[crimedog]

testbed in sig, mulitplier was kept at 14x and dividers were used for higher speeds. my prommy needs repair so i can't (if i wanted to) run prime95 or the like without overloading my cooling, so superpi 1m was all that was used to test stability.

PQI Turbo pc2-5400 2x1gb 4-4-4-12

4-4-4-12 4:5
1.8v 363mhz
1.9v 368mhz
2.0v 368mhz
2.1v 368mhz
2.2v 368mhz
2.3v 372mhz
notes: i only gained 1mhz when switching to cas5. i didn't want to test 5-5 timings because it's too sloppy for me.

4-3-3-10 1:1
1.8v 275mhz
2.1v 281mhz
2.3v 281mhz

4-2-2-8 would not boot, even at 200mhz 2.3v

Centon 2x512mb with Micron D9DQW (pic attached so you can see what d9 looks like)

I only tested this ram at 2.3v.

4-2-2-8 3:4 = 338mhz
5-2-2-8 3:4 = 346mhz
5-3-3-10 2:3 = 460mhz

Mushkin PC2-6400 2x512mb 5-3-3-8
Thanks Sentential for your results. He tells me these were stable results, with no active cooling. He also has a P5WD2-Premium.

5-4-4-10
1.9 474mhz
2.0 484mhz
2.1 502mhz
2.2 516mhz
2.3 524mhz

5-3-3-8 3:5
1.9v 452mhz
2.0v 466mhz
2.1v 492mhz
2.2v 514mhz
2.3v 522mhz

5-2-2-8
2.2v 348mhz
2.3v 358mhz

Sorry for the short review, I just wanted to see what these sticks can do. Hope it's helpful, I hate the lack of reviews out there with a decent testbed.

[d94]

BUMP for one of the few ddr2 reviews around
ESP for this RAM! which is $189 @ monarch

[Sentential]

Originally Posted by crimedog

PQI Turbo pc2-5400 2x1gb 4-4-4-12

4-4-4-12 4:5
1.8v 363mhz
1.9v 368mhz
2.0v 368mhz
2.1v 368mhz
2.2v 368mhz
2.3v 372mhz

That PQI is definatly very good for the money, but I do think our PC5300 warrents the cost as we do use *much* better chips

[crimedog]

guys please send me any ddr2 reviews you have in a similar format i'd love to add them on. i do want to make sure everyone has a good motherboard though.

[Alchemy1]

Nice reviews crimedog. That OCZ stuff looks amazing, but it is too rich for my blood.

[batboy]

I'm rather impressed with what I'm seeing from Mushkin lately. That Centon certainly did better than I thought too.

Here is my max tightest timing for 2X512 OCZ EB rev2 PC2-4200

3-2-2-6 3:4
2.2v 360.9 MHz

Here is my 2X512 Kingston HyperX PC2-5400

Tightest 3-3-3-8 3:4
2.0v 291.6 MHz

Max 4-4-4-12 3:4
2.0v 400.0 MHz

Note 1: the KHX test was done last December when I was still nervous about running too much voltage. I should redo my tests at 2.2v, bet it would do better.

Note 2: Both pairs of RAM were tested in my "green" rig listed in my signature.

[xtatdsm]

Corsair 5400UL (now playing)

DDR2 685 with 3-3-2-7 (2.1 volts) 100% stable
DDR2 928 with 4-3-2-4 (2.3 volts) same as above (this is my 24/7 setup)

OCZ Gold DDR2 800 PC2-6400 (sitting on the desk)
@DDR2 770-800 with 5-5-5-10 @ 1.95 volts 100% stable

any thing faster than that cause stabilty problem, the same thing with any tighter timing. I spent like a month testing these memory sticks to get better performance than the stock setting and I simply failed.

[Ross]

More Mushkin XP6400 info...

Volt modded P5WD2 Premium. 2.227Vddr is about what you'd get with a stock board and 2.30Vddr BIOS setting. All Vddr is measured on the board. I wasn't really excited about going for 2.6V, but 2.47V went off without a hitch and 2.6V is nowhere near the highest I've seen run through DDR2 Definitely not worth that high given the gains unless you're going for bragging rights on speed. MCH voltage measured 1.690V actual.

My main "test" was boot clean and run a SuperPi 1M without errors...I wasn't sitting here all night to run memtest on each

|------------------------------------------------------|
| Timings |- 2.227 Vddr -|- 2.476 Vddr -|- 2.597 V ddr |
|---------|--------------|--------------|--------------|
| 3-2-2-4 | DDR675 (4:5) | DDR760 (3:4) | DDR773 (3:4) |
| 3-2-3-4 | DDR715 (3:4) | DDR780 (2:3) | DDR795 (2:3) |
| 4-2-3-4 | DDR917 (3:5) | DDR1000(1:2) | DDR1040(1:2) |
| 5-3-3-4 | DDR1000(1:2) | DDR1060(1:2) | DDR1080(1:2) |
| 5-3-3-8 | DDR1000(1:2) | DDR1060(1:2) | DDR1080(1:2) |
|------------------------------------------------------|

Ratios make a noticeable diff in the max speed. 2:3 absolutely sucks and in some cases it would boot or not boot at a specific DDR speed depending what ratio was used to achieve it. For instance: 305FSB 3:5 for DDR1020 would boot and bench, but 255FSB 1:2 for DDR1020 would crash right after POST.

5-3-3-8 did little for speed over 5-3-3-4 other than possibly make it a tad more stable at the same speed. In fact, I tried seriously sloppy latencies (5-5-5-15) which did NOTHING to increase speed...pretty much whatever they run at 5-3-3-8 is what it will run.

4-2-2-4 didn't seem to work no matter what and for some reason I completely skipped trying 4-3-3-4, sorry.

You didn't really think I was going to get that close to DDR1100 and not get it did you?


Mushkin 6400 5-3-3-4 DDR1100. See the DDR freq / timings. This took 2.64Vddr *cough*. I wasn't going any higher and unfortunately, with 14x @ 330FSB for DDR1100 it just crashed out once it reached the desktop or that would've been some SICK benches. Maybe one day I'll get the cajones to try 2.70-2.75Vddr and see if it will bench at 330FSB, but for now, it's back to running comfortably at 2.10V.

[crimedog]

way to go ross

[Ross]

Thanks

I am dying to bench DDR1100 @ 330FSB. If I get drugged up enough at my dentist appointment today maybe when I get back

[crimedog]

if you're not drugged up enough to 2.7v maybe you should try 5-4-3

[RangerXLT8]

Jesus 2.64v is insane! Your lucky you didn't melt the chips out from under the heatspreaders. Otherwise congrats on DDR2-1100.

[batboy]

Sheesh, and I've been nervous about running my DDR2 at 2.2v.

[speed bump]

Its all in the cooling, although I would worry about running more than 2.2v everyday.

[Ross]

Yeah, I was sweating bullets when I tried it I NEVER would've tried it if I hadn't seen much more from other braver (crazier?) people. I wasn't as concerned about 2.4-2.5V. I had taken a stab at 2.30-2.35V prior to that post and the memory doesn't even get warm...I have plenty of fans around the mobo for tests like that

Like I said, other than maybe bragging rights, the gains with voltages like that really aren't worth risk. Lower speeds with tighter timings bench almost as well as high DDR speeds and looser timings...within reason. I'd run way less speed to not have to resort to the 2:3 divider...it sucks at all FSBs/speeds/timings/voltages.

The board/mem actually did DDR1017 with stock Vddr and MCH voltages maxed (like 2.22V and 1.68V actuals respectively) and more Vc than I ran for those tests. Dividers also play a role in max stable DDR speeds and those test were done at lower FSB than I usually run (and hence dividers I don't normally see). Like I noted, 3:5 @ 305FSB DDR1020 booted and benched, but 1:2 @ 255FSB DDR1020 would just get passed POST and die. If anything, you'd think the higher FSB would be more unstable from the extra stress. That only leaves dividers since nothing else had changed (timings/any voltages).

I guess long story short: I plan to stick with 2.2ish Vddr and DDR1000-1020 since anything more really doesn't really gain much.

[xtatdsm]

Ross you are right about deviders and stability. I have been trying to get the most out of my 5400UL for the past couple of days with no luck (I was running 3:4) I tried all kindda voltage and timing with no luck. then out of the blue I actually thought I should try lower devider (yeah run the memory at higher speed) with loose timing, so I went for 2:3, and it was stable, so I started tighting the timing and testing untill I found the sweet spot, and that was at DDR 924 with timing 4-3-2-4 at 2.3 volts. I am pretty sure that this is stable enough for my use (ran memtest for 8 hours and the windows based memtest for 3 hours or so with no errors).

so for me running with the above setting gives me the best stabilty with decent performance.

[Ross]

Yeah, I knew some would be better than others performance-wise, but was really shocked to see same speeds work or not at all because of the divider. I wonder if the Mushkin XP6400 and the ULs use the same chips? Funny that my max at ~2.3V BIOS with 4-3-2-4 is about DDR920 too. I wish my BIOS had tighter timings...3-2-2-4 is at good as it gets.

Nice clocks at stock Vc. Are you using SetFSB? I want a 660 so bad I can taste it

[crimedog]

Originally Posted by Ross

I wonder if the Mushkin XP6400 and the ULs use the same chips? Funny that my max at ~2.3V BIOS with 4-3-2-4 is about DDR920 too. I wish my BIOS had tighter timings...3-2-2-4 is at good as it gets.

yeah they do, mushkin has a brainpower pcb and i don't know what the corsair uses.

[Ross]

Thanks for the heads up crimedog

[Sentential]

Originally Posted by RangerXLT8

Jesus 2.64v is insane! Your lucky you didn't melt the chips out from under the heatspreaders. Otherwise congrats on DDR2-1100.

This type of DDR2 from Mushkin will not overheat. I have seen people run with his voltage 24/7 and not see any ill issues. They take voltage as good, if not *better* than BH5

As for the timings you are correct and your results echo what I have seen. Unfortunatly this version does not do well with Cas 4 which is why i suggested Cas 5 because they scale *quite* a bit better with such as you have noted. I ran my set at 2.3v @ 1040ish daily with no ill effects

The limiting factor on these chips is not TRCD as one would think, instead its the Cas latency. They should theoritically clock even better with Cas 6 but I could never get my mobo to boot with it.

Glad you like em cuz I sure enjoyed testing them