLK past 222MHz in 99.99% of cases. If anything, you will simply end up making the system very unstable in addition to quite likely corrupting your drives, if not immediately over time for sure. It most certainly does not increase system stability nor does that review even imply that.

Unfortunately, unless you have an extremely BCLK friendly chip you are pretty much out of luck doing this on the R2E, and OCing PCI-E will be a useless effort that will give you more headaches than anything.

Like i said before, your words mean nothing intill you can back it up with some hard facts, like Dino said, it has been the case with previous chipsets, both intel and nvidia to give better numbers when benching, for everyday use, i agree, overclocking your pci-e bus is pointless, but not when you are trying to get high numbers you need every little fps.

[dejanh]

Like i said before, your words mean nothing intill you can back it up with some hard facts, like Dino said, it has been the case with previous chipsets, both intel and nvidia to give better numbers when benching, for everyday use, i agree, overclocking your pci-e bus is pointless, but not when you are trying to get high numbers you need every little fps.

dejan overclocking pcie bus does allow u to go higher bclk

but this also depends on ioh/ich voltage

and certain cards like the 9000 series for nvidia their clock gen is based on pcie clock. clear example would be 9800gx2.

also if u want to see the relationship of the board clockgen lookup in set fsb.
when u hit a certain bclk hang.
when u reboot try then again before increasing to that bclk that hang. increase the pcie clock.

for 3dmark benching
example 9800gx2 756/1654/2200 on pcie 100
is the same as 702/1584/2200 on pcie 110. requires less voltage also.

try it.

PS GAR u crack me up kekekek
feel like putting that on a siggy

Read the second quote there. I did not say that PCI-E makes no difference, I said that you can get the same effect by OCing your graphics card. OCing PCI-E will basically OC your graphics bus as well so you get higher performance. Same can be done by overclocking the graphics on their own without causing problems for the rest of the system. OCing PCI-E is plain dumb, and very much pointless.

Now, on the note of higher BCLK being possible by OCing the PCI-E, you will first and foremost need a BCLK friendly chip, and then and only then will you have any use of pushing up the PCI-E to get higher BCLK. Not to mention that you also need to have a half-decent board.

[ZCommander]

Wow, really guys, overclocking with this mobo is so challenging compared to gigabyte EX58-Extreme!

with EX58 I could reach 3.8 with CPU Vcore of 1.35v only.
The R2E on the other hand is at 3.48 with CPU Vcore of 1.375 !!!!!!!!!!

I'm currently testing with Prime 95 at 3.48x GHz I wanted to reach 3.6 but it seems I'll have to pump more voltage which I don't like!

[Jonny Manc]

Currently stressing my i7 920 D0 FPO#3845B071:

LinX max mem, 5x, passed.
Stressing for the night with P95 small ftt.

Settings:

BCLK: 214
Multi: 21 (4494Mhz)
Memory: 1284Mhz 8-8-8-20-1T (will oc the memory tomorrow)
Uncore: 2782
QPI: 7704

Vcore: 1.3625
CPU PLL: 1.88
QPI/DRAM: 1.375
IOH: 1.365
IOH PCI: 1.696
ICH: 1.3
ICH PCI: 1.6

Turbo: off
Hypertreading: on

Stressed temps are:
Core's: 75~80 C
NB: 61 C
SB: 58 C

Let's hope it's still running in the morning...

Gonna keep an eye on this as I have the same stepping D0 sat on my desk, along with an EX58-UD5....waiting on some GSkill and my exams to fini.
Please let me know how you get on :P

[GAR]

Read the second quote there. I did not say that PCI-E makes no difference, I said that you can get the same effect by OCing your graphics card. OCing PCI-E will basically OC your graphics bus as well so you get higher performance. Same can be done by overclocking the graphics on their own without causing problems for the rest of the system. OCing PCI-E is plain dumb, and very much pointless.

Now, on the note of higher BCLK being possible by OCing the PCI-E, you will first and foremost need a BCLK friendly chip, and then and only then will you have any use of pushing up the PCI-E to get higher BCLK. Not to mention that you also need to have a half-decent board.

Nevermind dude, seems like nothing gets through to you enless it comes from your own brain. To each his own. The point is we like to overclock everything, if i could overclock my phone i would do it, if i could overclock my PS3 id do it.

[GAR]

Anyone try the W3520 on this board yet? which memory speeds are available? hows the overclocking?

[dejanh]

Anyone try the W3520 on this board yet? which memory speeds are available? hows the overclocking?

W3520 has locked multipliers for DRAM/Uncore/QPI. They appear to be hard locked on the chip so far...

In terms of overclocking, you can expect to get about 4.4GHz out of the chip pretty safely.

[jcniest5]

Well, I see that we disagree on OCing the PCI-E. I guess there are two sides to it. 1) Raising the PCI-E clock actually helps with getting bclk to go a little higher. 2) If you're doing that just to OC your video card, then there's very little benefit contrary to actually OCing the video card. I think we have to see both sides to come to an agreement. Each has its own benefit.

[GAR]

W3520 has locked multipliers for DRAM/Uncore/QPI. They appear to be hard locked on the chip so far...

In terms of overclocking, you can expect to get about 4.4GHz out of the chip pretty safely.

That sucks. I might as well search for a D0 920 then.

[dinos22]

W3520 has locked multipliers for DRAM/Uncore/QPI. They appear to be hard locked on the chip so far...

In terms of overclocking, you can expect to get about 4.4GHz out of the chip pretty safely.

you sure about that man
have you tested it

[dejanh]

you sure about that man
have you tested it

Yes. I saw a testing session and I can tell you that only two multipliers were available for DRAM and there was no adjustments for Uncore and QPI frequency. I also have a PM in my inbox from an owner of a W3520 who confirmed these findings for me on an R2E.

Peter (Shamino) has been dealing with this for the Classified as well afaik. Not 100% sure on the hard-locked portion as it could be that the locks can be circumvented by BIOS, but the parts I've seen so far were all locked and they were retail chips running on R2E and Classified board.

[maGSImum]

Gonna keep an eye on this as I have the same stepping D0 sat on my desk, along with an EX58-UD5....waiting on some GSkill and my exams to fini.
Please let me know how you get on :P

As said, 214x21 was 5x LinX max-mem stable.
I started Prime95 for the night, this failed on 1 of the 8 threads after 3.25 hours.

Because the weather in the Netherlands has changed, running the i7 on 4500Mhz with an ambient of 22-23 degrees lets it run all the way up to 83 degrees on the hottest core, which is a bit to much for a good night sleep in my opinion.
I still have to lap the bastard since it isn't flat at all.

So now i'm testing my max BCLK, best result so far :
227x19 (4313)
1.275 Vcore
15X LinX stable
PCI 103 (This unlocked 2Mhz of BCLK for me)
Stressed temps now are 63-69 degrees.

[dopestuff]

As said, 214x21 was 5x LinX max-mem stable.
I started Prime95 for the night, this failed on 1 of the 8 threads after 3.25 hours.

Because the weather in the Netherlands has changed, running the i7 on 4500Mhz with an ambient of 22-23 degrees lets it run all the way up to 83 degrees on the hottest core, which is a bit to much for a good night sleep in my opinion.
I still have to lap the bastard since it isn't flat at all.

So now i'm testing my max BCLK, best result so far :
227x19 (4313)
1.275 Vcore
15X LinX stable
PCI 103 (This unlocked 2Mhz of BCLK for me)
Stressed temps now are 63-69 degrees.

have you tried higher pci-e clocks than that or are you gonna??
what cooling are you using?? (components?)
HOLLAND!

[Brama]

Yes. I saw a testing session and I can tell you that only two multipliers were available for DRAM and there was no adjustments for Uncore and QPI frequency. I also have a PM in my inbox from an owner of a W3520 who confirmed these findings for me on an R2E.

Peter (Shamino) has been dealing with this for the Classified as well afaik. Not 100% sure on the hard-locked portion as it could be that the locks can be circumvented by BIOS, but the parts I've seen so far were all locked and they were retail chips running on R2E and Classified board.

First the bad news: I own a Xeon W3520 and I confirm that with BIOS 1306, there are the locks described.

Last the good news: a lot of guys with Xeon W35xx assure that on DFI and Gigabyte MB they are fully unlocked as i7.

[Brama]

So now i'm testing my max BCLK, best result so far :
227x19 (4313)
1.275 Vcore
15X LinX stable
PCI 103 (This unlocked 2Mhz of BCLK for me)
Stressed temps now are 63-69 degrees.

Which QPI voltage did you use?
Did you touch any other voltage/setting to arrive at 227 MHz of BCLK?

[BLHealthy4life]

I just had D0 arrive earlier today and installed it a few hours ago...

I thought it was my C0 that wouldn't allow me to go over 200bclk stable, however, I cannot get over 200bclk with this D0.

I am very confused.

For my C0 I had the following settings:

200x21
1.448 Vcore
1.508 VTT
1.336IOH
1.68vDIMM @ 2005 7-8-7-60-1N

I cannot run the D0 over 200bclk even at these voltages which should be PLENTY for a D0.

I've even dropped the voltages to see if it would allow me to go over 200bclk and it hasn't helped...

EDIT: Used setting in post #1851 and I'm at 214x21.

[GAR]

Ok, i got my D0 Xeon W3520, it does 4.2ghz with 1.24v!!! amazing cpu, the multi's are locked, but all i needed was 1600mhz memory which is unlocked, all in all a good $300 purchase.