***Asus Rampage III Extreme Owners Thread***

[donmarkoni]

This troubles me from SiSanda (latest version)

Can others verify if they see same on other channels?
Thx

And what does CPU Tweaker says about it? Pick each channel in drop box and see (and report please).

[p3gaz_001]

hey guys, wich voltage range do you people need for the 980x @ 4.5ghz ht on? actually i'm using 166x28 with 1.42



yes i use energy saving options since i'm using 4.5ghz as daily use.

[AceBaran]

Hey guys,

Having a little problem here.. Bought a GTX580 put it in pcie lane 1, connected gtx260 in lane 3 and everything runns nicely except the gtx260..
Ive connected it right @ psu..

Any idea ?

[p3gaz_001]

did you switch on the pci-ex 3 lane?

[AceBaran]

did you switch on the pci-ex 3 lane?

Omg this is embarrassing its OFF, i just checked

Tnx man !

[p3gaz_001]

Omg this is embarrassing its OFF, i just checked

Tnx man !

don't worry , it happens

[Andreezy]

hey my system went back to low GFlops and the only way i can run LinX all memory test and have reg Gflops is HT OFF, i had this exact 12GB Kit working with HT ON and 2x Uncore.
so i am stuck now with a CAP on Gflops and no matter what i try it doesnt seem to fix the problem.
i tested each mem module individually and on every mem slot on the motherboard.
while testing each module on single channel i found the optimal peak Gflops ~50gflops @1600mhz and they seem to be stable on RTL's of 54,54,55, so i marked the number on the modules and keyed them into the bios.
then, each module was stable with ~50Gflops per slot on single channel.
i then proceeded to test on dual channel and saw 80Gflops same RTL's same bios settings.

i then proceeded to insert the 3rd module to try and have the most optimal settings on the system and when i test with linx using all memory i only get 54/58Gflops Peak i tried upto 3,600mhz uncore and the most it gets its 58/60gflops......

i pretty much had it with this motherboard, i swear im gonna make it meet the wall.
how can the i7 GulfTown chip move 80Gflops on dual channel and still move about the average 80gflops on triple channel.
seems like some memory dont like all that many core/threads and or intel ed up on the gulfie chips.

[spoolindsm127]

What do we have here?

[zoson]

hey my system went back to low GFlops and the only way i can run LinX all memory test and have reg Gflops is HT OFF, i had this exact 12GB Kit working with HT ON and 2x Uncore.
so i am stuck now with a CAP on Gflops and no matter what i try it doesnt seem to fix the problem.
i tested each mem module individually and on every mem slot on the motherboard.
while testing each module on single channel i found the optimal peak Gflops ~50gflops @1600mhz and they seem to be stable on RTL's of 54,54,55, so i marked the number on the modules and keyed them into the bios.
then, each module was stable with ~50Gflops per slot on single channel.
i then proceeded to test on dual channel and saw 80Gflops same RTL's same bios settings.

i then proceeded to insert the 3rd module to try and have the most optimal settings on the system and when i test with linx using all memory i only get 54/58Gflops Peak i tried upto 3,600mhz uncore and the most it gets its 58/60gflops......

i pretty much had it with this motherboard, i swear im gonna make it meet the wall.
how can the i7 GulfTown chip move 80Gflops on dual channel and still move about the average 80gflops on triple channel.
seems like some memory dont like all that many core/threads and or intel ed up on the gulfie chips.

I retested many settings today with 12gb and 18gb memory. I could not break 63gflops with 3600mhz uncore.

AIDA shows the memory bandwidth is fine... I'm not sure this is directly related to the third channel... I lose performance after 9gb... not 8gb... so the third channel has to be partially in use...

It is a strange problem... I'm not very worried though, my performance seems fine even after starting many virtual machines and using all my memory.

I started a discussion for this issue here:
http://www.xtremesystems.org/forums/...d.php?t=271333

[Andreezy]

I retested many settings today with 12gb and 18gb memory. I could not break 63gflops with 3600mhz uncore.

AIDA shows the memory bandwidth is fine... I'm not sure this is directly related to the third channel... I lose performance after 9gb... not 8gb... so the third channel has to be partially in use...

It is a strange problem... I'm not very worried though, my performance seems fine even after starting many virtual machines and using all my memory.

I started a discussion for this issue here:
http://www.xtremesystems.org/forums/...d.php?t=271333

man i quit on trying HT ON, i settled to run it HT OFF.
i was really getting mad cause it was working before and now its no go..


i dont like the fact that when you load it with lots of ram, it does not reach the same full temp as a >9gb problem load.
ex. when you load a 5Gb problem size it reaches 83C on hottest core, and when you load <9GB it lingers 10C lower than then 5gb problem size test.

AIDA is a Single Threaded Bench Test, i used i7TurboGT to monitor the
multi's and the core/thread when troubleshooting.

[angel_dust]

Hi
Can someone please drop me in a link to where I can get the Marvell 9128 Driver 1.0.0.1051 WHQL for the R3E.
Cheers

[skuldarin]

i dont like the fact that when you load it with lots of ram, it does not reach the same full temp as a >9gb problem load.
ex. when you load a 5Gb problem size it reaches 83C on hottest core, and when you load <9GB it lingers 10C lower than then 5gb problem size test.

i had the same thing for the longest time... no idea what i did to change that but now i get the complete opposite. at first when i saw it change, i thought it was gonna scale based on how much mem was used... but it wouldn't change to higher temps at any setting until i used full mem. i can use 50mb less than full mem and temps are around the same of 10c lower... use full mem and it rises.

[Andreezy]

i had the same thing for the longest time... no idea what i did to change that but now i get the complete opposite. at first when i saw it change, i thought it was gonna scale based on how much mem was used... but it wouldn't change to higher temps at any setting until i used full mem. i can use 50mb less than full mem and temps are around the same of 10c lower... use full mem and it rises.

yeah, that sounds interesting.
def something is going on, you are on the i7 GT (980X) bandwagon with same
and or similar problems.

do some benchies and contribute if you can in zoson's thread.

[skuldarin]

^^
i tried to be nice to him as i don't harbor simple conflicts, even when i share what actually helped on my end (being each system will always be different) i still get spat at so... ohwell

as i stated in his thread, the only thing that got HT to get the high gFlops that i was getting wih HT turned off was by having CR on mem at 1 instead of 2. in doing so though, having CR at 1 wont allow Linx to pass full mem test. it'll pass at 99.9% full mem test... but add a few MB's for full and it stops on error every time, go just below and it passes every time with the same 96-ish gFlops i get with HT turned off (cpu at 4.53ghz on both settings). i've done just about ever adjustment possible just as Zoson had done with no good results for full mem testing.

for now i've kept HT off as i don't use any programs other than CAD that uses more than about 4 cores at once... plus summer temps are coming around and with HT on at 4.5ghz, at 65-70*F currently cpu temps at full load are in low 70's C across the board and the average summer temp in my area ishigh 80's to low 90's F constantly which i fear may leave my temps too high for daily comfert.

EDIT: then again, the temps i'd get are roughly the same as the full load temps i got with my 4.7ghz in my sig if i kept HT on. my laptop's got a 2.53ghz extreme chip in it (can't remember off the top of my head, part of the Q9 series i think) and i OC it to 3ghz (in an Asus VPw90 laptop, it's got awesome cooling) and when i run LinX on it, i don't let it do more than 3 roungs cause it climbs through the 90's lol and i've ran it OC'd every day since i got it back in 07-08? so if anything i'll just run HT off and do 4.7... if only 1.55 vcore wasn't so degrading

[Andreezy]

thats why zoson yells @ you......
he asked for proof and support for the bug that some of us Gulftown
users have and you become a storyteller........no pun
please dont get mad, im just saying it in more plain terms.

you need to document your stories and let us see the facts...

dont get mad, i know maybe you want to help and contribute and
are also very anxious/young.
we are all smarties here... just slow down and absorb more before you contribute , you have a great system
keep sharing your info with us with better documentation.

[tistou77]

No problem. Sorry for late reply. Here you go:

4700/4000/3600/2000

Target CPU Frequency		4686MHz
Target DRAM Frequency		1992MHz

LN2 Mode:			Disabled***
QPI Loadline Calibration:	Enabled***

Sync Mode			[Enabled] ** Gone since BIOS 0003/1301
AI Overclock Tuner		[Manual]
OC From CPU level Up		[Auto]
2D Performance Booster		[Disabled]
CPU Ratio Setting		[33.0]
CPU Turbo Power Limit		[Disabled]

>CPU Configuration----------------------------
CPU Ratio Setting		[33.0]
C1E Support			[Enabled]
Hardware Prefetcher		[Enabled]
Adjacent Cache Line Prefetcher	[Enabled]
MPS and ACPI MADT ordering	[Modern]
Intel(R) Virtualization Tech	[Enabled]
CPU TM Function			[Disabled]
Execute Disable Bit		[Enabled]
Intel(R) HT Technology		[Enabled]
Active Processor Cores		[All]
A20M				[Disabled]
Intel(R) SpeedStep(TM) Tech	[Enabled]
Performance/Watt select		[Power]
Intel(R) C-STATE Tech		[Enabled]
C State package limit setting	[Auto]
C1 Auto Demotion		[Enabled]
C3 Auto Demotion		[Enabled]
----------------------------------------------

BCLK Frequency			[142]
PCIE Frequency			[100]
DRAM Frequency			[1992]
UCLK Frequency			[3986]
QPI Link Data Rate		[6833]

Memory Recheck			[Disabled] ** BIOS 1301
Memory Configuration Protect	[Disabled]

>DRAM Timing Control--------------------------
CAS# Latency		      7	[7]
RAS# to CAS# Delay	      7	[7]
RAS# PRE Time		      7	[7]
RAS# ACT Time		     20	[20]
RAS# to RAS# Delay	      3	[3]
REF Cycle Time		     60	[60]
WRITE Recovery Time	      9	[9]
READ to PRE Time	      5	[5]
FOUR ACT WIN Time	     20	[20]
Back-To-BackCAS# Delay	      0	[Auto]

Timing Mode		     1N	[1N]
Round Trip Latency on CHA    53	[53]
Round Trip Latency on CHB    55	[55]
Round Trip Latency on CHC    57	[57]
tRL Channel A			[Auto] ** BIOS 0003
tRL Channel B			[Auto] ** BIOS 0003
tRL Channel C			[Auto] ** BIOS 0003

WRITE To READ Delay(DD)	      8	[Auto]
WRITE To READ Delay(DR)	      8	[Auto]
WRITE To READ Delay(SR)	     19	[Auto]
READ To WRITE Delay(DD)	      8	[Auto]
READ To WRITE Delay(DR)	      8	[Auto]
READ To WRITE Delay(SR)	      8	[Auto]
READ To READ Delay(DD)	      7	[Auto]
READ To READ Delay(DR)	      6	[Auto]
READ To READ Delay(SR)	      4	[Auto]
WRITE To WRITE Delay(DD)      7	[Auto]
WRITE To WRITE Delay(DR)      7	[Auto]
WRITE To WRITE Delay(SR)      4	[Auto]
----------------------------------------------

CPU Differential Amplitude	[Auto]
CPU Clock Skew			[Auto]
IOH Clock Skew			[Auto]
------------ Extreme Engine Digi+ ------------
Digi+ PWR Mode			[T-Balanced]
PWR Volt. Control		[Auto]
Load-Line Calibration		[Full]
CPU Voltege OCP			[Disabled]
CPU PWM Frequency		[Auto]

Extreme OV			[Enabled]*
Extreme OC			[Auto]

CPU Voltage		1.468	[1.47500]
CPU PLL Voltage 	1.349	[1.35150]
QPI/DRAM Core Voltage	1.376	[1.38750]
DRAM Bus Voltage	1.746	[1.74900]

>DRAM REF Voltages----------------------------
DRAM DATA REF Voltage on CHA	[+0mV]
DRAM CTRL REF Voltage on CHA	[+0mV]
DRAM DATA REF Voltage on CHB	[+0mV]
DRAM CTRL REF Voltage on CHB	[+0mV]
DRAM DATA REF Voltage on CHC	[+0mV]
DRAM CTRL REF Voltage on CHC	[+0mV]
----------------------------------------------

IOH Voltage		1.151	[1.15275]
IOH PCIE Voltage	1.508	[Auto]

ICH Voltage		1.111	[Auto]
ICH PCIE Voltage		[Auto]

*QPI Voltage OCP		[Enabled]
*IOH Voltage OCP		[Enabled]
*DRAM Voltage OCP		[Enabled]
*QPI PWM Frequency		[Auto]
*IOH PWM Frequency		[Auto]
*DRAM PWM Frequency		[Auto]

---------- Spread Spectrum Control -----------
CPU Spread Spectrum		[Disabled]
DRAM Spread Spectrum		[Disabled]



***on-board jumper dependant

you always have the bios 0003 or is the 1301?

thanks

[donmarkoni]

you always have the bios 0003 or is the 1301?

thanks

Yeah, 0003 all the way. Tried 1301, but the other one "felt" better.

[tistou77]

Yeah, 0003 all the way. Tried 1301, but the other one "felt" better.

ok, I retest the 0003, because since the 12xx and 1301, there is a loss of performance with my SSD (I think it comes from the bios)

thanks

[cx-ray]

The question I want to ask is what is the best RAID stripe size to use with two Crucial RealSSD C300 256GB drives in RAID0 and will windows 7 automatically
set the stripe size to the best for me when I create my drive partitions using the windows 7 DVD or would I have to set the stripe size manually?

If I have to set the stripe size manually, what is the best method of doing this and does the stripe size have to be set before createing any drive partitions or after?

Thanks for responding.

You have to set the stripe size in the Intel or Marvell (do not recommend that controller) ROM config screen. First enter the BIOS to enable RAID for either controller. Then during boot you'll be prompted to enter Ctrl I for the Intel or Ctrl M for the Marvell setup screens.

I strongly recommend that you just enable the Intel RAID. Use 128K stripe for RAID 0 on the C300 256GB drives. Also, if you're not using the Marvell controller for anything else, completely disable it.

[Sylver123]

You have to set the stripe size in the Intel or Marvell (do not recommend that controller) ROM config screen. First enter the BIOS to enable RAID for either controller. Then during boot you'll be prompted to enter Ctrl I for the Intel or Ctrl M for the Marvell setup screens.

I strongly recommend that you just enable the Intel RAID. Use 128K stripe for RAID 0 on the C300 256GB drives. Also, if you're not using the Marvell controller for anything else, completely disable it.

Thanks mate for your response.

I tested RAID on the Marvell Controller and people were right, the performance using this controller is poor when using two Crucial RealSSD C300 256GBs and so is not worth it compared to keeping them setup as single drives.

This morning, I was planning on giving the Intel Controller a test with RAID using my two Crucial SSDs but realized unforchantly that the Modified RE3 1301 bios I am using from the thread post below breaks RAID for the Intel Controller.

http://www.xtremesystems.org/forums/...&postcount=432

[NathObeaN]

I am thinking of selling my Rampage III Extreme, along with the full board EK waterblock (black/nickel). How much are these things worth second hand now days?

If anyone is interested, message me of course, but I am not 100% set on selling it yet, I want to find out how much it's worth first.

I haven't hardly used the board and it's features and I have hardly overclocked it simply because I haven't had the time. Shame tbh as I love my setup, but it seems a waste to have this ultra high end board and not use the features which I paid for. So I will probably sell it on and swap out for a SB instead as I am trying to cut back on the electricity costs too.

Obviously I would end up selling my i7 930 + 6GB G.Skill Tridents too but I will sell them seperately most likely, so how much do you think the REIII is worth people?

Thanks.

[Sylver123]

Hello!

Well I have managed to set up RAID0 using my two RealSSD C300 256GBs on the Intel Controller of my RE3 and I can tell you, there is a massive difference in performance on the Intel Controller compared to the Marvell Controller.

Can anyone please tell me whether this performance results are prittay good or not and is there anyway I can improve the performance more?

I have the latest Intel Rapid Storage Technology drivers & application installed and I am using a 128K stripe size.

I have also made sure to enable write-back cache in the Intel application.


Thanks for responding.

Sylver123.

[Tergon]

Won't need them similar system, same problem with the motherboard tray, fully configed system, dual raid SSD, as well as 2x750 raid, 2x580 EVGA SC, 990x@4448, 12 gig dominator ram, fully stable in both games and benchmarks, without the Morlex connected.

Cheers

I will be running 2 580's and a RAID card. It would be nice if ASUS gave a little more detailed information in their manual.
Many thanks for the info!

[Tergon]

Loving it, very fast and quite a cool system, HW monitor at boot up is 10C on the first core, recently relapped the TRUE that might be why, also was having overclocking issues with new build, crashing and stuff nothing was stable, reseated the TRUE with a thinner coat of AS5 and we are good to go. Been reading this thread since the beginning, four days ago, I am all caught up, still on page 175 or so. I few tips that helped.
QPI by default is on Auto is too high, moved to 1.25
PLL voltage to high as well went down to 1.65
IOH voltage up to 1.25.

This gave me a sold overclock to 4448 with a 990x,
Multi 31
BLCK 143
PCIE 102
Dram 2006
Uncore 3011
PWRMode T-balence
LLC Full
Vcore 1.35
PLL 1.65
QPI 1.25
Dram Bus 1.65
IOH 1.25

Thanks all for the help full hints posted will be following along.

Asus Rampage Extreme 3 BIOS 1301
Board reseated with AS5
Intel 990xEE@4.4 Batch # 3045A752
ThermalRight 120 extreme with AS5
Corsair Dominator GT 12gig 9,10,9,27 2054 MKz
Corsair 2xF120 SSD RAID on ICH10R, 2x750G Seagate
Evga 2x580 SC in SLI
Temjin TJ10 case
6xSythe 1600 FDB 120mm fans
Corsair AX1200
Logitech X-540 5.1
Asus VE278Q 1920x1080
Windows 7 SP1 64bit

[Andreezy]

nice chip you got there, regardless it being only 2gb test under LinX.
~1.38Vcore is very nice @ +4.4Ghz, you need to retest under LinX and select all memory to fully test your hardware.
if you test again with all memory, please let us know