• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

FEATURED Marathon Season IV December : SuperPi 32M

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Woomack | 6700K | SS | 5m 58.614s

22.jpg

These 2x8GB kits can't run high at CR1. For some reason also can't pass 1800MHz. Maybe mobo's fault.
 
Have a little ln2 left and a free evening so am hoping to get a sub in for this using my g3258 before the deadline. Here's my current best with crappy ram timings and no wazaa so hoping to improve a little.

Question...some say to use ram in single channel if dual isn't happening. That said...all the top subs are in dual mode so it appears to make a real difference? So...was only able to get dual channel working with stock (psc 2600 preset) with loose/sloppy subtimings. Each stick runs absolutely wicked tight (2666 psc preset plus many, many tightened subtimings) so the question is...which would be faster?
- dual channel, 2600 loose subtimings
- single channel, 2666 super tight subtimings

Whatcha think?
 
What I usually do is see how fast I can get the sticks then tighten the timings until I find my best result. Though, this takes time and you may want to try testing on ambient first. Also, I've never compared dual and single channel to see how much difference there was.
 
Thanks man. Spent a lot of time testing each individual stick and have 6 that do the 2666 crazy tight thing. In dual channel mode, though...pairs (of the best sticks) won't get close to even getting into windows with that speed or timings. Definitely hear ya' on the 'it takes time' thing....when I look at the difference between the sloppy subtimings and crazy tight ones, there are at least 10 subtimings involved (and many differ by more than 1 step) so no idea how long it'd take to figure this mess out. :/

Guess I'm gonna have to bother Woomack again.

Since it'll take longer than the deadline, guess I'll run it in dual chan as best I can then get one run in with single chan and see which is best.
 
Have a little ln2 left and a free evening so am hoping to get a sub in for this using my g3258 before the deadline. Here's my current best with crappy ram timings and no wazaa so hoping to improve a little.

Question...some say to use ram in single channel if dual isn't happening. That said...all the top subs are in dual mode so it appears to make a real difference? So...was only able to get dual channel working with stock (psc 2600 preset) with loose/sloppy subtimings. Each stick runs absolutely wicked tight (2666 psc preset plus many, many tightened subtimings) so the question is...which would be faster?
- dual channel, 2600 loose subtimings
- single channel, 2666 super tight subtimings

Whatcha think?

What's wazaa? I looked it up but I don't really understand what it does or how it's used.
 
Ohhh...dude...you really need that! Not sure of all the details under the hood but it more or less copies a huge file from one drive to another right before you run spi. It'll shave significant time off your subs (amount saved varies by platform but with haswell it's even more significant that prior setups).

Search for 'spi tweaker' (I think ocx did it). Start the tweaker app and get spi all set up to run including killing explorer, etc. In the tweaker, select your spi drive and your 'extra' drive, click 'clean' and then 'run'. As soon as it's finished and closes, use your keyboard to start spi.
 
Wazza is simply a way to clean up RAM. If you move large amount of data from one drive to another ( usually can be partition ) then OS will load that data to RAM and then will delete it so next operation will be able to save data in that empty space without losing time on random search of empty memory cells.
The only problem is how Windows is working ... it's not always cleaning RAM fast enough so regardless what wazza are you using ( manual or with some software ) sometimes you have to run it couple of times. You will see the results. On newer platforms it's not giving as good results as on older so I'm simply not wasting time on that unless I really want that couple of ms less.

Now maybe about that single/dual channel RAM. Dual channel will be much faster in Spi 32M. Even if clock will be much lower. There is one simple way to tell which option will be faster. Run maxxmem or AIDA64 memory benchmark and check read bandwidth. Change memory settings and compare it.
In most benchmarks what counts is only memory read and latency. Perfect balance between these two gives best results.
 
I tried again. This damn chip hates 5Ghz+. Even with one core 1.7 wouldn't get it through Spi32m.
Damn :(
 
LOL you're part of the benching team mad read the LBB!

- - - Updated - - -

I tried again. This damn chip hates 5Ghz+. Even with one core 1.7 wouldn't get it through Spi32m.
Damn :(
SP at some point temperature wise it doesn't matter how much voltage you're giving it it's not going to take it. Get it colder and I bet you'll get it over 5.0.
 
LOL you're part of the benching team mad read the LBB!

- - - Updated - - -

SP at some point temperature wise it doesn't matter how much voltage you're giving it it's not going to take it. Get it colder and I bet you'll get it over 5.0.

I wish I had more tubing. That way I could put the Blackice rad out the window!!
 
Back