Crucial Ballistix DDR3-1600 / PC3-12800

[Jaco]

wtf is that last XMP profile ?
4-1-1-16 31T 4.55v

[eva2000]

I really like this thing crucial does. No fancy packages, just performance.

BTW, nice review

thanks

Yeah weird XMP profile there.. I know i won't be enabling XMP with X38 boards with such memory - 4.55v vdimm

[saaya]

Thanks for the link

saaya, haven't got that in depth so far as Blitz Extreme low FSB did limit me to <485FSB... so just starting out with P5K3 Deluxe and 530FSB headroom.

So many ways to tweak on Asus P35 DDR3 you have static read and write delay, transaction booster levels, 1T/2T and 266FSB/333FSB strap so there's so many ways to reach your desired speeds

transaction booster is the same as perf level i think.
static read and write delay? does it have such a big impact on performance?

[eva2000]

static read/write helps tad more bandwidth and lower memory latencies but can quickly add to memory/system instability as well

TB is performance level or tRD AFAIK hehe

[saaya]

hmmm gotta try that out, thanks
ill check again but im pretty sure that when i played with transaction booster it only changed perf level. i prefer to change it in windows tho, so i left it untouched

[eva2000]

Changed from E6600 L709A958 to E6850 L723A347 cpu for more tests.

System:

• Intel Core 2 Duo E6850 L723A347
• Corsair Nautilus 500 H20
• Asus P5K3 Deluxe 0604 official bios
• 128MB Gainward FX5200 PCI
• 2x1GB Crucial Ballistix PC3-12800 DDR3-1600 8-8-8-20 at 1.8v rated dual channel kit
• 74GB WD Raptor 8MB
• LiteOn 165H DVD Burner
• Sony FDD
• 1KW PCP&#37;C SLI PSU
• WinXP Pro SP2 Luna Royale Blue themed with nLite + RyanVM updates

3500Mhz - 7x500FSB 5:8 800Mhz 8-8-8-20 1T at 1.8v vdimm
Stock Crucial Ballistix PC3-12800 speed and timings but with 1T enabled @1.8v vdimm.



Single and Dual Super Pi 32M



Everest Ultimate 4.10.1142 beta

Latency result bug in Everest ? When you run it individually it's lower latency then via the tool ?



Prime95 v25.4 short 32min Blend test





Sandra 2008 XII - Buffered and Unbuffered





Winrar 3.70 Trial

[eva2000]

4104Mhz - 8x513FSB 3:5 855Mhz 7-6-5-15 1T at 2.0v vdimm
Now to max out my E6850 on Corsair Nautilus 500 water cooling, seems it hit same max for 32M as when on Blitz Formula with same cpu cooling - 4104Mhz at 1.632v on Blitz Formula in 22-24C ambients while 4104Mhz at 1.672v on P5K3 Deluxe in 28-32C ambients (no more winter cool nights in OZ )

Didn't try lower than 2.0v vdimm for this session as I wanted to make sure memory wasn't the weak point for now.



Single Super Pi 32M

Bloody fast - 15.907 seconds faster than my Blitz Formula DDR2 run below!



DDR2 run at 11m 53.672s



Everest Ultimate 4.10.1142 beta

41.1ns latency!



Sandra 2008 XII - Buffered and Unbuffered

10200+ MB/s Buffered and 7000+ MB/s Unbuffered!





Winrar 3.70 Trial

[pH(x)]

Nice work and nice benches, nice RAM as well!

We appreciate the SS

[saaya]

Latency result bug in Everest ? When you run it individually it's lower latency then via the tool ?

hmmm what tool do you mean?
the window that shows all memory bench results at once?

[eva2000]

Nice work and nice benches, nice RAM as well!

We appreciate the SS

thanks

hmmm what tool do you mean?
the window that shows all memory bench results at once?

yup tool as in read/write/copy/latency report differs from individual latency reported result

[rra]

Yip as usual Goerge excellent testing and review always enjoy your posts nice one.

[andyddr]

Sigh, my TeamGroup 1600's clocks to 870MHz with the exact same timings as your sticks (7-6-5-15) but I can only get 8K on Sandra buffered. Board is a Gigabyte P35C-DS3R. The board doesn't offer 1T command rate though, but surely that can't make a 2K difference?

[saaya]

yup tool as in read/write/copy/latency report differs from individual latency reported result

just talked to tamas, he said its already fixed in the next release, just a minor bug about displaying the result not actually a bug of the bench it seems

[eva2000]

Yip as usual Goerge excellent testing and review always enjoy your posts nice one.

thanks glad you like

Sigh, my TeamGroup 1600's clocks to 870MHz with the exact same timings as your sticks (7-6-5-15) but I can only get 8K on Sandra buffered. Board is a Gigabyte P35C-DS3R. The board doesn't offer 1T command rate though, but surely that can't make a 2K difference?

also slightly depends on cpu clock frequeny so you would get higher sandra buff at 9x400FSB 1:2 than you would at 7x400FSB 1:2

just talked to tamas, he said its already fixed in the next release, just a minor bug about displaying the result not actually a bug of the bench it seems

Awesome, would save me having to run both methods and screenshot them side by side hehe

[andyddr]

Sorry for the massive pic!

Just thought I'd show you what I'm facing here Will pop my E6850 that does 550MHz in tonight and see if I can get a better score.

Thanks Eva.

[saaya]

andyddr, what chipset voltage are you using?
try more and see if that helps

[eva2000]

Sorry for the massive pic!

Just thought I'd show you what I'm facing here Will pop my E6850 that does 550MHz in tonight and see if I can get a better score.

Thanks Eva.

guide to better and smaller screenshot images http://i4memory.com/showpost.php?p=61904&postcount=3

Not sure what's going on with your board as i never used that board. 1T gives me an extra ~200MB/s memtest86+ v1.70 bandwidth over 2T so no it won't be reason for 2000+ MB/s difference in sandra bandwidth I don't think. What cpu speed was that 8k buffered run at ? 3050Mhz ? try at 9x 440FSB or higher

[andyddr]

Thanks for the input guys. Seems like the problem here is infact the motherboard's poor optimization for DDR3. I've tried the E6850 @ a higer FSB and only scored 300MB/S more. Add to that I cannot set to 1T in the bios. Should have done a bit more homework on this board (X-bit has an article up on the poor performance I just mentioned) before I bought it. See Newegg has just added X38's to their pricelist though. Great timing!

[saaya]

ohhh your on a hybrid ddr2 ddr3 board right?
Yeah, those all suck afaik... but i didnt know they perfom that much worse at the same clocks 0_o weird