CellShock DDR3-1800 8-7-6-21 Testing!

[camouflage]

I think with the Cellshocks I can break the 32M-10min-barrier soon:



Of course the RAM is not maxed out here - but very fast !!!

[SeLecT]

what vmch?

I don't know the real Voltage, but it was default + 0,325V. I will test with lower vmch in the next days.
Yesterday evening i tried to come under 12 s in 1M, but it failed....

With higher Vcore the system freezes...

4GB @ 935Mhz @ 8-8-8-21

[saaya]

+ 0,325V... default is 1.25v so this should be 1.575v... thats really too low for p35, whats the highest you can set?

[SeLecT]

+0,375 is the highest voltage...




http://www.xtremesystems.org/forums/...light=P35T-DQ6


Edit:

SuperPi 32M @ 4GB @ 800Mhz @ 7-6-5-15

[BenchZowner]

Finally decided on what to do with the round-up, and started the main tests today.

Here are some findings for this kit:

Mainboard: Asus P5K3 Deluxe ( BIOS 0810 )

Tightest stable timings for DDR3-1800 1.8V ( Stock Main Timings & Freq ):
[ in BIOS order ]
1T Command Rate
8-7-6-21-1-48-4-4-1-E-D-E(0)

Tightest stable timings for max 5-5-4-15 tests @ 2.25V & DDR3-1535:
[ in BIOS Order ]
1T Command Rate
5-5-4-15-1-36-3-4-1-E-D-E(0)

More tests...

Tightest stable timings for max 6-5-5-16 tests @ 2.25V & DDR3-1783:
[ in BIOS order]
1T Command Rate
6-5-5-16-1-48-4-4-1-E-D-E(0)

Not wanting to say something without being 100% sure, but it looks like this kit doesn't like 7-6-5-21 timings that much...max stable at 7-6-5-21 is probably around 1850...we'll find out soon ( currently progressing with the tests )

[saaya]

and with 876? can you get any higher than 1850 with that?

what about 665-18 tightest timings?

[BenchZowner]

and with 876? can you get any higher than 1850 with that?

what about 665-18 tightest timings?

Are you holding your breath baby ?

Tightest stable timings for max 8-7-6-21 tests @ 2.25V & DDR3-2052(!):
[ in BIOS Order ]
1T Command Rate
8-7-6-21-7-60-7-7-5-D-D-E(0)

{ note: this one is memtest only...gotta check Windows stability! }

Tightest stable timings for max 6-6-5-18 tests @ 2.25V & DDR3-1804:
[ in BIOS Order ]
1T Command Rate
6-6-5-18-1-48-4-4-1-E-D-E(0)

Tightest stable timings for max 7-7-6-21 tests @ 2.25V & DDR3-2022:
[ in BIOS Order ]
1T Command Rate
7-7-6-21-7-60-7-7-3-D-D-E(0)

{ note: this one is memtest only as well..gotta check windows stability! }

Break time...and other work time as well...

p.s. the roundup will be ready soon! finally!

[webwilli]

My little results.
Next step: DDR3-2000

Asus P5E3 Deluxe
VDIMM 1,90 (Bios)
Chipset Intel X38
fsb/dram 1:2
Cellshock DDR3-1800
Speed 900,0 MHz
Size 2 x 1024

Timings

CAS# Latency (tCL) = 7
RAS# to CAS# Read Delay = 6
RAS# to CAS# Write Delay = 6
RAS# Precharge (tRP) = 5
Precharge Delay (tRAS) = 15

Others Timings

All Precharge to Activate = 6
Refresh Cycle Time (tRFC) = 50
Performance Level = 6
Write to Precharge Delayed = 15
Write to Read Delayed (tWTR) = 15
Act. to Act. Delayed (tRRD) = 2
Read to Write Delay (tRD_WR) = 8
Read to Precharge (tRTP) = 2
All Precharge to Refresh delay = 3
Refresh Period (tREF) = 16383T
Command Rate (CR) = 1T





Asus P5E3 Deluxe
VDIMM 2,00 (Bios)
Chipset Intel X38
fsb/dram 1:2
Cellshock DDR3-1800
Speed 950,0 MHz
Size 2 x 1024

Timings

CAS# Latency (tCL) = 7
RAS# to CAS# Read Delay = 6
RAS# to CAS# Write Delay = 6
RAS# Precharge (tRP) = 5
Precharge Delay (tRAS) = 15

Others Timings

All Precharge to Activate = 6
Refresh Cycle Time (tRFC) = 60
Performance Level = 6
Write to Precharge Delayed = 17
Write to Read Delayed (tWTR) = 16
Act. to Act. Delayed (tRRD) = 2
Read to Write Delay (tRD_WR) = 8
Read to Precharge (tRTP) = 2
All Precharge to Refresh delay = 6
Refresh Period (tREF) = 16383T
Command Rate (CR) = 1T

[saaya]

very usefull information benchzowner and webwilli!
thanks for sharing, cant wait to see that roundup

[Achill3uS]

join da club guys
i love this ram's

some results with,

Asus P5K3 Deluxe (still unmodded)
Wolfdale E8400 watercooled 1.59v
CellShock DDR3 1666MHz 6-6-5-15-1-48-4-4-1-DDE(0) 1T 2.10v

32M



Everest



can't wait to the next sub-zero test

[Achill3uS]

I think with the Cellshocks I can break the 32M-10min-barrier soon:

[IMG]

Of course the RAM is not maxed out here - but very fast !!!

damn Frank, that cpu is killer
can you tell the fpo?

[SeLecT]

Achill3uS, can you run with DDR2-1000?

btw nice results!

[Achill3uS]

Achill3uS, can you run with DDR2-1000?

btw nice results!

yep, I had run the sticks on 1000mhz, but only for a cpuz valid, benchmarks are so unstable i think its mobo limit (maybe the 1.70 vmch was low), but i didnt tryd yet with more vdimm...

1000mhz 7-6-5-18 1T 2.20v

thanks

[SeLecT]

SuperPi 1M is unstable too?

But 1000Mhz @ 7-6-5-18 ist nice... Will you test with another mobo?

[3oh6]

yep, I had run the sticks on 1000mhz, but only for a cpuz valid, benchmarks are so unstable i think its mobo limit (maybe the 1.70 vmch was low), but i didnt tryd yet with more vdimm...

1000mhz 7-6-5-18 1T 2.20v

thanks

yup, my P5K3-Dlx needed more vMCH for 1000MHz to be stable for single 32M at even 8-7-6. i still haven't been able to get any kits over 960MHz dual32M stable though. the Maximus Extreme has taken care of that though so i am confident it is a motherboard limit.

one quic question Achill3uS, when your listing your settings, are you listing the voltage set in the BIOS for vDIMM or measured with a DMM? TIA

[Achill3uS]

SuperPi 1M is unstable too?

But 1000Mhz @ 7-6-5-18 ist nice... Will you test with another mobo?

here is your 1M
validation
1000mhz 7-6-5-18 1T
still 2.20v



my mate jugeen has got p5e3 ws pro, and tested with CellShocks, but for wolfdale i dont think x38 is a good choice...

[Achill3uS]

yup, my P5K3-Dlx needed more vMCH for 1000MHz to be stable for single 32M at even 8-7-6. i still haven't been able to get any kits over 960MHz dual32M stable though. the Maximus Extreme has taken care of that though so i am confident it is a motherboard limit.

one quic question Achill3uS, when your listing your settings, are you listing the voltage set in the BIOS for vDIMM or measured with a DMM? TIA

set in the bios (not measured), thanks for the confirm by the limitation, the mobo is unmodded till i get back my soldering iron
btw, how much vmch give you for 24/7 ? I've seen hipro using 1.9v on Blitz Extreme

[3oh6]

set in the bios (not measured), thanks for the confirm by the limitation, the mobo is unmodded till i get back my soldering iron
btw, how much vmch give you for 24/7 ? I've seen hipro using 1.9v on Blitz Extreme

thanks for the info.

well, this motherboard doesn't see 24/7, it is just used for benching and memory clocking. with that said, the last month has seen it running pretty much 24/7 running Prime, HCI Memtest, SPI, and looping 3DMark 01/06 for long term stability testing on memory kits.

i have had the vMCH at 1.71v~1.81v for most if not the entire time. at some points it has been up to 2.00v but for the long term stability it has stayed around the 1.71v~1.81v range. i am just using a Noctua NC-U6 for cooling with a 2300RPM 120mm fan over the memory blowing on the NB as well.

[maverik-sg1]

In an asus P5K3 board what BIOS setting refers to the term vMCH and what would it need to be (assuming the board is volt modded to adjust higher than BIOS normally allows) to get the best from the RAM?

Great stuff this cellshock.

[BenchZowner]

The NorthBridge Voltage (Vnb) maverik-sg1.
1.70V ( max available with a non-modded P5K3 ) is ok for daily usage.
Up to 1.9V should be ok with efficient chipset cooling.

[maverik-sg1]

Nice one BenchZowner, thanks for info.

On the P5K3 I have northbridge modded so I can set for benching with 1.91v and a nice 80mm fan blowing on NB, and one on each of the MOSFETs sinks all should be well with benching for a few hours or more?

Might try and do something tomorow or monday on this

[BenchZowner]

Yeah it should be more than fine for benching hours

[SeLecT]

here is your 1M
validation
1000mhz 7-6-5-18 1T
still 2.20v

http://img174.imageshack.us/img174/2...mxtwinzrp8.jpg

my mate jugeen has got p5e3 ws pro, and tested with CellShocks, but for wolfdale i dont think x38 is a good choice...

Thanks! Its nice!

In a few days we will test the p5e3 ws pro with LN2 and a QX9650. Of course with these Cellshocks too

(A closer look to saaya: Blue Kit? )

The NorthBridge Voltage (Vnb) maverik-sg1.
1.70V ( max available with a non-modded P5K3 ) is ok for daily usage.
Up to 1.9V should be ok with efficient chipset cooling.

I hope i will get my Blitz Extreme back soon, then i will try to reach the 1000Mhz. But i need a CPU for this

[saaya]

I've seen hipro using 1.9v on Blitz Extreme

afaik he used above 2v but not much above 2v i suspect he noticed it doesnt scale above that.

jody, did your p5k3 still scale up to 2v?

for wolfdale i dont think x38 is a good choice...

hmmm why not? it oced better on p35 boards for you?

(A closer look to saaya: Blue Kit? )
I hope i will get my Blitz Extreme back soon, then i will try to reach the 1000Mhz. But i need a CPU for this

sorry robert... im still in my hometown up in the north and a couple of blue kits together with your blitz board are still in southern germany.
I didnt think you would need it... sorry... my plan was to go back to the south after new year, then pick up the hardware and head over to munich and fly to ces.

happy 2008 to all of you!!! great to have so much usefull info in one thread

achill3us, please forward my greetings to jo!
Hope his gf is feeling better!

[3oh6]

jody, did your p5k3 still scale up to 2v?

i wouldn't use the word scale but 2v was necessary for some of the crazy things i was doing above 500FSB with tight PLs on the 266 strap. i went from 490FSB PL6 1:2 w/ 1.85v to needing 1.95v for 500FSB PL6 1:2 and then 2.01v for this run at 515FSB PL6 1:2 266 strap:

4639MHz / DDR3-2062 SPi 32M = 10m 16.750
click for full size including memset...


so i guess it scaled in a sense but it was more like leaped and bound as the FSB went up with the stoopid tight PL on the 266 strap.

dropping PL and i can get away with 1.76v for single 32M at 500FSB 1000MHz 8-7-6 on 200 or 333 strap, 1.81v for 266 strap i think...can't find my notes from that testing right now.