[klaxian]

Hello all. I have searched extensively on the forums and on Google and I can't find much information about my new OCZ Reaper HPC's OCZ3RPR1600C6LV4GK. They are rated for 1600MHz 6-8-6-24 1.65V, but I have them running at 1524MHz 5-7-5-20-27-1T 1.71V right now. I have only stress tested them for a few hours (will do more overnight), but there have been no errors so far.

What timings are others using for this or similar RAM? What about the more obscure timings that no one seems to mention? How far should I be able to push it? These have pretty big heat sinks/pipes. What is the maximum safe voltage for them? I heard that OCZ encourages overvolting with EVP, but I couldn't find any definitive information.

Thanks!

 

[xd_1771]

Oh my--I certainly don't have the same RAM but near 1600 at CAS5-7-5 impresses me thoroughly

 

[hesho]

umm.. THAT'S CRAZY!!!

Geez, my comp doesn't even boot up at CAS5.. i tried lol.

 

[klaxian]

Thanks guys

I let prime95 and LinX run for 10+ hours last night. I am stable at 1524MHz 5-7-5-16-24-1T with these modules, but I needed 1.73V to do it. I read that this kit is under warranty up to 1.8V and some have pushed them to 2.2V so I'm not too worried about that. I wonder if I should try to push them further, and what to change first.

 

[hesho]

btw, i hate u...

er.. i mean, very nice!!! Btw, u may want to run prime a bit longer. I was testing 6-7-5-18 on mine and i always got an error at about the 14 hour mark (drove me nuts!!!). Figure i'd let u know.

 

[klaxian]

Thanks for the tip. I probably did run for about 14 hours overall, but I stopped and changed the test after the first couple hours. I do [email protected] on this machine 24/7 (when not stress testing) so any possible instability that remains should be caught by that eventually.

Nice system by the way

 

[xd_1771]

Do it.
See if you can snatch 5-6-5 timings at 1.8V

 

[klaxian]

Quote:

Originally Posted by xd_1771 Do it. See if you can snatch 5-6-5 timings at 1.8V

Sorry, 5-6-5 at 1.77V not possible.

I'm trying to keep the voltage at 1.73V or below, but I am working on upping my clock speed to 1583MHz with a 2970MHz CPU/NB. Everything was going well until my computer turned itself off after about 10 runs of LinX. No BSOD, no freeze, just turned off. Temperatures looked fine at the time. Is this just another indication of instability?

 

[xd_1771]

Quote:

Originally Posted by klaxian Sorry, 5-6-5 at 1.77V not possible. I'm trying to keep the voltage at 1.73V or below, but I am working on upping my clock speed to 1583MHz with a 2970MHz CPU/NB. Everything was going well until my computer turned itself off after about 10 runs of LinX. No BSOD, no freeze, just turned off. Temperatures looked fine at the time. Is this just another indication of instability?

Most likely yes; you may have upped to the point where the motherboard power regulators exceeded their rated output and forced a shutdown, but with an 8+2 power regulator phasing on the 890FXA-UD5 I doubt that's the reason...
I sometimes get random restarts when my memory is unstable; a shutdown may be the same thing for you

 

[redhat_ownage]

Quote:

Originally Posted by xd_1771 Oh my--I certainly don't have the same RAM but near 1600 at CAS5-7-5 impresses me thoroughly

easily impressed much? xd1771... i would have expected more from you
i am dissapoint

5-6-5 wont work try 6-6-6 1.71V

 

[hesho]

isn't a random reboot with no bsod a cpu issue? generally lack of voltage?

btw, what memory divider r u running? I assume 5.33? (3:8)?

 

[klaxian]

Quote:

Originally Posted by hesho isn't a random reboot with no bsod a cpu issue? generally lack of voltage? btw, what memory divider r u running? I assume 3.75? (3:8)?

It wasn't a reboot. The computer just turned off. I've stress tested my CPU speed/voltage a ton at this point and I fold 24/7 without problems. In fact, I can run almost up to 4.1GHz at 1.425V stably, but I wanted to be extra cautious since I'm folding.

I'm not sure of the ratio but I'm running 5.33 x 286 = 1524MHz.

 

[RagingCain]

Hey Klaxian, look at my setup for the full break down of what I am running (all memory timings.)

On the simple stuff, for the 1090T thuban (I am sure you are the same) you need a minimum of CPUNB 1.225v for at least an 8 hour stability with mem at 1600 MHz, thats directly from RyderOCZ and Tony from OCz Tech forums. The memory AT stock needs a minimum of 1.705v (1.71v is the closest option I have on my 790FXTA-UD5.)

My memory is indeed slower than yours, but the OCZ boys told me to whenever possible keep the timings the same using dual channel and 8GB without pushing Bank Cycle Time too low (tRP + Bank Cycle is a good estimate.) So you could try 6-6-6-20-1-26 @ 1600 which is benchmarkingly better than 5-7-5-20-1-(insert other memorytimings) - 27 @ 1540. I only use multiplier.

Keep your NB under 2800 MHz while tightening up your timings and take the CPU down to 3.6. Just my advice.

Its not that the quality is bad on these sticks, its that they are extremely picking with EVERY setting available, and mine are fully stable at 3.6 GHz, but 100% unstable with the same settings at 3.8 GHz. I find it hard to use 200 MHz off of HTT/FSB, and getting these stable. I am only able to be stable on Prime95 Blend for 8 hours and 11 minutes on my OC / timings / voltage, but its not any more stable on Jedec believe it or not. Does pass MemTest86+ 4.00 though, so its not really a huge deal to me. I am a gamer though with an occasional [email protected] or [email protected] run, but not much anymore. Folding is very accuracy dependent I know.

Loose timings for me = unstable... go figure. I thought I was stuck with some bad ram, till someone suggested 7-7-7-20-1 for me... just tightened the other timings to get most performance while staying at 8+ hour blends.

 

[xd_1771]

1.225V seems correct, I needed to up to that from 1.15V in order to attain full stability in IntelBurn at stock speeds & rated RAM settings; other tests have responded the same with full stability.
The memory voltage may vary, though nominally on AMD systems an increase in voltage may not needed for memory stability depending on your kit (unless you run a C2 processor in which case it may indeed be necessary) - upping the CPU-NB voltage and sometimes even the NB voltage usually takes care of that.

 

[RagingCain]

Quote:

Originally Posted by xd_1771 1.225V seems correct, I needed to up to that from 1.15V in order to attain full stability in IntelBurn at stock speeds & rated RAM settings; other tests have responded the same with full stability. The memory voltage may vary, though nominally on AMD systems an increase in voltage may not needed for memory stability depending on your kit (unless you run a C2 processor in which case it may indeed be necessary) - upping the CPU-NB voltage and sometimes even the NB voltage usually takes care of that.

Be very wary with upping Mobo NB voltage. Do it as a last resort. My stock voltage on NB was 1.100, and 1.210 was enough to blow out the 4gb (2of4 x 2gb) RAM. However, it was RMA-ble because I was following OCz instruction to try and increment NB up to 1.275 and CPU-NB to 1.300.

 

[xd_1771]

You upped it to 1.2 and it blew out your RAM? Strange considering this is the motherboard northbridge with no connection to your RAM whatsoever unless going through the CPU. Upping the voltage to 1.2 has been known to stabilize alot though - twice before it worked when I've recommended it.

 

[RagingCain]

Just a cautionary tale, to be honest I personally think my motherboard is slightly unstable to be honest and I am stuck with it for now.

However, I believe the Northbridge chipset interfaces with the physical DIMM but the memory freq / access cycles are controlled/regulated by the CPU's IMC.

Thats why bumping up the voltage on DDR3 and NB is sometimes recommended.

I think its a little fuzzy without official documentation, but I have heard of other people overvolting NB thinking its CPU NB to 1.50v and have had no permanent damage, hence I was probably just unlucky.

It also could of been the memory itself was just waiting to go bad as well, I wasn't worried about it once OCz techy said just RMA it.

I haven't touched the NB voltage since though ; )