2GB OCZ PC2-9200 FlexXLC DDR2-1150 @670Mhz

[eva2000]

More DDR2 memory goodness thanks to the folks at OCZ Technology. OCZ sent me a pair of 2GB FlexXLC PC2-9200 DDR2-1150 memory modules http://www.ocztechnology.com/product...lexxlc_edition to play with and they arrived today!

The OCZ PC2-9200 FlexXLC came in a new plastic packaging housing the memory as they now have 4 plastic heat sealed tabs on it - mine were intact. So you know if the memory is opened before or sealed from manufacturer (brand spanking new).

The heatsink is tall from below pic, from gold contact to tip of heatsink = 6.5 cm and actual heatsink itself is 5.75cm tall! The barb size seems to be for 1/4" ID tubing as the barb diameter measures ~6mm.





What's so different about these memory modules you ask ? Straight from product page http://www.ocztechnology.com/product...exxlc_edition:

• The new OCZ FlexXLC (Xtreme Liquid Convention) heatsink delivers superior heat dissipation via the integrated hybrid copper and aluminum liquid injection system. The FlexXLC module was engineered with this unique “flexible” design to give enthusiasts the unparalleled option to run the modules passively or water cooled. The concurrent use of both technologies (water-cooling) promotes maximum heat dissipation and pushes thermal management of memory modules one step further to keep up with the ever-increasing frequency demands. FlexXLC modules seamlessly co-migrate with any system upgrade to liquid cooling.
• The PC2-9200 series is engineered with an 8-layer PCB with thermally conductive power and ground planes to guide heat away from the ICs and maintain the necessary signal integrity for high-end modules.
• OCZ EVP (Extended Voltage Protection) is a feature that allows performance enthusiasts to use a VDIMM of 2.4V ± 5% without invalidating their OCZ Lifetime Warranty.

Unfortunately, no photo snap shots from me as yet as my Nikon 5700 went to digital camera heaven http://i4memory.com/showthread.php?t=3802

System:

• Intel Core 2 Extreme X6800 L627A872 B2 Step 6
• Swiftech Apogee/Radiical Triple Fan Single Pass Rad/MCP655 + Kayl 100W Custom PSU
• Asus 680i Striker Extreme 0402 bios (Thermalright HR-05/80x38mm Thermaltake fan on NB, Thermaltake Extreme Spirit II on SB)
• 128MB Gainward FX5200 PCI
• 2x1GB FlexXLC PC2-9200 (cooled with 3x 60x25mm Sunon 23.5cfm MagLev fans sitting on modules)
• 4x80GB Hitachi 7K80 SATAII Raid 0 on SATA3 to SATA6 ports
• Pioneer DVD Burner
• 600W OCZ EvoStream

Update: April 3, 2007

I got my new digital camera yesterday, Fujifilm Finepix S9600 to replace my Nikon 5700 who's CCD sensor died

Fujifilm Finepix S9600's super macro mode of 1cm really hard to get to work as it disables flash (unlike on my Nikon 5700) so photos need more natural light I think. So used the normal macro mode of 10cm to take the OCZ PC2-9200 FlexXLC photos.

Unfortunately, tried using Fujifilm Finepix S9600's image stabilization mode at first (flash disabled automatically) but found out the truth it just cheats by driving up the ISO to 800 causing alot of noise. So for non-flash photos use Paintshop Pro XI's noise reduction to fix it up.

[eva2000]

Initial Memtest86+ v1.70 Results

Doing some quick memtest86+ v1.70 (2 full pass + 10x loops of test #5) to see how they do at default 5-5-5-18 timings with same sub timings manually set as my OCZ PC2-8000 Titanium Alpha VX2 of 5-22-6-8-Auto-7-42-Auto.

Memtest86+ v1.70 & Overclocking using default main timings 5-5-5-18

Settings used:

AI Tuning: Manual
Nvidia GPU Ex: Disabled
LinkBoost: Disabled

FSB - Memory Clock Mode: Unlinked
x FSB - Memory Raio: N/A
FSB (QDR), Mhz: 1480 or 1600 (means 370FSB or 400FSB)
Actual FSB (QDR), Mhz: 1480 or 1600
MEM (DDR), Mhz: 1151, 1163, 1184, 1202, 1234, 1269, 1295 or 1200, 1245, 1258, 1280, 1300
Actual MEM (DDR), Mhz: 1150.8, 1162.9, 1184, 1202, 1233.4, 1268.6, 1295 or 1200, 1244.1, 1257.1, 1280, 1300

Vcore Voltage: 1.425 or 1.475v set (1.32v or 1.36v bios - 0505 and higher bios fix the vcore issue but 0402 bios more stable for both my E6600/Q6600 cpus)
Memory Voltage: 2.30v to 2.55v
1.2v HT Voltage: 1.3v
NB Core Voltage: 1.50v to 1.65v
SB Core Voltage: 1.50v
CPU VTT Voltage: 1.3v
DDRII Controller Ref Voltage: AUTO
DDRII Channel A Ref Voltage: AUTO
DDRII Channel B Ref Voltage: AUTO

tCL (CAS Latency): 5
tRCD: 5
tRP: 5
tRAS: 18
Command Per Clock (CMD): 2 clock (2T)

Advance Memory Settings
tRRD: 5
tRC: 22
tWR: 6
tWTR: 8
tREF: AUTO
tRD: 7
tRFC: 42
Async Latency: AUTO

CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled
MCP PCIE Spread Spectrum: Disabled
SATA Spread Spectrum: Disabled
LDT Spread Spectrum: Disabled

CPU Internal Thermal Control: Disabled
Limit CPUID MaxVal: Disabled
Enhanced C1 (C1E): Disabled
Execute Disable Bit: Disabled
Virtualization Technology: Disabled
Enhanced Intel Speedstep Tech: Disabled
LDT Frequency: 5x

PCIEX16_1 Frequency (Mhz): 100
PCIEX16_2 Frequency (Mhz): 100
PCIEX16_3 Frequency (Mhz): 100
SPP<->MCP Ref Clock, Mhz: 200

I undervolted vdimm to 2.3v in bios (shows 2.34-2.35v in bios) at 1.5v NB volts unless otherwise stated and managed so far:

• 2.300v = PC2-1151 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.300v = PC2-1163 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.300v = PC2-1184 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.300v = PC2-1202 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.300v = PC2-1234 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED

To boot even higher with 2x1GB FlexXLC PC2-9200 at PC2-1269 and above needed to raise NB volts from 1.5v to 1.65v

NB volt = 1.65v

• 2.325v = PC2-1269 5-5-5-18 5-22-6-8-Auto-7-42-Auto = FAILED, Test #5 - 2 errors at 2nd pass
• 2.350v = PC2-1269 5-5-5-18 5-22-6-8-Auto-7-42-Auto = FAILED, Test #5 - 10 errors at 7th pass
• 2.350v + SB raised to 1.55v = PC2-1269 5-5-5-18 5-22-6-8-Auto-7-42-Auto = FAILED, Test #5 - 6 errors at 8th pass
• 2.375v = PC2-1269 5-5-5-18 5-22-6-8-Auto-7-42-Auto = FAILED, Test #5 - 4 errors at 31st pass
• 2.400v = PC2-1269 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.425v = PC2-1295 5-5-5-18 5-22-6-8-Auto-7-42-Auto = FAILED, Test #4 - 1 error at 1st pass, Test #5 - 12 errors at 1st pass
• 2.450v = PC2-1295 5-5-5-18 5-22-6-8-Auto-7-42-Auto = FAILED, Test #5 - 10 errors at 6th pass
• 2.475v = PC2-1295 5-5-5-18 5-22-6-8-Auto-7-42-Auto = FAILED, Test #5 - 6 errors at 6th pass
• 2.500v = PC2-1295 5-5-5-18 5-22-6-8-Auto-7-42-Auto = FAILED, Test #5 - 10 errors at 11th pass
• 2.500v = PC2-1295 5-5-5-18 5-22-6-8-Auto-8-42-Auto = FAILED, Test #5 - 1 error at 1st pass

Remember 680i chipsets don't have real unlinked memory dividers, so at 370FSB, the above memory clocks are what I have to work with from the memory dividers available to me at 370FSB. So to play with other memory clock possibilities we have to adjust FSB somewhat

Change from 9x370FSB to 9x400FSB with following results at 1.5v NB volts unless otherwise stated:

• 2.300v = PC2-1200 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.300v = PC2-1244 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.325v = PC2-1257 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED with 1.60v NB volts
• 2.400v = PC2-1280 5-5-5-18 5-22-6-8-Auto-7-42-Auto with 1.65v NB volts = FAILED Test #5 - 4 errors at 4th pass
• 2.425v = PC2-1280 5-5-5-18 5-22-6-8-Auto-7-42-Auto with 1.65v NB volts = FAILED Test #5 - 2 errors at 7th pass
• 2.475v = PC2-1280 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED with 1.65v NB volts looped for 32 minutes error free
• 2.525v = PC2-1300 5-5-5-18 5-22-6-8-Auto-7-42-Auto with 1.65v NB volts = FAILED Test #4 - 1 errors at 1st pass, Test #5 - 2 errors at 1st pass
• 2.525v = PC2-1300 5-5-5-18 5-22-6-8-Auto-7-42-Auto with 1.70v NB volts = FAILED Test #5 - 2 errors at 1st pass
• 2.550v = PC2-1300 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED with 1.70v NB volts

Final Memtest86+ v1.70 Overclocking Results
with default main timings 5-5-5-18

So overall results at default main timings of 5-5-5-18 with same sub timings used for my OCZ PC2-8000 Titanium Alpha VX2 default sub timings are as follows for quick Memtest86+ v1.70 test procedure:

• 2.300v = PC2-1151 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.300v = PC2-1163 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.300v = PC2-1184 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.300v = PC2-1200 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.300v = PC2-1202 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.300v = PC2-1234 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.300v = PC2-1244 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.325v = PC2-1257 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED with 1.60v NB volts
• 2.400v = PC2-1269 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED
• 2.475v = PC2-1280 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED with 1.65v NB volts looped for 32 minutes error free
• 2.550v = PC2-1300 5-5-5-18 5-22-6-8-Auto-7-42-Auto = PASSED with 1.70v NB volts

[eva2000]

Only had time today to boot into windows and test PC2-1300 speeds = 650mhz 5-5-5-18 5-22-6-8-Auto-7-42-Auto at 2.55v bios set vdimm.

CPUZ Validation


CPUZ Tabs


Everest Ultimate Bandwidth & Latency


Super Pi v1.50
1M = 13.969s
2M = 34.063s
4M = 1m 18.969s
8M = 2m 55.313s
16M = 6m 23.594s
32M = 13m 48.453s

[eva2000]

Slow pi times due to very loose timings but we can tighten those up later

That's it for now.. more tests to come



Update:

670mhz 5-5-5-18 6-33-6-11-AUTO-7-48-Auto at 2.55v

Next stop is finding the modules max Super Pi 1M results with default 5-5-5-18 timings at 2.55v but with sub timings closer to the default rated ones listed on OCZ site http://www.ocztechnology.com/product...exxlc_edition:



I decided to use sub timings of 6-31-6-11-AUTO-7-48-Auto and boot from 654mhz 5-5-5-8 into windows. tRC max value in bios is 31, so to get to tRC of 33 I set it manually in Memset v3.2.

Then I used clockgen to raise memory clocks to 670mhz 5-5-5-18 6-33-6-11-AUTO-7-48-Auto at 2.55v bios set vdimm





DDR2-1340 5-5-5-8 at 2.55v vdimm

[Charles Wirth]

Good stuff, are you using water cooling to cool? Since the ram is close to the drive bays they should have a small 5.25 kit to make a loop.

[eva2000]

Nah just air cooling ram with 3x 60x25mm Sunon 23.5cfm MagLev fans (no dead center)

[theelectic]

Interesting - according to the SPDTool screenshot, the FlexXLC is EPP programmed (though not marketed that way), and has a single EPP profile for DDR2-1142 CAS5 @ 2.1V!

[zitmatt]

optimums score...
these ram are pc6400 or pc9200...
cpuz says 6400 because???
thanks

[theelectic]

these ram are pc6400 or pc9200...
cpuz says 6400 because???

CPUZ is reporting what's programmed in the SPD. And what's programmed in the SPD is a medium "safe" speed that will boot in most motherboards. Then you go into the BIOS and change it to DDR2-1150 @ 2.35V yourself. That way you don't have to have a set of cheapo slow RAM to put into the motherboard just so you can change the BIOS and plug the faster RAM in.

[Kunaak]

how about someone finally actually watercooling these...
no one does, and thats what they are meant for...

[RyderOCZ]

how about someone finally actually watercooling these...
no one does, and thats what they are meant for...

http://www.bleedinedgesupport.com/oc...2&postcount=12

[dinos22]

http://www.bleedinedgesupport.com/oc...2&postcount=12

Sweet as Mr.Mutabi (from a movie hehe)

[andyOCZ]

Hmm, I need some samples of this! That water cooled rig looks might sweet.

[dinos22]

Hmm, I need some samples of this! That water cooled rig looks might sweet.

what do you mean RAM lol

or sample of watercooling

[eva2000]

how about someone finally actually watercooling these...
no one does, and thats what they are meant for...

Probably because logistically it's more of a pain if you're testing only 1 pair of FlexXLC and you have several motherboard/rigs across the house to test the memory in hehe :o

But I'll definitely get around to it eventually

[Kunaak]

so many bubbles...

I am far more interested in seeing watercooled results then just ram results.
we all know 1300 mhz isnt all that hard for the 680i assuming you have some quality ram. what I want to see is how the watercooling might bring that ram to the next level.

[eva2000]

so many bubbles...

I am far more interested in seeing watercooled results then just ram results.
we all know 1300 mhz isnt all that hard for the 680i assuming you have some quality ram. what I want to see is how the watercooling might bring that ram to the next level.

In due time... only had the memory for less than 12hrs and well you do need to find out it's oc's on air first to compare with water

[pumbertot]

nice going there George. you lucky bastard!

[gundamit]

In due time... only had the memory for less than 12hrs

Huge amount of info considering. NB voltage scaling up a bit with the dimm voltage ... duly noted, but wondering about NB cooling. Just stock heat pipe mess? Sealed factory sticks suggesting random results, albeit in the hands of a master?

[eva2000]

Huge amount of info considering. NB voltage scaling up a bit with the dimm voltage ... duly noted, but wondering about NB cooling. Just stock heat pipe mess? Sealed factory sticks suggesting random results, albeit in the hands of a master?

Updated first post with more cooling details

Asus 680i Striker Extreme 0402 bios (Thermalright HR-05/80x38mm Thermaltake fan on NB, Thermaltake Extreme Spirit II on SB)

I suppose factory sealed does mean random results....

More results..


670mhz 5-5-5-18 6-33-6-11-AUTO-7-48-Auto at 2.55v

Next stop is finding the modules max Super Pi 1M results with default 5-5-5-18 timings at 2.55v but with sub timings closer to the default rated ones listed on OCZ site http://www.ocztechnology.com/product...exxlc_edition:



I decided to use sub timings of 6-31-6-11-AUTO-7-48-Auto and boot from 654mhz 5-5-5-8 into windows. tRC max value in bios is 31, so to get to tRC of 33 I set it manually in Memset v3.2.

Then I used clockgen to raise memory clocks to 670mhz 5-5-5-18 6-33-6-11-AUTO-7-48-Auto at 2.55v bios set vdimm





DDR2-1340 5-5-5-18 at 2.55v vdimm

[erwinz]

very nice..

DDR2 memory guru..

[pumbertot]

maybe just being suspicious but im sure OCZ could pretest a great set then seal them for shipping to you. but just saying, im sure they are indeed a random pair. man think if theres better ones out there.

[Reg-Edit]

I got myself the OCZ Flex PC9200 RAM to

I use it on A Asus Commando.
But I have A problem and maybe some of you can help me ?

I manage to get the ram @ 650Mhz 5-4-4-12 @ 2.55 volt
When I push it A little more I can't run Spi 1M
It stops the first loop (Does not start the first loopt) , but at that moment I can run ATItool 3D view whit no problems. That isn't possible I guess ..
If the system is so unstable that he cant ever run the first loop of Spi, than it never can show 3D view of ATItool.

I dont come higher than that, but I don't think its my RAM.
I used over 3 Volt and even then I can't come over 650Mhz 1M stable.

Strange thing is ..
If I get my RAM 16M stable and I want to run the 32M, then also he doesn't finish or even start the first loop !
Is int it that Spi must stop (If its not stable enough) at loop ~12 or so and not the first loop ?


Is it my motherboard ?

I cant come over the 650Mhz even not @ Cas 5-5-5-15 or 6-6-6-18 and more voltage ..

Used divider's are 2:3 and 1:2, both the same

[eva2000]

pumbertot, highly doubt that .. i've seen better results online not as thorough as mine though heh.

Not sure if it's Asus 680i Striker 0402 bios issues but my FlexXLC ain't liking CAS4 much, CAS5 is fine though. To investigate, I've shove the OCZ PC2-9200 FlexXLC into my Asus P5B Deluxe rig for a whirl at CAS4

Reg-Edit not sure mate never used Commando before and I do know 3.0v vdimm is WAAAAAAAAAAAY too much vdimm to use even for water cooled modules... and I've never gotten more than 590mhz 4-4-4-5 with my other micron d9xxx based modules on Asus P5B Deluxe so your 650mhz is exceptionally good for 965P chipset...

Basically, you're unstable so drop mem clocks down until you can pass whatever tests you want to run.

[Quentin]

More Vmch on P965 can help for dual channel @ 600+

On my P5B Deluxe with OCZ 2Gb PC2-8500 SLI Ready Edition @ 2.45v

Vmch : 1.55v = Ram @ 630Mhz SuperPi 1M
Vmch : 1.65v = Ram @ 646Mhz SuperPi 1M

But in Single Channel when try Maxscreen Frequency :

Vmch : 1.45v = Ram @ 720 Mhz Cpu-z
Vmch : 1.55 or 1.65v = Impossible to hit 710 Mhz